Beruflich Dokumente
Kultur Dokumente
Structures and
Materials III
WITeLibrary
Home of the Transactions of the Wessex Institute.
Papers presented at High Performance Structures and Materials III
are archived in the
WIT eLibrary in volume 85 of WIT Transactions on
The Built environment (ISSN 1743-3509).
The WIT eLibrary provides the international scientific community with immediate
and permanent access to individual papers presented at WIT conferences.
http://library.witpress.com
Transactions Editor
Carlos Brebbia
Wessex Institute of Technology
Ashurst Lodge, Ashurst
Southampton SO40 7AA, UK
Email: carlos@wessex.ac.uk
E Alarcon
Universidad Politecnica de Madrid
Spain
C Alessandri
Universita di Ferrara
Italy
S A Anagnostopoulos
University of Patras
Greece
E Angelino
A.R.P.A. Lombardia
Italy
H Antes
Technische Universitat Braunschweig
Germany
D Aubry
Ecole Centrale de Paris
France
D E Beskos
University of Patras
Greece
J J Bommer
Imperial College London
UK
F Butera
Politecnico di Milano
Italy
P G Carydis
National Technical
University of Athens
Greece
J Chilton
University of Nottingham
UK
S Clement
Tranport System Centre
Australia
M C Constantinou
State University of New York at Buffalo
USA
G Degrande
Katholieke Universiteit
Leuven
Belgium
A De Naeyer
Universiteit Ghent
Belgium
W P De Wilde
Vrije Universiteit Brussel
Belgium
J Dominguez
University of Seville
Spain
F P Escrig
Universidad de Sevilla
Spain
M N Fardis
University of Patras
Greece
C J Gantes
National Technical
University of Athens
Greece
L Gaul
Universitat Stuttgart
Germany
Y Hayashi
Nagoya University
Japan
M Iguchi
Science University of Tokyo
Japan
L Int Panis
VITO Expertisecentrum IMS
Belgium
W Jager
Technical University of Dresden
Germany
C M Jefferson
University of the West of
England, Bristol
UK
D L Karabalis
University of Patras
Greece
E Kausel
Massachusetts Institute of
Technology
USA
K Kawashima
Tokyo Institute of Technology
Japan
A N Kounadis
National Technical
University of Athens
Greece
W B Kratzig
Ruhr Universitat Bochum
Germany
A A Liolios
Democritus University of
Thrace
Greece
J W S Longhurst
University of the West of England,
Bristol
UK
J E Luco
University of California at
San Diego
USA
L Lundqvist
Unit for Transport and Location Analysis
Sweden
M Majowiecki
University of Bologna
Italy
G D Manolis
Aristotle University of Thessaloniki
Greece
G Mattrisch
DaimlerChrysler AG
Germany
F M Mazzolani
University of Naples "Federico II"
Italy
K Miura
Kajima Corporation
Japan
G Oliveto
Universit di Catania
Italy
E Oate
Universitat Politecnica de
Catalunya
Spain
A S Papageorgiou
Rensselaer Polytechnic Institute
USA
G G Penelis
Aristotle University of
Thessaloniki
Greece
A M Reinhorn
State University of New York at Buffalo
USA
F Robuste
Universitat Politecnica de
Catalunya - LAMOT
Spain
C W Roeder
University of Washington
USA
J M Roesset
Texas A & M University
USA
M Saiidi
University of Nevada-Reno
USA
F J Sanchez-Sesma
Instituto Mexicano del
Petroleo
Mexico
S A Savidis
Technische Universitat Berlin
Germany
J J Sendra
Universidad de Sevilla
Spain
Q Shen
Massachusetts Institute of Technology
USA
A C Singhal
Arizona State University
USA
P D Spanos
Rice University
USA
C C Spyrakos
National Technical
University of Athens
Greece
H Takemiya
Okayama University
Japan
I Takewaki
Kyoto University
Japan
E Taniguchi
Kyoto University
Japan
J L Tassoulas
University of Texas at Austin
USA
M A P Taylor
University of South Australia
Australia
R Tremblay
Ecole Polytechnique
Canada
R van Duin
Delft University of
Technology
Netherlands
A Yeh
The University of Hong Kong
China
M Zador
Technical University of
Budapest
Hungary
R Zarnic
University of Ljubljana
Slovenia
HIGH PERFORMANCE
STRUCTURES AND MATERIALS III
CONFERENCE CHAIRMAN
C.A. Brebbia
Wessex Institute of Technology, UK
High Performance
Structures and
Materials III
Editor:
C.A. Brebbia
Wessex Institute of Technology, UK
C.A. Brebbia
Wessex Institute of Technology, UK
Published by
WIT Press
Ashurst Lodge, Ashurst, Southampton, SO40 7AA, UK
Tel: 44 (0) 238 029 3223; Fax: 44 (0) 238 029 2853
E-Mail: witpress@witpress.com
http://www.witpress.com
For USA, Canada and Mexico
Computational Mechanics Inc
25 Bridge Street, Billerica, MA 01821, USA
Tel: 978 667 5841; Fax: 978 667 7582
E-Mail: infousa@witpress.com
http://www.witpress.com
British Library Cataloguing-in-Publication Data
A Catalogue record for this book is available
from the British Library
ISBN: 1-84564-162-0
ISSN: 1746-4498 (print)
ISSN: 1743-3509 (on-line)
The texts of the papers in this volume were set
individually by the authors or under their supervision.
Only minor corrections to the text may have been carried
out by the publisher.
No responsibility is assumed by the Publisher, the Editors and Authors for any injury and/or
damage to persons or property as a matter of products liability, negligence or otherwise, or from
any use or operation of any methods, products, instructions or ideas contained in the material
herein.
WIT Press 2006
Printed in Great Britain by Cambridge Printing.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system,
or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or
otherwise, without the prior written permission of the Publisher.
Preface
This book contains the edited papers presented at the Third International
Conference on High Performance Structures and Materials, held in Ostende,
Belgium in 2006.
The Conference was dedicated to honouring the distinguished career of
Professor Patrick De Wilde, from the Free University of Brussels, Belgium, on the
occasion of his retirement.
The Conference addressed issues involving advanced types of structures,
particularly those based on new concepts or new types of materials. This responds
to the need to develop a generation of new materials that are suitable for high
performance structures which can easily resist a wide range of external stimuli and
react in a non-conventional manner.
Professor De Wilde, whose achievements are summarised in the following
pages, was the originator and driving force behind this Conference series since it
started.
The Editor
Ostende, 2006
Dedication
Professor W. Patrick De Wilde
This book is in homage to the long and
distinguished career of my friend and
colleague Professor Patrick De Wilde from
the Vrije Universiteit Brussel (VUB).
I was fortunate enough to meet Patrick
for the first time many years ago when we
were both young researchers. This was
on the occasion of the Seminar on Finite
Elements that I organised in Southampton
in 1970. He was then a Research
Assistant at the Department of Applied
Continuum Mechanics of VUB and I had
just been appointed a lecturer at the
Department of Civil Engineering at the
University of Southampton. Since then we kept in constant contact and this led me
to a better appreciation of his outstanding scientific career and the depth of his
knowledge and imagination. Patrick always brought a new and fresh perspective to
the most challenging engineering and scientific problems.
His academic career has been centred at VUB where he obtained his PhD in
Applied Sciences in 1976, rising from lecturer to Head of the Laboratory of Structural
Analysis and to full Professor in 1984. He also served with distinction as Chairman
of the Research Council and Vice Rector of VUB, and is now about to retire from the
position of Head of the Department of Mechanics of Materials and Constructions.
Patrick has always been characteristically generous with his knowledge,
lecturing part time at many institutions in Belgium and abroad. An appointment
that reflects his range of interests was as a Professor at the Higher Institute of
Architecture Victor Horta, the link between architecture and engineering being
always close to his heart. He also contributed substantially to the enhancement of
training and research capabilities in South American and African countries to mention
but a few of his international activities.
Contents
Section 1: Conceptual design and structural analysis
(Special session organised by W. P. De Wilde)
Conceptual design of lightweight structures:
the role of morphological indicators and the structural index
W. P. De Wilde ......................................................................................................3
Influence of dynamic loads on the optimum design of trusses
J. Van Steirteghem, W. Ponsaert, W. P. De Wilde & Ph. Samyn........................13
Modular grid-based design concept for fibre reinforced composite shells
E. De Bolster, H. Cuypers, W.P. De Wilde & J. Wastiels...................................21
Influence of stiffness constraints on optimal design of trusses
using morphological indicators
T. Vandenbergh, W.P. De Wilde, P. Latteur, B. Verbeeck,
W. Ponsaert & J. Van Steirteghem......................................................................31
Variations in form and stress behaviour of a V-shaped membrane
in a foldable structure
M. Mollaert, N. De Temmerman & T. Van Mele ................................................41
Section 2: Composite materials and structures
A new composite material based on natural fibres and a thermoset:
technology, applications and properties
G. Wuzella ...........................................................................................................53
Experimental test of threaded steel rods glued-in hardwood with epoxy
D. Otero, J. Estvez, E. Martn & J. A. Vzquez.................................................63
Predicting the mechanical behaviour of large composite rocket motor cases
N. Couroneau ......................................................................................................73
Section 1
Conceptual design and
structural analysis
(Special session organised by
W. P. De Wilde)
Abstract
Firstly, but in no way exclusively, this paper addresses architectural engineers
facing critical design decisions in the phase of so-called conceptual design.
The resulting design must yield a structure showing a sound behaviour in both
the serviceability limit state (SLS) and in the ultimate limit state (ULS), also
meaning that three essential criteria should be satisfied: strength, stiffness and
stability (the latter including, if relevant, an acceptable dynamic behaviour).
However, the question very often remains open as to which of the three criteria is
overruling the other ones. This paper, a synthesis of the work of a research group
headed by the author, tries to show that a conceptual design methodology can be
developed, hereby using the concept of morphological indicators (originally
developed by P. Samyn, subsequently P. Latteur and within the research group)
and a structural index (introduced by Shanley). This methodology also gives an
answer to the question that very often arises when designing lightweight
structures: design for strength or design for stiffness? Examples in
subsequent papers, presented during this Conference and thus included in these
Proceedings, illustrate the methodology nowadays used by our students in
architectural engineering when designing and analysing their structures.
Keywords: conceptual design of architectural structures, strength, stiffness,
structural stability, structural vibrations.
Morphological indicators
MULTI-WARREN trusses
WARREN trusses
ARCH + TIE + SW
1,700
ARCH + TIE + PR
1,414
ARCH + TIE
ARCH + SW
1,155
ARCH + PR
0,816
ARCH
0,707
0,577
PR = push rods
SW = suspension wires
1,633
2,309
2,828
3,138
3,266
L/H
Figure 1.
He also shows that they can be expressed as a function of L/H: W = W(L /H)
and = (L /H). This allows one to draw, either analytically or numerically,
diagrams showing the values of the indicators in function of the slenderness, for
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
different typologies of structures. The recent book of P. Samyn [2], although not
exhaustive, contains an impressive amount of typologies that can be compared:
see fig.1 as an example of a comparison of different typologies through the
volume indicator: this diagram is taken from [9]. Similar graphs were developed
for the displacement indicator .
It is clear that P. Samyn see the title of his thesis hereby provided the
architect with a tool allowing to reach a suboptimal design at the stage of the
conceptual design. The fact that he is still using it today, and undoubtedly
producing designs of outstanding quality, proves the robustness of the tools he
introduced.
However, two major objections could be foreseen and they were very soon
subject of controversy. Indeed, the two indicators allow for a preliminary design,
achieving the required performances of strength and stiffness with a minimum
volume of material (a fully stressed design of statically determinate structures,
subject to classical load cases)...but what about (in)stability? It is clear that, in its
most simple form, the developed method does not take at least explicitly - into
account possible buckling phenomena. However it would be unjust to consider
that it completely overlooks the existence of this phenomenon: in both [1] and
(more detailed) in [2], P. Samyn shows that correction factors can be
computed, thus defining an increase of material consumption, and trying to
control the stability of the equilibrium.
Nevertheless, gradually we were convinced that conceptual design should
take into account the totality of the criteria to be satisfied by the structure:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2.
Eventually one also has to consider an indicator describing the dynamic
behaviour of the structure. This is achieved through the use of the indicator of
L
1
= f , , extensively discussed by J. Van
the first natural frequency =
H
Steirteghem in [10]. As one sees, there is a direct link with the indicator of
displacement . This indicator is also directly dependent on the buckling
indicator, which is not surprising: both resonance and buckling phenomena in
elastic systems are related with transformation of energy (compression into
bending or torsion for buckling, potential into kinetic for resonance). An
important observation can be made here: if one accepts all the simplifications
introduced by the concept of morphological indicators and if one has a closer
look at the indicator , one will notice that it is independent on the volume
indicator and thus on the mass of the system. This could bring one to the
conclusion that at least, at the stage of conceptual design there is little hope
that one can improve the dynamic behaviour of the structure through simple
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
addition of mass. This has been confirmed when designing e.g. slender and
elegant footbridges: one quickly comes to the conclusion that the penalty in
mass, in order to achieve an acceptable spectrum of resonance frequencies is
(very) much higher than the one necessary to control its buckling behaviour (see
e.g. [10]).
A very well known example, proving this observation, is the 350m long
Millennium Bridge, designed by Sir Norman Foster in collaboration with Arup
Ltd. See e.g. http://www.arup.com/millenniumbridge/, site from which we quote:
...There are two fundamental ways to limit dynamic excitation:
* Stiffen the structure, so the frequency of the bridge and our footsteps no
longer match;
* Add damping to absorb the energy.
It was concluded that stiffening the bridge to change its frequency was not a
feasible option. The bridge would need to be at least tenfold stiffer laterally to
move its frequency out of the excitation range. The additional structure required
to do this would dramatically change the appearance of the bridge.
It was decided to adopt a damping solution, either active damping or passive
damping. Active damping uses powered devices to apply forces to the structure
to counteract vibrations. Passive damping relies on harnessing the movements of
the structure to absorb energy. (end of quote).
This is only one of the many examples proving that one has to give necessary
attention to the dynamic behaviour of lightweight structures, if one wants to
design an acceptable lightweight structure! But this also proves that solving the
problem with addition of mass and/or stiffness is very seldom the best option. A
new challenge thus for designers!
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Alternate strategies
Probably one of the most challenging and promising alternatives to the design
strategy described in the previous sections consists in the use of new
algorithms, allowing a search for improved typologies, especially in the case of
complicated load conditions. Instead of looking for the optimal typology of
structure, given the span L and the loads F, which yields an optimal slenderness
L/H for the catalogued typology (e.g. a Warren truss), one can open the search
space to yet unknown typologies. This can be achieved using e.g. genetic
algorithms, evolutionary strategies, etc. Verbeeck et al. show in [14] that genetic
algorithms can be very powerful tools in optimisation processes, but above all
have shown in [15] that they can yield new, often unsuspected, typologies, better
suited for the given problem. Important is to mention that the objective function
these algorithms try to minimize is the volume indicator W, the other
morphological indicators reducing the search space by inducing side constraints.
Preliminary conclusions
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgements
Scientific research is very rarely the output of a single brain, and it has not been
different with the present study. I thus want to pay due respect to colleagues and
students. In the first place, two very important contributors to my evolving
insights in structural design and analysis: Philippe Samyn, eminent architect and
engineer, and Pierre Latteur, a structural engineer with a real feeling for
structural behaviour. They triggered my interest for morphological indicators,
allowing me to direct several theses at both Masters and Doctoral level. All of
the others will recognise themselves, but I want to underline the outstanding
research work of Jan Van Steirteghem, architectural engineer, who sort of filled
some gaps left by the excellent work of P. Samyn and P. Latteur, and also
initiated work in the field of dynamics of lightweight structures: he allowed me
to come back to my real passion for structural vibrations.
During the last years I have been also collaborating with one of my former
students, Marijke Mollaert, who finished her Ph.D in 1984 on tensile structures
and, since then, has become a colleague and always has remained a close friend.
I still spend a lot of time discussing new ideas with her and her students, all of
them involved in this exciting field of form-active tensile structures. Although
the discipline of designing these structures is very much different from what we
are doing in my research group, the ideas one can pick up during those informal
meetings are refreshing and really exhilarating. Thank you, Marijke, for your
friendship.
WIT Conferences were also regular meeting points, confronting me with new
ideas and insights: first CADCOMP, later evolving into these HPSM
Conferences, OPTI, but also Structural Repair and Maintenance of Historic
Structures. In particular, I want to mention my friends Santiago Hernandez (U of
La Corua, Spain) and Jerry Connor (MIT, USA), with whom it is an
enlightening experience to discuss all kinds of matters, even structures!
Research needs funding and scholarships and I thus also want to acknowledge
the Institutions which helped this research financially: Fund for Scientific
Research in Flanders (FWO), Fund for Applied Research in Flanders (IWT), the
VUB and several industrial partners.
Last but not least a personal word of gratitude for Carlos A. Brebbia, always
showing more confidence in my skills than I deserve, and allowing me to meet
such fine persons within the Board of Directors of WIT. It is now about 35 years
that we met for the first time and it is always with extreme pleasure that we see
each other, either in the Lodge or at his Conferences.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
11
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
Verbeeck B., Van Steirteghem J., De Wilde W.P., Samyn Ph., The Need
of Numerical Techniques for the Optimization of Structures Using
Morphological Indicators, Proceedings OPTI2005, WIT Press, 2005.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
13
Abstract
The Theory of Morphological Indicators allows a preliminary optimisation of
structures at the stage of conceptual design. Samyn and Latteur developed the
Indicator of Volume to determine the efficiency of structures at early design
stages. The main advantage of this approach is that we only need a very limited
number of parameters. Samyn establishes efficiency curves, with respect to
minimum volume of material, for trusses in which he neglects buckling. Latteur
establishes efficiency curves in which he accounts for buckling. The
displacements are usually checked afterwards to verify if the normative
constraints are not exceeded. In this paper we argue that for trusses, loaded
dynamically and with large spans, dynamics become the dimensioning criterion.
We use the Indicator of the First Natural Frequency to determine the first natural
frequency of trusses. We find that for fully stressed trusses this natural frequency
is usually near to the excitation frequencies of man induced and wind induced
vibrations. Therefore, we need to include dynamics in the optimisation
procedure. We show that for trusses with important spans very large stress
reductions are necessary to obtain acceptable natural frequencies. This stress
reduction comes at the cost of a very important and unacceptable increase of the
volume of material. We determine which typology (Warren, Howe, Pratt) is the
most efficient (minimal volume) with respect to dynamics. Moreover we show
that when dynamics is the dimensioning criterion, the influence of buckling on
the optimum design is negligible since an important stress reduction is necessary.
Finally we propose a work scheme that allows considering dynamics in the
Theory of Morphological Indicators and we provide an example.
Keywords: conceptual design, morphological indicators, trusses, dynamics.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06002
qEF
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
15
The external live loads are subdivided in non co-vibrating live loads Fl1 = cFl
and co-vibrating live loads Fl 2 = (1 c )Fl , with c the share of the non covibrating live loads in the total live loads.
Furthermore, we define the following combinations:
(1)
When calculating the first natural frequency of a structure, we take into account
the co-vibrating loads as vibrating mass. We express the total co-vibrating load
FD as follows:
FD = F0 + F p + Fl 2 = FP + (1 c )Fl
Fl1
Non co-vibrating part
of live loads
cFl
Fl2
Co-vibrating part
of live loads
(1-c)Fl
FD =
Fl2 + Fp + F0 =
zF
all co-vibrating
loads
(2)
Fl =
Fl1 + Fl2
live loads
FE =
Fl + Fp
external
loads
F=
FE + F0
total
loads
Fp
external permanent loads
FP =
Fp + F0 =
F0
self weight
These loads refer to the Ultimate Limit State (ULS) calculation of the structure.
For a dynamic analysis, we must consider loads in a Serviceability Limit State
(SLS). Hence, we can express FD in SLS
F
F
(3)
FD* = P + (1 c ) l
1.35
1.50
And we express the vibrating mass m *D of a structure as:
m *D =
FD*
g
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(4)
(5)
(7)
By substituting eqns (7) and (5) in eqn (6), the expression of the first natural
frequency becomes:
c
gk
(8)
f 1 = cor
2 z * F
With = F k the static displacement, eqn (8) becomes:
c cor
g
2 z *
or as a function of the Displacement Indicator:
c
gE
f 1 = cor
2 z * L
f1 =
(9)
(10)
2f 1
c cor
z * L
gE
(11)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
17
2
j
1
f =
2
(12)
We checked the results of this method numerically with Robobat. The results
obtained using eqn. (12) underestimate the numerical ones marginally
(maximum 10% for small slendernesses).
2.4 Indicator of First Natural Frequency
We can use the Indicator of the First Natural Frequency to determine the first
natural frequency of structures.
=1
L = 90 m
1,5
L = 70 m
L = 50 m
Z=1
= 20
0,5
=0
1
L[m]
5
f [Hz ]
11
7f [ Hz 12
]
f [Hz ]
16
17
f [Hz ]
15
14
10
11
Figure 1:
13
10
12
11
10
concrete
timber
10
ccor =1,10
f1 =
S355
20
z * = 0,44
= 1,0
S235
9
15
c cor ,1
2
E
z*
1
L
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
L as a
abscissa for different material, stress level and value of z * . Since we include the
material, stress level and value of z * in the frequency abscissa the first and
second quadrant must never be changed for this structure.
In this example we determine the first natural frequency of a steel Warren truss
with 10 panels and a span of 50m in which we can neglect buckling ( = 0 ) . For
this example we obtain a first natural frequency of 1,7Hz.
3.1 Scope
We can now study the behavior of trusses that are subjected to dynamic loads.
We consider three truss typologies (Warren, Pratt and Howe):
Warren
Pratt
Howe
We consider three truss typologies (Warren, Pratt and Howe) with:
Three different spans (L = 20m - 50m - 70m)
Three number of panels (n = 4 - 8 - 12)
Three values of z* (z* = 0,15 - 0,44 - 0,74)
Two different materials (timber - steel).
Moreover, we impose that the natural frequency of the structure must be
larger than 5Hz because the man and wind induced frequencies are usually
between 0Hz and 5Hz.
For trusses in which we neglect buckling we can express the constraint on the
first natural frequency as a constraint on geometrical slenderness. First, we use
eqn. (10) compute an upper bound on :
2
gE
c
= dyn = max
0 5 Hz < cor
2 5 z * L
(13)
dyn 2dyn (n + 2 )
2
L dyn + dyn (n + 2 )
<
<
H
(n + 2 2n )
(n + 2 2n)
dyn 2dyn (n + 1) L
<
(n + 1 2n )
H
<
19
(14)
dyn + 2dyn (n + 1)
(15)
(n + 1 2n )
Table 1:
f > 5Hz
=0
co-vibrating load
Material
z* = 0, 74
timber
Number of panels
L = 20m
L = 50m
L = 70m
=
=
=
Number of panels
L = 20m
L = 50m
L = 70m
=
=
=
=
=
=
steel
Timber
0,20
0,08
0,05
0,30
0,12
0,08
0,20
0,08
0,05
0,30
0,12
0,08
steel
timber
0,18
0,07
0,05
0,27
0,11
0,07
steel
4
0,34
0,13
0,09
0,88
0,35
0,25
0,34
0,13
0,09
0,88
0,35
0,25
0,31
0,12
0,09
0,81
0,32
0,23
12
0,16
0,06
0,04
z* = 0, 15
8
0,18
0,07
0,05
Number of panels
L = 20m
L = 50m
L = 70m
z* = 0, 44
4
0,18
0,07
0,05
L H =8
1,00
1,00
0,29
8
12
1,00
1,00
0,29
12
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
0,93
0,37
0,26
4.1 Conclusions
In this paper, we showed that we can include dynamics into the Theory of
Morphological Indicators. This comes at the cost of a larger number of
parameters to consider. However, we believe that in order to be useful we must
include the dynamic behaviour of (lightweight) structures. Moreover, we argue
that the number of parameters to consider are still small and are all available at
the stage of conceptual design.
We developed the Indicator of the First Natural Frequency, , to compute
the first natural frequency of structures. The expression of clearly shows that
the problem is not scale independent anymore.
We found that trusses with large spans, large values of z* and small
stiffness/strength ratio are sensitive to dynamic loads and need important stress
reduction to obtain an acceptable natural frequency.
4.2 Further research
In order to make the Theory of Morphological Indicators more powerful in terms
of dynamic behaviour, the following topics can be worked out in the future:
Stress reduction due to fatigue
Higher natural modes and frequencies
External damping devices (passif, actif, hybrid)
References
[1] Samyn, Ph., Etude compare du volume et du dplacement de structures
bidimensionnelles, sous charges verticales entre deux appuis vers un outil
dvaluation et de pr-dimensionnement des structures, PhD thesis,
Universit de Lige, 1999
[2] Latteur, P., Optimisation des treillis, arcs, poutres et cbles sur base
dindicateurs morphologiques Application aux structures soumises en
partie ou en totalit au flambement, PhD thesis, Vrije Universiteit Brussel,
2000
[3] Van Steirteghem, J., A contribution to the Optimisation of Structures Using
Morphological Indicators: (In) Stability and Dynamics, PhD thesis, Vrije
Universiteit Brussel, 2006
[4] Ponsaert, W., Het gebruik van trillingsdempers in balkconstructies, Master
thesis, Vriie Universiteit Brussel, 2005
[5] Fertis, D.G., Mechanical and Structural Vibrations, John Wiley & Sons,
New York, 1995
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
21
Abstract
When designing a modular system, two main aspects need to be considered: the
aspect of one building stone and the aspect of the overall structure. Both will be
studied here as a function of geometry, mechanical behaviour and historical
background (architectural design, engineering science, etc.).
The state-of-the-art of design principles, used in architectural designs over the
latest decades, will provide the basic knowledge for the generation of a new kind
of modular construction. Combining both the artistic point of view (aesthetics)
and the engineering point of view (calculations, FEM), a new design concept
will be generated: a system with modular hyperbolic paraboloid building stones,
connected to one another through hinged connections.
Keywords: architecture, modular construction, design concept, hyperbolic
paraboloid.
Introduction
Hyperbolic paraboloids
When designing a modular system, two aspects need to be considered: the aspect
of one building stone and the aspect of the overall structure, consisting of several
building stones. Since a modular system is only as strong as its components
(building stones and the connection between them), it is important to select them
with care.
A hyperbolic paraboloid (also called: hypar) building stone (see figure 1) has
many advantages over other types of building stones. Firstly, a hypar surface can
be applied over any foundation shape (rectangular, triangular, circular,).
Secondly, a hypar surface is a curved surface and has therefore better bearing
capacities than flat surfaces, because the forces can more easily be introduced in
an efficient way (i.e. through higher normal forces and lower bending moments).
Due to its anticlastic doubly curved shape, instabilities are less likely to occur:
the convex curve will stiffen the behaviour of the concave curve and vice versa
(see figure 1, on top).
Z
Y
(1)
23
Even the fabrication of a hypar surface is not as difficult as one would expect
from a three dimensional shape: a hypar surface is a doubly ruled surface and can
therefore be formed by two distinct independent families of straight lines (see
figure 1, below), which makes the formwork much easier. When manufacturing
hypar-sandwich panels with composite faces, an adaptable and reusable cable net
could be used as a reconfigurable mould. The net, which is largely based on the
ADAPTENT-concept of Hebbelinck [1], then provides an easy, fast, re-usable
and economical methodology in fabricating hypar-sandwich panels that are to be
used in modular structures (see figures 2 and 3).
Figure 2:
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Design principles
There are many basic concepts that underlie the field of design. All basic tenets
can be grouped into design elements and design principles. They are applicable
to all of the visual arts, amongst which architecture. The design elements are the
components or objects of the design: a line, a point, a surface, colour, etc. The
design principles on the other hand govern the relationship between the design
elements, giving them meaning. There are many additional concepts that are
related to the principles of design. The grid-concept is one of these techniques: it
provides a framework for all basic principles.
3.1 Grid
Geometric patterns have always been a fundamental aspect of human culture. In
nature, geometric order is rarely found. It is rather mans way to understand his
surroundings.
Science uses the grid as a technique to cope with complexity. The grid is the
basis for most techniques which sample and quantitatively analyse the real
world. In archaeology for example, archaeologists use three dimensional grids
as a baseline for complex digs. Excavation takes place within trenches or areas of
about 1 metre square, which are located within a site survey grid so that every
point on the excavation site can be given a horizontal coordinate. The
information, revealed through laborious excavation can then be gathered and
ordered in a rather simple manner.
Ecological sampling methods also make use of the grid-method. If one would
want to know what kind of plants and animals are in a particular habitat, and
how many there are of each species, it is usually impossible to go and count each
and every one present. Samples are taken instead. A method of random sampling
is used to map the area under survey and to overlay it with a numbered grid. In
short, the fundamental technique for filtering the complexity of the observable is
based on the quadrat and transect: small square or cubic areas (quadrats) are
examined in detail along a line or across a grid of evenly distributed sample
points (transect).
For reasons difficult to explain, the significance of the grid in modern and
contemporary architecture (and art) is just as important.
3.2 Modern and contemporary architecture
The grid is one of the oldest architectural design tools. A grid can help a designer
control the positions of built and space elements, making the layout task more
systematic. Especially in laying out plans for new towns and cities, the use of
grids enables the designers to make decisions at the urban scale, nevertheless
providing relative freedom at the block and lot scale for individual developers
and house designers.
Taking it one step further, the grid can also be applied to a single building.
This formal and structural approach to design was seen in the Swiss International
Style. The term International Style was first used in New York City in
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
25
Overall design
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
y
z
Figure 4:
The two-dimensional horizontal grid not only defines the overall design, but
also facilitates the prefabrication of each hypar surface (see section 2). In figure
4, it was chosen to work with a single regular grid to which each building stone
is related.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
27
MESHING
MODULARISING
Figure 5:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Modules
FEM
DESIGN SYSTEM
Object
Finite element
Figure 6:
Meshing
Object
Conclusion
The grid-system has always been an important design principle: not only in
science, art, etc, but also in architecture. Modern and contemporary architecture
have turned it into an undeniable part of building concepts. The evolution of
scientific knowledge has turned mankinds urge of controlling his surroundings,
and thus the grid, into an unmistakable part of building realisations.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
29
Acknowledgment
Financial support from the Institute for the Promotion and Innovation by Science
and Technology in Flanders (IWT-Vlaanderen) is gratefully acknowledged.
References
[1] Hebbelinck, S., A generating system for temporary, adaptable and reusable
nets and tensile structures, PhD-thesis Vrije Universiteit Brussel, Belgium,
2002
[2] Leland, M., About the modern Style, http://www.michael.leland.name/
modern/index.html
[3] Ibelings, H., Supermodernism: architecture in the Age of Globalization, Nai
Publishers, Rotterdam, 1998
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
31
Abstract
Within the framework of sustainable development we strive for structures with a
minimum volume of material. When we only consider criteria on resistance and
buckling, Samyn and Latteur prove that even at the stage of conceptual design a
clear hierarchy among the different truss typologies can be established. Up to
now, stiffness constraints - such as the upper limit on static displacements - were
not considered. However, an optimum obtained by minimising the volume, only
considering the strength criterion, often results in solutions which violate the
stiffness constraint(s). To avoid large displacements a stress level reduction can
be imposed. However, this comes at the cost of a significant volume increase.
With an optimisation process that involves the stiffness constraints at the stage of
conceptual design, an optimum can be obtained without the necessity to alter the
structure drastically afterwards, which partly annihilates the main objective of
minimal use of material. This approach compares the different truss types on a
new priority scale, generating new optima. This implicates a non-negligible
change in the truss choice at conceptual design stage. The solutions are logically
depended on the displacement criterions. This approach forms a first step to a
new design philosophy that considers all the stiffness constraints (static
displacements, resonance, local and global buckling) at conceptual design stage
and is called design for stiffness.
Keywords: morphological indicators, stiffness, strength, truss, steel, static
displacements, optimisation.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06004
Introduction
33
the truss choice at conceptual design stage. As expected, we show that the
solutions are dependent on the displacement criteria.
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Name
Specific
weight
S235
S275
S355
(N/m)
78500
78500
78500
Youngs
modulus E
(GPa)
210
210
210
Allowable
strength
(MPa)
235
275
355
E
894.6
763.6
591.5
(m)
2993.6
3503.2
4522.3
3.1 Strategy
We violate the displacement constraint for a large interval of the slenderness and
for common typologies of trusses. Figure 2 depicts the indicator of displacement
for Warren trusses with minimal W. We observe that for common values of
L ( 1 200 and 1 500 ) an upper value for exists which we cannot exceed.
This value of determines a validity interval for the slenderness.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
35
4.5
3.5
3
2.5
2
1.5
1
= 20
= 30
0.5
= 10
L/H
0
0
Figure 2:
10
15
20
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 3:
3.2 Results
In [2] Latteur selects the optimal (i.e. the lowest) indicator of volume W and its
corresponding slenderness L H opt for every value of the indicator of buckling
. We plot W and L H as a function of (Figure 3). This allows us to
37
efficiency curves for the indicator of volume as function of the slenderness for
= 0,10, 20, 30. = 0 corresponds to the theoretical value of Samyn [1], in
which buckling is neglected. Common values of vary between 10 and 30.
Between the curves of = 10, 20 and 30 linear interpolation is allowed,
certainly at conceptual design stage.
10
9
8
7
= 30
= 20
6
5
= 10
=0
3
2
1
L/H
0
0
10
12
14
16
18
20
0.9
=0
= 10
0.8
L/H
= 20
= 30
0.7
0.6
0.5
18
10
12
14
16
18
20
n opt
16
= 30
14
12
= 20
10
8
6
4
2
= 10
Figure 4:
L/H
=0
0
2
10
12
14
16
18
20
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
9
8
= 30
= 20
7
6
5,46
5
4
3
4,04
3,75
3,16
2,42
2
1
=0
0
0
Figure 5:
10
12
14
16
L/H
18
20
Practical example
We design a S 355 steel Warren truss with a span of 50m and a maximal height
of 4m. We must limit the maximal static displacement to 1 300 . If we only
consider the resistance criterion [1] the slenderness and the mesh number are the
only variables and we obtain a truss with a slenderness of 12,5 and 2 meshes
when we limit the volume of material. In the strategy of Latteur [2], in which
buckling is considered explicitly, more input information is needed to determine
the value of : the selected material, the section type of the bars and the total
load. S355 steel, tubes with thickness/diameter ratio of 0.04 and a total load of
2500kN , result in an indicator of buckling of = 22.62. We obtain an optimum
truss with 11 panels and a slenderness of 12,5. Finally, the constraint on the
static displacement does not influence the slenderness, on the other hand the
number of panels decreases to 5. For this truss we obtain W = 3.75. Since we
include more phenomena (buckling, static constraints), the indicator of volume
obtained by Samyn [1] increases (from 2.42 to 5.46). The solution obtained by
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
39
Latteur [2] also increases due to the constraint on the static displacement (from
3.16 to 4.04). Those results are shown in Figure 5 and clearly illustrate the
importance of considering stiffness constraints at conceptual design stage. Here
we obtain a gain in volume of 8% in comparison with Latteur [2] and up to 46%
with the result of Samyn [1].
Conclusions
We found that trusses with a large stiffness are composed of a small number of
panels. Even though a larger number of panels reduces the buckling length of
compression members, we observe that when we consider the constraint on static
displacements we obtain an optimal solution with a smaller number of panels [1].
Moreover, stiff trusses are usually stocky (typically L H between 1.5 and 5),
which confirms the results obtained by Samyn [1], guiding us towards small
slendernesses. We observe that these stiff trusses are often very light (especially
when the truss is not sensitive to buckling).
The more buckling sensitive the truss, the smaller the necessary stress level
reduction, since the material against buckling also provides stiffness against
displacements. This can be noticed on the curves by the larger s at larger
' s for the same slenderness. On the other hand, a stress level reduction against
displacements means a less buckling sensitive truss. This explains why at larger
slendernesses (the most compliant geometry), the curves of the different s
join. For very large L/Hs and/or very strict displacements constraints, we can
assume that the optimal structures become independent of .
The optimal stress level is not always = 1 . This clearly shows that fully
stressed design doesnt always give the lightest structure.
Finally, Howe and Pratt trusses do not present any advantage in comparison
with Warren trusses, since they not only need more material but are less stiff.
Further research
Acknowledgement
I would like to acknowledge the Institute for the Promotion of Innovation
through Science and Technology in Flanders (IWT Vlaanderen), which helps this
research financially.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1] Samyn P., Etude Compare du Volume et du dplacement de Structures
Isostatiques Bidimensionnelles sous Charges Verticales entre Deux Appuis.
Vers un outil dvaluation de prdimensionnement des structures (Tome 1
4), PhD thesis, Universit de Lige, Belgium, 1999.
[2] Latteur P., Optimisation et Prdimensionnement des Treillis, Arcs, Poutres
et Cbles sur Base dIndicateurs Morphologiques (Tome 1 4), PhD thesis,
Vrije Universiteit Brussel, Belgium, 2000.
[3] Orton A., The way we build now (Chapter 2). Building Structures, E & FN
Spon , pp. 42-44, 1994.
[4] Eurocode Steel/3, NBN ENV 1993-1-1:1992 pp. 49-52, 1992.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
41
Abstract
Adaptable tensile structures are often considered to be either completely opened
or completely closed. The current study is part of a research project studying
adaptable tensile structures which demonstrate stable behaviour within a wide
range of opened positions.
In this paper a simple V-shaped membrane is studied during the unfolding
process. Starting from an initially flat folded membrane, which is not
pre-tensioned, a slight curvature is obtained when it is unfolded due to the fact
that along the folding line a curved section is cut out of the fabric. The tension
introduced in the transverse direction implies a tension in the longitudinal
direction too.
Two cases are analysed: one with a high curvature in the diagonal cable (~5%
sag) and one with a low curvature (~1.3% sag). Based on computer simulations
the form and the tensions are verified for different opening angles. The
deformation under loading is checked for the shape with a low curvature of the
diagonal cable at an opening angle of 70. The results indicate that the membrane
could be used as a fabric roof. Further refined analysis is needed to be able to
implement the presented concept for real applications.
Keywords: tensile structure, formfinding, coated textiles, adaptable shelter,
foldable structures.
Introduction
Figure 1:
Parallel
sliding.
Figure 2:
Central
sliding.
Figure 3:
Circular
sliding.
Figure 4: Folded
Figure 5: Reference
Figure 6: Unfolded
configuration.
configuration.
configuration.
Curved
diagonal.
43
tensioned in the transverse direction, and due to the bi-axial behaviour of the
material, the longitudinal direction is tensioned too.
The opening angles ranging from 60 to 80 are the most appropriate (in terms
of covered area: see Fig. 11, Fig. 12) if one wants to use the unfolded tensioned
membrane as an architectural roof.
Covered
Figure 12: Covered area
area for an
for an opening
angle of 80.
opening
angle of 70.
H
arc AA
arc BB
Figure 13:
Table 1:
Case
S
=
AA
apex angle of
triangular
module
opening angle
of triangular
module
height of the
longitudinal
arc
arc in
longitudinal
direction
(along Xaxis)
arc in
transversal
direction
(along Yaxis)
span of the
longitudinal
arc
The height to span ratio of the diagonal for different forces in the
diagonal cable.
Elevation
Force
[kN]
Height
[m]
H/S
[%]
2.1453
0.0775
1.29
1.7173
0.0952
1.59
1.2895
0.1233
2.06
0.8622
0.1750
2.92
0.4359
0.3013
5.02
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
45
2.50
Force [kN]
2.00
1.50
1.00
0.50
0.00
1.29
1.59
2.06
2.92
5.02
Figure 14:
It is obvious that the smaller the force in the diagonal cable, the greater the
curvature of the diagonal arc will be (for the same pretension in the membrane of
0.02kN/m).
But on the other hand, the greater the curvature of the diagonal cable, the
higher the stresses in the transverse direction will be when unfolding.
Tension [kN/m]
0.50
0.40
0.30
0.20
0.10
0.00
10
20
30
40
50
60
70
80
Opening angle []
Figure 15:
From the calculations with a stiffness of 1kN in the internal net it can be
concluded that
- in the longitudinal direction the tension in the membrane remains at a
value of about 0.02kN/m
- in the direction perpendicular to the folding axis the tension increases
from 0.02kN/m to 0.44kN/m for larger opening angles.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
6.00
3.40
5.00
3.30
3.20
4.00
H/S [%]
Force [kN]
3.00
2.00
3.10
3.00
2.90
2.80
1.00
2.70
0.00
10
20
30
40
50
60
70
80
10
20
Opening angle []
Figure 16:
30
40
50
60
70
80
Opening angle []
Figure 18: A higher resultant force Figure 19: A lower resultant force
with higher membrane
with
lower
membrane
tensions at 70
tensions at 60.
3.2 Case 5: Lower curvature in the diagonal cable
The influence of the opening angle on the stresses in the membrane and the
diagonal cable is verified for a more realistic stiffness in the internal net of
150kN. The stiffness for the diagonal cable remains 5000kN. The 150kN per
0.25m corresponds to a value of 600kN/m valid for a typical PVC coated
polyester membrane.
From the calculations where the opening angle varies between 60 and 80
can be concluded that:
- in the longitudinal direction the tension in the membrane
varies from a value of about 0.6kN/m up to 2.2kN/m for
larger opening angles
- in the direction perpendicular to the folding axis the tension
increases from 0.4kN/m to 20.0kN/m for larger opening
angles
Similar to the membrane tensions in the longitudinal direction, the forces in
the diagonal cable increase up to a maximum value and then decrease due to
variation of the opening angle.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 2:
47
Elevation
Front view
40
2.50
Tension [kN/m]
Tension [kN/m]
80
2.00
1.50
1.00
0.50
0.00
60
65
70
75
80
25.00
20.00
15.00
10.00
5.00
0.00
60
Opening angle []
65
70
75
80
Opening angle[]
Figure 20: Tension in the longitudinal Figure 21: Tension in the transverse
direction due to variation
direction due to variation of
the opening angle case 5.
of the opening angle
case 5.
Table 3:
65
80
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Front view
30.00
25.00
20.00
15.00
10.00
5.00
0.00
5.00
4.00
H/S [%]
Force [kN]
3.00
2.00
1.00
0.00
60
65
70
75
80
60
65
Figure 22:
80
Front view
Elevation
Front view
Figure 25:
75
Figure 24:
70
Opening angle []
Opening angle []
Still for case 5, with a stiffness of 150kN in the internal net and 5000kN in the
diagonal cable, external loading will be applied for an opening angle of 70. Two
representative load cases have been chosen [5]: a uniform snow load of
0.4kN/m2 and an upward wind load of 0.6kN/m2. For the equilibrium form with
an opening angle of 70 without external loading the average tension in the
membrane is 2kN/m in the longitudinal direction and 4kN/m in the transverse
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
49
direction. The static analysis under external loading combines the displacement
of the fixed nodes according to the opening angle of 70 with the application of
load vectors for the selected load case on the equilibrium form without external
loading. A uniform snow load of 0.4kN/m2 has been applied. In this case the
maximum deflection is 2cm. In the transverse (load bearing) direction the
average tension increases from 4 up to 5kN/m.
Next an upward wind load of 0.6kN/m2 (suction) has been applied. In this
case a maximum deflection of 4cm occurs. In the longitudinal direction the
average tension increases from 2 up to 3.1kN/m.
Plan
Elevation
Front view
Figure 26:
General remarks
The simulations only give a first estimation of the structural behaviour: the
model does not take into account the shear stiffness of the coated fabric, the
material behaviour is considered to be linear and independent of the stress level
and frequently folding and unfolding will also influence the material behaviour.
For more precise results a refined model is required. Moreover, an optimised
configuration could be found by choosing a different stiffness or selecting
another boundary geometry in plan view.
Conclusion
For the same external loading on a membrane roof a more expressive double
curvature will imply lower tensions in the membrane. Nevertheless more and
more applications use a flat appearing curvature to cover box-like spaces. The
rhombus-shaped membrane elements will be used as cladding components in a
mobile foldable shelter consisting of articulated bars as the primary load bearing
structure [6]. Based on the simulation of a rhombus shaped membrane it is
shown that the unfolding of flat membrane pieces can, within a certain range of
opening angles, create a slightly curved tensioned membrane which can resist, in
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1]
[2]
[3]
[4]
[5]
[6]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 2
Composite materials
and structures
53
Abstract
The aim of this work was the development of a new composite material based on
natural fibres as reinforcement and an acrylic thermoset as binder to investigate
the material properties for various applications. All applications were made by
forming and curing the composite material in a hot forming press. In a first step
the technology for formation of such a composite material was tested: Based on
the nonwoven-process natural fibre mats were made, only composed of natural
fibres without binder and with different area weights. After that, different
technologies of application of binder and additives like hydrophobizing agents
were tested. In the second step the material properties (flexural and impact
strength, water uptake and moisture expansion, testing for climatic extremes and
sound-absorbing properties) were investigated for different applications as
function of the density of the pressed parts, the mixing ratio of binder and
additives. The following properties can be given as very special features of the
material: In comparison with composites made of usual thermoset binders the
fabricator has not to apply the resin immediately before processing. Secondly the
material can be stored easily, because the curing is started beyond 100C, at
lower temperatures the binder is thermoplastic. In comparison with composites
made of thermoplastic binders, a lesser fraction of binder is necessary and the
pressed parts have a very high flexural stiffness. Finally the critical properties
like the water uptake and moisture expansion were improved by the addition of
hydrophobizing agents.
Keywords: natural fibres, nonwoven, composites, acrylic resin, binding
technology.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06006
Introduction
Actual industrial demand for natural fibres has developed over the past few
years. Today, the use of natural fibres has become common in some applications;
the most important customer is the automotive industry. Technically speaking,
the use of natural fibres in automotive applications involves primarily pressmolded composites, produced by the pressing of a nonwoven mat with a binder.
Typical uses are in door panels, hat racks, and trunk liners. Two production
technologies are commercially employed:
a blend of natural and polypropylene fibres is processed into a nonwoven
mat and pressed into the desired shape under heat ("thermoplastic matrix");
nonwoven mats are coated with thermosets, such as epoxy resin or
polyurethane and moulded. The ultimate material is generated by
polymerization and hardening of the resin ("duromeric matrix").
The main reasons for the use of natural fibres are:
Weight reduction of 10 to 30%
Good mechanical and manufacturing properties
Possibility to manufacture complex structural elements from one
material in a single pass
Good performance in accidents (high stability, no splintering)
Superior environmental balance during material and energetic use
Occupational health advantages compared to glass fibres
No emissions of toxic substances
Overall cost advantage compared to conventional construction
The aim of this work is the development of a new composite material, based
on natural fibres as reinforcement but instead of a thermoplastic binder only a
thermoset is used to realise a new range of applications. Instead of the commonly
used thermosets like epoxy resins, polyurethanes or aqueous phenolic resins,
another thermoset is used, which can be processed easier and which corresponds
better to the proposition of an ecological awareness.
Due to the machines of the partner company the production of fibre mats is
based on the nonwovens technology: dry-laying staple fibres (the fibres are 1 to
20 cm or longer, but not continuous) by an aerodynamic web formation and
mechanically bonded by needle-punching. The term nonwoven is defined as a
manufactured sheet, web or bat of directionally or randomly oriented fibres,
bonded by friction, and/or cohesion and/or adhesion, excluding paper or products
which are woven, knitted, tufted stitch bonded incorporating binding yarns or
filaments, or felted by wet milling, whether or not additionally needled. The
fibres may be of natural or man-made origin. They may be staple or continuous
or be formed in situ [1, 6].
Materials
For composites composed of natural fibres, blends of natural fibres (e.g. flax and
kenaf or flax and hemp) are particularly interesting. The finer flax fibres impart
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
55
high stability, but because they impede complete soaking with thermosetting
binders, fractures may develop. Only the mixture with coarser fibres like kenaf
fibres achieves an optimum balance between stability and complete saturation
with the binder. In this project all nonwovens are a mixture of kenaf and flax at
the ratio of 50:50.
2.1 Kenaf fibres
Kenaf is the term for the plant Hibiscus cannabinus L., a warm season annual
closely related to cotton (Gossypium hirsutum L.) and okra (Abelmoschus
esculentus L.). The kenaf fibres are gained from the bark of the stem of the
plants. The staple fibres have a length of 40 60 mm and an average thickness of
25 m. Because of their mechanical properties they are used as reinforcing fibres
in a composite [2].
2.2 Flax fibres
Flax fibres are amongst the oldest fibre crops in the world and the use of flax for
the production of linen goes back 5000 years. Flax fibres are soft, lustrous and
flexible. They are stronger than cotton fibre but less elastic. Because of their
mechanical properties they are used as reinforcing fibres in a composite [3].
2.3 Matrix
The used matrix system in this work is a reactive aqueous acrylic resin, which is
free of phenol and formaldehyde. This one-component resin has good storage
stability and cross-links upon heating to 180 220C. After the impregnation the
fibre mats are dried. The binder begins to cure beyond 100C, and semi-finished
products impregnated with the acrylic resin therefore have excellent storage
stability in comparison to other thermosets (e.g. epoxy resins or polyurethane).
During the hot press step the impregnated mats are formed and the resin is crosslinked. Composite parts bound with the acrylic resin show high stiffness and
strength and meet the ecological requirements of the automotive industry. To
achieve the best mechanical properties and moisture resistance, press moulding
is carried out with tool temperatures of about 200 to 220C and cycle times of
one minute or less [4, 5].
2.4 Additives
In this project two types of additives are used to improve the properties of the
composite. The first is a borate, which contributes to avoid the creation of
mildew, because of the moisture content in nonwoven composed of natural
fibres. The second additive is a paraffin wax, which improves the water
resistance (water uptake and moisture expansion, influence of climatic extremes
on material properties) of the moulded part. In particular the type of paraffin wax
has to be stable against acid, because of the acidity of the used acrylic resin [5].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Technologies of binding
The first task of this work was to find a technology to apply the binder to the
natural fibres. In the following second task the found technology was adjusted to
find the optimised settings for the application. The application process of the
binder can be done in two ways. Either the naturals fibres are impregnated before
they are laid to fibre mats or the application is done subsequent to the
nonwovens process. All 3 tested technologies in this project follow the second
way of impregnation with the following pre-adjustments:
Used binders are liquid, especially aqueous binders
Additives are mixed together with the binder before impregnating
3.1 Foam impregnation technology
The binder is mixed to a foam by a mixing machine and applied to the rotating
rolls. The nonwoven, composed of natural fibres, is impregnated by pressure
between the roller clearance. After that the impregnated nonwoven is dried in an
oven and winded to a roll or cut into formatted pieces.
Figure 1:
57
two rolls. After that the impregnated nonwoven is dried in a vertical drying
tower and winded to a roll or cut into formatted pieces.
Figure 2:
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
The impregnated mats are moulded to composite parts and the following
properties complying with DIN standards are tested
Table 1:
Measured properties.
Property
Flexural strength [N/mm]
Flexural modulus of elasticity (MOE) [N/mm]
Water uptake [%]
Increase of thickness of samples due to moisture
expansion [%]
Impact strength [mJ/mm]
Standard
DIN EN 310
DIN EN 310
DIN 52351
DIN 52351
DIN EN ISO 179
90
80
70
60
50
40
30
20
10
0
25 % resin
25 % resin;
transverse
15 % resin
15 % resin;
transverse
15 % resin
0.1
0.3
0.5
0.7
0.9
1.1
15 % resin;
transverse
Density [g/cm]
Figure 4:
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
25 % resin
25 % resin;
transverse
15 % resin
15 % resin;
transverse
15 % resin
0.1
0.3
0.5
0.7
0.9
Density [g/cm]
Figure 5:
59
1.1
15 % resin;
transverse
Both flexural properties, the flexural strength and the flexural modulus of
elasticity, primarily depend on the density of the moulded composite. The higher
the density of the moulded part is the higher are the flexural properties. The
binder content has only a small effect on the flexural properties; for samples in
production direction and with a higher density the properties are higher, if the
binder content is 15% instead of 25%. In comparison to other composite
materials based on natural fibres the attention should be paid to the high values
of flexural properties for samples with a density of 0.9 g/cm and above: samples
in production direction have a flexural MOE between 7000 and 8000 N/mm and
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
350
300
250
15 % resin; Density
= 0.3 - 0.9 g/cm
200
15 % resin
150
25 % resin
100
50
0
0.1
Figure 6:
0.3
0.5
0.7
Density [g/cm]
0.9
1.1
40
38
35
30
25
25 % resin; Density =
0.6 - 0.9 g/cm
25
20
15 % resin; Density =
0.3 - 0.9 g/cm
14
12
15
10
5
0
10
15
20
25
30
Figure 7:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
61
content
Content
additive [%]
0
1.5
0
1.5
of
Water
[%]
58
22
31
16
uptake
Moisture
expansion [%]
30
14
13
6
coefficient of absorption
1.20
1.00
0.80
0.60
0.40
0.20
0.00
400
500
630
800
1000
1250
1600
2000
2500
3150
4000
5000
6300
frequency [Hz]
Default material
Sample 02; thickness = 5.5 mm
Sample 04; thickness = 3 mm
Figure 8:
Summary
References
[1]
[2]
[3]
[4]
[5]
[6]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
63
Abstract
This paper summarizes test results of an experimental study on threaded steel
rods epoxied into structural hardwood members.
Two test campaigns were carried out. In the first one 36 test specimens of
Tali sawn timber (Erythropleum ivorense) with threaded steel rods were tested.
Tests arose from the need to evaluate the behaviour of this type of joint because
of its use in the construction of hollow bars spatial mesh of laminated timber.
Three anchorage lengths and two bar diameters were tested. Moreover, two
different thicknesses of the glue line, 1mm and 3 mm, were investigated.
In order to complement the first tests a new experimental study with 180 test
specimens was carried out. In the second tests, double-sided pull-out tests of
Spanish chestnut timber saw specimens were performed. Small diameter
threaded steel rods (8, 10 y 12 mm) quality 8.8 (yield stress 640 MPa) were used.
In the paper the most significant results of both test campaigns are
summarized, as well as the comparison with the existing design formulae for of
this type of union.
Keywords: joint design, glued-in rods, adhesives for wood, hollow bars, spatial
mesh, destructive testing.
Introduction
Our team has been researching for a long time on space frames. Timber space
frames are a particular aspect of this research. Special attention has been paid to
the use of hollow sections in the making of this type of structure. Several patents
are the outcome of our pioneering work in this field.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06007
Figure 2:
The only rules references for the calculation of this type of unions appears
collected as informative annex in the Eurocdigo 5, according to:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
65
R ax,k = dequ Lg f vk
(1)
-0,2
f vk = 0,8 10-3 1,5
k d
(2)
(3)
(4)
with:
Rax,k
fws
fwl
k
d
Lg
Adhesive
thickness
Anchorage
length
Specimen
dimensions
axaxL
Lb
Lm
10 mm
d e
D
a
b
a
1
2
3
4
5
6
7
8
9
10
11
12
24
24
24
24
24
24
27
27
27
27
27
27
26
30
26
30
26
30
29
33
29
33
29
33
1
3
1
3
1
3
1
3
1
3
1
3
[mm]
200
200
250
250
300
300
200
200
250
250
300
300
[mm]
Edge distance
Hole diameter
Embedded rod
slenderness
Rod diameter
Series
Table 1:
L
d
a/2
d
75x75x210
8,33
75x75x260
10,42
75x75x310
12,50
90x90x210
7,41
90x90x260
9,26
90x90x310
11,11
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
1,56 d
1,67 d
67
The test sample was longitudinally drilled with an auger drill to the centre of
the section. The drill hole went across all the specimen length. The lower end of
the sample was then closed with a 10 mm thick piece of elastic material (elastic
stopper). The adhesive was blown to fill the drilled cavity with a mixing gun.
Afterwards, the threaded bars were inserted manually by continuous pressing and
rotating. The particular design of the test sample has been chosen in order to
enable the test to be carried out in an INSTRON mod. 8805, 1000kN one-sided
pull-compression universal dynamic press (Figure 3). The elastic stopper at the
lower end of the test sample allows the threaded bar to glide down when it
reaches its load limit. For test reliability purposes, the rod exposed end was
fastened to the load cell by means of a nut. Also, the test sample was placed on
the base plate with a 3 mm thick neoprene sheet in between, in order to take the
imperfections of the support.
A picture of the test device is shown in figure 3.
Figure 3:
Results
The INSTRON test machine has an electronic device that controls the
displacement of the load cell. It also interprets the results with a computer
system that gives the digitalized relationship between load, displacement and test
time-length. The graphic representation of results showed correlation between
stiffness of the three systems adhesives which were used in the second
experimental campaign.
More repeated failure modes were shear failure in timber, and timber
splitting. For specimens with the lowest diameter the main sample failure was
tensile failure in steel threaded rods.
Using Eurocode 5 and Riberholt formulae the theoretical values for each
series were calculated. Both experimental and theoretical values were plotted to
assess correlation between them.
One representative example of each experimental campaign is shown in the
next figures. Solid lines in the first one (Figure 6) represent theoretical values for
Elondo specimens, with a characteristic density of 865 kg/m3, a rod diameter of
24 mm and a glueline thickness of 1mm, with anchorage lengths from 150 to 350
mm. Crosses show experimental results, corresponding series 1, 3 and 5.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 2:
69
La
La
Li
Le
da
da
Lb
B1b1
L
Total rod
length
Cross-section
side
Outer rod
length
Lb
Timber
specimen
length
Le
Inner gap
length
Anchorage
length
Li
Adhesive
thickness
La
Hole diameter
Rod diameter
da
Series
[mm]
[mm]
[mm]
[mm]
[mm]
[mm]
[mm]
[mm]
[mm]
1a
1b
1c
1d
1e
2a
2b
2c
2d
2e
3a
3b
3c
3d
3e
8
8
8
8
8
10
10
10
10
10
12
12
12
12
12
10
10
10
10
10
12
12
12
12
12
14
14
14
14
14
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
60
90
120
150
180
60
90
120
150
180
60
90
120
150
180
60
90
120
150
140
60
90
120
150
140
60
90
120
150
140
180
270
360
450
500
180
270
360
450
500
180
270
360
450
500
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
180
210
240
270
300
180
210
240
270
300
180
210
240
270
300
42
42
42
42
42
60
60
60
60
60
72
72
72
72
72
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 5:
LOAD [KN]
400
300
EURO
OLT
RIBERH
200
100
150
CODE
200
250
300
350
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
71
LOAD [KN]
50
40
30
20
10
T
OL
ERH
RIB
DE
OCO
EUR
0
40
60
90
120
150
180
200
Figure 7:
Results show that more works needs to be done in joints made with rods
glued in high density hardwood timber. This works will allow one to adapt
formulae to this material characteristic and will allow one to study the influence
of edge distance. Our research team is working about it.
Conclusions
Several series of samples have been tested. The test samples were made with
threaded steel rods glued into two different kinds of hardwood timber. The rods
were glued using three different systems of two-component epoxy-base
adhesive.
The test samples were made varying the following parameters: glued
anchorage length, rod diameter and glue thickness
The obtained results have been compared to existing theoretical formulations.
The comparison showed a significant difference between the theoretical
predictions and the test results when high density hardwood is used.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgment
This research is sponsored by the Ministry of Science and Technology through
research project titled UNIONES METLICAS ENCOLADAS CON
ADHESIVOS EN BARRAS DE MADERA (Glued Anchored Timber Joints).
The financial support is gratefully acknowledged.
References
[1]
[2]
[3]
[4]
[5]
[6]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
73
Abstract
A method to develop finite element models of the rocket motor cases of a
strategic missile is presented. Based on the use of multivariate analyses, this
approach is made necessary given the impossibility to control all input data:
difficult characterization of filament wound materials, influence of
manufacturing and qualification processes, unknown fabrication parameters, etc.
An initial reference model is built up using preliminary data and theoretical
winding laws. Having compared the models predictions and available structural
test results, a sensitivity analysis is carried out to discuss the individual influence
of the input data on the accuracy of the predictions. The multivariate analysis
finally enables a global assessment of the parametric analysis results.
Keywords: rocket motor case, composite filament winding, finite element model,
variable scattering, multivariate analysis, test prediction.
Introduction
In addition to providing ground test facilities for the development of the future
generation of French strategic missiles, the Centre dAchvement et dEssais des
Propulseurs et Engins (CAEPE) is responsible for evaluating the degree of
performance, durability and security of the solid propellant rocket motors
constituting the missiles. The mechanical expertise work is carried out using
finite element models able to predict the stresses and strains of the mechanical
parts throughout the fabrication and service life.
Many difficulties arise when attempting to describe the mechanical behaviour
of the rocket motor cases. The filament-wound composite structures have
complex geometry and properties especially in the dome area. As the wound
layers are added on a cylindrical mandrel, the curvilinear path leads to a
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06008
2.1 Architecture
The solid propellant rocket motors corresponding to the two first propulsion
stages of the French strategic missiles share a common outer diameter (> 2 m).
Their structure is based on the same architecture and the same materials:
- The carbon-epoxy filament-wound pressure vessel, acting both as the
propellant tank and the combustion chamber,
- The two metal polar mountings located at the aft and front openings to
provide connection with the igniter and the nozzle,
- The two cylindrical skirts, made of both carbon-epoxy tissues and
circumferential windings, ensuring the connection with the rest of the
carrier,
- Rubber connections between the vessel and the skirts on the one hand, and
between the vessel and the polar mountings on the other hand.
2.2 Winding law
The composite envelope is constituted of a succession of circumferential
windings on the cylindrical part and satellite windings running between the two
openings. The laminate lay-up in the cylindrical area is rather simple to describe
with an assumed constant value of winding angle and direction of all plies.
Conversely, the complex dome geometry involves a rapid change in angle and
thickness along a meridian. Different netting theories [2] based on geometrical or
mechanical approaches describe this distribution. The planar theory used here
assumes that the fiber patterns lie in a plane which is tangent to the polar opening
at one end and tangent to the opposite side of the polar opening at the other end.
The winding angle calculated (Fig. 3a) with the planar theory at a given
location is applied as a constant value to all plies. The thickness distribution
(Fig. 3a) shows good agreement with the measured values (Fig. 3b). A
preliminary analysis was performed to investigate the effect of an evolution of
the winding angle in the thickness direction, as suggested by Park et al [3]. The
thickness of the first ply was calculated from the mandrel shape whereas the
subsequent plies were calculated with the updated shape. The difference in angle
between plies was particularly noticeable near the polar bosses where the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
75
Figure 1:
Figure 2:
Winding path.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
a) Winding angle
Figure 3:
b) Thickness
a) First stage
Figure 4:
b) Second stage
3D mesh using solid elements.
The model contains general boundary conditions for axisymmetry and full
displacement constraints on the frames to take account of the inter-stage skirts
which are not included in the models.
2.4 Evaluation of the models
Two load cases are selected to evaluate the finite element model performance :
internal pressure (Fig. 5a) and axial compression (Fig. 5b). The internal pressure
load case is corresponding to the pressure proof test carried out to accept the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
77
a) Internal pressure
Figure 5:
b) Axial compression
a) First stage
b) Second stage
c) First stage
d) Second stage
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
a) First stage
Figure 7:
b) Second stage
79
The different parameters above are modified one by one in the reference
model (Tab. 1). For each of them, a possible variation is assumed and each case
results in a new computation being performed. The matrix degradation is
modelled by a decrease of the transverse modulus of the hoop layers, the
winding laws are transformed by an offset or a multiplying coefficient, and the
initial geometry is modified by means of a prior pressurization of the model.
Table 1:
(1)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
The results of the multivariate analysis (Tab. 3) show, on the opposite of the
analysis performed on individual parameters, that no significant modification has
to be applied to the reference materials properties. The 1st stage model can be
optimized by a slight modification of the winding law which was initially
approached by a theoretical law. For the 2nd stage model, the best solution is
essentially obtained through the use of a prior deformation of the model with a
18 bar internal pressure.
Table 3:
The verification carried out with this new set of input data confirms a
significant improvement of the predictions for the internal pressure load case
(Tab. 4), for the 2nd stage model (Fig. 8b, d). For the 1st stage model (Fig. 8a, c),
the axial behaviour of the rear dome is much improved at the cost of a slight
degradation of radial displacements prediction. The initial calculations for the
compression load case (Fig. 7) are not affected by the modifications.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 4:
Figure 8:
81
a) First stage
b) Second stage
c) First stage
d) Second stage
Conclusion
This study shows the potential interest of multivariate analyses for the
development of mechanical models for structures with uncertainties concerning
the input data (difficult characterization of filament wound materials, influence
of the fabrication and qualification processes, unknown winding parameters,
etc.). The method presented here for two structures and two load cases results in
a considerable improvement of the predictions with a coherent modification of
the input data. This procedure can be broadened to a multiple load case analysis
with non-linear fits for each parameter.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
83
M. Sejnoha
& J. Zeman
Czech Technical University in Prague, Faculty of Civil Engineering,
Department of Structural Mechanics, Prague, Czech Republic
Abstract
A class of heterogeneous material systems often regarded as random or imperfect
composites is addressed in this paper. The literature now offers a number of contributions that open the way to the analysis of large material systems with complex microstructures while taking advantage of popular micromechanics based
approaches building on periodicity and first order homogenization techniques.
Until recently the attention has been mostly limited to rather classical material
systems such as unidirectional fibrous composites and textiles with emphasis on
various types of imperfections developed during fabrication process on both microscale (level of fiber bundles) and meso-scale (level of textile geometry). From the
basic mechanics point of view, however, it appears logical to exploit the essential
principles of the proposed procedures in bridging the gap between mechanical and
civil engineering applications. In this regard, historical masonry structures classified as systems composed of more than one material component serve as a typical
example of civil engineering applications, which may benefit from standard first
order homogenization schemes extended to account for possibly irregular arrangement of individual stone blocks. In this contribution, both groups of material systems will be treated on the same footing demonstrating the applicability of basic
homogenization techniques as well as similarities between various heterogeneous
material systems when referred to as random or imperfect.
Keywords: random and imperfect composites, textiles, masonry, periodic unit cell,
two-point probability function.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06009
1 Introduction
Techniques of numerical first-order homogenization have acquired a considerable
attention particularly in applications where detailed numerical analysis of complex heterogeneous material systems proved to be prohibitively expensive. Natural
assumption of the existence of periodic microstructure, e.g., periodic distribution
of fibers in the metal matrix or ceramic matrix composites [1], often serve as the
point of departure. The resulting homogenized or effective material parameters are
then used in large scale structural analysis. Unfortunately, in the vast majority of
real material systems the assumption of periodicity generally fails owing to the
presence of various types of geometrical imperfections usually developed during
fabrication. Giving up the benefit of periodic fields in such cases seems, therefore,
reasonable but definitely not very practical. Instead, the recently introduced concept of statistically equivalent periodic unit cell (SEPUC) [2, 3, 4], appears as a
suitable method of attack.
The leading idea of this approach is to replace a complex non-periodic microstructure by a certain periodic unit cell (PUC), which still optimally resembles the
original microstructure in a proper sense. Here, this objective is formalized as a
difference between appropriate statistical descriptors related to the original media
and the periodic unit cell, respectively. If the original microstructure and the periodic unit cell is described by an identical set of parameters, this problem coincides with reconstruction of random materials. If the periodic unit cell is described
by a substantially smaller number of parameters in order to reduce the problem
complexity, it leads to a problem of the best approximation within the selected
statistical descriptors.
It will be shown in subsequent paragraphs that the proposed technique is applicable not only to more or less classical types of composites represented here by
textiles, but also to more conventional class of material systems such as historical
masonry structures.
The principle idea is evident from Figs. 1-3 showing images of real material
or structural systems with corresponding simplified representative volume elements (RVE) presented in terms of certain statistically equivalent periodic unit
cells. When referring to textiles, Figs. 1 and 2, the crucial sources of imperfections
are attributed to generally random arrangement of fibers within the bundle crosssection (microlevel) and the waviness, misalignment and/or non-uniform crosssectional aspect ratio of individual bundles in the longitudinal direction (mesolevel). Qualitatively similar types of imperfections associated with irregular
arrangement of stone blocks, both in terms of block sizes and their location,
together with a variable thickness of the mortar phase are often encountered when
dealing with historical masonry structures. A typical example is plotted in Fig. 3(a)
showing a parapet wall of the Charles Bridge in Prague.
The stepping stone in the analysis of all systems is the possibility to replace
the original color images, Figs. 1(a)-3(a), by their binary counterparts, Figs. 1(b)3(b). The latter representatives of the true micro or meso-structures are further
exploited in the next section when deriving various statistical descriptors. MatchWIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
85
ing the material statistics of the real microstructure with those corresponding to
simplified periodic unit cell in the framework of a certain optimization problem
then allows for deriving the desired geometrical parameters needed in the construction of individual SEPUCs. Knowing the periodic unit cell then opens the way for
the derivation of effective elastic properties using the well known elements of first
order homogenization procedure briefly outlined in Section 3.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(1)
For the case of statistically homogeneous and ergodic media, information contained in the one-point probability function reduces to the volume fraction of
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
87
2
0
(i, j) Srs (i, j) ,
Srs
(3)
0
where Srs
is the two-point probability function related to the original microstructure while Srs stands for the two-point probability function of the idealized
unit cell. It can be shown that the objective function E is non-convex, multi-modal
and discontinuous due to the effect of limited bitmap resolution. Based on our
previous works, a stochastic global optimization algorithm based on combination
of real-valued genetic algorithms and the simulated annealing method, see [6],
is employed to solve this optimization problem. This approach was successful in
delivering the desired periodic unit cells for all material systems considered herein.
Figure 4: (a) Localized equivalent strain, (b) Mesoscopic response for 5-fiber and
10-fiber PUCs.
that the average of in the unit cell vanishes. The local stress fields in the PUC
are constrained by constitutive equations and equilibrium conditions
(x) = L(x) : (x),
div (x) = 0,
(5)
89
Figure 5: (a) Experimental setup, (b) Simulations for various material parameters of interface elements, (c) Experimental vs. numerical response, (d)
Mesoscopic response.
4 Conclusion
A rather general approach to the analysis of heterogeneous materials with either
random or imperfect microstructures was reviewed. The basic scheme assumes
formulation of a certain periodic unit cell statistically equivalent (up to two-point
probability function) to real material systems. It is expected that the periodic unit
cell being statistically similar (from the geometrical point of view) to real systems
will also provide similar (at best the same) mechanical response. The robustness of
this approach has been demonstrated through applications to rather different material systems varying from plane weave textiles over to natural stone masonry. A
special attention was further devoted to the derivation of effective fracture energy
of masonry systems to support applicability of homogenization techniques also to
quasi-brittle materials.
Acknowledgment
Grant No. 103/04/1321 and CEZ MSM
The financial support provided by GACR
6840770003 is gratefully acknowledged.
References
[1] Dvorak, G.J., Bahei-El-Din, Y.A. & Wafa, A., Implementation of the transformation field analysis for inelastic composite-materials. Computational
Mechanics, 14(3), pp. 201228, 1994.
91
[3] Sejnoha,
M. & Zeman, J., Overall viscoelastic response of random fibrous
composites with statistically quasi uniform distribution of reinforcements.
Computer Methods in Applied Mechanics and Engineering, 191(44),
pp. 50275044, 2002.
[10] Cervenka,
V., Jendele, L. & Cervenka,
J., ATENA Program Documentation
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
93
Abstract
This paper presents the experimental results of nine roof panels made of
Ferrocement. Two types of channel sections and one type of box section were
tested. All panels were 2m long, 470mm wide and 20mm thick. Channel type A
had side edge beams 95mm deep and channel type B had side edge beams 50mm
deep. The depth of the box section was 95mm. Thin hexagonal wire mesh was
used as reinforcement. The number of wire mesh layers was varied between two
to six. The wires were impregnated midway through the thickness of the panels.
The panels were tested for bending moment with simple supports. The main
variables studied were the number of wire mesh layers, the cross sectional shape
of the panel and the depth of edge beam. Tests revealed that all panels showed
acceptable strength for roofing systems. The increase in the number of wire mesh
layers leads to an increase in the flexural strength. The box section showed
strength similar to that of the channel section with 95mm edge beam. The
channels with 50mm deep edge beams showed strength much less than the ones
with 95mm edge beam and box section.
Keywords: Ferrocement, fibre reinforcement, slab panels, bending, box section
panels, channel panels.
Introduction
95
flexural load and increases the first crack load. Considerable reduction in crack
widths and spacing was observed with specimen with Ferrocement layers. The
performance of Ferrocement panels under normal, moderate and hostile
environments was investigated by Masood et al. [9]. They concluded that the
flexural capacity of the panel increases with the addition of fly ash. Considerable
deterioration of wire meshes fabric was observed due to sustained exposure in
saline casting and curing condition. Recently, Hago et al. [10] conducted 6
experimental tests to study the ultimate and service behaviour of Ferrocement
roof slab panels. The parameters studied include: the effect the effect of the
percentage of wire mesh reinforcement by volume and the structural shape of the
panels on the ultimate flexural strength, first crack load, crack spacing and loaddeformation behaviour. The results demonstrated that the monolithic shallow
edge Ferrocement beams with the panels considerably improves the service and
ultimate behaviour, irrespective of the steel layers used. Also, slabs with channel
sections supported larger ultimate loads and behaved better under service loads
than their flat slabs counterparts. Due to large deflections experienced by the
thin panels, large deflection theory was adopted in the analysis. Good agreement
was obtained between the theoretical and experimental ultimate loads using the
proposed mathematical model.
In this research, nine simply supported slab panels were tested for flexure.
The specimens were arranged in three categories based on the cross-section:
channel section type A, channel section type B and box section. The aim was to
study the effects of the shape of cross-section and the number of wire mesh
layers on the behaviour and ultimate capacity of the tested panels. The panels
were constructed manually in a simple manner, so similar panels can be
constructed and used as roofing system with almost no equipment needed.
Test program
Material used
Ordinary Portland cement and natural sand were used in making the Ferrocement
concrete in the ratio of 1:2 respectively with a water to cement ratio of 0.55. The
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Model
No.
Dimensions
(mm)
Depth of
edge beam
(mm)
No. of
steel
layers
% Volume
of steel
Ch2-A
Ch4-A
Ch6-A
Ch2-B
Ch4-B
Ch6-B
Box1
Box2
Box3
2100x470x20
2100x470x20
2100x470x20
2100x470x20
2100x470x20
2100x470x20
2100x470x20
2100x470x20
2100x470x20
95
95
95
50
50
50
95
95
95
2
4
6
2
4
6
2
4
6
1.36
1.60
1.76
1.36
1.57
1.77
1.24
1.43
1.62
Compressive Flexural
strength
strength
(N/mm2)
(N/mm2)
54.9
46.0
47.5
42.6
40.6
42.0
44.1
31.0
54.6
5.0
4.8
4.6
6.6
6.8
6.6
5.3
5.5
7.2
Test procedure
All slabs were tested for flexure. They were simply supported with a clear span
of 2000mm and test span of 600mm in mid-span. The load was applied as two
symmetrically arranged concentrated loads, using a spreader steel beam and a 5ton hydraulic jack. The load was measured using an electric load cell of 50kN
capacity and was applied in increments of 0.5kN. The slabs were painted using
white emulsion to assist in detecting the cracks. Deflection under the centre of
the slab was measured using Linear Variable Displacement Transducers
(LVDT). The load cell and LVDT were connected to a data acquisition system.
Surface concrete strains were measured using a digital DEMEC gauges. At each
load increment, careful search was made for cracks on all sides of the slab with
the aid of a magnifying glass and a powerful electric lamp. The crack spacing,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
97
the number of cracks, the extent of the cracked zone over the length of the slab
and the ultimate load were all noted. The failure load considered in this
investigation was the load value after which the panel ceases to resist additional
load or the load measured just before sudden collapse. Figure 4 shows typical
tested panel.
1R6
20m
1R6
95m
20m
1R6
470mm
1R6
1R6
1R6
50mm
1R6
20m
20mm
Figure 1:
470mm
Figure 2:
470mm
Figure 3:
95m
20m
1R6
55m
20mm
Along with each panel, six 100x100x100mm cubes were tested for
compressive strength and two 100x100x500mm prisms were tested for modulus
of rupture.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 4:
3.5
3.0
2.5
2.0
1.5
Channel A
1.0
Channel B
Box
0.5
0.0
2
4
No. of Mesh Layers
Figure 5:
Cracking load.
5 Experimental observation
5.1 Cracking load
Figure 5 shows that all sections cracked at loads close to each other when the
number of layers was 2. The box section cracking load was less than both
channel sections when the number of layers was increased to 4 or 6. In general,
as the number of wire mesh increases the cracking load increases in all shapes.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
99
8.0
7.0
6.0
5.0
4.0
3.0
Channel A
2.0
Channel B
1.0
Box
0.0
2
4
No. of Mesh Layers
Figure 6:
Failure load.
60.0
50.0
40.0
30.0
Channel A
20.0
Channel B
10.0
Box
0.0
2
Figure 7:
Maximum deflection.
5.3 Deflection
Figure 7 shows maximum vertical deflection at mid-span. It is clear that when
the number of wire mesh layers was 2, all panels had close to each other
deflection values. With exception to channel B, the maximum deflection was
decreased with the increase of the wire mesh layers from 2 to 4 and 6. Figure 8
shows that, in all panels, as the number of wire mesh increases the deflection
reduces for the same load. Figure 9 shows that there are no major differences in
the behaviour of channel A and box section but channel B behaved in a softer
manner, more deflection for same load.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Load(kN)
8
7
6
5
4
3
2
1
0
Ch-2A
Ch-4A
Ch-6A
10
20
30
Deflection(mm)
40
50
(a)
Channel B (50mm edge beam)
Load(kN)
5
4
3
Ch-2B
Ch-4B
Ch-6B
1
0
0
10
20
30
40
Deflection(mm)
50
60
(b)
Load(kN)
9
8
Box-2
Box-4
Box-6
5
4
3
2
1
0
0
10
20
30
40
50
Deflection(mm)
(c)
Figure 8:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
101
Load(kN)
Ch-2A
Ch-2B
Box-2
10
20
30
Deflection(mm)
40
50
(a)
4 w ire mesh layers
Load(kN)
6
5
4
Ch-4A
Ch-4B
Box-4
1
0
0
10
20
30
40
50
60
Deflection(mm)
Load(kN)
(b)
6 w ire mesh layers
9
8
7
6
5
4
3
2
1
0
Ch-6A
Ch-6B
Box-6
10
20
30
40
50
60
Deflection(mm)
(c)
Figure 9:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusion
Nine roof panels were tested for pure bending. Results from two types of channel
sections, channel A and channel B, differing in the depth of the edge beam
(95mm and 50mm) and one type of box section were compared. The number of
wire mesh layers was varied from 2 to 6. Results show that channel type A
behaved in a similar way of the box section with close to each other failure loads
and deflection while channel type B was softer regardless of the number of wire
mesh layers. All panels showed acceptable cracking and failure load for roofing
systems.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
103
Abstract
A finite element method (FEM) study of the actuation and sensing performance
of actuated fibre composites (AFC) is presented. The effect of non-continuous
fibres on the AFC performance was analyzed for actuation and sensing
applications. The results of the FEM analysis of AFC with non-continuous fibres
are compared to experimental results obtained in specimens with fibres damaged
by large deformation. A study of the change in the polarization state in the fibre,
resulting from the formation of a gap, was performed and its results were
incorporated into the FEM. The correlation between the available experimental
data and simulation results is discussed.
Keywords: PZTs, actuated fibre composites, AFCs, piezoelectric fibres, sensors,
smart rotor.
Introduction
Modern military and civilian aircraft structures require the use of monitoring
systems to identify and predict the health of critical components within the
aircraft. Health Prognostic Monitoring (HPM) is in the process of being
introduced to modern aircraft structures. This concept will allow the aircraft to
determine if the life of a critical component is coming to an end, and have the
overall system schedule the time and place for the component to be
repaired/replaced. Sensors are required to acquire the current state of strain,
temperature or other parameters of concern in specific areas of the aircraft,
which is a critical part of HPM. Several sensors show promise for these
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06011
)2 + f () d 3r + 2P02 3
p =1 q =1
d 3k k p k q ~
p (k )~q (k )*
3
2
(2 ) k
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
105
Figure 1:
(a) Domain patterns in the broken fibre in the initial state and (b)
after relaxation in fibres with a 6 cell long gap and (c) with a 9 cell
long gap. The polarization direction is shown by arrows.
Two Finite Element (FE) models were constructed. The first model, for the
simulation of sensing applications, consisted of an AFC subjected to a strain of
0.9 micro strains. The second model for the simulation of actuation applications
represented an AFC subjected to 900V. In both models the effect of gaps with
and without a depolarized field was studied. The models were constructed using
the material properties for PZT-5A fibres, an epoxy matrix and two Kapton
layers. The piezoelectric fibres in the AFC model have the following
piezoelectric parameters: d31 of 44010-12 C/N, d33 of 18510-12 C/N, and d24 and
d15 of 56010-12 C/N. Mechanical properties were represented by C11=120 GPa,
C33=110 GPa C44=21 GPa, C66=23 GPa, C12=75.2 GPa and C13=75.1 GPa [13].
The epoxy had a Youngs modulus of 2.58 GPa and Poissons ratio of 0.38. The
Kapton had a Youngs modulus of 2.5 GPa and Poissons ratio of 0.34.
3.1 Sensing AFC Model
Figure 2:
(a) The AFC model used for simulation of the sensing application
shown with an epoxy matrix, (b) inner components of the AFC
model.
The sensing AFC system was set to 2100m in length, 1200m in width and
260m in thickness. It consisted of four piezoelectric fibres of 254 m diameter
and an epoxy matrix contained between two Kapton layers, as shown in
Fig. 2(a). Figure 2(b) shows the same model with the Kapton removed and a
transparent epoxy matrix. An assumption was made that due to the presence of a
30 m gap, the regions of the fibre next to the gap with the length equal to the
fibre diameter, had a zero poling effect. This means that piezoelectric constants
in these regions were equal to zero; however, elasticity constants were the same
as in the rest of the fibre. A defined strain of 0.9 micro strains was applied in an
extension mode. The electric potential difference was measured between the top
and bottom of the fibres.
3.2 Actuation AFC model
The unconstrained actuation AFC system was set to 4600m in length, 1200m
in width, and 260m in thickness. It consisted of four piezoelectric fibres, an
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
107
epoxy matrix, and copper inter-digitized electrodes (IDE) that were covered by a
thin layer of epoxy. Figure 3(a) shows a view of the AFC model including the
epoxy matrix. Figure 3(b) shows the same AFC with the epoxy matrix made
transparent to show the fibres and the copper electrodes. PZT fibres with 250m
diameters were spaced 300m center to center. The AFC was cantilevered at one
end. The fibres had a varying polarization field along their length. The fibre
under the electrodes was modeled to have a zero polarization.
Figure 3:
The fibres between the electrodes have a polarization along either positive or
negative x-direction, as shown in Fig. 3(b). Switching of the polarization
direction permits the AFC to extend or contract, since the direction of the electric
field also changes from one electrode pair to another. This type of AFC structure
and actuation method were presented and studied in [56]. The four copper
electrodes are spaced 900m apart and are 200m wide and 10m thick. The
electrodes in Fig. 3(b) are labeled E1, E2, E3 and E4. The electric potentials for
E2 and E4 are set to zero, while E1 and E3 are set to electric potentials of 900,
1350, and 1800 volts for different simulation runs. The copper electrodes make
direct contact with the fibres through an angle of 60, as shown in Fig. 3(c). In
order to study the effect of broken fibres on the actuation performance of the
AFC, we created models with different numbers of gaps (0, 1, 2, 5, 10 and 20)
introduced into fibres. We used models in which such gaps were introduced
without changing the polarization state in the rest of the fibre volume, as well as
models in which gaps were accompanied by depolarized regions at the ends of
the fibres with a length equal to the fibre diameter.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 4:
As seen in Fig. 4, the electric potential difference through the thickness of the
fibre is close to zero in the gap and in the depolarized region near it. Such loss of
voltage can be a problem since discontinuities in the AFC tend to appear in the
proximity of the electrodes used to apply voltage to fibres [15]. If the same
electrodes were used for sensing, then they would detect a voltage corresponding
to a strain that is significantly lower than the actual strain. In such a case a dual
electrode systems can improve sensing performance. The first electrode group
may be used for poling of the piezoelectric fibres, while the second electrode
system would be used to monitor voltages produced by the piezoelectric fibres. It
should be noted that in the AFC model where gaps in the fibres were not
accompanied by depolarized regions, the width of the dead zone in the voltage
distribution was much smaller (approximately equal to the gap width) and the
loss of voltage less severe.
4.2 Results of actuation application analysis
The FEM analysis of the AFC under applied voltage (Fig. 5) showed that gaps in
fibres can reduce the actuation performance. If a fibre between two IDEs is
broken, then the electrical field is concentrated in the gap, and two pieces of fibre
between electrodes have a very low electrical field applied, producing no
contribution to the actuation. However, any significant reduction in the overall
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
109
Percentage of Displacement
actuation performance of the AFC is obtained only when a large number of gaps
is introduced effectively creating a short fibre composite. It is unlikely that such
a density of damage can be developed before overall failure. The FEM results
also indicate that there is very little difference between systems with
depolarization regions added to the gaps in fibres and systems with gaps only
(without depolarization regions).
100.00
80.00
60.00
40.00
20.00
0.00
0
10
Number of Gaps
15
20
Fully polarized
Depolarized fibers
(b)
Figure 5:
Acknowledgement
The authors gratefully acknowledge the financial support of Natural Sciences
and Engineering Research Council of Canada (NSERC).
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
Banks H.T., Smith R.C., Wang Y., Smart Material Structures - Modeling,
Estimation and Control. John Wiley & Sons, 1996.
Bent A. A., Active Fibre Composites for Structural Actuation, Ph.D.
Thesis, Massachusetts Institute of Technology, January 1997.
Hall S.R., Prechtl E.F. Smart Materials and Struct. 5 (1996) 26.
Flinn, E.D. Aerospace America 37 (1999) 40.
Odegard G.M., Modeling of Piezoelectric Polymer Composites National
Institute of Aerospace, Hampton, Virginia. NASA/CR-2003-212681NIA
Report No. 2003-10
Wilkie W.K., et al., NASA Langley Research Center Macro-Figer
Composite Actuator (LaRC-MFC): Technical Overview,
Brei D., Cannon B.J., Composites Science and Tech. 64 (2004) 245.
Sporn D., Schoenecker A., Mat. Res. Innovat. 2 (1999) 303.
Pettermann H.E., Suresh S., Int. J. of Solids and Struct. 37 (2000) 5447.
Wilkie W., High J., Bockman J., Reliability Testing of NASA
Piezocomposite Actuators, U.S. Army Research Laboratory and NASA
Langley Research Center, Hampton, Virginia, USA.
Wickramasinghe V.K, Hagood N.W., Smart Mater. Struct. 13 (2004)
11551165 PII: S0964-1726 (2004) 83334-2.
Melnykowycz M., et al., Performance of Integrated Active Fibre
Composites in Fibre Reinforced Epoxy Laminates. to appear in Smart
Materials Structures.
M. Martinez, et al., Finite Element Analysis of Actuated Fibre
Composites. Proceedings of CANSMART 2005, Toronto, Canada (2005)
231.
Semenovskaya S., Khachaturyan A.G., J. Appl. Phys., 83 (1998) 5125.
Wickramasinghe V.K, Hagood N.W., J. of Aircraft, 41 (2004) 931.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
111
Abstract
This paper provides an experimental analysis of timber-framed walls, coated
with carbon fibre-reinforced polymers (CFRP) strengthened fibre-plaster boards,
usually used as main bearing capacity elements in the construction of
prefabricated timber structures. The tensile strength of the boards is lower than
the compressive one and essentially lower than the strength of the timber frame,
therefore it is convenient to strengthen boards in their tensile diagonal direction
with high-strength materials in order to gain a higher capacity. It has been shown
that the inclusion of CFRP diagonal strip reinforcement on the load-carrying
capacity can be quite high and that it is maximized when the carbon strips are
connected to the timber frame. On the other hand, the ductility itself was not
significantly improved. The test samples proved an important distinction in
behaviour in timber frame-fibreboard connecting area, dependant on the
boundary conditions between inserted CFRP strips and timber frame.
Keywords: timber, frame walls, CFRP, fibre-plaster boards, experiments.
Introduction
FH
b
nb
FH =
Figure 1:
b
timber frame
(the studs)
FH ,tot
n
coating boards
Design models
behaviour of the joints between the sheet and the frame members is
assumed to be linear-elastic until failure,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
113
the frame members and the sheets are assumed to be rigid and
hinged to each other.
The influence of shear deformations in the fibreboard can be additionally
estimated by introducing the shear angle. Additionally, two models are presented
based on the assumption that the load-displacement relation of fasteners is
completely plastic. Kllsner and Lam [5] presented the walls load-carrying
capacity as a function of fasteners spacing along the upper horizontal timber
member assuming constant fastener spacing along all timber members.
Two simplified computational methods are given in the final draft of
Eurocode 5 [2] in order to determine the load-carrying capacity of the wall
diaphragm. The first simplified analysis Method A, is identical to the Lower
bound plastic method, presented by Kllsner and Lam [5]. This method defines
the walls shear resistance (Fv,d) as a sum of all the fasteners shear resistances
along the loaded edges in the form of:
-
bi
ci
s
Ff,Rd . lateral design capacity per fastener,
bi .. wall panel width,
s .. fastener spacing,
Fv ,d = F f ,Rd
1 for bi b0
ci = bi
b for bi b0
0
where b0 = h/2
(1)
(2)
Fv ,d = F f ,Rd
d fastener diameter,
(3)
( EI y )eff = E i I yi + yi Ai a i2 =
i =1
ntimber .
i =1
(E
I yi + E i
2
yi Ai a i timber
nboard
(E i I yi )board
(4)
j =1
As described, the FPB are usually a weaker part of the presented composite
system, because their tensile strength is evidently smaller than the wood strength
of all members in the timber frame. Thus, especially in multi-level buildings
located in seismic or windy areas, cracks in FPB usually appear. In these cases
the FPB lose their stiffness and therefore their resistance should not be
considered at all. Stresses in the timber frame under a horizontal loads are
usually not critical. There are several possibilities to reinforce panel walls in
order to avoid cracks in FPB:
- by using additional boards. The boards are usually doubled:
- symmetrically (on both sides of a timber frame),
- non-symmetrically (on one side of a timber frame),
- by reinforcing boards with steel diagonals,
- by reinforcing boards with carbon or high-strength synthetic fibres.
3.1 Strengthening with additional FPB
In [6] we presented the first possibility experimentally using additional FPB,
which gave higher elasticity of elements, whilst bearing capacity and especially
ductility were not improved in the desired range.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
115
A1*b
1 E
1
= t + s sin 2 cos A1s ,0
b
Gb
b
(5)
b* =
A1*b
1 E
1
= b + s sin 2 cos A1s ,0
t
Gb
t
(6)
In the above equations represents the angle of inserted steel diagonals with
the net area (A1s,0). A non-dimensional coefficient is shear cross-section
coefficient defined as a proportion between the shear and actual cross-sectional
area of the FPB with the shear modulus (Gb). Alongside the steel diagonals
influence these models enable simultaneous consideration of the fasteners
flexibility between the board and the timber frame and any appearing cracks in
the tensile area of the FPB. Un-reinforced panels can be computed using actual
dimensions of the FPB. Numerical results presented in [7] on diagonally steel
reinforced elements show good agreement with measurements performed on the
test samples.
3.3 Strengthening with diagonal CFRP strips
As the tensile strength of FPB is obviously lower than the compressive strength
and corresponding capacity of timber frame, the treated elements tend to fail
because the cracks are forming in the tensile area of the FPB, therefore this
tensile area could be reinforced with high-strength materials. This strengthening
concept is such that the composites would contribute to tensile capacity when the
tensile strength of FPB is exceeded. No FRP applications on the treated fibreplaster boards were found in the literature.
At, Et
yi
Ab , E b
t =1.5
Vz
9.0
4.4
9.0
9.0
ai = 58
b =125 cm
Figure 2:
The static model according to Figure 1 was used for all groups of test
samples. The samples were actually rotated by 900 according to Figure 1 and
they were therefore subjected to vertical force acting at the end of the elements
(Figure 3a).
The FPB were reinforced in the tensile diagonal area using
SikaWrap-230C strips (Sika [9]) made from carbon high-strength fibre
reinforced polymers of thickness 1.2 mm. Strips with different widths (300 or
600 mm) and of different boundary conditions were glued to the FPB.
The first group (G1) of three test samples was additionally reinforced with two
CFRP diagonal strips (one in each FPB) of width 300 mm which were glued on
the FPB using Sikadur-330 LVP. The strips were additionally glued to the timber
frame (Figure 3a,b) to ensure the transmission of the force from FPB to the
timber frame.
The second group (G2) of three test samples was additionally reinforced with
two CFRP diagonal strips of width 600 mm. The strips were glued on FPB and
to the timber frame as in G1 (Figure 3b) to ensure the transmission of the force
from FPB to the timber frame.
The third group (G3) of three test samples was additionally reinforced with two
CFRP diagonal strips of width 300 mm as in G1 but they were not glued to the
timber frame (Figure 3c).
Material properties for the test samples for all groups were the same (Table
1). Values for timber of quality C22 are taken from EN338 [10], the
characteristics of fibre-plaster boards from Knauf [8] and for carbon strips Sika
[9] data were used.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
a.)
117
b.)
c.)
Figure 3:
a.) G1; the static system, b.) G2; the CFRP strip is glued on the
FPB and additionally to the timber frame, c.) G3; the CFRP strip
is not glued to the timber frame.
Table 1:
E0,m
Timber
FPB
SikaWrap
fm,k
ft,0,k
fc,0,k
[N/mm2]
[N/mm2]
[N/mm2]
[N/mm2]
[N/mm2]
10000
3000
231000
630
1200
/
22
4.0
/
13
2.5
4100
20
20
/
m
[kg/m3]
410
1050
1920
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
It is evident that the elastic resistance (force forming the first crack) essentially
increased for all kinds of CFRP strengthened test samples, but mostly for
samples G3, where the CFRP strips were not fixed to the timber frame. The
CFRP influence was not so obvious at samples G1, where carbon strips of the
same dimensions were additionally glued to the timber frame.
On the other hand, when comparing the measured results of the crushing
force, a greater improvement can be noticed in the groups where the CFRP
diagonals were glued to the timber frame. Compared to the un-strengthened test
sample, the crushing force in samples G2 was increased by 78%. In samples G3
the crushing force practically coincided with a force forming the first crack, so
cracks hardly appeared at all, which is not a good solution to ensure better
ductility, necessary for seismic design.
For further analysis it is important to present measured maximal cantilever
deflections (w) (Figure 4) and slips () in the connecting area (Figure 5).
F[kN]
Fcr,3
Fcr,2
UNS
G1
G2
G3
Fcr,1
w[mm]
Figure 4:
119
F[kN]
Fcr,3
Fcr,2
G1
G2
G3
Fcr,1
[mm]
Figure 5:
Conclusions
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
121
Abstract
We have evaluated the structural integrity of a sandwich composite train roof
which can find a lightweight, cost saving solution to large structural components
for rail vehicles in design stages. The sandwich composite train roof was 11.45
meters long and 1.76 meters wide. The FE analysis was used to calculate the
stresses, deflections and natural frequencies of the sandwich composite train roof
against the weight of air-condition system. The 3D sandwich FE model was
introduced to consider the hollow aluminum frames jointed to both sides of the
sandwich train roof. The results shown that the structural performances of a
sandwich composite train roof under the loading conditions specified were
proven and the use of aluminum reinforced frame was beneficial with regard to
weight saving in comparison to steel reinforced frame. Also, we have
manufactured the prototype of sandwich composite train roof on the basis of
analysis results.
Keywords: aluminum honeycomb structure, train roof structure.
Introduction
123
Figure 1:
Figure 2:
Skin
Al. 5052
Core
Aluminum
honeycomb
core
Dimension
t=1.2mm
Properties
E=69GPa
=0.33
=2700(kg/m3)
E11=8.27MPa
E22=1.31MPa
E33=1276MPa
G12=0.0001GPa
G23=117GPa
G13=296GPa
12=0.75
23=0.0001
13=0.0001
=100(kg/m3)
Table 2:
Cases
1
2
3
4
5
Table 3:
125
Mode
1
2
3
4
5
6
7
8
9
10
Case 1*
1.87(B)
5.16(B)
8.65(T)
10.14(T)
16.8(B)
17.48(T)
25.13(B)
26.68(T)
35.14(B)
36.42(T)
Case 2
1.64(B)
2.22(T)
3.22(B)
4.52(T)
5.33(B)
6.98(T)
7.96(B)
9.64(T)
11.13(B)
12.58(T)
Case 3
1.25(B)
3.45(B)
5.68(T)
6.78(B)
11.25(B)
11.51(T)
16.87(B)
17.61(T)
23.64(B)
24.11(T)
Case 4
1.86(B)
5.12(B)
8.46(T)
10.05(B)
16.62(B)
17.10(T)
24.82(B)
26.11(T)
34.61(B)
35.66(T)
Case 5
1.86(B)
5.12(B)
8.44(T)
10.04(B)
16.60(B)
17.06(T)
24.77(B)
26.05(T)
34.5(B)
35.0(T)
Modal Analysis
Structural Analysis
Cases
1-1
1-2
1-3
1-4
1-5
1-6
1-7
1-8
Element
Shell63
/Solid45
Shell63
/Solid45
Jointing condition
Bonded contact
Coupled
set
Bonded contact
Coupled
set
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Stiffener materials
Aluminum
Steel
Aluminum
Steel
Aluminum
Steel
Aluminum
Steel
Figure 3:
Figure 4:
Table 5:
Mode(Hz)
1
2
3
4
5
6
7
8
9
10
Weight(kg)
Case 1-2
50.62
51.71
52.70
55.25
59.36
63.95
71.88
79.81
89.59
101.41
596.26
Case 1-3
64.33
65.07
66.25
69.12
73.76
78.93
87.29
95.55
105.85
117.34
393.58
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Case 1-4
50.24
51.38
52.38
54.93
59.04
63.65
71.56
79.52
89.30
101.13
596.26
127
Figure 5:
Max. deflection(mm)
Max. principal
stress(MPa)
Total weight(kg)
Case 2-1
Case 2-2
0.42
0.39
8.69
8.93
393.58
596.26
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Max. Deflection(mm)
Max. principal
stress(MPa)
Total weight(kg)
Case 2-1
Case 2-2
0.73
0.67
15.89
15.83
393.58
596.26
129
Figure 6:
Figure 7:
Figure 8:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusions
References
[1] K B. Shin, C G. Kim, C S. Hong., Correlation of accelerated aging test to
natural aging test on graphite-epoxy composite materials, Journal of
Reinforced Plastics and Composites 22(2003) 849-866.
[2] K B. Shin, S H. Hahn., Evaluation of the structural integrity of hybrid
railway carriage structures including the aging effects of composite
materials, Composite Structure 68 (2005) 129-137.
[3] J. Cantrill, R. Mableson. Development and prototyping of a lightweight
composite train roof, 14th COMPOSITE Workshop (2003).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
131
Abstract
A single-fiber pull-out model composite for an aramid/epoxy system was
specially designed to measure the stress distribution of the aramid fiber
embedded in the matrix using micro-Raman spectroscopy. The stress transfer
length of the fiber obtained was about 400-500m, which was equal to the result
of FEM analysis. Just after the initiation and propagation of the fiber/matrix
interfacial debonding, the fiber was broken, and the fiber in the matrix had the
axial tensile residual stress. The tensile residual fiber axial stress showed the
maximum at around the tip of the interfacial debonding. The stress was reduced
and became almost equal to zero after being immersed in deionized water at
80 C for 44h. This behavior agreed with the result of FEM analysis, in which
the friction coefficient was introduced in the fiber/matrix interface.
Keywords: interfacial properties, micro-Raman spectroscopy, pull-out test,
aramid fiber, friction.
Introduction
2-6
12
Cut off
12
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
133
(b) Shape and dimensions of pull-out specimen. All dimensions are in mm.
Figure 2:
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 1:
135
Mechanical properties of the fiber and resin [6-8]: (a) Epoxy resin
(Epikote828 + Epicure Z), (b) Fiber (Kevlar 49).
(a)
Youngs modulus
2.48 GPa
Poissons ratio
0.4
6010-6 C-1
(b)
Youngs modulus
Poissons ratio
Coefficient of thermal expansion
EL
129.6 GPa
ET
2.49 GPa
LT
0.62
TT
0.31
-210-6 C-1
5910-6 C-1
Shift
Intensity a.u.
10
Figure 5:
0 GPa
2 GPa
1600
1400
1200 1 1000
Raman shift cm
Raman spectra of a Kevlar 49 single fiber without loading and with loading of 2
GPa tensile stress are shown in fig.5. The increase in tensile stress resulted in a
clear shift of the spectrum to a lower wavenumber. In this study, a strong band
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Peak wavenumber cm
1614
1612
1610
1608
1606
1604
1602
Figure 6:
1
2
Tensile stress GPa
The stress distribution of the axial tensile stress of the fiber during the pullout test was shown in fig.7 with the result of FEM analysis. The applied nominal
stress to the fiber is also plotted in the figure. The origin point of the distance
along the fiber is the meniscus point from which the fiber was embedded in the
epoxy matrix. The fiber at the positive value of the distance along fiber was
embedded in the resin matrix, whereas the fiber at the negative value
corresponded to the unembedded free fiber. The stress distribution was measured
when the applied stress was 0, 0.47, 0.94, 1.41, 1.89 and 2.36 GPa. The stress of
the embedded fiber decreased gradually along the fiber from the meniscus point,
which means that the stress was transferred from the fiber to the matrix.
Considering from the fact that the fiber axial stress at the distance of 400 m and
that of 500 m were almost equal, the stress transfer length was about 400-500
m for the Aramid/epoxy system adopted in our study. This behavior is agreed
with the result of FEM analysis.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Before debonding
2.36GPa
1.89GPa
1.41GPa
0.94GPa
0.47GPa
Applied stress
0GPa
Mesured
with Raman FEM
spectroscope
Embedded fiber
Figure 7:
137
200
0
200
400
Distance along fiber m
Fiber axial stress along the fiber before debonding with the results
of FEM analysis.
When the applied stress was increased to 2.65 GPa, the fiber/matrix
interfacial debonding was initiated and propagated, and at the same time the fiber
was fractured. After this interfacial debonding, the stress distribution in the resin
was again measured. The stress distribution after the fiber/matrix interfacial
debonding differed from that before the debonding, which is shown in fig.8.
Although external stress was not applied to the fiber because of fiber fracture, the
fiber axial stress in the debonded region had the residual tensile stress and had
the maximum at the distance of around 200 m. The position of the interfacial
crack tip was 180 m, which was confirmed after Raman measurement by using
optical microscope. This crack tip was located almost at the position of
maximum residual stress. This residual stress was considered to be caused by the
holding of the previous stress by the fiber/matrix interfacial friction acting on the
debonded interface. This speculation was supported by the fact that the friction
stress plays an important role in the propagation of the fiber/matrix interfacial
debonding and the propagation rate of interfacial debonding slowed down and
retarded when the a constant fatigue load was applied to the specimen [2].
After debonding, this specimen was kept in dry air (relative humidity: 25%)
for 144 h and after that it was immersed in deionized water at 80 C for 44 h.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 8:
Embedded fiber
200
0
200
400
Distance along fiber m
Conclusions
139
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 3
Natural fibre composites
(Special session organised by
T. Katayama and H. Takagi)
143
Abstract
Nowadays, there is a serious environmental problem with waste disposal etc. In
fiber reinforced composite, glass fiber reinforced composite is difficult to reuse
and recycle. Instead of glass fibers, natural fibers such as bamboo kenaf, jute,
and hemp fibers are focused on for environmental friendliness. The mechanical
properties for some natural fiber reinforced composites have been studied. The
number of papers for green composite is greatly increased.
In this study, jute fibers are used as reinforcement because of the huge
production and low cost. Polypropylene is used as a matrix because of recycling
properties and cost. The strength and creep properties for Jute Fiber Reinforced
Plastics (JFRP) are examined.
In Japan, natural fibers have been used as fishing nets for a long time.
Astringency of a persimmon is used for hemp fiber as a reinforcement and
binder. So, the surface treatment is conducted for the fiber using the astringency
of a persimmon. The mechanical and creep properties for jute fiber reinforced
composite and the effect of surface treatment are investigated.
From the results, the following conclusions are obtained.
The treatment using the solution of astringency of persimmon is effective to
increase mechanical properties. In the case of using the treatment, some densities
are effective and 25% solution is most effective. For the creep properties, there is
an effect of surface treatment to creep strain. The effect is large when the load is
small and the effect emerges in the initial stage.
Keywords: jute fiber, polypropylene, surface treatment, strength, creep.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06015
Introduction
Fiber Reinforced Plastics (FRP) is widely used for airplane, automobile parts,
fishing pole and any other engineering products. But, recently there are some
environmental problems, especially the disposal of glass fibers could not be
burned out. So, natural fibers such as jute, bamboo, hemp and flax are focused as
reinforcement [1-5] In Japan, third international workshop for green composite
was held in 2005. Many researchers are interested in this issue.
Jute is focused from the viewpoint of cost. And it is easy to take jute from
general market. So Jute Fiber Reinforced Plastics (JFRP) is used in this study.
The static tensile tests for this composite are conducted in order to examine
strength and stiffness etc. Natural fiber has been used for fishing net for a long
time especially in Japan. Many chemical treatment methods have been used to
the natural fibers. Among them, the astringency of a persimmon has been used
for the improvement the face of natural fibers. So, the surface treatment method
is conducted for this natural fiber using an astringency of a persimmon. So,
mechanical and creep properties for the natural fiber reinforced composite and
the effect for astringency of a persimmon are investigated.
2.1 Specimen
Plain woven jute fiber is used as reinforcement. The directions of jute fibers are
parallel and perpendicular to a load direction. Polypropylene resin (Shin Kobe
Electric Inc. PP-N-AN) is used as matrix. The geometry of specimens is
referenced to JIS (Japanese Industrial Standard) 7054. The length of the
specimen is about 200mm. The breadth and thickness are about 15mm and
2.5mm respectively. For surface treatment specimen, the plain woven fiber is
laid in the solution of an astringency of a persimmon.
2.2 Static tensile test for jute fiber
Shimadzu autograph tensile test machine (AG-IS) is used for static tensile test.
The test condition is referenced to JIS7054. The surfaces of the composite are
observed using Scanning Electron Microscope (SEM) to confirm the effect of
surface treatment.
2.3 Creep test for jute fiber
Creep Tester 100LER (Toyo Seiki. Co.) is used for this creep test. The test is
continued by a failure of specimen. When a specimen does not fail, the creep test
stops at 100hours. Three tests can be conducted simultaneously. Creep extension
and strain are measured. This test is referenced to JIS7087. The creep load is
used for the consideration of static strength.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
145
Untreated
Load(N)
200
Treated(50%)
150
100
50
0
0
Strain(%)
Figure 1:
Figure 2 shows the stress-strain curves for jute fiber composite. The solution
densities for surface treatment are 0, 25, 50, 75, 100%. For the viewpoint of
stiffness, 25% solution is most effective. For untreated specimen, the stiffness is
lower than that of 25% specimen. But the strength is not low. For 100% solution,
the strength and stiffness is worst. So it is understood that the treatment using an
astringency of a persimmon is effective for the composite, but the densities are
important. For 100% solution, it seems that the fiber has some damage on its
surface. The knee point stress is important for its use as application. The knee
point stresses for 25% and 50% are almost same. The knee point stresses for 0%
and 75% are the same. The knee point stress for 100% solution is obviously
lowest. From this figure, 50% solution may seem to be as effective as 25%
solution.
The photograph of jute fiber is shown in Fig.3. From the surface of treatment,
there is much resin on the surface. It is understood that the adhesive strength on
the surface is improved. And, the fiber is not straight. It seems that the fiber has
ductile properties. So, the stress-strain curve has knee point. And high stiffness is
obtained from the surface.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2:
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
147
0.04
200N
300N
Strain
0.03
0.02
0.01
0
0
Figure 4:
20
40
60
Time
80
100
Table 1 shows the creep strains for 25% treated and untreated composite
specimen. From this table, it is understood that the effect of surface treatment is
emerged at lower load. In the case of 300N, the creep strains seem to be same.
From that result, there is obviously an effect of surface treatment. But when the
load is bigger, the effect is hidden. In the case of 300N, the shapes of the curves
are the same. On the other hand, in the case of 200N, the difference is emerging
in the transit region. So, the effect of surface treatment is bigger at the initial
stage for lower load. At the initial stage, there is the friction on the surface of
jute fibre. So, it is thought that the effect of this region is greater.
Table 1:
Untreated
1.4
2.4
Treated
1.2
2.4
Strain
0.03
300N
0.02
0.01
0
0
Figure 5:
20
40
60
Time(h)
80
100
0.04
Strain
0.03
0.02
200N
300N
0.01
0
0
Figure 6:
20
40
60
Time(h)
80
100
So, the
Conclusions
In this study, the static and creep properties of jute fiber reinforced composite are
examined and the effects of an astringency of a persimmon are also examined.
As a result, following conclusions are obtained.
(1) There is an effect of astringency of a persimmon to static tensile properties.
(2) The effect is dependent on the solutions densities. When the density is 25%,
it is more effective.
(3) For creep properties, there is an effect of surface treatment to creep strain.
The effect is big when the load is small. The effect emerges in the initial
stage.
(4) The effect of surface treatment using only water is not obtained. On the
other hand, in this case, the creep strain becomes big.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
149
References
[1]
[2]
[3]
[4]
[5]
D. Nabi Saheb and J.P. Jog, Natural Fiber Polymer Composite: A Review,
Advances in Polymer Technology, Vol.18, No.4, 351 (1999).
H. Takagi and Y. Ichihara, JSME International Journal, Series A, 47,551
(2004).
K. Goda et al., Proc. of the 2nd Int. Workshop on Green Composites,
96(2004).
B. Singh et al., Composite Science and Technology, 60,571-589(2000).
J. Gassan, at. al., Composite Science and Technology, 60,2857 (2000).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
151
Abstract
This paper deals with the development of resinless, unidirectional bamboo
fibre-reinforced eco-composites. Resinless bamboo composites were fabricated
by the hot-pressing of unidirectional bamboo fibres extracted by a steam
explosion method. The effects of forming conditions on their mechanical
properties were investigated. The forming temperature was changed every 20C
from room temperature to 180C, and the forming pressure used was 10 MPa
and 50 MPa. Except for the conditions below 60C, their tensile strength and
fracture elongation decreased with increasing forming temperature. Average
tensile strength and Youngs modulus of the resinless bamboo composites
fabricated at 120C and 50 MPa were 322 MPa and 37 GPa respectively.
Keywords:
natural fibre, bamboo, resinless, binderless, unidirectional
composites, tensile strength, modulus.
Introduction
Biodegradable composites that are also known as green composites are mostly
made from natural fibers and a biodegradable resin. Therefore the composites are
one of the most environment-friendly materials, because they are finally
biodegradable, and thus can be resolved into carbon dioxide and water through
the ideal decomposition stage. Hence, many researchers have been conducting
the research on biodegradable composites, and especially on their mechanical
properties [1-7]. However, most of the biodegradable resins are more expensive
than conventional plastics, such as PP and PE. In addition, extra energy is needed
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06016
Experimental method
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
153
100
Figure 1:
Sample
10
Aluminium tab
30
Under the moulding condition of R.T. and 10 MPa, bamboo fibres did not adhere
to each other; therefore, it was impossible for this sample to carry out the tensile
test. However resinless bamboo composites can be fabricated by hot-pressing at
temperatures over 80C. The colour of surface changed from dark brown to
black with increasing the moulding temperature. This colour change might be
attributed to the chemical reaction of lignin comprised in bamboo fibres.
Figure 2 shows typical stress-strain curves of resinless bamboo composites
moulded at 50 MPa as a function of moulding temperature. The initial slope of
the curve (corresponding to Youngs modulus) becomes steeper with increasing
the moulding temperature, and simultaneously tensile strength also increases.
The stress of composites moulded below 180C gradually decrease in the latter
part of the deformation. This phenomenon should be responsible for the
cumulative fibre fracture. However, the stress-strain curve for the composites
moulded at 180C becomes a straight line with the same steep slope as 120C. At
the same time, fracture behaviour is brittle, and fracture elongation becomes
considerably small value less than 0.005. In the case of moulding pressure of 10
MPa, almost the similar stress-strain curves are obtained as show in fig. 3.
The relationship between tensile strength of bamboo composites and moulding
temperature is shown in fig. 4. The tensile strength reaches a maximum value of
about 300 MPa at 120C at both of the moulding pressures. It can be seen from
this graph that the bamboo composites become brittle at temperatures over
120C, and bamboo fibres do not adhere to each other at temperatures below
120C.
The relationship between Youngs modulus and moulding temperature is
presented in fig. 5. Youngs modulus initially increases with the moulding
temperature, and then levels off at temperatures above 120C. This rise in
Youngs modulus is consistent with the increase in the density of samples. This
is because modulus as well as stress also increases with the moulding
temperature.
Figure 6 shows the relationship between fracture elongation and moulding
temperature. The fracture elongation is almost constant below 120C, and
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
500
R.T.
80C
120C
180C
Stress (MPa)
400
300
200
100
0
0
Figure 2:
0.01
0.02 0.03
Strain
0.04
0.05
500
80C
120C
180C
Stress (MPa)
400
300
200
100
0
0
Figure 3:
0.01
0.02 0.03
Strain
0.04
0.05
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
155
600
10MPa
50MPa
500
400
300
200
100
0
Figure 4:
150
200
50
100
o
Moulding temperature ( C)
80
60
40
20
0
Figure 5:
10MPa
50MPa
50
100
150
200
o
Mouding temperature ( C)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
2.5
1.5
1.0
0.5
0
Figure 6:
10MPa
50MPa
2.0
50
100
150
200
o
Moulding temperature ( C)
Conclusions
Acknowledgement
The authors greatly acknowledge the partial financial support provided by the
president of The University of Tokushima in 2005.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
157
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
159
Abstract
To solve the problem of the large environmental burden in the disposal of FRP
(Fibre Reinforced Plastics), natural fibre and biodegradable resin have received a
lot of attention as easily degradable materials in the natural environment. To
enlarge the usage of the composites based on natural fibre and biodegradable
resin, cost reduction and enhancing its strength are essential subjects. In this
study, in order to develop the environment friendly biodegradable composites
that have high performance of strength, rigidity and productivity, the non-woven
stacking method was proposed and the influence of moulding conditions on
impregnating property and mechanical property of Jute Fabric Reinforced
Thermoplastic composites (JFRTP) based on PLA non-woven fabric was
discussed. It takes several minutes for PLA to impregnate into fibre bundles after
the material reaches preset moulding temperature due to its high melt viscosity
and twist of fibre bundles. The properly moulded JFRTP specimen reached about
115 MPa in bending strength, which is comparable to GFRTP with 20 wt % long
fiber made by injection moulding.
Keywords: green-composite, natural fibre, jute fibre, PLA, non-woven fabric,
compression moulding, bending property.
Introduction
Experimental procedure
10mm
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
50m
10mm
(a)
Figure 2:
161
(b)
Figure 3:
Meltblown process.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 4:
Table 1:
Laminated condition.
Molding pressures and times.
Pressure (MPa)
Time (s)
Reinforcing
fabric
1.0
20
40
60
2.0
3.0
120 180
4.0
300
600
3.1 Impregnation
The relationship between density and moulding time is shown in fig. 5. Before
60 s, the density increases drastically irrespective of the moulding pressure. After
300 s, however, it become stable and the density under moulding pressure of 3.0
MPa almost equal to that under 4.0 MPa.
Figure 6 shows the cross section of the moulding product under the moulding
pressure of 1.0MPa, whose results shows more clearly the difference of
moulding time on the impregnating state than other moulding pressure. Before
180 s, many visible voids are observed. After 300 s, however, matrix is
impregnated into fibre bundles. This observation corresponds with the results of
measured density change shown in fig. 4. Considering the fact that the higher
moulding pressure than 3.0 MPa and longer moulding time than 300 s dont
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
163
Density (g/cm )
1.6
1.4
1.2
1.0
0.8
Theoretical Density
1.0 MPa
2.0 MPa
3.0 MPa
4.0 MPa
0.6
0.4
0.2
0.0
Figure 5:
Figure 7 shows the temperature change at the centre of the material during
moulding under the moulding pressure of 1.0MPa. After 10 s from the start of
moulding, the material reaches the melting point of PLA (140 C) and after 60 s
it becomes stable. In spite of the higher temperature than the melting point of
PLA after 10 s, it takes almost 300 s for PLA to impregnate into the fibre
bundles. The high melt viscosity of PLA and the twist of jute fibre bundles are
considered to be the reason for this.
3.2 Bending property
Figures 8 and 9 show the influence of moulding condition on the bending
strength and the bending modulus, respectively. Before 300 s, the bending
strength and modulus drastically increase but after that they become stable. This
behaviour is similar to the result of density shown in fig.5. From the viewpoint
of the mechanical properties, moulding pressure of 3.0MPa and moulding time
of 300 s are optimum moulding value.
Figure 10 shows the stressdisplacement curve of bending test of the
specimen moulded under pressure of 3.0 MPa. In the case of specimen moulded
for 300 s and 600 s, the bending stress is decreased rapidly after reaching the
maximum value and they fractured in brittle manner. This behaviour is one of
the evidence that the resin was impregnated into the fibre bundles.
Figures 11 and 12 show the comparison of the bending strength and modulus
between our JFRTP and other materials. Injection moulded composite of glass
fibre reinforced polypropylene with 20wt% long fibre and jute fibre reinforced
polypropylene with 51wt% long fibre (Tanaka et al. [4]) are shown in these
figures. The PLA/JF product in our study has better bending properties than
GFRTP with 20 wt % long fiber.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
500m
500m
(a) 20s
(b) 40s
500m
500m
(c) 1min
(d) 2min
500m
500m
(e) 3min
(f) 5min
500m
(g) 10min
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Molding Temperature ( )
Figure 7:
165
100
Temperature
1.0MPa).
profile
200
300
400
Molding Time (s)
during
500
moulding
600
(moulding
pressure:
140
120
100
80
60
1.0 MPa
2.0 MPa
3.0 MPa
4.0 MPa
40
20
0
Figure 8:
100
600
Conclusions
6.0
5.0
4.0
2.0
1.0
0.0
Figure 9:
1.0 MPa
2.0 MPa
3.0 MPa
4.0 MPa
3.0
100
600
140
120
20s
40s
60s
120s
180s
300s
600s
100
80
60
40
20
0
Figure 10:
3 4 5 6 7 8
Displacement (mm)
9 10
140
120
100
80
60
40
20
0
PP/GF 20wt% PP/JF 51wt% PLA/JF 53wt%
Long Fibre [4] Long Fibre [4] Plain Weave
Figure 11:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
167
8
7
6
5
4
3
2
1
0
PP/GF 20wt% PP/JF 51wt% PLA/JF 53wt%
Long Fibre [4] Long Fibre [4] Plain Weave
Figure 12:
Acknowledgement
This study was partially supported by the Academic Frontier Research Project on
NewFrontier of Biomedical Engineering Research of Ministry of Education,
Culture, Sports, Science and Technology.
References
[1]
[2]
[3]
[4]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
169
Abstract
The utilization of unused forest products is beneficial, and therefore the
application of natural fibers to FRP products has been proposed. Bamboo fibers
have attracted particular attention because bamboo has the fastest growth rate
among various types of renewable natural fibers. Moreover, bamboo fibers have
high specific strength and stiffness appropriate for structural materials. In the
present study, we explain a new method designed to obtain high-quality bamboo
fibers by effectively end-milling them with a machining center (MC) using
numerical control (NC) and an automatic tool changer (ATC), which enables the
manufacture of a wide variety of products. NC programs encoding a spiral tool
path were used for the bamboo pipe in order to obtain bamboo fibers effectively.
The fiber length and diameter were evaluated at various feed rates, cutting
speeds, and depths of cut in the fiber direction. Microscopic observation showed
that the length of the fibers obtained can be controlled by selecting the depth of
cut along the bamboo fiber direction and the diameter can be controlled by
adjusting the feed of the end-mill center, which is determined by spindle speed
and feed speed of end-mill under a constant end-mill diameter. Thus the desired
fiber shape can be controlled with consistently high accuracy using end-milling
by adjusting the cutting conditions. Moreover, using spiral tool path end-milling
with a straight cutting edge tool can efficiently acquire high quality straight
bamboo fibers with no thermal damage.
Keywords: bamboo fiber, end-mill, machining center, spiral tool path.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06018
Introduction
Recently, industrial products made of fiber reinforced plastic (FRP) such as glass
fiber-reinforced plastic (GFRP), are being utilized extensively because they have
high specific tensile strength (that of GFRP is as high as steel) but are
significantly lighter. The use of such materials may have a negative
environmental impact because of problems associated with their disposal or
destruction. Conversely, the utilization of unused forest products is beneficial to
the environment, and therefore a method of acquiring and applying natural fibers
to FRP products has been proposed [1-7]. Bamboo fibers, in particular, have
attracted attention because bamboo has the fastest growth rate among the various
types of renewable natural fibers, and is grown throughout Japan. Moreover,
bamboo fibers have high specific strength and stiffness as appropriate for
structural materials [8, 9]. Bamboo fibers have been obtained from the wood-like
plant through various methods, such as crushing or heat steaming the bamboo
stems. Using these methods, however, it is difficult to control the diameter,
length and other dimensions, as well as to obtain large quantities of a consistent,
uniform shape as necessary for high-quality fiber components in industrial
materials [10]. This paper attempts to obtain high quality bamboo fibers by endmilling with a machining center (MC) as an alternative to previous conventional
methods. The MC makes it possible to manufacture many kinds of products and
the proposed method boasts some advantages over conventional methods. The
shape of the bamboo fibers can be controlled because all the machining center
processes are controlled by digital program. Moreover, the spiral tool path
should produce uniform bamboo fibers and allow accurate control of the shape
of fibers by variation of the cutting conditions.
Experimental method
2.1 Material
Naturally growing Mousouchiku bamboo was used as the base material. The
bamboo was cut using a metal saw into 100 mm-long bamboo pipes in order to
remove the bamboo joints. The bamboo pipes were then set on the table in an
MC and end-milled. The waste bamboo obtained was used as the bamboo fibers
used in the experiments.
2.2 Machining equipment and conditions
A ROBODRILL (Type: T14iDs, FANUC Co., Ltd.) was used as the MC for
milling. The end-mill tool used in the experiments was a square type with two
straight cutting edges, and made of high-speed steel without coating films on the
tool surface. The end-mill diameter was 6 mm. Table 1 shows the machining
conditions. The tool followed a spiral path from the perimeter to the center with
a cutting depth of 50 m per one cycle in the radius direction of bamboo, as
shown in Fig. 1.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 1:
171
Machining conditions.
End-mill
Bamboo pipe
Figure 1:
Bamboo pipe
Figure 2:
h = f r sin( en + )
(2)
Here, the depth of cut in the radius direction per cycle Rv is very small
compared with the end-mill radius r and the bamboo pipe radius Rb. As a result,
the cutting engagement angle en is extremely small, so as to be nearly zero.
Therefore, eqn (3) can be obtained from eqn (1) using the approximation
sinenen.
(3)
L = r en
Eqns (4) and (5) can be obtained by the geometrical relation of the triangle
that consists of the bamboo pipe center point, the end-mill center point, and the
sharp end of the end-mill cutting edge, as shown in Fig. 2.
(4)
r sin en = (Rb + Rv ) sin
(5)
Eqn (6) is obtained from the relation that the ratio of the feed of the cutting
edge fr and the feed of the end-mill center fen is equivalent to the ratio of the
distance from the spiral path center to the end-mill center point and to the sharp
end of the end-mill cutting edge.
f en
R +r
= b
fr
Rb + Rv
(6)
Here, the eqn (7) is conducted by eqns (2), (4), (5), and (6).
h = f en sin en
(7)
Therefore, eqns (8) and (9) can be obtained by replacing the parameters
utilized in this study shown in Table 1 into eqns (3) and (7).
r Rv
r
(8)
Rv
R
2 v
r
r
(9)
L = r arccos
h=
F
S Z
15
Theoretical
200
100
Fiber diameter ( m)
300
10
5
0
100
200
300
400
100
200
300
400
Cutting speed (m/min)
100
Aspect ratio
80
60
40
20
0
0
0.5
Fiber diameter
Fiber length
Aspect ratio
0.4
0.3
0.2
0.1
0.0
100
200
300
400
Cutting speed (m/min)
173
100
200
300
400
with
cutting
speed
V = 2r S
F
f en =
S Z
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(10)
(11)
End-mill
Bamboo pipe
37 C
40 C
Cutting
temperature
as
monitored
(fen=0.05 mm/tooth, Ad =10 mm).
67 C
by
thermography
The results verify that the diameter of the end-milled bamboo fiber decreases
with cutting speed until about 200 m/min though obtained fiber diameter has
good agreement with theoretical one under low cutting speed. However, above
this speed, the fiber diameter remains almost constant. On the other hand, the
fiber length is almost constant, at 10 mm, for any cutting speed. As a result, the
aspect ratio increases with cutting speed until about 200 m/min and then plateaus
at a fiber diameter of about 150 m at the cutting speeds above 200 m/min.
Higher cutting speed with greater feed speed improves the machining
throughput. However, the coefficient of variation for the fiber diameter, fiber
length and aspect ratio show a similar tendency of increasing with cutting speed.
Therefore, a cutting speed of about 200 m/min is suitable in order to obtain high
efficiency machining and maintain low variation. It seems that the bamboo fiber
diameter is controlled by cutting speed under about 200 m/min.
On the other hand, the increased cutting heat generated may damage the
bamboo fiber in end-milling; the cutting temperature increases with cutting
speed. The cutting temperature in steel rises to 800 C or higher. The melting
points of bamboo pipe components of cellulose, hemi cellulose, and lignin are
about 240, 180, and 420 C, respectively. The temperature in bamboo cutting was
monitored by thermography as a non-contact method. It can be seen that the
temperature in bamboo cutting also increases with cutting speed as shown in
Fig. 4. However, up to a cutting speed of 200 m/min the temperature grows to no
more than about 70 C. It can be concluded that under these conditions bamboo
fibers end-milled from bamboo will sustain no thermal damage.
3.2 Influence of cut of axial depth along the fiber direction on acquired
bamboo fiber quality
Figures 5 shows the microscopic images of end-milled bamboo fibers under
various axial cut depths. It can be seen that uniform, straight fibers with no
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
175
5mm
(a) Ad =2.5 mm
Figure 5:
(b) Ad =5 mm
(c) Ad =10 mm
200
bamboo
100
10
5
0
5
10
15
Axial depth of cut (mm)
5
10
Axial depth of cut (mm)
5
10
15
Axial depth of cut (mm)
15
100
Aspect ratio
from
15
Theoretical
Fiber diameter ( m)
300
end-milled
0.5
Fiber diameter
Fiber length
Aspect ratio
0.4
0.3
0.2
0.1
0.0
0
10
15
Figures 6 shows the changes in fiber shape with axial depth of cut under a
constant feed of the end-mill center of 0.05 mm/tooth and a cutting speed of
188 m/min, the conditions determined to be optimum for acquiring high quality
bamboo fibers efficiently as described in section 3.1. Moreover, L and h are
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
15
500
400
300
200
Ad=2.5mm
Ad=5mm
Ad=10mm
Theoretical
100
0
Fiber diameter ( m)
Ad=2.5mm
Ad=5mm
Ad=10mm
10
5
0
0.0
0.1
0.2
0.3
Feed of end-mill center
(mm/tooth)
0.0
Aspect ratio
100
0.1
0.2
0.3
Feed of end-mill center
(mm/tooth)
80
60
40
20
0
0.0
0.1
0.2
0.3
Feed of end-mill center
(mm/tooth)
Change of bamboo fiber shape with feed of the end-mill center and
the depth of cut in fiber direction (fen=0.05 mm/tooth,
V=188 m/min).
It is clear that the fiber diameter measured shows almost constant value,
which is almost half of theoretical one, with the axial depth of cut as shown in
Fig. 6(a). On the other hand, the fiber length increases with the axial depth of cut
as shown in Fig. 6(b). The result shows that the fiber length agrees very well
with the axial depth of cut, as the data points lie almost exactly on a line of slope
1 (Fig.6(b)). As a result, aspect ratio increases with axial depth of cut as shown
in Fig. 6(c). Fibers with a large aspect ratio are desirable to confer superior
material properties to the FRP. It may be effective to increase the axial depth of
cut in order to obtain high quality bamboo fibers. However, the coefficient of
variation of fiber diameter, length and aspect ratio seem to have minimum values
at an axial cut depth from 5 to 8 mm. Therefore, it seems effective to set the axial
cut depth from 5 to 8 mm in order to acquire high quality bamboo fibers.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
177
Conclusions
End-milling using a machining center that follows a spiral tool path was tried for
the production of high quality bamboo fibers. The shape of the fibers end-milled
from bamboo pipe was evaluated under various cutting conditions, giving the
following results.
(1) End-milling with a machining center that follows a spiral tool path is
effective for obtaining high quality, straight bamboo fibers with no thermal
damage.
(2) It is possible to accurately control the bamboo fiber length by adjusting the
axial depth of cut.
(3) A higher cutting speed can provide bamboo fibers with smaller diameter.
However, the decrease in bamboo fibers diameter with higher cutting speed
saturates at a certain cutting speed.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgement
This research is supported by the Research and Development Center for
Engineering Application of Bamboo Resources at Doshisha University.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
Ogawa K., Hirogaki T., Aoyama E. and Katayama T., Data mining of
optimum condition to acquire bamboo micro-fiber with mechanical
method, High Performance Structures and Materials II, WIT PRESS,
pp.441-450, 2004.
Jana S. C. and Prieto A., On the development of natural fiber composites
of high-temperature thermoplastic polymers, Journal of Applied Polymer
Science, Vol. 9, No. 28, pp.2159-2167, 2002.
Li H. and Sain M. M., High stiffness natural fiber-reinforced
polypropylene
composites,
Polymer-Plastics
Technology
and
Engineering, Vol.42, No.5, pp.853-862, 2003.
Luo S. and Netravani A. N., Interfacial and mechanical properties of
environment-friendly green composites made from pineapples fibers and
poly (hydroxybutyrate-co-valerate) resin, Journal of Materials Science,
Vol. 34, No. 15, pp.3709-3719, 1999.
Ogawa K., Hirogaki T., Aoyama E., Tomioka Y., and Shiomi T., Shape
Control of Fibers End-milled from Bamboo with a Machining Center,
Proc. of The 3rd International Conference on Leading Edge
Manufacturing in 21st Century, pp.565-570, 2005.
Thwe M. M. and Liao K., Environmental effects on bambooglass/polypropylene hybrid composites, Journal of Materials Science,
Vol. 38, No. 2, pp.363-376, 2003.
Martikka H. and Katajisto J., Study of Natural Fiber Reinforced
Biodegradable Composite Materials for Designing Optimally Sustainable
Products, Proc. of the 23rd Riso International Symposium on Materials
Science, pp. 251-258, 2002.
Takagi H. and Ichihara Y., Effect of fiber length on mechanical properties
of green composites using a starch-based resin and short bamboo fibers,
JSME International Journal, Series A: Solid Mechanics and Material
Engineering, Vol. 47, No. 4, pp.551-555, 2004.
A. Varada RaJulu, S. Allah Baksh, G. Ramachandra Reddy and K.
Narasimha Chary, Chemical Resistance and Tensile Properties of Short
Bamboo Fiber Reinforced Epoxy Composites, Journal of Reinforced
Plastics and Composites, Vol. 17, No. 17, pp. 1507-1511, 1998.
Seema Jain, U. C. Jindal and Rakesh Kumar, Development and fracture
mechanism of the bamboo/polyester resin composite, Journal of Materials
Science Letters, 12, pp. 558-560, 1993.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
179
Abstract
Characteristic behaviors of plain-woven Carbon Fiber Reinforced Plastics
(CFRP) at cryogenic temperature were investigated under static and cyclic
loading. The test results were compared with those of Glass Fiber Reinforced
Plastics (GFRP). Tensile behaviors of monofilaments were also evaluated. The
test results showed that two knee-points appeared in the stress-strain curve of
GFRP under static load, while that of CFRP was almost linear. Both strength and
failure strain of the CFRP at cryogenic temperature were lower than those of
CFRP at room temperature, because the epoxy resin was brittle at cryogenic
temperature. At cryogenic temperature, the knee-point was also shown in the SN curve of GFRP of the fatigue test but that was not shown in the S-N curve of
CFRP. The elastic modulus of CFRP at cryogenic temperature suddenly
decreased in the final stage of fatigue, while such change was not observed at
room temperature. The thermal fatigue test where no cyclic loads were applied to
the specimen was also conducted in order to investigate the damage progression
due to temperature change. After the thermal fatigue test, the residual strength of
the CFRP and GFRP were measured at room temperature. The change in residual
strength was explained by the difference of the coefficient of thermal expansion.
This paper also mentioned that, at cryogenic temperature, the stiffness reduction
under cyclic loading was related to the local thermal stress by thermal cyclic
fatigue. It should be said that the carbon fiber was failed with accompanying
critical crack propagation of the matrix due to the brittle of resin at cryogenic
temperature.
Keywords: fiber reinforced plastics, cryogenic temperature, coefficient of
thermal expansion, residual strength, stiffness reduction.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06019
Introduction
Fiber reinforced plastics (FRP) are widely used for the structural component due
to their low coefficient of thermal expansion as well as high specific strength and
stiffness [13]. The FRPs are also expected in use at harsh environment, for
example outer space and liquid fuel tank for disposable rocket. The GFRP (Glass
Fiber Reinforced Plastics) has been partly used at low temperature of a ship, and
Sakurai [4] discussed the strength of GFRP for them. However their use is
limited but aluminium alloy is still used for many applications. To apply the
FRP, for example, for the fuel tank instead of the aluminium alloy, behaviors of
FRP such as CFRP (Carbon Fiber Reinforced Plastics) and GFRP under the
cryogenic temperature should be cleared.
The purpose of this study is to investigate the mechanical properties of CFRP
at room and cryogenic temperature using liquid nitrogen, in comparison with
those of GFRP. The mechanical properties of carbon and glass monofilaments at
room and cryogenic temperature were also shown. The coefficients of thermal
expansions were measured for FRP as well as the epoxy resin for polymer
matrix. The tensile tests of CFRP and GFRP were conducted at room and
cryogenic temperature. The tension-tension cyclic loading was applied to the
CFRP and GFRP with hydraulic servo testing machine having nitrogen cryostat.
Thermal fatigue tests (without loading) were also conducted in which
environmental temperature was alternated between room and cryogenic
temperature. The residual strengths of the damaged specimens were measured at
room temperature after the thermal fatigue test.
Aluminium
R70
Cut off
20
8
80
150
Adhesion
t = 2 mm
Figure 1:
Monofilament
Grip area
25
Diameter 6.2-8.3m
Experimental procedure
181
(a)
Figure 2:
(b)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Tmax (oC)
23
(oC)
(oC)
Materials Diameter
[m]
Temperature
[oC]
23+2 (Room Temp.)
CF
240
3503
1.34
269
3168
1.22
68.5
1330
1.70
67.7
2650
3.10
6.08.5
-196 (Cryogenic Temp.)
23+2 (Room Temp.)
GF
7.814.1
-196 (Cryogenic Temp.)
183
5m
x7.0 k
x7.0 k
5m
5 m
9.0 k
7.0 k
5m
500
Stress[MPa]
800
Stress[MPa]
600
RT
CT
600
400
200
400
300
200
RT
CT
100
0.0
0.5
1.0
Strain[%]
1.5
2
3
Strain[%]
Stress level
=max/st [%]
CFRP
RT
639.1
CT
630.0
GFRP
RT
294.6
CT
564.5
Stress[MPa]
60
45
30
RT
CT
15
0
0.0
0.5
1.0
1.5
2.0
Strain[%]
Figure 7:
185
temperature to room temperature. Figure 8 and 9 show the S-S curves of CFRP
and GFRP, respectively after thermal fatigue test, in which the thermal stress
was applied to the specimens before the static test.
Residual strengths [13] of CFRP and GFRP were almost equal to their each
static strength even after 100 cycles of thermal fatigue loading. However the
residual elastic modulus of CFRP was decreased by the cyclic thermal stress.
There was the large difference in the coefficient of thermal expansion of carbon
fiber and that of pure epoxy resin. Therefore the epoxy resin in CFRP should be
damaged during the thermal fatigue test [14], because the magnitude of thermal
free strain was larger than the failure strain of pure epoxy resin at cryogenic
temperature as shown in section 3.3.
Table 4:
Materials
CFRP
Tmin (oC)
23
-60
700
300
600
250
500
Stress[MPa]
Stress[MPa]
Epoxy resin
Tmax (oC)
400
300
200
0 cycle
100 cycles
100
0
200
150
100
0 cycle
100 cycles
50
0
0.5
1
1.5
Strain[%]
0.5
1.5
Strain[%]
On the other hand, in Fig. 9, the residual elastic modulus of GFRP was equal
to the static elastic modulus. This is explained by that the resin in GFRP was not
damaged at cryogenic temperature, because the coefficient of thermal expansion
of glass fiber was almost equal to that of pure vinyl ester resin.
3.5 S-N curves of CFRP and GFRP
The S-N curves of CFRP and GFRP at room and cryogenic temperature are
shown in Fig. 10 and 11. CFRP had long durability at cryogenic temperature in
high cycle fatigue compared with that at room temperature. The fatigue lives of
GFRP at cryogenic temperature were also longer than those at room temperature.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
700
650
600
550
500
450
CT
RT
400
100 101 102 103 104 105 106 107
Number of cycles to failure Nf
Figure 10:
The knee-point appeared in the S-N curve of the GFRP at cryogenic temperature.
The maximum stress at the knee point in S-N curve was almost coincident to the
stress at knee point under static load at cryogenic temperature. It should be said
that the stress at static knee-point is a threshold in cyclic fatigue loading which
determines the damage progression mode in fatigue of GFRP.
The elastic modulus decay diagrams of CFRP was shown in Fig. 12, where
=90% at room and cryogenic temperature. The elastic modulus of CFRP at
cryogenic temperature suddenly decreased in the final stage of fatigue, while
such change was not observed at room temperature. Same result was observed in
the paper [15]. At the cyclic number to failure in the condition of room
temperature, the sharp decreasing in elastic modulus occurred at cryogenic
temperature.
600
500
400
300
200
100
CT
RT
0
100 101 102 103 104 105 106 107
Number of cycles to failure Nf
Figure 11:
120
100
80
60
40
RT
CT
20
0
100
101
102
103
104
105
Number of cycles N
187
120
100
80
60
40
RT
CT
20
0
100
101
102
Number of cycles N
103
RT
CT
Conclusion
(1) The fatigue lives of CFRP and GFRP at cryogenic temperature were higher
than those at room temperature.
(2) At cryogenic temperature, the magnitude of thermal strain of pure epoxy
resin was higher than the failure strain of the pure epoxy resin.
(3) At cryogenic temperature, the epoxy resin adhered carbon fiber remained on
the fracture surface of CFRP applied cyclic loading (=90%), while debonding
was found around carbon fiber on the fracture surface of CFRP at room
temperature.
(4) At cryogenic temperature, the elastic modulus of CFRP decreased in the final
stage of fatigue, while that of GFRP decreases after the first cycle of fatigue.
(5) At cryogenic temperature, the stress at static knee-point of GFRP was the
threshold in cyclic loading, which determined the damage progression mode in
fatigue.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgement
This study was supported by the Ministry of Education, Culture, Sports, Science
and Technology for RCAST (Doshisha University).
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
189
Abstract
As interest in studying environmental issues has grown, the mechanical
properties of eco-friendly materials have been studied. In our research, a card
machine that enables raw materials to loose and to mix uniformly was used.
Fibrous polypropylene and cotton or ramie fiber were used as raw materials
since many textiles and apparel are made of such fibers. From a viewpoint of
reuse, the loosing technique of the card machine can be effective. We designed
anisotropic composites by using the card machine followed by the hot press
process. The tensile, flexural and compression properties were investigated.
Ramie fiber reinforced polypropylene (RP) was stronger than cotton fiber
reinforced polypropylene (CP); however, breaking elongation of composites of
CP is superior to that of RP because of the character of natural fiber. The density
of composites applied to the loosing technique also affected the strength of
composites. It was clear that density is an important parameter to describe the
performance of a green composite. Moreover, it was found that the initial
fracture in flexural behavior is affected by the compression strength of
composites using the loosing technique.
Keywords: cotton fiber, Ramie fiber, loosing technique, tensile property, flexural
property, compression property, density.
Introduction
The increasing concerns for environmental issues and the growing eco-friendly
society direct universities and makers to development of Green composites
[1-4]. Green composites should be friendly for both environment and human
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06020
Experiment
2.1 Material
Polypropylene (PP), a matrix resin, is a fibrous resin with a melting point of
165. Cotton and ramie fibers were used as reinforcement. Length of cotton
fiber was 30 mm, and ramie fiber was 80 mm.
2.2 Production of composite
2.2.1 Loosing method
Loosing of fibrous PP resin and natural fiber was performed by a card machine.
In mixing of the fibrous resin and the natural fiber, it is important to distribute
the fiber uniformly in resin. Therefore, a loosing fibrous resin sheet and a loosing
fiber sheet were separately made by the card machine before mixing fibrous rein
and fiber.
2.2.2 Molding method
Specimens were manufactured by a vacuum compressed molding method as
shown in Fig.1. In this research, molding conditions were selected as shown in
Table 1.
Pressure
Heater
Puressre plate
Vacuum bag
Spacer
Sealant
Composites
Heater
Pressure
Figure 1:
Vacuum pump
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 1:
Molding condition.
30
50
10
15
Molding temperature []
Molding time [min]
191
210
10
80
RP50
SR50
RP30
SRP30
70
Stress [MPa]
60
50
40
CP50
CP30
30
PP
20
10
0
0.0
1.0
Figure 2:
2.0
Strain [%]
3.0
4.0
80
70
SRP( = 0 )
Stress [MPa]
60
RP( = 0 )
50
SRP( = 90)
CP ( = 0 )
40
30
PP
PP
20
10
0
0.0
1.0
2.0
3.0
4.0
Strain [%]
Figure 3:
193
80
R50
70
R30 PP
60
C50
50
40
30
C30
20
10
0
0
Figure 4:
Table 2:
Displace me nt [mm]
PP
C30
C50
R30
R50
65
1.9
55
2.2
64
2.7
69
4.4
66
3.8
100
90
80
70
LD C30
LD C50
LD R30
LD R50
HD C30
HD C50
HD R30
HD R50
PP
60
50
40
30
0.60
0.70
0.80
0.90
1.00
1.10
Density [g/cm ]
Figure 5:
Tensile streng th [M Pa ]
70
60
50
LD C30
40
LD R30
LD C50
30
LD R50
HD C30
20
HD C50
10
HD R30
HD R50
0
0.60 0.70 0.80 0.90 1.00 1.10 1.20
3
Density [g/cm ]
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
195
3.4 Difference of fracture surface between low & high density material
Fig.7 shows SEM photograph of tensile fracture surface of low density and high
density CP, respectively. There are many pull out single fibers in both, however
these forms are different. A notable point is the difference among their lengths.
In the high density composite, fibers were evenly broken shortly. Therefore, it
showed relatively flat fracture surface. It was also observed in the fracture
surface of CP.
The form of the tensile fracture surface is determined by the relative degree
and friction force between fiber and resin. If the ratio of tensile strength to
interfacial shear strength is relatively small, then the fracture surface is flat [6].
In this research, high pressure was applied to manufacture high density
composites. Fatal damage caused by high molding pressure decreased the
potential of reinforcement by the natural fiber. Therefore, it is thought the
massive damage on the crossing point of fibers made high density composites
low strength.
142m
142m
point. At Knee point, composite reaches compression yielding point, and then
relationship between flexural stress and displacement starts non liner behavior.
Fig.9 shows flexural stress at Knee point in flexural behavior.
50
45
40
35
30
25
20
15
10
5
0
Low de nsity
High density
PP
C30
C50
R30
R50
Material
Flexural stress
Figure 8:
. Elastic
. Compression yielding
. Tensile fracture
Knee point
X-head displacement
Figure 9:
50
197
Low density
45
High density
40
35
30
25
20
15
10
5
0
PP
C30
C50
R30
R50
Material
Figure 10:
Conclusion
Cotton fiber and ramie fiber reinforced polypropylene were manufactured using
loosing technique. The experimental data showed that:
(1) Loosing technique enables to uniform fiber orientation in both cotton fiber
reinforced polypropylene and ramie fiber reinforced polypropylene.
(2) Long fibers (fiber length=80 mm) were easier than short fiber (fiber
length=30 mm) to orient natural fiber uniformly in ramie fiber reinforced
polypropylene.
(3) There was a positive correlationship between mechanical properties such as
tensile strength and flexural strength and density in low density materials.
(4) Damage of natural fiber by high pressure caused weakness of tensile
strength of composites.
(5) Initial fracture in flexural behavior of loosing natural fiber reinforced
polypropylene was affected by compression strength. Therefore,
improvement of compression strength is necessary to obtain higher
performance of loosing natural fiber reinforced polypropylene.
Acknowledgements
Authors would like to express special thanks to CHISSO Ltd. for providing
fibrous PP resin in this study. This study was supported by the Academic
Frontier Research Project on New Frontier of Biomedical Engineering
Research of ministry of education, culture, sports, science, and technology.
References
[1]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 4
Material and mechanical
characterisation
201
Abstract
The influence of the loading speed on mechanical response of structural
materials can be accounted by means of strain-rate sensitivity parameters. The
aim of the present work is to show a numerical technique based on an inverse
approach to determine strain-rate sensitivity parameters of steels for car body
constructions. This technique is based on the numerical simulation of a simple
test according to the ASTM D5420/96 standard by means of a finite element
explicit code. The test consists of a falling tup with a spherical head impacting
on a thin sheet. Some experimental tests are conducted at different speeds, from
quasi-static to impact loading conditions, on a specimen made of XE280P steel.
A series of simulations are performed, changing the strain-rate sensitivity
parameters in each run according to a genetic algorithm strategy. The strain-rate
parameters that lead to the best fit of the experimental load-displacement curve
with the numerical result are the assumed material characteristic parameters. The
Cowper-Symonds and Johnson-Cook strain rate models have been taken into
consideration.
Keywords: strain-rate sensitivity, optimisation, genetic algorithm, finite element
analysis.
Introduction
Since the second half of the past century, research in the automotive industry
focused on safety improvement. Increasing customer interest about passive
safety and the ever stricter regulations, both in the US and EU, pushed towards
more reliable vehicle structure design and a deeper insight into material
behaviour. The finite element method applied to crash simulations with explicit
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06021
Quasi-static uniaxial tensile tests and a series of simple bending tests at different
strain rates were performed to characterise the XE280P steel sheets and to
identify parameters of models. Both types of test are necessary because they
involve different mechanisms of deformation and help the stability and
completeness of the identification.
The uniaxial tensile tests were performed by means of a general purpose
hydraulic testing machine (DARTEC HA100). The usual dog bone shaped
specimens were manufactured from a laminated sheet of 1.5 mm thickness. The
obtained true stress vs. true strain curve is fundamental for the implementation of
the material behaviour in FE codes and was used for a first direct
characterisation (fig. 1).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
203
600
500
400
300
Elastic modulus:
Yield stress:
Maximum stress:
Failure stress:
200
100
0
0,00
Figure 1:
0,05
0,10
180 GPa
280 MPa
517 MPa
396 MPa
0,15
0,20
True strain [-]
0,25
0,30
30
Force [kN]
25
v = 6,26 m/s
20
15
v = 0,1 mm/s
10
5
0
0
Figure 2:
10
15
Stroke [mm]
20
25
Figure 3:
Beyond the yield stress, the static plastic behaviour of the material is
modelled, as usual, by the following Hollomon formulation:
s = A + B pl n
(1)
where pl is the plastic component of the total strain, while A, B and n are
material constants. For the dynamic behaviour two strain rate sensitivity models
have been considered. Both models modify the static stress-strain curve by
adding a multiplying factor depending on the strain rate and both models are
characterised by the presence of two parameters that have to be identified. The
Cowper-Symonds model has the following formulation:
1p
= s ( ) 1 +
D
(2)
where s() is the static stress and D and p are material constants. The JohnsonCook model includes a further factor to account for the temperature dependence
that is not considered in this work:
= s ( ) 1 + C ln
0
T
1
Tm
(3)
205
The identification of the static and strain rate sensitivity parameters was
performed by means of an optimisation technique. A set of simulations were
executed, changing the values assigned to the material characterising parameters
in each run according to a genetic algorithm strategy. For each run a complete
generation of a defined number of models was analysed. The parameters that
lead to the best fit of the numerical result with respect to the experimental result
have been selected as the optimal material model parameters. In particular the
following objective function to be minimised was used:
N
(y
exp,i
y mod,i
)2
i =1
(4)
where parq are the parameter values to be identified, yexp,i and ymod,i are the
experimental and numerical responses respectively, and N is the number of curve
points. The objective function was evaluated on the stress-strain curve for the
uniaxial tensile test and on the force-displacement curve for the bending tests. In
both cases curve data up to the maximum of, respectively, the stress or the force
were considered.
The genetic algorithm is defined by the mechanisms of selection,
recombination and mutation. The selection is used to choose the parent models
and is performed randomly by means of probability functions derived from the
model fitness: the higher is the model fitness the higher is the probability to be
chosen as parent of the new generation. The model fitness is defined as the
difference between the value of the objective function for the worst model of the
generation and the value of the objective function for the model itself. The
selection is limited to the models that constitute the 50% of the cumulated
fitness, decreasingly ordered. The recombination is the main feature of the
genetic algorithm. It defines the son models by combining different parent
models:
x a = (1 ) x + x ; x b = x + 1 x
(5)
Two son models (xa, xb) are generated from two parent models (x, x). and
are random numbers generated by a normal probability function with null
500
B=491,678 MPa
Experimental curve
300
7-9
9-11
11-13
13-15
15-17
17-19
19-21
21-23
23-25
25-27
0,36
0,32
0,28
200
n [-]
Stress [MPa]
0,40
400
0,24
0,20
100
0,05
Figure 4:
0,10
True plastic strain [-]
0,15
130
145
160
175
190
205
220
235
250
265
280
295
0,16
0
0,00
0,20
0,12
A [MPa]
The bending test performed at low speed can also be used for the static
parameters identification. In this case different and not uniform mechanisms of
deformation are involved and FE simulations are needed to have the numerical
force-displacement curve to be compared with the experimental one.
600
Experimental curve
Model gradient method
Model genetic algorith
25
20
Stress [MPa]
Force [kN]
30
15
10
5
0
0
Figure 5:
10
Stroke [mm]
15
20
500
400
300
Experimental curve
Model gradient method
Model genetic algorith
200
100
0
0,00
0,05
0,10
0,15
0,20
207
method that could lead to the same well simulated curves with different
parameter values. Moreover it shows that the FE model of the bending test could
be not sufficiently accurate, both for what concern the mesh refinement and for
what concern the through thickness integration.
600
Experimental data
Model gradient method
Model genetic algorith
25
20
Stress [MPa]
Force [kN]
30
15
10
5
0
0
Figure 6:
10
Stroke [mm]
15
500
400
300
Experimental data
Model gradient method
Model genetic algorith
200
100
0
0,00
20
0,05
0,10
0,15
0,20
The use of the complete structural model (fig. 3(b)) for the identification
leads to the results shown in fig. 6. A more accurate solution is obtained: the
material model fits better the experimental stress-strain curve of the tensile test,
while keeping the same quality of fitness for the force-displacement curve of
bending test. All the identified parameters are summarised in table 1.
Table 1:
A [MPa]
B [MPa]
N [-]
F [MPa, kN]
Uniaxial tensile
test
Genetic Gradient
241.7
213.5
482.5
491.7
0.3298
0.2857
8.328
8.186
Bending test
Simplified model
Genetic Gradient
238.7
305.2
499.0
521.1
0.5913
0.9116
0.361
0.179
Bending test
Complete model
Genetic Gradient
282.4
273.1
532.1
528.4
0.5323
0.4967
0.173
0.185
F(p1 , p 2 ) =
1
2
(F
exp1,i
Fmod1,i
i =1
N1
N2
)2
+
1
2
(F
exp 2,i
Fmod 2,i
i =1
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
N2
)2
(6)
6,0
5,5
5,0
4,5
4,0
3,5
3,0
2,5
2,0
1,5
1,0
0,5
0,0
Objective F [kN]
Objective F [kN]
Minimum of generation
Maximum of generation
1
Generation
10
11
12
13
10
9
8
7
6
5
4
3
2
1
0
14
DYNAMIC
Gradient
method
Model: Cowper-Symonds
Iterations
Figure 7:
Table 2:
D [s-1]
p [-]
F [kN]
Complete model
Genetic
Gradient
4987
3013
1.619
1.503
1.817
1.822
Simplified model
Genetic
Gradient
2399
3006
1.329
1.416
1.187
1.209
Conclusions
209
35
DYNAMIC
30
Genetic
algorithm
Force [kN]
25
Model: Cowper-Symonds
20
15
Experimental curve - h=1 m
Experimental curve - h=2 m
Simulation curve - h=1 m
Simulation curve - h=2 m
10
5
0
0
Time [ms]
Figure 8:
Table 3:
0
Complete model
Genetic
Gradient
Simplified model
Genetic
Gradient
1.00
1.00
[s-1]
0,01149
1,375
C [-]
F [kN]
0,01223
1,372
0,01028
1,329
0,01024
1,330
35
DYNAMIC
30
Genetic
algorithm
Force [kN]
25
Model: Johnson-Cook
20
15
Experimental curve - h=1 m
Experimental curve - h=2 m
Simulation curve - h=1 m
Simulation curve - h=2 m
10
5
0
0
Time [ms]
Figure 9:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgement
The authors kindly acknowledge the European Commission that, by funding the
ADVANCE project through contract GRD1-200-25914, made this research
activity possible.
References
[1]
[2]
[3]
[4]
[5]
[6]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
211
Mechanical characterisation of a
viscous-elastic plastic material, sensitive to
hydrostatic pressure and temperature
V. D. Le1, M. Caliez1, M. Gratton1, A. Frachon1 & D. Picart2
1
Abstract
This paper deals with the characterization of the static mechanical behaviour of
an energetic material. Due to the constituents (crystals and a polymeric binder),
the behaviour is influenced by the pressure, the temperature and the strain rate.
The temperature, considered varying slowly, is a parameter and the
computational problems are uncoupled. Therefore, a complete experimental
protocol and a model have been developed. Inspired from the Visco-Scram
model, the behaviour is described using a general Maxwell model in which all
the branches are affected by an isotropic damage. The first branch takes into
account elastic-plastic behaviour. The yield stress is given by a parabolic
criterion, characterized using compressive, tensile and tri-axial tests. The
hardening is isotropic and the plastic flow rule is nonassociated. The other
branches are viscoelastic. A genetic algorithm is used to optimise the viscoelastic
parameters, previously obtained using DMA measurements. Comparisons
between the model and experiments are proposed for different temperatures,
strain rates and pressures. At last, a user material subroutine has been developed
in Abaqus Standard and finite element computations of the Brazilian test are
compared to the experimental response.
Keywords: energetic material, parabolic plastic criterion, genetic algorithm,
DMA, viscoelasticity, Isotropic damage, Brazilian tests.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06022
Introduction
The material is made of organic and energetic crystals mixed with a few
percentage of a polymeric binder. After an isostatic compaction forming process,
the material has a small porosity of a few percent. Samples can be machined in
several geometric shapes, which are more than ten times the length of the
material heterogeneity.
In order to survey the possible aging of this material, an accurate
determination of the mechanical properties has to be done. Unfortunately, this
material being available in small amounts, the characterization must be made
using a reduced number of standard tests. An unusual experimental procedure is
proposed in this paper for this kind of material. When monotonic loading paths
were used to determine for example the influence of the strain rate, each sample
is submitted to complex loading paths including relaxation, recovery and cyclic
conditions.
The observation of loading-unloading diagram on figure 1 shows some of
the main features of the material at room temperature, and entails specific
arrangements for the mechanical tests. 1) Hydrostatic pressure sensitivity: to
consider it, an initial hydrostatic loading path (0 MPa, 5 MPa and 10 MPa) is
made before the run of a uniaxial compression load. 2) Viscosity: different strain
rates (5.10-6 s-1 to 10-3 s-1) have been used to observe such effect. The parameters
of the viscoelastic part of the behaviour have been determined using a DMA
apparatus (Dynamic Mechanical Analysis). 3) The plastic strains are determined
using relaxation and recovery delays. 4) Initial elastic behaviour: standard tests
made in various material directions show an initial isotropic behaviour.
5) Damaging: systematic cyclic loading-unloading programs have been
performed. 6) Dispersion: to ensure a minimum statistical validity, each loading
program is repeated five times.
The temperature is considered as a parameter in the model. The material
been temperature dependent, compressive and tensile tests are perform at four
different temperatures: 5C, 20C, 35C and 50C. The figure 2 shows the effect
of the temperature on monotonic compressive experiments.
The main models available in the literature ([13]) for this kind of material
have been developed for transient dynamic behaviour and are not adapted for a
quasi-static study. For example, the influence of the pressure is omitted, even as
the difference of behaviour observed in tension and compression loading paths.
The Maxwell model we have chosen (fig. 3) is close to the constitutive law
proposed in [2]. Several damageable viscoelastic branches and one damageable
elastoplastic branch are used. The main difference with the Bennett and coworkers model is the presence of the elastoplastic branch.
The determination of the plastic (resp. viscoelastic) behaviour is described
in the second (resp. third) part of this paper. In particular, a genetic algorithm has
been used to optimize the determination of the viscoelastic parameters. The
fourth part deals with the damage rule. It can be noted that the characterization of
the plasticity, the damage and the viscosity are uncoupled. The damage rule is
assumed to affect also the viscoelastic branches.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
213
Lastly, some comparisons are presented between the model response and
the available experimental data.
Figure 1:
Damageable
Ep
E1
Ej
En
Figure 3:
1
j
n
The test procedure is described in [4]. The first stage (when this one exist)
of the loading program is a hydrostatic loading phase. Then, five or six uniaxial
loading-relaxation-unloading-recovery cycles are done. The test is driven by one
of the two longitudinal gages. Relaxation times have been defined to guarantee
an almost complete relaxation of viscous stresses. All the tensile tests and the
compressive tests have been done for several temperatures (5C, 20C, 35C and
50C).
Some compressive DMA experiments have been realized using samples of
50 mm long, and a cross section of 4x4 mm2. A small initial preload of 10 m
and a strain amplitude of 5 m are used to stay in the viscoelastic domain.
The range of frequencies going from 0.004 to 40 Hz, the strain rate ranges from
2.10-6 s-1 to 2.10-2 s-1.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Elastoplasticity
The macroscopic stress tensor is defined as the sum of the stress of each
branch:
ep
j =1
(1)
where the subscript ep (resp. j) denoted the elastoplastic branch (resp. the jth
viscoelastic branch). An additive decomposition is assumed for the strain
mechanical part (superscript m) and the strain thermal part (superscript th):
= m + th .
(2)
An additive decomposition of the mechanical strain of each branch is also
assumed between: an elastic part (superscript e) and a plastic (superscript p)
part (resp. viscous) (superscript p or v):
p
elasto-plastic branch,
(3)
m = eep + ep
= ej + vj visco-elastic branches.
(4)
2.1 Thermo-elasticity
The elastic part is supposed linear damageable and given by the following
equation:
E
ep = (1 d ) ( ) : eep ,
(5)
Temperatures
Initial Young modulus Eep (MPa)
5C
3400
20C
2900
35C
2700
50C
2450
215
adaptations of this model were then developed for various applications [68]. As
the forming process of the material is an isostatic compaction up to a pressure of
200 MPa (which amplitude will never be reached in quasi-static applications),
the criteria is supposed open on the hydrostatic negative axis. Open threshold are
usually derived from Mohr-Coulomb, Mises-Schleicher [9], Drucker-Prager [10]
and more recently Hoek-Brown formulations [11]. Raghava et al. [12] applied
the Mises-Schlechers threshold to polymers. The evolution of this criterion is
described by two hardening variables, associated to tensile and compressive
response. Lastly, a unified model is proposed by Aubertin and Li [5] in order to
reproduce all the kinds of criterion (elliptic, parabolic, hyperbolic).
Due to a lack of data about the nature of the hardening mechanisms, an
isotropic hardening parameter, denoted k, is introduced in the model. Then, a
saturation of the hardening mechanism at the maximum stress is taken into
account in the model. A nonlinear plasticity criterion reproduces the evolution of
the yield stress (fig. 4). The following relation is used:
f ( Q , P , k ) = eq k = 0
where Q = (1 3 )
with
eq = Q 2 +
k2
X (k )
P , (6)
and P the pressure. The set of yield curves is completely defined as soon as the
function X(k) and the hardening law are defined.
The following guidelines help for the determination of the function X(k).
First, it is assumed that the yield curves do not cross themselves in the P-Q
plane, each one being embedded in those of higher levels, all of them being
embedded in the extreme curve. This is a necessary - but not sufficient condition for the phenomenon to be governed by a unique state variable which is
the isotropic strain-hardening parameter. Elementary algebra shows that the
following relation satisfies the previous assumption:
X (k ) = X 0 + ( X m X 0 )
k k0
km k0
(7)
where the parameter X0 = 1,5 MPa, positioning the summits of the initial
parabola, is supposed not being temperature depending, and the three parameters
Xm, k0 and km are temperature depending. The Xm and km parameters are
determined using the ultimate yield stress curve relating the maximum stress
states in the P-Q plane. The values are given table 2.
Table 2:
Temperatures
Xm (MPa)
k0 (MPa)
km(MPa)
c1
c2
20C
1,620
0,55
3,5
500
100000
35C
1,610
0,45
3,26
500
100000
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
50C
1,592
0,35
2,82
500
90000
Figure 4:
Figure 5:
For the hardening law, the following hyperbolic relation is used to interpolate the
data:
1
k ( p ) = k0 + ( km k0 ) 1
2
1 + c1 p + c2 p
where c1 and c2 are two parameters temperature depending (table 2).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(8)
217
eppV
eppD
p
with eppV = Tr ( ep
)
and
pD
pD
: ep
eppD = 3 ep
, (9)
pD
ep
being the deviatoric plastic strain rates.
ep
=
+
3
3Q
3
I .
,
(1 + 2 )
(10)
being the plastic multiplier and the plastic flow direction being normalized.
The dilatancy is characterized from the experimental results as a constant
( = 0,15). As a result, a nonassociated plastic law is justified.
Viscoelasticity
j = 10
( A1
j 1
n 1
+ A2
((
and E j = A3 + A4 e
A5
j 1
n 1
1 .
(11)
Then, the number of unknowns decreases to five (A1 to A5). These relations allow
reproducing at the same time the storage and the loss modulus.
3.2 Genetic algorithm
The previous set of parameters is used here to determine the bounds of each
parameter. We are in the case of a combinatory optimisation problem where a
large number of solutions could be suitable. We have chosen to perform an
inverse identification of the viscoelasticity parameters directly from the
experimental tests. Classical optimisation methods, like conjugated gradient,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Obj =
i =1
( ie ( t ) is ( t ) ) . Li ,
(12)
nb exp pts
L
i =1
219
Therefore, for the tests at 5C, 35C and 50C, only the parameters A1 and
A2 have to be identified with the genetic algorithm. The results are presented in
the table 3.
Table 3:
Temperature
5C
Tests com- Compressive,
puted from: one strain rate
A1
2,78
A2
2,63
A3
A4
A5
20C
All
3,98
1,64
160,31
16,15
1,21
35C
50C
Tensile/Compressive, Tensile/Compressive,
one strain rate
one strain rate
2,15
2,62
2,60
2,32
Isotropic damage
E0 E
E0
(13)
where Eo and E are the initial and current Youngs modulus. A constant
Poissons ratio is assumed here.
Figure 6:
1
,
( I ) > + + d 2 .1d = d 1 . sup < max
N
N
1 + d 3 sup < max
I >+
N
=
1
3
I
to
N
time
time
(14)
I =1 to 3
The damage rule is reported in the fig. 6. One can see that the model
response is identified using the compression measurements. The hydrostatic data
are not taken into account because the pressure stops an eventually growth of the
microcavities. The introduction of a second damage mechanism, as proposed in
[19], to model the tensile damage is in study. The identified values are the same
for all the temperatures: d1 = 0,5, d2 = 0,75 and d3 = 75.
The constitutive law has been implemented in the finite element code
Abaqus/standard. The model is compared to experimental data with unloading
cycles to access to the plasticity level and the damage level. Those seem quite
well reproduced even if the transversal model response does not present enough
damage in compression (fig. 78). This observation can be associated to an
anisotropic damage (which has been neglected here).
The plasticity branch reproduces very well the pressure effect that causes
the difference between tensile and compressive responses (fig. 7, 9, 10). The rate
effect is also quite well reproduced on the two compressive tests even if the
unloading curves do not present the same nonlinearity (fig. 78). In the same
manner, the viscous effects in the triaxial test with 10 MPa of confinement
pressure (fig. 10) is underestimated. Certainly, these phenomenons is associated
to an internal friction in the material or a viscosity pressure dependency. The
pressure dependency is certainly the reason of the discontinuity of the Ai
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
221
parameters at 20C (table 3), temperature for which all the tests are computed
and in particular the triaxial ones.
Figure 9:
Figure 11:
Conclusion
Acknowledgement
The authors address a special thanks to J.L. Brigolle for its contribution to this
study, especially for the realization of the experiments.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
223
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
225
Abstract
In load carrying applications one is mainly interested in the stiffness properties
of the structural component. Elastic and shear moduli supplemented with section
properties render the stiffness properties of the component. Moduli of isotropic
materials are well known and well documented. However, this is not the case for
composite materials.
The developed procedure belongs to the group of mixed numerical
experimental methods. The method makes use of modal data to determine the
effective orthotropic material properties of composite beams. Modal reference
data is experimentally obtained from the beam at hand. The other modal data set
is obtained from a finite element model of the same beam. The orthotropic
material properties, also called parameters, in the finite element model are then
modified in such a way that both sets of modal data match. If those two sets
match, the virtual model has in a global sense the same mass and stiffness
properties as the real model.
A program written in FEMtools is applied to five test cases. Results are
discussed.
Keywords: inverse methods, FEA, orthotropic material properties, composite
materials, modal data.
Introduction
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Input Geometry
227
Model Update
Modified fem model is shown.
Reference physical test model is shown.
MAC-matrix fem versus test is shown.
Correlation tracking curve is shown.
Modified Material Properties listed.
No Successful
Identification
Successful Identification
From a mathematical point of view, the difficulty with model updating is that the
relation between output vector and the parameter vector is nearly always nonlinear. This means that updating the parameter values from an initial value to a
final value has to be done iteratively. The value of the output for some new
parameter values can be estimated with a Taylor expansion. The Taylor series
can be cut off after the linear term or can be cut after some higher order terms.
Figure 1 illustrates the mathematics involved in model updating.
The matrix [S ] that appears in the linear Taylor term is called the sensitivity
matrix. It contains the partial derivatives of the output components for the
different parameter values. The success of model updating is highly dependant
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 1:
Figure 2:
{p}0
229
[W]y
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
231
y i y i (p j p j ) y i (p j )
=
p j
p j
The sensitivities discussed so far are absolute sensitivities. This means that they
use the units of the response and parameter value. The absolute sensitivities can
be made independent of the units used for the response and parameter values.
They are then referred to as normalized sensitivities. A normalized sensitivity
shows the percentage change of the response value for one percent change of the
parameter value. The element ij of the matrix [S n ] can be written as
Sij(n) =
y i p j
p j y i
({ } { })
)=
({ } { })({ } { })
2
MAC(analyt , exp
After model updating, a MAC value can be calculated between the updated fem
model and the physical test modal. If no mode switch occurred during model
updating this MAC matrix is a diagonal matrix.
5.3 Aspects of convergence
CCABSOLUTE: Average value of weighted absolute relative differences
between predicted and reference resonance frequencies.
CCMEAN: Average value of weighted relative differences between predicted
and reference resonance frequencies.
In model updating, the above correlation coefficients are interpreted as an
objective function that needs to be minimized. With each iteration loop the
values of the correlation coefficients will be verified to check if a convergence
criterion is satisfied. The following criteria are used:
(1) The value of the reference correlation coefficient is less than an
imposed margin:
CCt < 1
(2) Two consecutive values of the reference correlation coefficient are
within a given margin.
CC t +1 CC t < 2
The iteration loop in model updating will be stopped as soon as one of these tests
is satisfied. Default values for 1 and 2 are used namely 0.08.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
233
65 mm
26.5 mm
3.18 mm
19 mm
3.18 mm
3.18 mm
63.5 mm
Figure 3:
Symmetric U-profile.
Ex
Initial
Method used
value
to obtain value
(N/mm)
22000
Measured
Ey
11800
Measured
Gxy
5000
Literature
vxy
0,3
Literature
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Mode 1: 32
Mode 2: 61
Mode 3: 116
Mode 4: 140
Mode 5: 197
Mode 6: 199
Mode 7: 207
Mode 8: 217
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Mode 9: 221
235
Mode10: 235
Figure 4:
Sensitivity matrix.
Conclusions
Initial values
[N/mm]
Ex
Parameter
selection
Mode
Fem data based on
shape
initial values
switch
versus Exp. data [Hz]
occur
Fem Exp. Description
Y N
value value
mode
32
32.5
1 torsion
1 complex
bending Y
116 137.5 1 bending Z
2 complex
140 146.9
bending Y
Final parameter
values [N/mm]
Ex
Ey Gxy Vxy
61 66.25
The program needs initial values for orthotropic material properties, before
identification can start. A deviation of 25 % given on estimated initial values,
results in the same final updated values for the parameters. Hence, the final
updated results are almost not sensitive to a deviation of initial values.
The program can also be used to study the influence of the length of the
structure on the possibility to identify certain parameters. Consequently, an
optimal length can be determined for which the first (two) frequencies are very
sensitive to a change in value of a preferred parameter. Making it possible to
identify orthotropic material properties in a more easy and structured manner.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
237
References
[1] Modal Testing: Theory and Practice, DJ Ewins, Research studies press
[2] Engineering Vibration, DJ Inman, Prentice Hall
[3] Stress analysis of fiber-reinforced composite materials, MW Hyer, McGrawHill
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
239
Abstract
The application of FE simulation in manufacturing processes and virtual
prototyping increases every day. In order to allow accurate simulations, correct
constitutive models are needed as input to the FE software. A modern and
promising way to identify the material parameters in those constitutive models is
inverse modeling. Full-field measurement is a suitable way to get the
necessary experimental data. The technique has many advantages such as large
information contents, non-contacting measurement, and versatile size of
observation region, among others. However, there is no standardization yet for
this kind of measurements. Therefore, there are many disagreements among
researchers about how to design DICT experiments and how to correctly collect
the data from DICT experiments. This paper will concentrate on discussing the
key points of those problems as well as presenting some work experience with
the DICT.
Keywords: inverse method, FEA, full field measurement, digital image
correlation.
Introduction
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
241
during the service life of the construction or during the manufacturing process,
e.g. Single Point Incremental Forming process (SPIF).
The main problem of adopting more complicated tests in the past was always
hindered by the fact that complex displacement fields simply could not be
measured. In recent years however, an increasing number of important
developments in the field of full-field displacement measuring have been
presented. Moreover, modern measurement equipment, as there are Flow
Induced Birefringence, Electronic Speckle Pattern Interferometry, Digital Image
Correlation techniques, have become commercially available and more
accessible than they used to be in the past. Furthermore, more information about
material parameters can be caught from one experiment, so called full field
measurement.
In the current article, an example of full-field measurement by means of
digital image correlation technique (DICT) will be presented. This technique
exploits a good prospect for inverse modeling since it has some important
advantages, for instance non-contacting measurement, large observation region,
etc. However, there is no standardization for such kind of measurements.
Therefore, there are many disputations about how to design DIC experiment and
how to correctly collect the data from DIC experiment. In the following the
influence of subset size, step size and strain window size on the final strain
computation will be discussed in detail. Some experiences for DIC experiment
will be introduced for sharing the knowledge.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2:
The CCD cameras use a small, rectangular piece of silicon, which has been
segmented into arrays of individual light-sensitive cells, also known as photo
sites. Each cell constitutes one element of the whole picture and is called a
pixel. Every pixel stores a certain grey scale value varying from 0 to 255, in
agreement with the intensity of the light, reflected by the surface of the tested
specimen. Thus, an image can be looked at as a matrix in which every element
represents the grey value of the corresponding pixel.
In order to run the correlation algorithm, the image is divided in a number of
subsets. A subset represents a part of the whole image. The size of a subset can
be varied by the user and the choice of a good subset size depends on the
deformation. As was already mentioned, this technique uses a random speckle
pattern that is simply sprayed onto the surface of the object or that is offered by
the texture of the specimens material. The objective is to obtain an image with a
varied and distinctive pattern, which enables the correlation algorithm to trace
the subsets of the reference image in the deformed images [3, 4]. The concept
behind the DICT-software 2D matching algorithm is that the distribution of grey
values in a rectangular area (subset) in the picture taken of the specimen in the
undeformed state, corresponds to the distribution of grey values of the same area
in the picture taken at the deformed specimen. In this way the motion and the
deformation of the subsets during the deformation of the object are determined.
This will finally lead to a displacement field for the whole area of interest. Some
additional calculations allow to extract strain data from this displacement field.
Figure 3 shows such a subset in the undeformed (left) and deformed (right)
configuration.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 3:
243
140
50
280
15
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
During the post processing of the DIC experiment, the area-of-interest (AOI)
should be selected. Following that, subset size (pixel) and step size (pixel) are
chosen for calculating the displacement field. The strain window size is chosen
for the strain calculation by Vic 3D software. One image, taken when force
applied on the specimen reached a value of 40kN, is selected to study the
influence of Subset, Step and Strain window size on the strain value.
Figure 5:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
245
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 7:
247
Figure 8:
Figure 9:
Conclusion
In the present paper, the feasibility and reliability of data collection from fullfield optical measurement is studied. It has been shown that the size of subset,
step and strain window have a large influence on the accuracy of the measured
displacements and calculated strains. Unsuitable subset, step size and strain
window size will cause experiment data either underestimated or scattered.
In the post processing of DIC experiment, generally the larger the subsets are,
the better the results are. However, large is a relative concept. It is important that
the subset size is chosen in accordance with the expected deformations. It is clear
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1] Shiro Kubo, Inverse problems. Atlanta Technology Publications, ISBN #1883793-01-07
[2] Sol H., Oomens C.W.J., Material identification using mixed numerical
experimental methods. 1997 Kluwer Academic Publishers. pp 1-9
[3] Synnergren P., Sjdagk M., A stereoscopic digital speckle photography
system for 3-D displacement field measurements. Optics and Lasers in
Engineering, 31, pp 425-433, 1999
[4] Fricke-Begemann T., Burke J., Speckle interferometry: three-dimensional
field measurement with a single interferogram. M., A stereoscopic digital
speckle photography system for 3-D displacement field measurements.
Applied Optics 40(28), pp 5011-5022, 1 October 2001
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
249
Abstract
In the past ten years many new processes for making foamed metals, mostly
aluminium or aluminium alloys, have been developed. As a matter of fact,
closed-cell aluminium foam offers a unique combination of properties such as
low density, high stiffness, strength, and energy absorption capability.
One of the main differences in the mechanical behaviour of cellular materials
with respect to classical homogeneous materials such as solid metals is that foam
failure is not independent from a hydrostatic state of stress. Therefore, it is not
possible to describe the failure surface from a single, usually uniaxial, test but it
is necessary to perform tests with different combination of deviatoric and
hydrostatic stress components.
Within the European Project APROSYS, whose main objective is increasing
the safety of all road-users, the mechanical behaviour of aluminium foam, under
multiaxial loading, was investigated by the authors. In this paper the results of
the hydrostatic and hydro-compression experimental tests are reported. From the
results of these tests, it has been possible to obtain the yield locus of the
aluminium foam in the deviatoric-hydrostatic stress components space.
Keywords: aluminium foam, hydrostatic, hydrocompression.
Introduction
It is well know that the mechanical behaviour of a cellular material, like foam, is
not independent from a hydrostatic state of stress as, on the contrary, it happens
for classical homogeneous materials such as metals. Therefore, it is not possible
to describe the failure surface from a single simple uniaxial test (in tension or
compression if there is a different behaviour) but it is necessary to perform tests
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06025
Experimental tests
Hydro-compression
Hydrostatic
compression
= atan(r)
1
Figure 1:
R
R
A
p = A / B
dev = A R
hyd =
hyd
A + 2 R
3
r = dev / hyd
Hydrostatic and hydro-compression tests area in the deviatorichydrostatic stress components space.
251
compression tests were carried out changing the ratio p in order to obtain
different hyd - dev combinations. As shown in fig. 1, combining in different
ways the axial stress A and the radial stress R it is possible to investigate the
highlighted area of the hyd - dev plane. The limits of this area are the pure
hydrostatic test (A = R) and the pure compression test (R = 0).
2.1 Testing device
Performing hydrostatic and hydro-compression tests on a metal foam is a
challenging task. The foam specimen have to be completely separated from the
fluid used to load it; for this reason it was covered with a latex sheath in order to
avoid seepage of the fluid in the specimen. The cover have to be enough strong
to not break during the test and, at the same time, it have to be very light and thin
in order to not change the test results. Furthermore in the hydro-compression
tests it is necessary to control the axial and the radial pressure separately. For all
this reasons a testing device (fig. 2) was designed and built expressly to perform
these tests. The test chamber, containing the foam specimen, was filled with a
fluid (mixture of water and glycol) and mounted directly on the fixtures of a
general purpose hydraulic testing machine DARTEC HA100 controlled with a
DARTEC 9600 electronic unit.
Figure 2:
In the hydrostatic tests the specimen was put on the vertical rod connected to
the testing machine. This rod could move inside the test chamber in the axial
direction changing the volume of the chamber and, consequently, the pressure of
the fluid inside the chamber, i.e. the pressure on the specimen.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Top-bottom
253
Longitudinal
Transversal
Figure 3:
The properties of the specimens used for hydrostatic compression and hydrocompression tests are reported in the following tables.
Table 1:
SPECIMEN
CODE
Diameter
Height
DENSITY
(kg/dm)
AXIAL
DIRECTION
H-B2-1
41.2
41.8
0.66
Top-Bottom
H-B2-2
40.8
39.7
0.66
Top-Bottom
H-B3-1
40.8
45.5
0.64
Transversal
H-B3-2
40.8
45.3
0.37
Transversal
H-B4-1
40.8
41.2
0.27
Transversal
H-B4-2
41.0
41.0
0.36
Transversal
H-B4-3
41.0
40.7
0.40
Transversal
Table 2:
SPECIMEN
CODE
Diameter
Height
DENSITY
(kg/dm)
SPECIMEN AXIAL
DIRECTION
HC-B2-1
40.6
40.5
0.40
Top-Bottom
HC-B2-2
40.8
39.2
0.34
Top-Bottom
HC-B2-3
40.6
40.2
0.57
Top-Bottom
HC-B3-1
40.8
45.3
0.54
Transversal
HC-B3-2
40.8
45.3
0.47
Transversal
HC-B3-3
40.8
45.0
0.38
Transversal
HC-B4-1
40.8
45.4
0.60
Transversal
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Experimental results
30
H-B2-1 0.66
H-B2-2 0.66
H-B3-1 0.64
H-B3-2 0.37
H-B4-1 0.27
H-B4-2 0.36
H-B4-3 0.4
25
20
15
10
5
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 4:
255
(a)
Figure 5:
(b)
Cross section of H-B3-1 (a) and H-B2-2 (b) specimen after the
tests.
80
S-B1-1 0.36 Uniaxial
70
60
50
40
30
20
10
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Volumetric strain
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
SPECIMEN
Density
(kg/m3)
p = R / A
r = dev / hyd
HC-B2-1
0.40
p = 0.658
r = 0.445
HC-B2-2
0.34
p = 0.366
r = 1.111
HC-B2-3
0.57
p = 0.658
r = 0.445
HC-B3-1
0.54
p = 0.658
r = 0.445
HC-B3-2
0.47
p = 0.366
r = 1.111
HC-B3-3
0.38
p = 0.658
r = 0.445
HC-B4-1
0.60
p = 0.366
r = 1.111
H-B3-1
0.64
p = 1 (hydrostatic)
r = 0
H-B3-2
0.37
p = 1 (hydrostatic)
r = 0
Examining the experimental data (axial and radial stress-strain curves), it was
possible to notice a different behaviour in the axial and radial directions. This
behaviour is shown clearly in the fig. 7, where the results of the hydrostatic test
performed on specimen H-B3-2 are reported. This test was performed with the
experimental device for the hydro-compression tests so it was possible to
evaluate the axial and the radial characteristics separately.
Radial strain (-)
0
0.05
0.1
0.15
0.2
8
Axial stress
Radial stress
7
6
7
6
0.05
0.1
0.15
0.2
Figure 7:
In fig. 7, it is possible to notice that, even if the load ratio p = 1 was kept at a
constant level during the test, the axial and the radial yield stress were quite
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
257
different. This means that the ratio between the axial yield stress and the radial
yield stress was different from the ratio p at which the experimental test was
carried out. This event is rather clear in the hydrostatic tests, while in the hydrocompression tests it occurred only some times and not in a so evident way.
However, due to this phenomenon, it was not possible to use the experimental
data directly to obtain the failure locus because it was not clear which one of the
yield stresses should have been chosen. In order to overcome this problem the
following procedure was used. First of all, the hydrostatic stress-volumetric
strain curve was calculated from the axial and radial stress-strain curves through
the following equations:
hyd = (A + 2 R) / 3, vol = A + 2 R
(2)
Then the yield stress for each test was evaluated considering the hydrostatic
stress-volumetric strain curve. In order to remove the effect of the density on the
results, each yield stress, calculated as described above, was divided by the
theoretical yield stress evaluated through eqn (1). Therefore a normalised yield
yield-adim stress was obtained
Finally, the normalised deviatoric and hydrostatic stress components were
obtained combining the following equation:
R = p A, hyd = (A + 2 R) / 3, dev = | A R |
(3)
These components are plotted in the dev hyd (fig. 8) in order to obtain the
failure locus.
1.8
Uni-axial compression test
dev normalised
1.6
1.4
1.2
HC-B4-1
1
HC-B3-2
0.8
0.6
HC-B3-1
HC-B3-3
H-B3-2
H-B3-1
0.4
0.2
0
0
0.25
0.5
0.75
hyd normalised
Figure 8:
Failure locus.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
1.25
Acknowledgements
The financial support of the European Commission by means of the APROSYS
project and Mr. Dirk Lehmhus of Fraunhofer-Institute for Applied Materials
Research (IFAM) are gratefully acknowledged.
References
[1] Gibson, L.J, Ashby, M.F., Cellular solids: structure and properties (Second
Edition), Cambridge University Press, 1997.
[2] Collins, J.A., Failure of materials in mechanical design, John Wiley & Sons,
1980.
[3] Khan, A.S., Huang, S., Continuum theory of plasticity, John Wiley & Sons,
1995.
[4] Integrated Project on Advanced PROtection SYStems (APROSYS). On line:
www.aprosys.com.
[5] Ashby, M.F., Evans, A., Fleck, N.A., Gibson, L.J., Hutchinson, J.W.,
Wadley, H.N.G., Metal foams A Design Guide, Butterworth Heinemann,
2000.
[6] Ehlers, W., Mullerschon, H., Klar, O., On the behaviour of Aluminium
Foams under Uniaxial and Multiaxial Loading, Verlag Mit, 1999.
[7] Deshpande, V.S., Fleck, N.A., Isotropic constitutive models for metallic
foams, Journal of the Mechanics and Physics of Solids, Pergamon, 1999.
[8] Deshpande, V.S., Fleck, N.A., Multiaxial yield behaviour of polymer foams,
Acta Materialia, Pergamon, 2001.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
259
Abstract
In the automotive industry a lot of energy is put into the development of
lightweight auto body structures that are able to outperform the classic
structures. For these purposes tremendous advances have been made in the field
of multi-phase steels. Complex steel grades have been developed with
exceptional mechanical properties: they combine high strength values (yield
strength, tensile strength, etc.) with an excellent ductility. TRIP steels
(TRansformation Induced Plasticity steels) show these properties pre-eminently.
To guarantee a controlled dissipation of the energy released during a crash,
knowledge and understanding of the impact-dynamic material properties is
essential. In this paper the results are presented of an extensive experimental
program to investigate the strain rate dependent mechanical properties of
different TRIP steels. The influence of different alloying types (Al, Si, SiAl, etc.)
on the static and dynamic stress-strain behaviour is investigated. A split
Hopkinson tensile bar set-up was used for the experiments. Microstructural
observation techniques such as different optical methods, SEM and XRD were
used to reveal the mechanisms governing the observed high strain rate
behaviour. From the results it is clear that the excellent mechanical properties are
not only preserved at higher strain rates, but still improve. The influence of the
alloying elements is comparable in the static and dynamic case: aluminium tends
to increase the elongation level of the material, whereas silicon improves the
stress that is achieved.
Keywords: high strain rate behaviour, split Hopkinson tensile bar, TRIP steel,
alloying elements.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06026
Introduction
Figure 1:
SEM-micrograph showing the multiphase microstructure of a nondeformed CMnAl-TRIP steel (magnification: x 5000).
261
2.1 Materials
A key parameter for the TRIP effect is the stability of the meta-stable austenite
which is mainly determined by the austenite particle size and the composition,
especially the carbon content is an important parameter (Itami et al. [3]).
Alloying elements such as silicon, aluminium and phosphor in TRIP steels principally added to inhibit carbide precipitation during the second stage
isothermal holding temperature in the production process of TRIP steels - also
significantly influence the thermodynamic stability of the austenite phase. The
influence of the alloying elements is assessed by the study of four different TRIP
steel grades: CMnAl-, CMnSi-, CMnSiAl- and CMnSiAlP-TRIP. Specific care
was taken to keep the same carbon content for each steel grade. Table 1 lists the
composition of the low alloy TRIP steels.
Table 1:
Steel grade
CMnAl-TRIP
CMnSi-TRIP
CMnSiAl-TRIP
CMnSiAlP-TRIP
C
0.24
0.25
0.25
0.20
Mn
1.61
1.67
1.70
1.56
Al
1.54
0.78
0.69
0.29
Si
0.091
1.28
0.55
0.38
P
0.006
0.012
0.011
0.012
Silicon strengthens the ferrite phase considerably, but high Si-contents results
in an adherent FeO.SiO2 oxide layer on the sheet surface, which generates
surface defects on the hot rolled sheet and which is difficult to remove by
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2:
According to the one-dimensional wave theory and the assumption of a uniaxial and homogeneous stress and strain in the specimen, the stress, strain and
strain rate in the specimen can be written as follows (Kolsky [5]):
(t)=
EbAb
U -U
2C
t ( t ) , ( t ) = ob ib =- b
As
Ls
Ls
( ) d ,
r
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
263
Results
Figure 3:
= K n
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(2)
Figure 4:
Figure 5:
265
Figure 6:
Figure 7:
In dynamic conditions the strain rate has limited influence on the material
properties. If these dynamic properties are compared to properties after static
deformation on the other hand, an important difference can be observed. Table 2
shows the mechanical properties of the CMnAl-TRIP steel after static (strained
at a constant strain rate of 10-4 s-1) and after dynamic deformation. An important
increase in both yield and tensile strength can be noticed when comparing static
to dynamic conditions. The uniform elongation however shows limited changes,
as well as the energy dissipation until 10% deformation.
Table 2:
Static
704 s-1
1246 s-1
1898 s-1
Tensile
strength,
MPa
Uniform
elongation, -
506
622
646
672
689
786
800
814
0,265
0,264
0,301
0,252
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Energy
dissipation at
10%
deformation,
106 J/m3
55,91
59,16
61,39
64,99
267
This can be explained when looking at the X-ray diffraction (XRD) patterns
in order to reveal the mechanisms governing the complex mechanical behaviour.
In figure 8 XRD patterns of non-deformed specimens are compared with patterns
of specimens after static and dynamic tests. The measurements are performed on
a Siemens D5000 diffractometer. Only the ferrite (b.c.c.) and austenite (f.c.c.)
peaks are considered in the measurements. Diffractograms were obtained in the
19-40 2-range using a filtered molybdenum K radiation.
Figure 8:
Conclusions
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
Bleck, W., Using the TRIP effect - the dawn of a promising group of cold
formable steels. Proceedings of the International Conference on TRIPAided High Strength Ferrous Alloys, ed. B. C. De Cooman, pp. 13-24,
2002.
Davies, E. D. H. & Hunter, S. C., The dynamic compression testing of
solids by the method of the split Hopkinson pressure bar. J. Mech. Phys.
Solids, 11, pp. 155-179, 1963.
Itami, A., Takahashi, M. & Ushioda, K., Plastic Stability of Retained
Austenite in the Cold-Rolled 0.14-Percent-C-1.9-Percent-Si-1.7-PercentMn Sheet Steel. Isij International, 35(9), pp. 1121-1127, 1995.
Girault, E., Mertens, A., Jacques, P., Houbaert, Y., Verlinden, B. & Van
Humbeeck, J., Comparison of the effects of silicon and aluminium on the
tensile behaviour of multiphase TRIP-assisted steels. Scripta Materialia,
44, pp. 885-892, 2001.
Kolsky, H., An investigation of mechanical properties of materials at very
high strain rates of loading. Proc Phys Soc Lond Sec B, 62, pp. 676-700,
1949.
Stout, M. G. & Follansbee, P. S., Strain rate sensitivity, strain hardening,
and yield behavior of 304L stainless steel. Journal of Engineering
Materials and Technology - Transactions of the ASME, 108(4), pp. 344353, 1986.
Samek, L., De Cooman, B. C., Van Slycken, J., Verleysen, P. & Degrieck,
J., Physical Metallurgy of Multi-Phase Steel for Improved Passenger Car
Crash-Worthiness. Steel Research International, 75(11), pp. 716-723,
2004.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
269
Abstract
The construction of dams using Roller Compacted Concrete (RCC) is a relatively
new technology, which has rapidly developed in recent years. To increase the
safety factor in designing the various horizontal construction joints in RCC
dams, an experimental research using interlayer cement grout has been carried
out to improve the bond strength. A device with the ability of exerting the
normal pressure in the direct shear method has been designed and used. 144 tests
were carried out to assess the effect of interlayer cement grout along with
exerting the normal pressure. The tests include different proportions of water to
cement of the interlayer grout (0.5, 0.75, 1) and different values of normal
pressure (0, 5, 10, 15 kg/cm 2 ) in the ages of 7, 28 and 90 days. The results
indicate that the use of interlayer cement grout in the range of the normal
pressure exerted gives higher bond strength compared to the state of no grout. By
reducing the ratio of water to cement of the grout, the bond strength increases.
Keywords: Roller Compacted Concrete, horizontal construction joint, interlayer
cement grout, bond strength, normal pressure.
Introduction
The RCC dams are built in relatively thin layers, with a thickness of
approximately 30 centimeters placed, spread and compacted over each other;
therefore a large number of horizontal joints are formed in the dam. The quality
of the concrete at the joint surface is of great importance to ensure the integration
of layers. But in practice, this surface has a lower quality compared to the
concrete mass; and a lower tensile and shear strength and a higher permeability
is expected [1].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06027
RCC materials
2.1 Cement
The heat generation control is of great importance in selecting the cement type.
Heat generation is typically controlled by using pozzolans and slags, therefore
Portland cements including Type , and V can be used [3]. In the present
research, exports cement Type of the Isfahan cement factory was used with the
chemical analysis given in Table 1. The quantities of C 3 S, C 2 S, C 3 A, C 4 AF
and insoluble residue in this cement are respectively 46.13, 27.38, 9.61, 9.98 and
0.44%.
Table 1:
Chemical analysis of the cement used in the RCC mix design (%).
SiO 2
Al 2 O 3
Fe 2 O 3
CaO
MgO
SO 3
Na 2 O
K2 O
21.68
5.72
3.28
63.53
1.75
1.63
0.2
0.52
2.2 Pozzolans
Ninety percent of the dams built until 1998 contain some kind of additional
material such as slag or pozzolan, and in only 10 percent of the cases, the cement
has been used individually to make RCC [3]. To replace a portion of the cement
by pozzolan (25%) a natural pozzolan powder was used in the RCC mix design.
The chemical analysis is given in Table 2.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 2:
271
SiO 2
Al 2 O 3
Fe 2 O 3
CaO
MgO
SO 3
Na 2 O
K2 O
57.35
18.2
6.23
1.45
1.93
3.5
1.72
2.3 Aggregates
Considering that the aggregates fill 75 to 85% of the RCC volume [4], Their
properties are an effective factor on fresh and hardened RCC. The gravel is
crushed river aggregates with a maximum size of 25.4 millimeters in accordance
with ASTM C33, and the sand is natural sand with 4.2% of the material passing
No. 200 sieve (non-plastic) in accordance with ASTM C33. The gravel/sand
ratio in the RCC mix design is 55 to 45. The grading of the coarse and fine
aggregates is given in Table 3.
Table 3:
Coarse
Agg.
Fine
Agg.
Agg.
size (mm)
137
12.5
9.5
4.75
2.36
1.18
Passing
Percent
92.3
52.1
28.2
1.3
0.0
0.0
Agg.
size (mm)
4.75
2.36
1.18
0.6
0.3
0.15
Passing
Percent
96.5
87.1
66.9
42.8
27.7
12.2
Table 4:
Water
19.0
Tests program
The above materials, cubic RCC specimens with 15x15x15 cm dimensions were
made using the optimum mix design [5] given in Table 4. The specimens were
made in two layers and each layer was compacted for 15 seconds using the
electric vibrating hammer (ASTM C1435).
The specimens are cured and protected by common laboratory methods to
carry out shear strength tests at the ages of 7, 28 and 90 days.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
and 15 kg/cm ).
3.1 Test method for determining the bond strength
Considering that the construction joint can be referred to as an imposed crack
(failure), for determination of the RCC bond strength according to previous
researches [6], a device with a similar mechanism to the soil direct shear
apparatus was designed and used. The RCC specimen was placed in the frame
(figure1). The frame consists of a fixed plate and a moving plate. The normal
pressure is exerted according to the mechanical theory of high strength bolts and
by tightening the bolts on the moving plate. The device was then placed under
the pressure jack (with the pressure being exerted in the same direction as the
contraction joint). Finally, two sliding plates were placed tangent to the joint and
on opposite sides of it. By exerting the pressure on the jack, shear failure would
occur at the joint (figure 2).
Figure 1:
Analysis of results
273
ages of specimens). Any test where no grout is used has the least bond strength,
and where the w/c ratio is equal to 0.5 the most bond strength is obtained. By
decreasing the w/c ratio, the bond strength increases. This matter can be
observed more significantly in the diagrams of figure 4 (at the age of 28).
Figure 2:
Table 5:
Age
(days)
28
90
w/c
ratio
No grout
1
0.75
0.5
No grout
1
0.75
0.5
No grout
1
0.75
0.5
Normal
pressure
= 0
2
kg/cm
16
17.2
17.6
19.4
21.8
23.6
23.3
24
25.4
25.3
27.4
28.5
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
no grout
R2 = 0.9419
w /c of grout=1
R2 = 0.945
w /c of grout=0.75
w /c of grout=0.5
R = 0.9496
no grout
R2 = 0.9595
w /c of grout=1
w /c of grout=0.75
5
10
15
Norm al Pressure(kg/cm 2)
Figure 3:
20
w /c of grout=0.5
38
36
no grout
R2 = 0.9099
34
32
w /c of grout=1
R = 0.9147
30
w /c of grout=0.75
w /c of grout=0.5
28
26
no grout
R2 = 0.8853
24
w /c of grout=1
R2 = 0.9386
22
w /c of grout=0.75
20
0
Figure 4:
5
10
15
Norm al Pressure(kg/cm 2)
20
w /c of grout=0.5
The RCC shear strength parameters which include the interlayer cohesion (C)
and the angle of interlayer friction () have been derived for each age and each
proportion of w/c. The results are given in table 6.
4.2 Effect of specimen age on the bond strength
For all normal pressure values and proportions of w/c, on average, the 7-day
bond strength is approximately 77.7% of the 28-day bond strength. Also, the 90WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
275
day bond strength has an increase of 10.1% compared to the 28-day bond
strength (figure 6).
40
R2 = 0.9328
38
no grout
36
w /c of grout=1
34
w /c of grout=0.75
R2 = 0.8959
32
w /c of grout=0.5
R2 = 0.8868
30
28
no grout
R2 = 0.9007
26
w /c of grout=1
24
w /c of grout=0.75
22
0
Figure 5:
5
10
15
Normal Pressure(kg/cm2)
Specimen
Age (days)
7
28
90
w /c of grout=0.5
20
w/c ratio
No grout
0.5
0.75
1
No grout
0.5
0.75
1
No grout
0.5
0.75
1
Interlayer
Cohesion, kg/cm
16.44
18.51
18.28
16.87
21.32
25.25
24.16
23.39
25.06
26.98
26.46
26.41
Interlayer
friction angle
33.7
32.96
33.58
32.41
35.14
35.4
36.2
34.9
36.2
36.4
37.3
36.6
4.3 Effect of w/c ratio of interlayer grout and specimen age on shear
strength parameters
The interlayer cohesion is intact with the bond strength, resulting that the
maximum and minimum cohesion occur at w/c=0.5 and state of no grout used,
respectively (figure 7). The angle of interlayer friction has no specific trend, and
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
its maximum and minimum values occur at w/c=0.75 and w/c=1.0, respectively
(figure 8).
34
32
30
28
w /c=-
26
w /c=1
24
w /c=0.75
22
w /c=0.5
20
0
10
20
30
40
50
60
70
80
90
100
Age(days)
Figure 6:
Cohession (kg/cm2)
30
28
26
24
22
w/c=-
20
w/c=1
18
w/c=0.75
16
w/c=0.5
14
0
10
20
30
40
50
60
70
80
90
100
Age(days)
Figure 7:
Conclusions
In this research, a device with the ability of exerting the normal pressure was
designed to assess the RCC bond strength. The analysis of the overall results
obtained from the optimum mix design, different w/c ratios (0.5, 0.75, and 1) and
2
277
Friction(angle)
38
36
34
w /c=w /c=1
32
w /c=0.75
w /c=0.5
30
0
Figure 8:
10
20
30
40
50
60
Age (days )
70
80
90
100
The minimum bond strength was obtained in the tests where no grout was
used, and where the interlayer grout is used, the bond strength would decrease
whilst the w/c ratio is increased.
The variation of the interlayer cohesion including all proportions of w/c and
also the state of no grout used is intact with the variation of the bond strength,
whereas the angle of interlayer friction does not follow any special pattern.
References
[1]
[2]
[3]
[4]
[5]
[6]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
279
Abstract
High performance materials are rarely used as 100% compound, therefore most
of the time when reference is made to these categories of materials it is as a
matter of fact a reference to a system comprising a well-combined ingredients
exhibiting together high performances in use. Fibres, micro-fibres and now nanofibres are definitely part of the high performance reinforcing ingredients, which
in a system largely contributes to the performance in use of the designed system.
High performance fibres, and more specifically aramids for their outstanding
heat and mechanical properties, although most often used as reinforcing
compound, are also used more uniquely and directly in situations where
mechanical and thermal protections are required. It is therefore useful as a
preliminary grounding to review these types of materials in terms of their
properties. Then their attributes and potential contribution alone or in a system
become more specific and can be extrapolated more easily. The mechanical
performance and more specifically the cut resistance of articles made of high
performance materials are generally measured against several norm
specifications. Those are not designed to predict the real in use performance of
the considered protective articles. This gap is even larger when the cut resistance
of the compounding fibre increases. Although relationships exist between the
various cut performance levels obtained with the ASTM, ISO and EN norms
(respectively American, International and European standards), there is a need
for a more fundamental understanding and interpretation of the data. The ASTM
and ISO approaches remain the most suitable methods for that purpose. Various
parameters, such as the high performance material and fibre composition, the
effect of kinetic such as the cutting speed and the speed profile, the force
distributions, the artefacts generated by various surface coatings are presented
and discussed aiming at improving the interpretations of the data generated
through the norm procedures. Therewith, significant progress in the
understanding of the cut fundamentals and the norm harmonization are made
available to the users of high performance fibres, such as KEVLAR.
Keywords: high performance materials and fibres, textile structures, coated
textile materials, mechanical protection, cut mechanism, cut fundamentals,
KEVLAR.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06028
Introduction
281
Nomex
The discovery in 1965 of high tenacity, high modulus fibres from liquid
crystalline solutions of synthetic para-aromatic polyamides led to the commercial
production of KEVLAR aramid fibre by DuPont Co in 1971[9]; the
corresponding chemical formula is given below:
KEVLAR
283
organic fibres and fibre elongation is lower. Aramid fibres can be woven on
fabric looms more easily than brittle fibres such as glass, carbon or ceramic.
They also exhibit inherent resistance to organic solvents, fuels, lubricants and
exposure to flame.
The superimposed structures, such as the crystallites, the fibrils and the skincore boundaries, are definitely unique attributes, which can be partially tailored
through the fibre process engineering.
2.4 Mechanical properties
Typical stress-strain curves of different KEVLAR fibres are provided in
fig. 2(a), which clearly outlines the modulus increase from KEVLAR 29 to
KEVLAR 149. There are differences between KEVLAR 49 and KEVLAR
29. The respective moduli, brought by various spinning conditions and posttreatments performed on para-aramid precursors are generally considered as
intermediate between those of graphite, boron and glass fibres. The linear stressstrain behaviour of para-aramids is special compared to most man-made fibres
which tensile behaviours are depicted in fig. 2(b).
Figure 2:
Creep is measured either by the length variation under tension or by the stress
decrease at constant gage length. Para-aramids, which exhibit little creep, differ
significantly from other highly oriented polymeric fibres, such as HMPE fibres,
which can break after several days under intermediate load due to their high
creep properties associated with a stress slip of molecules compared to a
structure tightening in the case of para-aramids. The temperature, the load
relative to the fibre ultimate strength, the water content and other parameters
affect creep.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
285
Figure 3:
287
The analysis of the influence of PVC dots, melted and cured onto the knitted
surface to enhance the grip properties of the fabric or the gloves, was performed
using the EN and the ASTM procedures. This further illustrates some of the
specific features of the two methods. Fig. 5 provides the resulting data outlining
NL1, which is the load, in g, provoking the cut through of the specimen for a
blade motion of 25.4 mm.
The motion of the blade on the sample tested has been studied in terms of
speed using the ASTM and ISO blade calibration material, i.e. a 1.6 mm
neoprene layer, and 100% para-aramid felt, which characteristics are indicated in
fig. 6(a) and 6(b).
As depicted in fig. 7 a load cell, linked to a computer, was placed in the
balanced arm holding the blade in order to estimate the force opposing the blade
motion during the ASTM testing. Therewith the ratio of this resisting force to the
applied load force has been calculated for various materials, such as neoprene,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
289
100% KEVLAR knitted gloves and another HCP material. Results are
summarized in fig. 8
Figure 5:
PVC dots influence on the EN (top table) and the ASTM results
(top table and curves).
Figure 6(a): Blade speed influence on the cut performance of the calibration
material.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 6(b): Blade speed influence on the cut performance of a 100% paraaramid textile.
Figure 7:
291
more easily accessible sliding and friction forces. It is not our intention to
pretend that a direct liaison can be established between these two physical
phenomena but, in view of the lack of published data on the subject matter, it is
valuable to bring another piece of comprehension to this scientific matter of
relevant value to personal protection. During an accidental cut the sliding
attributes of the personal protective equipment are determinant.
Figure 8:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 10:
a reasonable puncture resistance towards sharp but not too thin perforators
the amount of material per unit volume, which the cutting threat has to go
through
the deformability of this material and the tendency for this material to
maintain a high density under pressure still keeping its flexibility
the blade dulling effect of the material being cut. Glass is a good example.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
293
the eventual surface or bulk phase changes occurring during the cut either
due to local heating under friction or heating.
.
the water or liquid retention and absorbency of the protective barrier.
Fibres with skin-core structures are generally more protective. The thickness
of the skin is a positive attribute. Para-aramid fibres are cut resistant.
Rather brittle materials tend to be less effective. On the other hand ductility can
be a good attribute. Steel and glass tend to support these two statements.
As exhibited in fig. 9 the design of blade profiles, leaving apart the steel
selection and processing, remains an art. For example what are the criteria taken
into account to make an effective knife cutting edge to cut tomatoes versus bread
versus meat versus cheese, versus wood, versus textile, etc.
When one observes a few cutting edges, it seems, as depicted in fig. 10, that
the sequence of -1- material blockage-densification and compression followed by
-2- the effective sliding and cut of the previously compressed materials and -3the relaxation and separation of the already cut material is a sequence which
may be taken into account to better understand the selection of a large proportion
of cutting edge profiles. That densification-cut-relaxation sequence repeats itself
along the blade as the cutting operation proceeds. The threshold compression
ratio at which cut can progressively take place depends on the fibre mobility, the
deformation/cut speed, the local humidity as well.
3.3.1 Blade degradation during EN testing
Using the EN norm, the degradation effect of the materials on the blade cutting
edge was already demonstrated by Lara [21] with a series of test performed on
fibreglass reinforced glove samples. According to this study, a fraction of cycle
a cycle being represented by one blade full rotation counter-clockwise and its
reverse rotation clockwise - was initially sufficient to cut the cotton calibration
fabric. The same blade, after running for 50 revolution cycles on the HCP glasscontaining material, then cut the reference cotton fabric after 25 cycles compared
to 0.8 cycles initially. This shows that less than a cycle was necessary to dull the
blade.
Glass fibres are known for their dulling effect. A major concern shall be
outlined regarding products, which have a dulling effect. Shall they be
considered as effective high cut performance materials?
A partial answer is established by the fact that the ASTM/ISO norms are
more and more used for these types of materials. These norms being less
sensitive to the dulling effect since one spot of the blade is in contact with the
sample to cut only once. The EN norm recent revisions suggest the use of the
ISO testing in such cases. This is a significant step and improvement.
3.3.2 Speed effect on cut resistance
In which proportion the blade speed can affect one material more than another.
For example figs. 6(a), 6(b) compares the speed effect, on a neoprene rubber
layer and a KEVLAR sample stamped out of a knitted glove, for a speed
variation ranging from 14 to 28 mm/s.
Such a variation affects much more the neoprene load-to-blade displacement
curve than the load-to-blade displacement relationship of the para-aramid
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
295
blade orientation and displacement, and force application point. This force is
generated by weights placed on a plate mounted on a lever arm assembly in a
way that doubles the resulting force applied to the specimen. The test apparatus
also consists of a motor-driven balanced arm, which holds the blade.
In order to measure the co-linear force component resisting to the
displacement of the blade, a 50 N (or 200N) load cell, type S2 from HBM
Germany, was installed in the arm holding the blade as per the diagram of fig. 7.
This load cell was connected to a computer interface, module MP55 from HBM
Germany. The resisting force, F2 , was recorded during the cutting process
using the CATMAN software from HBM Germany. The normal force, F1 ,
applied onto the blade edge is assumed to be directly proportional to the double
weight force value, which is an approximation since during the displacement,
and due to the thickness of the sample, the blade cutting edge does not remain
strictly perpendicular to the sample contacting point. This is depicted in fig. 7.
For the neoprene calibration material the resisting force, F2 , is about 2.2
Figure 11:
Figure 12:
Figure 12 outlines the similitude of the forces involved in the case of cutting
versus the ones involved in the case of sliding. An amazing and revealing level
of concordance can be observed, which means that one could definitely relates,
to a certain extend, the surface wear and friction with the cut mechanism, which
is by far more complex. This goes beyond the scope of this study, although
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
297
Conclusion
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgements
The authors wish to express their gratitude to Andr Miret-Casas for his
meticulous technical assistance and enthusiasm. Portia Yarborough, chairman of
the F23 ASTM committee brought valuable comments in her review. DuPont
fibre production centre in UK was instrumental to the independent evaluation
and analysis of the materials tested.
Special thanks are also due to Andr Courgey from the European Automotive
and Transportation Institute in France; his contribution to the wear and friction
study is especially recognised.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
299
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
301
Abstract
Experimental methods designed to study the bulk of materials do not necessarily
detect the changes taking place at the surface of the crystallites. For example,
divalent tin-containing materials appear to be stable at ambient conditions in air,
provided they are not hygroscopic, X-ray powder diffraction shows only the
peaks of the expected tin(II) phase. However, we have observed that the
Mssbauer spectrum of polycrystalline samples contain, in addition to the
expected tin(II) peak(s), a small peak at 0 mm s relative to CaSnO3 at ambient
conditions, that can be attributed only to tin(IV) coordinated by oxygen. A
detailed study of this phenomenon has shown that Mssbauer spectroscopy is
quite sensitive for detecting thin layers of oxide at the surface of crystallites of
tin(II). This phenomenon has been exploited for the study of spontaneous
oxidation of various tin(II) fluoride and chloride-containing materials, some of
these fluorides being the highest performance fluoride-ion conductors known to
date. It was observed that passivation is quite efficient in the fluorides, and in
the chloride fluorides that have all their tin(II) covalently bonded. On the other
hand, the materials containing a mixture of covalently bonded tin(II) and the Sn2+
stannous ion namely, the Ba1-xSnxCl1+yF1-y solid solution, show a higher rate of
oxidation, which is highly dependent on the method of preparation and the
composition parameters, x and y.
Keywords: passivation, oxidation, disordered phases, fluorite-type structure,
Mssbauer spectroscopy, X-ray diffraction, ionic conductivity, BaClF.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06029
Introduction
The MF2 fluorides of large divalent metals (M = Ca, Sr, Ba & Pb) have the well
known fluorite type structure, with a cubic unit-cell, space group Fm3m, and a
cubic coordination of the metal ion (fig. 1) [1]. This contrasts with SnF2 tin(II)
fluoride (stannous fluoride) that has a molecular structure; it is made of Sn4F8
tetramers and the Sn-F bonds are strongly covalent [2]. When SnF2 and MF2 are
combined together, some new materials are produced, some of which are
ordered, others disordered. The MSnF4 compounds have a M M Sn Sn
order parallel to the c axis of the unit-cell (fig. 1) [3]. Tin and lead belong to
group 14, and therefore they have four valence electrons, and in the +2
suboxidation state, the 5s2 (Sn) or 6s2 (Pb) electrons are unused and form a nonbonding electron pair, also called a lone pair. When bonding is ionic, the
orbitals are not hybridized and the lone pair is located on the ns native orbital
that has spherical symmetry, unless it is distorted by polarization, and a quite
regular coordination is observed, and the lone pair is said to be non-stereoactive
since it does not modify the coordination. In fluorite-type PbF2, lead has a
cubic coordination, therefore its lone pair is located on the native 6s orbital and
bonding is ionic, like in BaF2 (fig. 1). In tetragonal MSnF4, the coordination of
M is a distortion of the MF8 cube of the fluorite structure, with in addition, four
longer interactions, to form an overall MF4F4F4 where the bond length
increases as follows: M-F (from M-F-M bridges) < M-F (from M-Feq-Sn
bridges) < M-F (from M-Fax-Sn bridges), where Feq and Fax form equatorial (4)
and axial (1) bonds with tin, respectively (fig. 1). For the sake of clarity, the BaF bonds are not shown on figure 1. The lead coordination in PbF2, is still
ionic since Sr2+ and Ba2+ take the same coordination in SrSnF4 and in BaSnF4,
respectively. In contrast, tin forms covalent bonding in MSnF4, with a very short
axial Sn-F bond, and the lone pair is located on a hybrid orbital, in transposition to F (fig. 1). The tin lone pair is said to be stereoactive since it changes
drastically the tin stereochemistry (lower coordination number, highly distorted
coordination). A consequence of the M/Sn order and of the stereoactivity of the
tin lone pair is that the lone pairs cluster in sheets that are cleavage planes, which
make the structure very highly two-dimensional. Examples of disordered
structures are PbSn4F10, the M1-xSnxF2 solid solution (M = Ca and Pb), and MSnF4 (M = Ba and Pb) obtained by ball-milling. The disordered phases have
an Fm3m cubic unit-cell like the fluorite-type MF2, and their diffraction pattern
shows no peak splitting or additional low angle peak that would indicate the
presence of a lattice distortion or a superstructure. Therefore, Sn and the other
metal M are fully disordered on the metal ion site of the MF2 structure (fig. 2).
The M bonding is ionic and the coordination is cubic with some disordered local
distortions in the neighborhood of tin, while tin bind to only one of the faces of
the F8 cubes, with its stereoactive lone pair pointing towards the center of the
cube, and thus its bonding is covalent. BaClF crystallizes in the PbClF structure
(space group P4/nmm). Its structure is also a tetragonal distortion of the fluoritetype BaF2, with layers of F and Cl ordered parallel to the c axis of the unit-cell,
making it a layered structure, in contrast with the three-dimensional cubic
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
303
structure of BaF2 (fig. 1). We have recently found two methods for substituting
substantial amounts of Ba by Sn in BaClF, and also some F by Cl (y>0), or some
Cl by F (y<0), making it a doubly disordered Ba1xSnxCl1+yF1-y solid solution. Xray diffraction shows that Ba and Sn are fully disordered on the Ba site of the
BaClF structure. In addition, while the two anionic sites remain ordered, for y>0
the y Cl in excess are disordered with the remaining (1y) F on the F site,
whereas for y<0, the (y) excess F are disordered with the remaining (1+y) Cl on
the Cl site. However, the electronic situation at tin cannot be assessed by X-ray
diffraction. All the tin(II)-containing fluorides studied here are high performance
fluoride ion conductors (about three orders of magnitude higher than the
conductivity of the corresponding MF2).
Figure 1:
All these materials appear stable in air at ambient conditions, and X-ray
diffraction shows no sign of oxidation to tin(IV). The present study shows that
tin-119 Mssbauer spectroscopy is an excellent method for detecting minute
amounts of tin(IV) oxide in tin(II) compounds, and that all materials show at the
minimum some surface oxidation, and some passivate very well while other do
so less efficiently.
Experimental
Fluoride materials synthesis was carried out either in aqueous medium or at high
temperature under dry conditions. Tetragonal -PbSnF4 was obtained by
precipitation on addition of an aqueous solution of lead(II) nitrate to a solution of
SnF2. Ca1-xSnxF2 precipitated when a solution of calcium nitrate was added to a
solution of SnF2 at high Ca/Sn ratio, or by leaching, on stirring, CaSn2F6 in a
large amount of water. The synthesis of Pb1-xSnxF2 and BaSnF4 was carried out
under nitrogen in sealed copper tubes heated at 500 oC and quenched. Barium
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2:
BaF2 + xSnF2 +
1+ y
2
(1)
305
Transmission
1
0 .9 6
0 .9 2
(a )
0 .8 8
Transmission
0 .9 9 5
0 .9 7 5
0 .9 5 5
(b )
0 .9 3 5
-8
-6
-4
-2
Ve l o c i t y ( m m /s )
Figure 3:
(A)
Figure 4:
(B)
(C)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
307
Transmission
(a)
0.93
0.86
Transmission
0.97
(b)
0.94
0.91
0.88
0.85
-8
-6
-4
-2
Ve locity (mm/s)
Figure 5:
Transmission
0.99
0.98
0.97
0.96
0.95
(a)
(d)
0.94
0.93
-8
-6
-4
-2
Velocity (mm/s)
Transmission
0 .9 8 5
0 .9 6 5
( A M - 2 3 4 , X = 0 .8 6 0 , a g e d f o r 4 8 .0 m )
0 .9 4 5
(b)
0 .9 2 5
-8
-6
-4
-2
V e l o c i t y i n ( m m /s )
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
309
tin(IV) peak is weaker than the tin(II) peak, however, it is stronger than that of
stoichiometric chloride fluorides.
(AM-1180, x = 0.05, y = 0.10)
0.999
0.994
Transmission
Transmission
0.995
0.99
0.985
0.984
(a1)
(a2)
0.98
0.979
0.99
0.99
0.98
0.98
Transmission
Transmission
0.989
0.97
0.96
(b1)
0.95
0.97
0.96
(b2)
0.95
0.94
0.94
Transmission
Transmission
0.995
0.988
(c 1)
0.99
(c 2)
0.985
0.98
0.978
-8
-6
-4
-2
-8
-6
-4
Velocity (mm/s)
Figure 7:
119
-2
Velocity (mm/s)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusion
Acknowledgements
This work was made possible by the support of Concordia University and the
Natural Science and Engineering Research Council of Canada. Grateful thanks
are also due to the Procter and Gamble Co. (Mason, Ohio) for supporting our
Mssbauer laboratory.
References
[1]
[2]
[3]
311
Abstract
A doubly disordered Ba1-xSnxCl1+yF1-y solid solution has been prepared. X-ray
diffraction shows that Ba and Sn are fully disordered on the Ba site of the BaClF
structure. In addition, while the two anionic sites remain ordered, for y>0 the y
Cl in excess are disordered with the remaining (1-y) F on the F site, whereas for
y<0, the (-y) excess F are disordered with the remaining (1+y) Cl on the Cl site.
119
Sn Mssbauer spectroscopy showed the Ba1-xSnxCl1+yF1-y solid solution is a
truly unique material. Indeed, in addition to the rarity of being a doubly
disordered solid solution over a wide range of chemical composition, it also
exhibits a variable mixture of ionic tin(II) (Sn2+ stannous ions, with a
non-stereoactive lone pair) and covalently bonded tin(II) (with a stereoactive
lone pair). A new material, Ba2SnCl6, has been prepared and its crystal
structure solved. It contains octahedrally coordinated Sn2+, with a perfectly
non-stereoactive lone pair, for the first time in a chloride. This makes the lone
pair a potential charge carrier.
Keywords: disordered phases, BaClF structure, Mssbauer spectroscopy, X-ray
diffraction, mixed conductivity.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06030
Introduction
Figure 1:
Experimental
The BaClF phase precipitate was obtained when aqueous solutions of barium
chloride and tin(II) fluoride were mixed, at a X = BaCl2/(SnF2 + BaCl2) molar
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
313
ratio above 0.85, when barium chloride was added to tin(II) fluoride (BaSn),
or the other way around (SnBa). Synthesis under dry conditions was
performed when mixtures calculated according to eqn (1) were heated for three
days under nitrogen in sealed copper tubes at temperatures varying from 350 to
800 oC.
1 2x y
2
BaF2 + xSnF2 +
1+ y
2
(1)
The X-ray powder diffraction pattern of the precipitate (fig. 2b) has the same set
of peaks as BaClF (fig. 2c), at the same positions, and therefore, it seems that the
product obtained is simply BaClF. This seemed confirmed by Search-Match
that resulted in BaClF being the only reasonable match. Although the match was
reasonably good (Similarity Index SI>150 and all 15 peaks matched), a close
examination of the peak position revealed a small, but significant shift, shown by
arrows on figure 3a, relative to unsubstituted BaClF (fig. 3b). The shift of all
peaks is consistent with a 0.91% decrease of the a parameter and a 0.97%
increase of c, while the unit-cell volume decreases by 0.86% and the c/a
tetragonal distortion increases by 1.89%. The change of unit-cell parameters
suggests a change of chemical composition. A logical change was the
substitution of a fraction of Ba by Sn, to give non-stoichiometric Ba1-xSnxClF.
Chemical analysis was performed to determine the tin molar fraction x, and it
showed that the compound is also non-stoichiometric for the anions, such that
the formula is best written as Ba1-xSnxCl1+yF1-y (x = 0.15, y = 0.11), to show it is
a substituted BaClF phase, as shown by X-ray diffraction.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(a)
220
212
104
211
202
103
003
112
200
102
110
101
Intensity
001
002
Intensity
(b)
(c)
5
10
15
20
25
30
35
40
45
50
55
60
Deg. 2 theta
Figure 2:
Preparations carried out by the dry method yielded a material that also gives
the BaClF diffraction pattern (fig. 2a). A wide double non-stoichiometric solid
solution was obtained: 0x0.25, -0.15y0.25. The inversion of the relative
intensities of the (002) and (110) peaks could be in part due to the change of
chemical composition. It is likely that it is mostly due to preferred orientation
favoring the (00l) peaks in the latter, i.e. the material tends to cleave parallel to
(a,b). This is consistent with the crystal structure (fig. 1): the long Ba-Cl bonds
(Cl atoms at the corners of a square) are the weak links. In addition, figure 3
shows that the peak shift in the solid solution prepared by the dry method is
negligible, and it was found it changes very little with the composition
parameters x and y. The different unit-cell parameters of the solid solution
prepared by the two methods shows that something differentiates them from one
another, although both are doubly disordered BaClF structures. For these
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
315
110
110
101
(c)
(a)
20
002
101
Intensity
(b)
110
002
101
Intensity
002
25
30
Deg. 2 theta
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(b)
Figure 4:
317
electric field gradient (efg) at the Sn2+ site. It results no splitting of the nuclear
levels, and therefore only one nuclear transition takes place (1/2 3/2), and a
single Mssbauer absorption line is observed. This contrasts with the case of
covalent bonding, where the tin lone pair is located on one of the hybrid orbitals
(stereoactive lone pair). This lone pair is highly axial and therefore generates a
large efg, (Vzz)val. In addition, the stereoactive lone pair distorts very strongly
the tin polyhedron of coordination, which generates another term of the efg,
(Vzz)latt. The overall effect is the splitting of the first excited state which results
in two nuclear transitions, and the Mssbauer spectrum is a doublet, with a large
quadrupole splitting (fig. 4b). In addition, the higher the valence s electron
density acting at tin, the higher the isomer shift , i.e. the line shift relative to a
CaSnO3 standard. As can be seen from figure 4, this makes the isomer shift of
Sn2+, 4.0-4.1 mm/s (5s electron density concentrated around the tin atom) higher
than that for Sn(II)cov, 2.5-3.5 mm/s (5s electron density pulled away in hybrid
orbitals).
The ambient temperature spectrum of precipitated Ba1-xSnxCl1+yF1-y is a
single line at 4.07 mm/s (fig. 5a1) which shows that tin is present in the form of
Sn2+ stannous ions. For the Ba1-xSnxCl1+yF1-y solid solution prepared in dry
conditions, the Mssbauer spectrum is a function of the x and y composition
parameters. For a high tin content (x = 0.225), the spectrum is also a single line
at ca. 4 mm/s if Cl/F>1 (y = 0.25), and therefore bonding is ionic (fig. 5b1). For
the same x value and Cl/F<1 (y = -0.15), the tin(II) region of the spectrum is a
very highly asymmetric doublet, which is actually due to the sum of a doublet
and a single line. Since for fluorides, the high velocity line of a doublet is at
about 4 mm/s, it adds to the single line of ionic tin to makes a spectrum that
looks like a strongly asymmetric doublet. Therefore, at high tin content, and
when there is more fluorine than chlorine, the Ba1-xSnxCl1+yF1-y solid solution
contains a mixture of Sn2+ ions and covalently bonded tin(II). The line at ca. 0
mm/s in the spectra of figures 5c1 and 5c2 is due to a tin(IV) impurity generated
by surface oxidation of the particles. This is common in tin(II) compounds. The
ambient temperature spectra have a highly unequal intensity, and strangely
enough, the poorest in tin is the strongest (fig. 5a1). In addition, the spectrum of
figure 5b1 is much weaker than that of figure 5c1, even though they contain
exactly the same amount of tin. The line intensity is a function of the recoilless
fraction fa, which is itself a function of bond strength.
All the Mssbauer observations make it possible to build a model of tin
bonding in Ba1-xSnxCl1+yF1-y solid solution: (i) at low tin concentration (small x),
like in the precipitated solid solution, due to dilution tin is isolated from other tin
atoms, the solid network is determined by the size of the Ba2+, Cl- and F- ions,
and since tin occupies a barium site, it takes the same mode of bonding, i.e. ionic
(figs. 6a); (ii) at high tin content (x = 0.25), the tin concentration is high enough
to have some tin clustering, and in an excess of fluorine (y = -0.15)each tin atom
is connected by bridging fluorine, and this result in covalently bonded tin with a
stereoactive lone pair (fig. 6b). The sample of figure 5c1 contains a mixture of
the two models of figures 6a and 6b. However, at the same tin concentration (x
= 0.25), in the presence of an excess of chlorine (y = 0.25), the excess Cl is likely
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
0.93
0.98
Transmission
Transmission
0.98
X= 0.896,
Ba->Sn, T = RT
0.88
X= 0.896,
Ba->Sn, T = 17K
(a)
(a1)
(a
)
0.94
(a(b)
2)
(a
0.83
0.78
-8
-6
-4
-2
Velocity (mm/s)
Transmission
0.96
T r a n s m is s io n
0.995
0.99
(b1)
(x = 0.225,
0.225, y = 0.25, T
x=
y = 0.250, T = RT
(b1)
0.92
x= 0.225,
(x
y == 0.225,
0.250,y = 0.25, T
= 17 K)
T = 17K
0.88
0.84
0.8
-8
0.985
(b(b22))
-6
-4
-2
Velocity (mm/s)
0.995
T r a n sm issio n
0 .9 5
0.99
0.985
0 .9
x =x=
0.225,
y = - 0.15, T = RT
0.225,
y = -0.150,
T = RT
0.98
0.975
(c1)
(x =x=
0.225,
y = - 0.15, T= 17
0.225,
K)
y = -0.150,
T = RT
0 .8 5
(c1)
(c2)
(c2)
0 .8
0.97
-8
-6
-4
-2
-8
-6
-4
Velocity (mm/s)
Figure 5:
119
-2
V e lo c it y ( mm/ s )
Sn Mssbauer spectrum at 293 K (a1, b1, c1) and 17 K (a2, b2, c2)
of samples: (a1 & a2) AM-1273 (precipitated at ambient conditions,
X = 0.896, BaSn), (b1 & b2) AM-1270 (dry method, x = 0.225, y
= 0.250), (c1 & c2) AM-1148 (dry method, x = 0.225, y = -0.150).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
319
(a)
(b)
Figure 6:
Figure 7:
Conclusion
Acknowledgements
This work was made possible by the support of Concordia University, the
Natural Science and Engineering Research Council of Canada and the universit
Mentouri. Grateful thanks are also due to the Procter and Gamble Co. (Mason,
Ohio) for supporting our Mssbauer laboratory.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
321
References
[1]
[2]
Flahaut, J., Les structures de type PbFCl (EOI) et type anti-Fe2As (C38)
des composs ternaires deux anions MXY. J. Solid State Chem., 9, pp.
124-131, 1974.
Muntasar, A., Preparation, characterization and properties of novel
materials in the BaCl2/SnF2 system, PhD thesis, Concodia University,
Montreal, pp. 39-89, 2002.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
323
Abstract
Energy dissipation and fatigue properties of nano-layered thin films are less well
studied than bulk properties. Existing experimental methods for studying energy
dissipation properties, typically using magnetic interaction as a driving force at
different frequencies and a laser-based deformation measurement system, are
difficult to apply to two-dimensional materials. We propose a novel experimental
method to perform dynamic testing on thin-film materials by driving a cantilever
specimen at its fixed end with a bimorph piezoelectric actuator and monitoring
the displacements of the specimen and the actuator with a fibre-optic system.
Upon vibration, the specimen is greatly affected by its inertia, and behaves as a
cantilever beam under base excitation in translation. At resonance, this method
resembles the vibrating reed method conventionally used in the viscoelasticity
community. The loss tangent is obtained from both the width of a resonance
peak and a free-decay process. As for fatigue measurement, we implement a
control algorithm into LabView to maintain maximum displacement of the
specimen during the course of the experiment. The fatigue S-N curves are
obtained.
Keywords: nanolayered thin films, fatigue, loss tangent.
Introduction
Figure 1:
Experimental
325
Agilent Technologies, Inc., Loveland, Colorado, U.S.A.). The free length of the
piezo actuator was 10mm. The multilayer specimens were mounted onto the top
of the piezo actuator with a set of aluminium clamp. The weight of the
aluminium clamp was 235.4 mg. The deflection of the specimen and aluminium
clamp was monitored by a fibre-optic measurement system (MTI-2000
FotonicTM sensor, MTI Instruments, Inc., Albany, NY, U.S.A.) with the probes
labelled Fibre 1 and Fibre 2 in the figure. The location of Fibre 1 was chosen to
be close to the fixed end of the specimen to avoid signal overload due to large
deflection. Lock-in amplifiers (Model SR830, Stanford Research Systems, Inc.,
Sunnyvale, CA, U.S.A.) were connected with the fibre-optic probes to reduce
noise in the displacement signals. Experiments were performed under moderate
vacuum conditions typically about 50 mtorr to reduce air damping and achieve a
reasonably high Q factor. Since our primary interest is to observe the initiation of
fatigue cracks, not crack propagation, we define fatigue failure when a 10 Hertz
reduction in specimens resonant frequency is observed.
Cu/Nb multilayers, with individual layer thickness of 40 nm and consisting
of 500 bilayers of Cu and Nb with a total thickness of 40 m, were prepared by
magnetron sputtering deposition on silicon wafers that were cut about a half way
through thickness by a wafer saw into a pattern of 2 mm long and 1 mm (or 0.5
mm) wide rectangular strips on the wafer surface. A typical cross-sectional TEM
image of the Cu/Nb multi-layers is shown in Figure 2 along with selected area
diffraction pattern. Note the polycrystalline structure of the layer with in-plane
grain size on the order of the layer thickness. The multi-layer exhibited a strong
Kurdjumov-Sachs orientation relationship: {111}Cu // {110} Nb, and <110>Cu
// <111>Nb. The parallel {111}Cu and{110} Nb planes formed the interface.
Figure 2:
1 l 4 m 4 n 4 2m 2 n 2 2l 2 n 2 2m 2 l 2
=
+
+
+
+
+
,
F1
F2
F3
E E1 E2 E3
(1)
where
2
1 C22C33 C23
1 C11C 33 C132 1 C11C22 C122
=
,
=
,
=
,
E1
E2
E3
2 2 (C12C13 C11C23 ) 1
=
+
C44 ,
F1
2 2 (C13C23 C33C12 ) 1
=
+
F2
C55 ,
327
actuator before mounting the aluminium clamp and fatigue testing. Fatigue
damage such as extrusions and intrusions are not observed around the crack on
the surface where maximum stress was exerted. A SEM image of annealed Cu
fatigued by the resonant frequency device is shown in Figure 4 (b). This sample
was fatigue tested at a stress after several thousand cycles. It can be seen that two
cracks were initiated at the edges of specimen, and propagated inwards. The
bright regions along the cracks are slip-related intrusions and extrusions.
160
4800
140
Resonant frequency, Hz
Directional E, GPa
Cu in (111)
120
Nb in (110)
100
4000
80
3200
2400
1600
800
<100>
60
-100
Figure 3:
-50
0
, deg
50
100
(a)
10
(b)
(a)
Figure 4:
4
6
Length, mm
(b)
The S-N curves of several materials, quenched and tempered (550F) 4340
steel, the 40 nm Cu/Nb multilayers, mild steel, 6061-T6 and annealed Cu, are
shown in Figure 5. The nanolayered Cu/Nb exhibits a fatigue limit at about 450
MPa. It can be seen that the high fatigue endurance of the 40 nm Cu/Nb
multilayers correlates well with its high tensile strength of around 1400 MPa. We
obtain the ratio of fatigue endurance limit to ultimate tensile strength to be about
0.35 for the 40 nm Cu/Nb multilayers [9].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
4340 Steel
Quenched and Tempered (550F)
1000
40 nm CuNb
multilayers
Al 6061-T6
100
Annealed Cu
100
10
10
10
10
10
10
Figure 5:
tan =
0 3
(2)
Here 0 is the resonant frequency of the specimen and the frequency span at
half maximum intensity. The other approach is to extract loss tangent from a free
decay process, as follows.
tan =
1
A
ln 0 .
k Ak
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(3)
329
Cycled
-3
15
0
-2
-4
L=11.3 mm
-6
-5
380
Figure 6:
Virgin
20
Amplitude, volts
25
-8
385
390
Frequency, Hz
395
400
tan = 3.3 x 10 -3
50
100
150
(a)
200
Time, ms
250
300
(b)
Conclusions
The resonant frequency method was constructed and shown to be efficient for
mechanical fatigue and energy-dissipation investigations of self-supported thinfilm materials. Cu/Nb multilayers with 40 nm individual layer thickness exhibit
over an order of magnitude greater fatigue strength compared to bulk Cu. No slip
band intrusions and extrusions were observed near the fracture surface. The ratio
of the fatigue endurance limit to ultimate tensile strength was around 0.35,
consistent with similar empirical scaling observed in other materials. Loss
tangent of the nanoscale films was about 4 x 10-3. However, potential parasitic
damping from the cantilever supports requires further study.
Acknowledgements
This research was supported, in part, by the Department of Energy, Office of
Science, Office of Basic Energy Sciences. YCW acknowledges support from the
LANL Directors postdoctoral fellowship.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
Clemens, B.M., Kung, H., & Barnett, S.A. (1999). Structure and strength
of multilayers. MRS Bulletin 24(2), 20-26.
Misra, A., Zhang, X., Hammon, D., & Hoagland, R.G. (2005b). Work
hardening in rolled nanolayered metallic composites. Acta Materialia 53,
221-226.
Frost, N.E., Marsh, K.I., & Pook, L.P. (1974). Metal Fatigue. Clarendon
Press, Oxford, UK.
Suresh, S. (1998). Fatigue of Materials. Cambridge University Press,
Cambridge, UK.
Bathias, C., & Paris, P.C. (2005). Gigacycle fatigue in mechanical
practice. Marcel Dekker, New York.
Wang, Y. C., Hoechbauer, T., Swadener, J. G., Misra, A., Hoagland, R.
G., Nastasi, M., Mechanical fatigue measurement via a vibrating
cantilever beam for self-supported thin solid films, submitted (2005).
Love, A.E.H. (1944). A Treatise on the Mathematical Theory of
Elasticity. Dover Publications, New York.
Hirth, J. P. and Lothe, J. (1982). Theory of Dislocations, Wiley, New
York.
Wang, Y. C., Misra, A. and Hoagland, R. G., Fatigue properties of
nanoscale Cu/Nb multilayers, submitted (2005).
Lakes, R.S. (2004). Viscoelastic measurement techniques. Review of
Scientific Instruments 75(4), 797-810.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
331
Abstract
In this paper, orientation maps of consecutive serial sections are collected in an
automated manner by a Focused Ion BeamScanning Electron Microscope
(FIB-SEM) outfitted with an EBSD system. Micro-Imager, a program developed
in this work, uses the 2D EBSD maps to define microstructural features such as
grains and grain boundaries. Parameters used to characterize microstructure are
also calculated by Micro-Imager for every section. The statistical measurements
of each section are compared to assess variability in the microstructure. The 2D
sections are reconstructed into a volume by Micro-Imager3D, another program
developed in this work. Statistics analogous to those measured in 2D are
calculated and compared to the expected distributions predicted by the 2D
measurements coupled with stereology. As a result, quantitative descriptions of
microstructure are made and improvements over conventional methods are
yielded. Information about individual constituents allows correlations between
distributions to be derived. The correlations drawn allow models to account for
aspects of microstructure that have classically been overlooked. The resulting
3D grain structure serves as a realistic model microstructure.
Keywords: quantitative characterization, equivalent representation, modelling.
Introduction
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
333
Figure 1:
Statistical analysis
(A)
(B)
Figure 2:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
335
(A)
Figure 3:
(B)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 4:
337
Figure 5:
Figure 6:
Plot showing the values for average grain area and maximum grain
area for each slice.
Figure 6 tracks the average grain area and the maximum grain area on every
slice. One can observe that the average grain area appears essentially constant
for every slice, but there is noticeable variation in the maximum. The data
shown in this section further illustrates the potential need for more detailed
microstructural analysis than simply the average of a parameter. Often, the
average value for a particular parameter appears constant from slice-to-slice, but
can vary markedly in the extreme values. If only the average value of a feature
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 7:
Figure 8:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
339
Figure 9:
5 Conclusion
Material characterization is a key in the prediction of material function. The
ability to understand the microstructure is the first step in being able to optimize
parameters that inevitably lead to superior materials. This study has introduced
both experimental and computational procedures that enable new techniques to
the forefront of microstructural analysis. This paper presents a snapshot of the
representation and analysis methodology as it exists currently. The focus of this
work is to continue to improve the representation of microstructures to create
realistic computational models for prediction of material capabilities.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
341
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
343
Abstract
Using the Kolsky method, the authors studied dynamic diagrams of uniaxial
compression of aluminum alloys AD-1, AMg-6, and D-16. These alloys are
applied in aviation, space technologies, shipbuilding and many other industries.
Alloys similar to the above mentioned alloys have been studied abroad as well
(for example, aluminum alloys 1100-0, 5182, 6061-T6); however in general data
on dynamic deformation of the mentioned aluminum alloys have been obtained
only at normal temperature. In this work, we performed systematized research of
the aluminum alloys at strain rates of 2001400s-1 and temperatures of 25
250o.
Keywords: Kolsky method, compression, temperature, strain rates, aluminum
alloys, temperature and velocity dependencies V-0.2, model of strength.
Introduction
Research results
Using the Kolsky method, the authors studied dynamic diagrams of uniaxial
compression of aluminum alloys AD-1, AMg-6, and D-16. These alloys are
applied in aviation, space technologies, shipbuilding, and many other industries.
Available data on these alloys are mostly data obtained at static velocities of
loading [4]. There are some data obtained under dynamic loading, but only for
normal temperature [58]. Alloys similar to the above mentioned alloys have
been studied abroad as well (for example, aluminum alloys 1100-0, 5182, 6061T6), strain diagrams have been obtained for them as well, but they are also for
normal temperature [911]. In this work, we performed systematized researches
of the aluminum alloys at strain rates of 2001400s-1 and temperatures of
25250o. Samples in as-delivered condition having sizes 88mm were
studied. In each experiment, diagrams of dynamic compression were obtained,
and yield strengths -0.2 were determined. The diagrams were plotted in the
stress intensity strain intensity coordinates. Recalculation was performed
using the dependencies taking account for change of the effective Poissons ratio
in the area of elastic-plastic transition, eqns (1) and (2):
i=/[1-ln(1+)]2
(1)
i=2(1+)ln(1+)/3
(2)
In eqns (1) and (2) =0.5-0.5(1-2)/(). Fig 1 shows averaged diagrams ii of the alloys. To make the figure simpler, the diagrams are given for each
material at temperatures of 25 and 250o (the diagrams at 150o are located in
the middle). Experiments revealed that strain hardenings of alloys D-16 and
AMg-6 were approximately the same, and it was higher than that of alloy AD-1.
D-16 is the hardest, and AMg-6 and AD-1 are located in the decreasing order
(see Fig 1).
The yield strengths temperature dependencies -0.2=f(T) and velocity
dependencies -0.2=f() are shown in Fig 2 and Fig 3. The same as in Fig 1, the
dependencies in Fig 3 are given at temperatures of 25 and 250o. These
dependencies have generally the linear character. It should be noted that the
velocity dependence -0.2 does not evidently result from curves 7, 8 (for AMg-6)
and curves 11, 12 (for AD-1) from Fig 1. However, in Fig 3, where results of
each experiment are pointed, one can see, nevertheless, the weak dependence of
-0.2 on for the mentioned materials. It follows from Fig 2 and Fig 3 that the
strongest change of -0.2 occurs in D-16 with growth of and , and the weakest
change occurs in AD-1.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
i, MPa
345
1
5
2
400
3
7
6
8
9
200
10
11
12
0
0
0.1
0.2
-0.2, MPa
300
200
2
3
4
100
0
5
6
0
100
, 0
200
2
3
200
100
0
5
6
0
400
800
1200
, s-1
347
Modeling
(3)
p
p
Here i - intensity of plastic strains, i - intensity of rate of plastic strains, -
i = A f 1 ( iP ) f 2 (iP ) f 3 (T )
(4)
ip
i = A 1+ a ( ) 1+b ln p 1T k
i0
p n
i
(5)
r
, m melting
m
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
process
compression
extension
compression
extension
compression
extension
, MPa
200
130
43.5
30
240
220
a
2.45
2.40
2.8
2.8
2.28
2.28
n
0.4
0.4
0.335
0.330
0.28
0.28
b
510-4
510-4
210-4
210-4
1.210-4
1.210-4
m
2.55
2.50
3.6
3.6
3.3
3.3
k
2
2
2
2
2
2
,oK
298
AMg-6
423
523
298
AD-1
423
523
298
D-16
423
523
,s-1
250550
0.1870.12
11001300
360620
10201300
190530
12001400
200370
0.2100.08
0.1690.09
0.1900.14
0.1590.12
0.175
0.0550.04
0.201
0.155
0.169
0.170
0.177
0.058
8001200
230290
9001200
330600
9201100
0.0670.03
0.0490.03
0.053 0.03
0.0450.02
0.0490.02
0.065
0.051
0.056
0.043
0.049
310500
0.3100.09
0.320
8001100
280520
8001100
280550
9001050
0.3380.1
0.2630.14
0.2880.045
0.2450.09
0.2650.11
0.340
0.280
0.300
0.245
0.263
test
Table 3:
349
Material
,oK
AMg-6
298
AD-1
298
D-16
298
, s-1
640800
12001450
210370
9101100
230370
6501130
0.1410.05
0.1560.08
0.0400.08
0.0430.08
0.2850.05
0.3030.014
0.147
0.152
0.040
0.045
0.288
0.300
Conclusion
Figure 4: Experimental (different form points) and calculated (solid and dotted
lines) - diagram of compression and extension of aluminum alloy
AMg-6 at strain rate of 103 s-1. 10=298o (compression), 2
0=423o (compression), 30=523o (compression), 40=298o
(extension).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
351
Abstract
Plasma sprayed molybdenum coatings have been applied to friction parts in the
automotive field because of their high wear resistance. Clarification concerning
the significance of mechanical properties and the residual stress state of the
coating is important to improve the performance of the coating. However there
are some difficulties in the measurement of mechanical properties of the coating,
especially Youngs modulus, because it has complex structures including
lamella, pores, micro cracks and so on. Youngs modulus is required to
determine the x-ray stress constant for x-ray residual stress measurements. Strain
gauges are often used to measure the strains of a component. If glue is applied to
thermal spray coatings to attach a strain gauge, it might be that it penetrates into
the pores and becomes solidified, hence incorrect strains will be measured.
In this research, firstly, Youngs modulus and x-ray stress constant, K, for
molybdenum thermal spray coating were determined experimentally by
four-point bending tests. The effects of quick-drying glue on strain measurement
using a strain gauge were investigated. Secondly, the residual stresses of the
coatings were measured using x-ray diffraction. Six types of substrates with a
different thermal expansion coefficient (TEC) were prepared. The materials
used, in ascending order of TEC, were Super Invar, molybdenum, SCM440,
SK5, SUS304 and A6063. Molybdenum powder was sprayed on the substrates in
air with various thicknesses.
The following results were obtained. (1) The effect of quick-drying glue on
the mechanical strain measurement on the sprayed coating was negligible. (2)
The Youngs modulus of the coating was lower than that of the commercial
molybdenum sheet. (3) Linear strain response against applied mechanical loads
was observed in the case of the polished coating surface. (4) The x-ray stress
constant K of the coating and the bulk were almost the same. (5) Residual
stresses on the coating surfaces were of tensile type on all substrates used in this
study. (6) Tensile residual stresses of the coatings increased with the decreased
thermal expansion coefficient of substrate.
Keywords: x-ray stress measurement, residual stress, plasma spraying,
molybdenum coating, thermal expansion coefficient, quick-drying glue, strain
response.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06034
Introduction
Experiment procedure
Super
Molybdenum SCM440
SK5
SUS304
A6063
Invar
ASTM
AISI
Molybdenum AISI 4140 SK50H AISI 304 ISO AlMg0.5Si
F1684
JIS
0.5
4.9
11.8
11.8
17.3
24.3
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 2:
Spraying material
Mo powder
Method
APS
Current
800 A
Voltage
26 V
Spraying distance
Maximum thickness
353
100 mm
300 m
Experimental procedure
tC AS ES C +1 EC Z S
S
EC =
C
+1 Z S tC AC
S
(1)
from the theory of composite beam. Where t is the thickness, A is the crosssection area and Z is the geometrical moment of area, respectively. Two suffixes
C and S mean coating and substrate. Coating surfaces are postulated as uniaxial
stress.
Furthermore the coating was broken by bending just after cooling with liquid
nitrogen. The fractured surface was observed by a Scanning Electron Microscope
to clarify the penetration depth of quick-drying glue.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
K=
Table 3:
app
M
M=
(2 )
(sin2)
(2)
Characteristic x-ray
X-ray wavelength
Diffraction plane
Diffraction angle 2
K filter
Peak determination
Stress determination
sin2 range, step
Detecter
Tube voltage, current
Peak count
Diameter of incident x-ray
V-K
0.250483 nm
Mo211
154.267 deg
Ti
Center of FWHM
2 -sin2 method
0.0 - 0.6 , 0.1step
PSPC
30 kV, 8 mA
2048
3 mm
355
Youngs modulus of the coating was measured using three specimens with
different conditions. Tensile and compressive stresses were applied to as-sprayed
or polished surface, and thicknesses of substrates were 3mm and 5mm. As a
result measured EC was about 235 GPa on an average. EC seems to be lower than
that of bulk molybdenum (323 GPa [5]) because of the porosity of the coating.
Then Youngs modulus of quick-drying glue was 3.2 GPa. The penetration
depth of quick-drying glue was about 10% of the coating thickness, and the
Youngs modulus of the glue was about 1% of the coating. It means that the
effect of quick-drying glue on mechanical strain measurement on the sprayed
coating was negligible.
Coating surface
Coating surface
50m
As-sprayed
Figure 1:
50m
Glue applied
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-400
-300
-200
-100 -0.1 0
-0.2
Bulk : Bulk molybdenum
-0.3
Sub3 : Substrate t=3mm
-0.4
Sub5 : Substrate t=5mm
Tens. : Tension
-0.5
Comp. : Compression
-0.6
-0.7
Bulk Tens.
-607 MPa/deg
Bulk Comp.
-547 MPa/deg
Sub3 Tens.
-536 MPa/deg
Sub3 Comp.
-630 MPa/deg
Sub5 Tens.
-523 MPa/deg
Sub5 Comp.
-530 MPa/deg
100
200
300
400
200
9 passes
4 passes
2 passes
1 pass
0.5 pass
150
B
100
50
0
0
10
15
20
25
357
Conclusion
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
359
Abstract
Laser peening without pre-coating has been applied to water-immersed
specimens of high tensile strength steel. In order to understand the generation
process of compressive residual stress, specimens with various laser irradiation
patterns, i.e., single spot, line scanning and area scanning, were prepared.
Detailed distributions of residual stress on the specimens were measured using
synchrotron radiation. Large tensile residual stresses, which might be caused by
the thermal effect of the laser pulse, were observed in the center region on the
single spot. It decreased towards the edge of the spot, and changed to
compression around the spot. The compression became larger with the increasing
pulse numbers irradiated on the same spot. In the line scanning, a tensile residual
stress was observed in the final spot of the line, which decreased and changed to
compression as the distance from the final spot increased. The residual stress on
the area scanning was compression as a whole. It was estimated that the
compressive residual stress in the area scanning would be generated from the
overlapping effect of the compressive field made around each laser spot. The
residual stress generated by laser peening without pre-coating is considered to be
the superimposition of the tensile and the compressive components due to
thermal effect and plastic deformation, respectively.
Keywords: laser peening, surface residual stress, single pulse, line irradiation,
laser pulse density, x-ray stress measurement, synchrotron diffraction.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06035
Introduction
Fig. 1 shows the fundamental process of laser peening. Laser beam focused by a
lens is irradiated on a specimen surface in water. The material surface layer
becomes plasma if the power density of the laser pulse at the material surface
exceeds the threshold of the ablation of the material. The inertia of water
prevents the expansion of the plasma, and the energy of the plasma concentrates
in a narrow space. The resulting plasma pressure becomes 10100 times higher
than that in air and it reaches 1 to 10 GPa [8]. The shock wave is generated by
this pressure, and it propagates in the material. Plastic deformation occurs near
the surface of the material due to the dynamic stress of the shock wave, and the
material is hardened through the process. The compressive residual stress is
generated by the elastic restraint from the un-deformed region. The generation of
the compressive residual stress has been explained basically by this mechanism,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
361
since the depth profile of the residual stress by laser peening is well reproduced
by elastic-plastic dynamic simulations.
Laser pulse
Lens
Plasma
Water
Compression
Test sample
During laser irradiation
Figure 1:
Experimental procedure
Si
Mn
0.13
0.25
0.92
Table 2:
0.011 0.001
Mo
Nb
Cr
0.39
0.02
0.04
0.87
Elongation [%]
23
30 m
Figure 2:
Mirror
Window
Water jacket
Figure 3:
15
20
27
18
18
80
Figure 4:
Fig.4 illustrates the single spot specimen where laser pulses irradiated at the
same point 1, 4, 10, and 40 times. The line and area scanning were performed as
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
363
shown in Fig.5 (a) and (b). The lines with pulse densities of 1, 5, 10 and 100
/mm were prepared in the line scanning, changing the scanning speed. In the area
scanning, an area of 10 10 mm2 is irradiated by scanning laser pulses twodimensionally, as shown in Fig.5 (b). The coverage was 800% which was
defined in the same way in shot peening. The laser irradiation conditions are
summarized in Table 3.
17
17
(D)100 /mm
(C)10 /mm
28
(B) 5 /mm
28
(A) 1 /mm
Figure 5:
30
30
Material
Pulse energy
Spot diameter
Pulse duration
Coverage
Pulses irradiated per
unit length
Single pulse
HT1000
215 mJ
1.0 mm
8 ns
800% ( Fig,5 (b) )
1 /mm (A)
5 /mm (B)
10 /mm (C)
100 /mm (D)
1, 4, 10 and 40 pulses
X-ray source
Material
Diffraction plane
Wavelength
Diffraction angle
Detector
Irradiated area
X-ray stress constant, K
Synchrotron radiation
KEK, PF BL3A
HT1000
- Fe211
0.228 [nm]
154 [deg]
PSPC
0.2 [mm] or 0.5 [mm]
-353 [MPa/deg]
Experimental results
Y
X
Figure 6:
Fig. 7 shows the surface residual stress distributions along X-axis of Fig. 6.
Large tensile residual stresses of about 600-800 MPa were observed in all the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
365
laser spots with 1 to 40 pulses. The tensile residual stresses decreased toward the
edge of laser spot, and changed to compression in the outside of the spot. The
compression was about -200 MPa in the single pulse spot. When the number of
laser pulses increased, the compression increased at 0.3mm outside from the
edge of laser spot. The maximum compressive residual stress reached to -600
MPa for 40 pulsed spot.
Laser spot
1000
800
600
400
200
0
-200
-400
-600
-800
1 pulse
4 pulses
10 pulses
40 pulses
Figure 7:
0.5
1.5
Final spot
5 pulses/mm (B)
10 pulses/mm (C)
X
: Stress measured positions, 0.5
mm
Figure 8:
4.2 Surface residual stress distributions in laser irradiated line and area
The interaction of adjoined laser spots was examined using the line-scanned
specimen. Fig.8 shows the positions of residual stress measurement on the laser
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
800
1 pulse/mm (A)
5 pulses/mm (B)
10 pulses/mm (C)
100 pulses/mm (D)
600
400
200
0
-200
-400
0
0.5
1.5
2.5
Discussion
The residual stress at the center of the single spot was tensile even if laser pulses
were irradiated 40 times at the same position (Fig. 7). Meanwhile when the laser
pulses were scanned in a line as shown in Fig. 9, tensile residual stresses were
rapidly decreased with increasing distance from the final spot. These facts mean
the overlapping of laser pulses at the same position dose not generate
compression at the center of laser spot, on the other hand the off-center
overlapping of laser pulses decreases the tension as shown for the lines A, B and
C in Fig. 9. The reason of this phenomenon is considered that the compressive
residual stress region around laser spot overlaps with tension region at the spot
center of the adjoining spot by off-center overlapping.
When the irradiation density was increased to 100 pulses/mm (Line D), the
residual stresses changed to compression. However, although the irradiation
density per unit length of the area irradiation was 3.2 pulses/mm, the residual
stress was compression. These facts mean that an off-center overlapping of laser
scanning lines is also important to generate higher compressive residual stresses.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
367
Our laser peening process does not use any pre-coating. Therefore, the material
surface is exposed to high temperature plasma, and the resulting surface residual
stress depends mainly on the balance of the compressive stress component
generated by cold working due to high-pressure laser plasma and the tensile
stress component generated by rapid cooling due to thermal effect. Thermally
affected depth is only several micrometers [9] and the effect seems to be reset
pulse by pulse. However, the plastic deformation caused by cold working can be
accumulated [10]. Therefore, the tensile component might be constant and the
compressive component increases when the laser pulse density increases. Thus,
the whole stress level is changed to the compression side, when the number of
laser pulses increases, as shown in Fig. 10.
Residual stress
th
Tension
1
r = th + sw
Compression
sw
r: Surface residual stress on laser peened area
th: Tensile stress component generated by thermal effect
sw: Compressive stress component generated by
compressive residual stress around laser spot
Figure 10:
accumulation
of
Conclusion
Overlapped single laser pulse, line irradiation and area irradiation were
performed on high tensile strength steel, HT1000 and the residual stress
distributions near the irradiated spot, line and area were measured using a
synchrotron radiation source. The results obtained through the experiments are
summarized as follows:
(1) Residual stress was tension in the single spot even if 40 laser pulses were
irradiated at the same position. Compressive residual stress was observed in the
outside of the spot.
(2) Off-center overlapping of laser pulses generates the compressive residual
stress on the surface of HT1000 because the compressive stress region at the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgement
This work has been performed under the approval of the Photon Factory
Program Advisory Committee of the High-Energy Accelerator Research
Organization, Japan (Proposal No. 2003G032).
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
369
Abstract
In this paper we have used the Hlder exponent to characterise the scale of
roughness and to study the scale effect of high spatial frequencies on elastic
contact between solids. The mathematical approach shows that the Hlder
exponent of roughness is a sophisticated tool for modelling realistic surface
roughness at different scales of observation. The incidence of Hlder exponent
on the prediction of pressure and displacement is studied in elastic contact
between a smooth plane and rough surface.
Keywords: Holder exponent, scale effect, elastic contact.
Introduction
islands on the earths surface relief, which follows the power law N a 2 ,
where N is the total number of islands whose area larger than a, and D is the
fractal dimension of its coast line ( 0 < D < 1 ).
In this work we have used the Hlder exponent to characterize the scale of
roughness and to model the effect of roughness scale in contact mechanics.
In the second part we have studied, through experiments, the mechanism of
asperity deformation using two numerical approaches based on geomorphologic
characterization and fractal geometry using the concept developed by
Mandelbrot.
The majority of the methods used to determine the relationship between normal
load and displacements, except finite elements methods, are based on the
assumption of elastic solid, homogeneous and semi - infinite The basic
assumptions necessary to use the continuous equations of elasticity:
The slopes of roughness are weak;
The contact area is small in front of other dimensions.
With zero loads, surfaces of the two bodies in the immediate neighbourhoods
of the point of contact can be comparable with paraboloids of equations:
z1 = A1x2 + A2xy + A3y2
(1)
1
1
1
1
.
+
+
+
R1 R '1 R2 R '2
z1 =
371
1
1 2
1 2
1 2
y
x +
y , z2 = ' x22 +
' 1
'' 1
'' 2
R
R
2 R1
2 R1
2
2
2
(2)
The geometry of the contact between two unspecified solids is thus brought back
to the problem of the contact between two paraboloids:
h( x, y ) = Ax 2 + By 2 =
1 2
1 2
x +
y
2 R1
2 R2
(3)
1 i 2
wi ( x , y ) =
Eequ
Ac
( x x ')
+ ( y y ')
(4)
S1
Z1
S2
Figure 1:
w2(x,y)
w(x,y)
Z2
Contact
region
Rgion
du
Rgion
ducontact
contact
w1(x,y)
So if one calls w(x, y), the difference between the M1(x point, y, z1) of the
surface of the solid 1 (S1) and the M2(x point, y, z2) of the surface of solid 2
(S2), the geometry enables us to write the following relation:
w( x, y ) = w1 ( x, y ) + w2 ( x, y ) =
x2
y2
2 R1 2 R2
(5)
where w1(x, y) and w2(x, y) are relative displacements on the two solids, R1et R2
the principal radii of curvature of the two solids.
The determination of displacement in a point of surface due to the field of
pressure is carried out by the superposition of the influence of all the efforts.
Displacement at the point ij can then be written in the form:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(6)
P(x,y)
P(x,y)>0
P(x,y)=0
Figure 2:
C ijkl p kl + h( x, y ) = 0
(7)
If one consider the pressure as an input signal and displacement like a response
of exit by the following relation to a dimension (Johnson 1985):
w( x) = h( x) p( x)
(8)
This equation is well defined in the space field in the form of a product of
convolution: If the pressure applied is form
p( x) = B0 sin( 2x / )
(9)
The obtained displacement has the following expression (Johnson 1985)
w( x) =
(1 2 )
B0 sin( 2x / ) + cst
E
(10)
what gives like transfer function of the system in the field of the space
frequencies:
H ( ) =
2(1 2 )
E
(11)
and in the field of the space frequencies one has like response
W ( ) = H ( ) P( ) =
2(1 2 ) P( )
(12)
In addition, by using the properties of the Fourier Transform one can write that
the derivative of displacement can be written in Fourier domain as
W ( ) = iW ( )
by replacing in the equation of the pressure one can write:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(13)
P( ) =
F ( ) =
373
E
W ( )
i 2(1 2 )
(14)
E
W ( )
i 2(1 2 )
(15)
2
2(1 ) P ( x , y )
(16)
2
2
where = x + y represent a space frequency.
The determination of the field of pressure p(x, y) is determined by opposite
transform of Fourier for each level of bringing together :
-1
p ( x, y ) = TF
2
2(1 )
W ( x , y )
(17)
The spectrum of pressure is thus spread out between the low and high
frequencies:
vx =
1
1
1
1
, vy =
mm-1 , hx =
mm-1
, hy =
Lx
Ly
2x
2y
with Lx, Ly , the size of the surface, x , y the sampling steps in the x and y
directions, which initially define rough surface before contact.
For normal macroscopic imposed load F, one numerically brings closer two
surfaces and to solve the equation for each level of bringing together . For each
normal position of the rigid plan, one determines contact pressures p(x, y) by
using the direct and inverse Fourier transform of the contact equations.
Convergence is reached when the imposed effort is equal to the sum of the local
efforts calculated for each iteration:
F = pij ( x, y ) xy
i, j
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(18)
As the details of roughness: Z(x) depends on the length scale, we assume each
realization of Z(x) to be a continuous, but non-differentiable function. It means
that the presence of any small roughness elements may prevent us from reaching
a satisfactory limit of
(Z ( x + ) Z ( x) ) as 0
(19)
Z ( x + ) Z ( x) H , 0 < H < 1 0
(20)
(Z ( x + ) Z ( x) )2
2 H 0<H<1, 0
(21)
(Z ( x + ) Z ( x) )2
2H
, 0 < H < 1, 0
(22)
stands
for the usual norm in this space. This is not a new condition; it is proven for all
( Z ( x + s ) Z ( x))
sH
(23)
Z ( sx ) = s H ( Z ( x )
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(24)
375
Z ( sx )
are statistically
sH
indistinguishable.
Voss (1985) showed that the fractal dimension is related to parameter H by
the relation D = E+1-H where E is the dimension of the Euclidean space RE.
These functions, which correspond to a non-stationary random process with
stationary increments, present many other interesting properties. For instance
their power spectrum S() presents a well-defined decrease following the power
law
S ( ) =
= 2H + E
2H +E
(25)
S (k ) =
1
k
2H +E
S ( )
(26)
If one makes the analogy with the relation between the spectrum of pressure and
the spectrum of the slopes of displacements which vary in the same space
frequency band, one can notice that the law of power spectrum of pressure
depends on the Hlder exponent as following
P ( ) =
E
W ( )
i 2(1 2 )
P (k ) =
1
k
2H +E
P( )
(27)
(28)
This relation, shows well the effect of the exhibitor of Hlder on the spectrum of
pressure, and consequently on the local singularities of pressures and
displacements which will induce local plasticity.
(a)
Figure 3:
S1
S2
S3
Figure 4:
(b)
(c)
(a1)
(b1)
(c1)
(a2)
(b2)
(c2)
(a3)
(b3)
(c3)
The following table (Table 1) regroups the whole of contact parameters below
(Bearing surface area and mean contact pressure) of the three different rough
surfaces.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 1:
377
Surface roughness
S1 (H=0.1)
S2 (H=0.5)
S3 (H=0.8)
These results (Table 1) show that the roughest surface (S1) which has the
lower Holder exponent is the more resistant in contact. Indeed, it presents the
lowest bearing area surface (11.66%) and the most important mean pressure
(3.40 GPa). This proves the narrow relation between punctual pressure fields
singularity and the Holder exponent indicator of roughness.
Conclusion
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
Lopez J., Hansali G., Zahouani H., Leboss J.C., Mathia T., 3D Fractalbased
characterisation
for
engineered
surface
topography,
Int.J.Mach.Tools Manufact, Vol 35, pp.211-217, 1995.
Mandelbrot B., Stochastic models for the Earths relief, the shape and the
fractal dimension of coastlines, and the number-area rule for islands,
Proceedings of the National Academy of Science, Vol 72, pp.3825-3838,
1975.
Richardson L.F., The problem of contiguity: An appendix of statistics of
deadly quarrels, General Systems Yearbook 6, pp.687-689, 1961.
Mandelbrot B., How long is the coast of Britain - Statistical self-similarity
and fractional dimension, Science, Vol 155, pp.636-638, 1967.
Brown S.R. and Scholz C.H., Broad bandwidth study of the topography of
natural rock surfaces, J. Geophys. Res., 90, pp. 12575-12582. , 1985.
Felder J., Fractals, Plenum Press, New York, 1988.
Mandelbrot B. and Van Ness J.W, Fractional Brownian motions,
fractional noises, and applications, SIAM Review 10, 422-437, 1968.
Russ John C., Fractal surfaces, Plenum Press, New York, 1944.
Saupe D. and Peitgen H.O., The science of fractal images. SpringerVerlag pp.82-91, 1988.
Zahouani H., Vargiolu R., Fractal models of surface topography and
contact mechanics, Vol28, N 4-8, pp 517 534, 1998.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 5
High performance
concretes
381
Abstract
This paper presents results from an experimental investigation carried out to
study the potential use of copper slag as fine aggregate on the strength of both
normal and high strength concrete. Concrete mixtures were prepared using
different proportions of copper slag as partial and full replacement of fine
aggregate. The percentage of copper slag added by weight ranged between 10100% of sand used in concrete.
For each concrete mixture, six
150mmx150mmx150mm cubes, three 300mmx150mm dia. cylinders and three
100mmx100mmx500mm prisms were cast. Density, compressive, tensile and
flexural strengths were determined at 28-day of curing. Cube compressive
strength was also determined at 7-day of curing. Results demonstrated that there
is general an increase in the density and workability of both normal and high
strength concretes as copper slag quantity increases. Also results showed that
the compressive strength of concrete is generally improved, compared with the
control mix, with the increase of copper slag up to a certain copper slag content
beyond which the strength generally reduces. Mixes with large copper slag
percentage showed signs of bleeding and segregation due to the significant
increase of workability.
Keywords: copper slag, fine aggregate, concrete, strength, workability, density.
Introduction
Aggregates are considered one of the main constituents of concrete since they
occupy more than 70% of the concrete matrix. In many countries there is
scarcity of natural aggregates that are suitable for construction whereas in other
countries the consumption of aggregates has been increased, in recent years, due
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06037
Research objectives
The main objective of this study was to investigate the use of copper slag as
partial and/or full replacement for fine aggregate in normal and high strength
concrete mixtures. The following were specific tasks:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
383
Materials
3.1 Cement
The cement used in this study was ordinary portland cement (OPC) purchased
from Oman Cement Company. This cement is the most widely used one in the
construction industry in Oman.
3.2 Coarse and fine aggregates
Coarse aggregates (i.e. 20 mm and 10 mm) and fine sand were purchased from a
nearby crusher in Al-Khoudh area, which are typically the same materials used
in normal concrete mixtures. The gradation test conducted on aggregates
showed that they met specifications requirements.
3.3 Copper slag
Copper slag is a by-product material produced from the process of
manufacturing copper. As the copper settles down in the smelter, it has a higher
density, impurities stay in the top layer and then are transported to a water basin
with a low temperature for solidification. The end product is a solid, hard
material that goes to the crusher for further processing. Copper slag used in this
work was brought from Oman Mining Company, which produces an annual
average of 60,000 tons.
3.4 Silica fume
The silica fume used in the production of high strength concrete was supplied
and added to the mix in a powder form (Elkem Emsac 500s).
3.5 Superplasticizer
In order to improve the workability of high strength concrete, superplasticizer in
the form of a polynaphthalene sulphonate-based admixture (conplast SP430) was
used. This had 40% active solids in solution.
Type
Cement
NSC
HSC
416
400
w/c
Wat
er
207
140
ratio
SP
l/m3
0.5
0.35
7.9
The properties of normal and high strength concretes in terms of density, slump,
compressive strength, tensile strength and modulus of rupture are shown in
Tables 2 and 3, respectively.
The effect of copper slag addition as replacement of fine aggregate on the
density of normal and high strength concretes is shown in Fig. 1. It can be seen
from Fig. 1 that there is general increase in the density of both normal and high
strength concretes with the increase of copper slag replacement in the concrete.
The density of concrete was increased by almost 5% for both concrete types.
This is mainly due to the higher specific gravity of copper slag which was 3.4
compared to fine aggregate which has a specific gravity of 2.8.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 2:
Mix
No.
Mix
Type
Control
(100% S)
10%CS +
90S
20%CS +
80%S
30%CS +
70%S
40%CS +
60%S
50%CS +
50%S
70%CS +
30%S
90%CS +
10%S
100%CS
+ 0%S
2
3
4
5
6
7
8
9
385
Strength (MPa)
(Fcu)+
(Ft)+
Density
(kg/m3)
Slump
(mm)
(Fcu)*
2524
50
33.2
44
7.7
2515
90
37.1
44.9
3.5
7.2
2540
90
39.6
48.5
3.7
7.2
2558
75
37
48.9
3.2
6.9
2550
145
36.2
48.1
3.8
6.5
2560
180
39.5
53.1
4.1
7.3
2601
185
37.2
46.6
3.6
6.3
2597
200
37.9
50.1
3.6
7.2
2653
210
38
45
3.4
5.9
(Fcr)+
S= Sand
Table 3:
Mix
No.
Mix
proportions
Control
(100% S)
10%CS
90S
20%CS
80%S
40%CS
60%S
50%CS
50%S
60%CS
40%S
20%CS
80%S
100%CS
0%S
2
3
4
5
6
7
8
S= Sand
+
+
+
+
+
+
+
+= Cured at 28 days.
Density
(kg/m3)
Slump
(mm)
(Fcu)*
Strength (MPa)
(Fcu)+
(Ft)+
(Fcr)+
2568
28
76.9
93.9
5.4
14.6
2530
28
79.6
99.8
5.2
13.6
2588
50
74.5
95.3
6.2
12.4
2586
125
64.8
79.6
4.6
10.8
2625
115
77.8
96.8
6.1
12.9
2658
128
69.0
83.0
4.8
11.1
2673
102
63.8
79.0
4.7
10.3
2700
150
63.4
82.0
4.4
10.1
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Normal concrete
High strength concrete
2650
2600
2550
2500
2450
2400
1
M ix No.
Figure 1:
Slump (mm)
250
Normal concrete
High strength concrete
200
150
100
50
0
1
M ix No.
Figure 2:
387
measured slump was 50mm and 28mm for normal and high strength concretes
respectively whereas for mixes 8 and 9, with 100% replacement of copper slag
for both NSC and HSC respectively, the measured slump was 210mm and
150mm for both types of concrete. This suggests that copper slag has much less
absorption characteristics than sand which explains the considerable increase in
workability as copper slag quantity increases. This increase in workability can
have beneficial effect on concrete in the sense that concrete mixes with low
water-to-cement ratios can be produced which can have good workability,
greater strength and improved durability than conventional concrete. However,
it should be noted that mixes with high contents of copper slag showed signs of
bleeding and segregation of which can have detrimental effects on concrete
performance.
7-days
20
28-days
10
0
1
Mix No.
Figure 3:
7-days
28-days
Mix No.
Figure 4:
Conclusions
This paper presented the results of a research study on the effect of using copper
slag as a substitute for fine aggregate in normal and high strength concrete
mixes. Normal and high strength concrete mixes were prepared with different
proportions of copper slag. Concrete mixes were tested for workability, density,
compressive strength, tensile strength and flexural strength. Results showed that
the increase in the copper slag content causes an increase in both workability and
density of normal and high strength concretes. Results also showed that the
compressive strength of concrete is generally improved, compared with the
control mix, with the increase of copper slag up to a certain percentage beyond
which the strength was slightly reduced. Mixes with large copper slag
substitution suffered from bleeding and segregation due to the increase in
workability. Generally it is concluded that copper slag can be used as fine
aggregate to produce concrete with good strength.
References
[1]
Shi, C., and Qian, J.; High Performance Cementing Materials from
Industrial Slags A Review. Resources, Conservation and Recycling,
29:195-207, 2000.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
[2]
[3]
[4]
[5]
389
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
391
Abstract
Fibers in concrete should primarily restrain the material from cracks during the
curing process and dehydration. For standard evaluation of the influence of the
fibers, the three or four point test is studied, both experimentally and numerically
on a straight beam. In tunnel lining the situation is slightly different. While the
three or four point test leads to the assessment of the structure being only
bended, in tunnel reinforcement not only this but also bending with normal force
has to be observed. Moreover, the curing is affected by the non-uniform
distribution of moistening, as from the side of the surrounding rock the measure
of wet is principally higher than on the face where ventilation speeds up the
dehydration. These two assumptions are considered and treated experimentally
and numerically. Because a real tunnel lining is too large, a scale model is used
and, consequently, more samples can be studied. Similarity rules and an
appropriate calibration have to be obeyed. This paper is principally focused on
the theory of possible coupled modeling of behavior of the one-sided moistened
tunnel lining.
Introduction
Recently, the author published several papers on behavior of steel FRC with
Dramix type fibers in underground structures, especially in tunnel linings, [1].
One-sided moistened concrete beams have been studied from the mechanical
point of view. Scale models were prepared and studied.
The process, which we deal with in this paper, starts with coupled modeling
targeted to description of one-sided moistened beams. Then arches are modeled
and the mechanical properties are taken from the straight beam behavior. From
the numerical point of view a linear system of equations for unconstrained
problem has to be solved and when restrictions are imposed on stresses a
nonlinear problem is to be observed. The solution of such a problem can be
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06038
The behavior of one-sided moistened FRC was studied on straight beam, since
curvilinear beam could be modeled in a cumbersome way. The samples were
prepared in an enameled basin, the tap water in which was pored, which was
permanently added to hold the water level at a proper elevation. The wetted
medium surrounding the sample simulated real situation on site. Consequently,
the sample was prepared from two kinds of layers: the first layer was built up
from high porously liquid-absorbing mass. This mass was separated from the
concrete (the sample itself) by a filter paper to ensure convenient removal of the
concrete slender element from the equipment for preparing the samples. As a
consequence of this structure the lower side of the samples was permanently
highly moistened. This was also the aim of the experimental study
The samples were built up by auxiliary timber frame; its thickness along three
sides had the value of the right layer of the samples. The fourth side was free in
the beginning of the layers creation. Then the first layer of the fiber reinforced
concrete was constructed and the rectangular frame was completed by gluing the
last part (the shape of it corresponded to the height of the layer). This stage of
building the experimental equipment is depicted in Fig. 1, where also wire
holders for sticking next molding are seen to create the following layers of the
concrete. In this figure still no water is pored, and the three remaining moldings
inside the form are still not yet put inside the form to create the first concrete
layer.
In order to create the thicknesses of the layers as accurately as possible, the
moldings were put into the form along all sides of the equipment, so that the
room for the first and the subsequent layers was exactly defined. In other words,
at each stage of layers creation, one planar frame was built up with the same
thickness of moldings (as desired).
A plateau with the set up of fibers was drawn in Corel Draw on hardened
paper of 200 g/m2 In order to hold the cold temperature, which could cause
temporary sticking of fibers to the paper, the temperature of the paper with the
lay-up of fibers was dropped by putting the system in a freezer. Frozen at 32
degrees of Celsius every paper was three times sprinkled to assure the adhesion
of fibers to the paper. The number of fibers was calculated in such a way that the
fiber volume ratio was 1 percent. In each layer are 270 fibers in 11 interfaces of
concrete layers (6 and 5 fibers are repeated in rows). The thicknesses of the
layers are 6 + 4 +...+ 4 + 2 mm.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
393
The concrete mixture was made of Knauf BP-8, the stones of which larger
than 0.8 mm were removed by filtering.
For 28 days in room temperature the slender beam was dried in such a way
that from the upper side a fan two times a day for 2 minutes was used from a
distance of 2 m. From below steady delivery of water was assured.
The final stage is seen in Fig. 2. The water was poured in the basin
immediately after the first layer of concrete was put in the frame. The sample
was not fully in a directly contact with the water, but along the interface of the
sample with the plateau the moistening of the lower layer was ensured. The
dimensions of the whole sample element were: 150 x 250 x 48 mm3. This sample
was cut into beams of a cross section 2.8 x 4.8 mm2, which were loaded by a
concentrated load in the middle of span of the beams. The ratio of thickness to span was about 1/3. From that point of view, the mechanical behavior was closer
to a plate.
Figure 1:
Figure 2:
The aim of this study was to discover how the situation would worsen during
the bending test comparing these samples with that without loading by the
moistening.
In Fig. 3, a relation applied force - deflection is depicted. Two cases of the
beams are compared: moistened and external moistening free plates are
considered. Both types of beams were cured 28 days and then the experiments
were accomplished.
From Fig.3 one can see a sudden drop at the value of loading about 1.8 kN in
the case of moistened beam and about 2.6 kN in the case of the wet-free beam.
These drops differ also in deflections, which were in the first case 0.18 mm and
0.22 mm in the second case. Sudden opening of a crack below the concentrated
load causes the drops of loading. After increase of loading peaks are reached.
While the externally moistened beam attains 3.8 kN, the moistening free beam
4.5 kN. After this, the beams still bear, but their bearing capacity lowers and
softening of the aggregate occurs. This is caused by a pulling of fibers out of the
concrete matrix. When the destruction of the beams is attained, it can be seen
that no steel fiber were disconnected, fibers were only pulled out of the concrete.
The different behaviors of both beams are caused by pore pressure, which is
stored in cured beams. In mathematical models this property is very easy to
describe. Pore pressure is a special case of eigenstrain tensor, , or eigenstress
tensor, , which both are expressed in the generalized Hookes law as:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
395
= L( ) or = L + ,
where is the stress tensor and is the strain tensor. L is the elastic material
stiffness matrix.
Another type of experiment was carried out on these beams. A time
dependent behavior of both types of beams is observed by bending with the same
disposition as in the previous experiments, see Fig. 4.
Figure 3:
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 5:
It was shown that the bearing capacity of the moistened structure is less than
that of beams created in the standard way. Also the creep seems to be more
important and larger in the case of moistened structure.
The clamped arch can be solved by Fourier method: the unknowns and the
loading are split into series in one direction and in the remaining direction the
ordinary differential equation is solved. The problem of clamped arch is divided
into solution of simply supported arch and the rotation at clamped edges is
annihilated by applying external bending moment. The solution of the latter case
can be found in [2]. The first problem is a generalization of the problem from
[3]. Results from similar cases can be seen in [4].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
397
z
r
rz
Lrr
L r
L
= rz
0
0
0
L r
L
Lz
0
0
0
Lrz
Lz
L zz
0
0
0
0
0
0
G r
0
0
0
0
0
0
Gz
0
r r
r r
zr zr
2( r r )
2( z z )
G rz 2( rz rz )
0
0
0
0
0
(1)
where the stress tensor was related with the strain tensor and the eigenstrain
tensor through the material stiffness matrix L, and Lij , G r , Gz , G rz are the
stiffness coefficients. Nine stiffness coefficients and six eigenstrains describe
this kind of anisotropy. Cylindrical orthotropy is characterized by the fact that
properties in the tangential, radial and axial directions are distinct; in other
words, the material is orthotropic in a Cartesian system which is located at any
point within the structure, with the three axes pointing in the axial, tangential and
radial directions respectively. If Lrr > L then the material is called radially
orthotropic, and if L > L rr then it is called circumferentially orthotropic.
In the special case of transversally isotropic solid which may represent the
average structure of fiber reinforced material, nine independent material
constants in (1) are related as:
Lrr = L ,
Lrz = Lz ,
G rz = Gz ,
Lrr - Lr = 2G r
r =
u
,
r
1 v u
+ ,
r r
r =
1 u v v
+
r r r
(2)
1
( r ),
E
1
( r ),
E
r r =
1
r (3)
G
1 v
u
u
r + Lr
+
r = Lrr ( r r ) + Lr ( ) = Lrr
r
r
1 v u
u
r
+ + L
= Lr
r
r
1 u
v v
r = G r r = G
+
r
r
r
r
(4)
(5)
r 1 r
+ r = 0,
+
r
r
1
+ 2 r = 0
+
r
r
(6)
Expanding two components of displacements into sine and cosine series and
denoting the k-th term by U r( k ) and U ( k ) yields
u r( k ) (r , ) = U r( k ) (r ) cos 2k ,
u( k ) (r , ) = U ( k ) (r ) sin 2k ,
(7)
u z( k ) = 0
where U r( k ) U ( k ) and U ( k ) V ( k ) are unknown functions of r which need to
be determined from the equations of equilibrium. In particular, the substitution of
(1) in the stress-strain relations gives stresses which, when substituted into the
equations of equilibrium in cylindrical coordinates, provide the following
equations for evaluation of U ( k ) and V ( k ) .
Similarly the eigenparameters will be expanded into series with the k-th
components:
r( k ) (r , ) = R ( k ) (r ) cos 2k , ( k ) (r , ) = F ( k ) (r ) cos 2k ,
r(k ) (r , ) = M ( k ) (r ) sin 2k
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(8)
399
r( k ) = U ' cos 2k , ( k ) =
1
(2kV + U ) cos 2k , r(k ) = 2k U + V ' 1 V sin 2k
r
r
r
(9)
(k )
2k
(10)
(11)
Substituting (10), (11) to (6) and using the last four relations, two equations for
unknown amplitudes of displacements read as:
M
F
3
8k 2 + 1
4k +
4k ( 3)
V ' ' V '+
V
U '+
U M '+
+ R '+ 2 = 0
2
r
r
r
R
1
r
(1 ) r
The last two equations lead to the solution, which defines the displacements
on each lamina. Four integration constants are determined from interfacial
conditions between adjacent layers. The angle determined from the above
approach is eliminated by introducing moments at the end points to get
conditions for clamped segment. The influence of such a moment to the segment
is taken from [1], where the approach is described in details.
Conclusions
Acknowledgment
This paper was prepared under financial support of GA AV R, project No. IAA
2119402.
References
[1]
[2]
[3]
[4]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
401
Abstract
As is well known, prefabrication procedures allow significant reductions in
construction time. In many countries, precast concrete bridges are customary and
made up of pretensioned prestressed girders. Both the compressive and tensile
strength of the concrete mix must be enough at transfer (which takes place
typically two or three days after concrete casting) in order to avoid excessive
prestress losses, and undesirable cracking. Many attempts have been made to
increase the spectrum of the usual high performance concretes to other concrete
types (i.e. normal strength concrete, lightweight concrete and recycled aggregate
concrete). Why are these special concrete types not used in the current
production of large pretensioned concrete girders? Technical problems have been
observed at the critical zones: development length, geometrical discontinuities,
supports, etc. A new technique has been developed by the authors to produce
hybrid pretensioned concrete elements, combining the usual high strength
concretes in the critical zones with other concrete types in the rest of the
elements.
Keywords:
prestressed concrete, precast concrete, pre-tensioning, pretensioned, pretensioned, high performance, high strength, prestressed, girders,
beams.
Introduction
Transfer the prestressing force to the beam during the service life.
Transfer length is defined (after Weerasekera [1]) as the distance over
which the strand should be bonded to concrete in order to develop the
effective prestress stress (fpe) in the strand (see Figure 1) .
Guarantee the anchorage of the prestressing force. The flexural bond
length is defined as the distance beyond the transfer length required to
achieve the prestress design stress (fps) during the service life (see
Figure 1). The development length is the sum of the transfer length and
the flexural bond length.
Accordingly two different zones are determined along these elements: the
external zones, where the bearing capacity is lower than the design capacity
(development length) and the inner zone, where prestress is fully effective.
Both the compressive and tensile strength of the concrete mix must be high
enough at transfer (which takes place typically two or three days after concrete
casting) in order to avoid excessive prestress losses, and undesirable cracking.
The stresses created by bond in concrete are so high that the use of high strength
concrete is usually required to prevent cracking along the transfer length.
Figure 1:
Many attempts were made to increase the spectrum of the conventional high
performance concretes to other concrete types (i.e. moderate strength concrete,
lightweight concrete and recycled aggregate concrete).
The use of lightweight concrete having similar strength compared to the
current high strength concrete, would allow either the reduction of the self
weight of these elements while maintaining their span length, or the increase of
the span length while maintaining the self weight.
On the other hand, a reduction of the requirements of the resistant properties
of concrete would imply a significant cost reduction. The use of normal strength
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
403
concrete is desirable, instead of the current high strength concrete. However this
possibility must be carefully studied, in order not to affect the performance of the
structure during the service life.
Environmental regulations are hardening in developed countries. The
extraction of aggregates is being penalized. In the future, a considerable increase
of the cost of natural aggregates will be experienced. Consequently, the
applicability of recycled aggregate concrete to the prefabrication processes must
be studied.
Why are these special concrete types not used in the current production of
large pretensioned concrete girders? In our opinion it is due to their high cost and
eventually to the technical problems at the critical zones: development length,
geometrical discontinuities, supports, etc.
Nowadays, the bearing capacity of the pretensioned concrete elements is
limited by the feasibility of introducing the maximum prestress in the cross
section due to the restrictions of minimum center-to-center strand distance as
well as on minimum cover.
Several attempts to use high strength lightweight concrete instead of the
current normal weight concrete, with the same design criteria have failed due to
longitudinal cracking along the ends of the girders (Vzquez-Herrero and
Martnez-Abella [2]). These cracks are generated by the high stresses in the
boundary between strands and surrounding concrete:
Normal stresses between strand and the surrounding concrete, caused by the
recovery of the strand diameter after transfer. As a result transverse tensile
stresses appear in the surrounding concrete (bursting stresses) along the
whole length of the pretensioned concrete elements.
Bond stresses along the transfer length. The resulting transverse stresses
which appear in the surrounding concrete are named splitting stresses.
Methods to enhance the confinement of the strands. With this aim transverse
reinforcement is disposed, either through stirrups or spiral reinforcement.
This procedure does not prevent cracking, but it certainly reduces crack
width, and prevents losing the confining effect. These methods imply a
considerable increase of operational costs.
Methods to reduce bond between active reinforcement and concrete by
different means, i.e.: reduction of the strand surface roughness, use of
lubricating substances, use of partially debonded tendons, etc. An adverse
effect is collateral to these measures: the increase of the development length
means a reduction of the bearing capacity of the girders at this zone.
Besides, it is difficult to exert and control the rate of bond reduction desired.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
The bending moments along the structure determine the situation of the
zones where the prestressing force is not constant, which means the
existence of bond stresses.
The stress concentration zones, such as supports, holes, zones where the
geometry changes abruptly,...
The objective concrete will be used in most of the concrete girders, but in the
critical zones, which will be executed using a high strength concrete mix that has
proved a good performance. The adequate performance of the objective concrete
in the non critical zones must be checked experimentally, in order to determine
the adequacy of the material for this use.
The production procedure is similar to current girders, but for the different
phases, which will be determined by the different critical zones. The boundary
surface between the different concrete types can be accomplished by the use of
any kind of formwork, or by the natural slope adopted by concrete.
405
2.
3.
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 4:
Prop.
UNITS
Concrete
specific
weight
KN/m3
Average
Average
Average
Average
Average
Average
concrete
direct
concrete
direct
split tensile modulus of
cylinder
tensile
cylinder
tensile
strength, at elasticity, at
strength, at
strength, at strength, at strength, at
2 days
2 days
2 days
2 days
28 days
28 days
MPa
MPa
MPa
MPa
MPa
MPa
RC
22
50
3.2
30441
3.3
67
>3.3
LC
17
48
2.7
19939
2.7
56
2.9
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
407
Strand type was Y1860 S7 (1860 MPa), with a diameter of 15.2 mm (0.6).
Strand was tensioned until a stress of 75% of the minimum ultimate stress was
attained. From the stressing process to the prestress force release 48 hours
elapsed. Transfer was performed through gradually by flame heating.
The prisms were continuously monitored during the transfer process and
afterwards. Transfer length was measured periodically after transfer, and its
values kept steady within the high strength concrete critical zones. The prisms
did not show any visible cracking for several months, but after 3-5 months
longitudinal cracks appeared along the lightweight concrete inner part of all the
hybrid specimens. Cracks progressed throughout the whole lightweight concrete
inner part.
Consequently, the tested lightweight concrete mix used is not able to
withstand the bursting stresses resulting from the strand diameter recovery which
takes place after transfer due to the Poisson effect. Although the performance of
the hybrid prisms is adequate for the first several months after transfer, durability
cannot be assessed throughout the whole service life of the structure.
Figure 5:
Conclusions
Acknowledgements
This research project has been sponsored by the cement and concrete firm
Corporacin Noroeste, and by the projects XUGA22801A97 and
CICYTMAT20001-0765 (Ministry of Science and Technology, Spain). The
authors appreciate the contribution of the engineers Mr. Humberto Vzquez, Mr.
Arturo Martnez, Mr. Marcos Fernndez and Mr. Juan Rabual, and of Professor
Dr. Pablo Rodrguez-Vellando for the revision of this text.
References
[1] Weerasekera, I.R., Transfer and Flexural Bond in Pretensioned Prestressed
Concrete, PhD Thesis, UMI: Michigan, 1991.
[2] Vzquez-Herrero, C., Martnez-Abella, F., Bond Properties of High Strength
Pretensioned Lightweight and Normal Weight Concrete, Proceedings of
Bond in Concrete, Budapest, pp. 791-798, 2002.
[3] Saudo-Herrera. F., Experimental study of the behavior of hybrid
pretensioned concrete elements using high strength concrete and lightweight
concrete, Undergraduate Research Project (in Spanish), University of La
Corua, 2003.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
409
Abstract
The moisture transport and storage parameters of high performance concrete
belong to the most critical parameters in designing and using complex reliability
based models for service life prediction of concrete structures. However, mostly
it is supposed that the material is compact without any significant cracks, which
is not always true. In this paper, the effect of thermally induced cracks on the
basic moisture transport and storage properties is analyzed. Experimental results
show that the presence of cracks in hardened high performance concrete
mixtures dramatically changes all measured hygric parameters. Also, the effects
of both microsilica and the size and presence of aggregates on moisture transport
and storage parameters are very important. The liquid water transport is found to
be affected by cracks, microsilica and aggregates in a much more significant way
than water vapour transport.
Keywords: high performance concrete, moisture transport and storage
properties, cracks.
Introduction
Materials
Basic tested material was high performance concrete C90/105 containing silica
fume suspension, microsilica, denoted as BI in subsequent text. For the sake of
comparison, the same concrete mixture C60/75 (denoted as BII) was prepared
without microsilica. In order to analyse the effect of aggregates on moisture
transport and storage parameters, two other mixtures based on BI and BII were
prepared. The first one (denoted as BBI and BBII) was cement mortar without
the 8-16 mm aggregate fraction. The second one was cement paste denoted as PI
and PII. Table 1 presents the composition of the studied cement mixtures.
Table 1:
Composition in g
**
SiO2
Aggregates
Woer- ***
CEM I
Lentan
suspen- 0-4
ment
4-8
8-16
52.5 R
VZ 33
sion
mm mm
mm FM 794
BI
480
72
664 207
995
7.74
2.58
BII
470
0
668 209 1001
5.17
2.35
BBI
480
72
664 1202
0
7.74
2.58
BBII
470
0
668 1210
0
5.17
2.35
PI
2346
352
0
0
0
38
12.67
PII
2348
0
0
0
0
26
12
*
SiO2 suspension water suspension consisting of 88-95% of SiO2 and
amounts of calcium oxide, magnesium oxide and nitrogen oxide.
**
FM 794 plasticizer on the basis of polycarboxylateether.
***
Lentan VZ 33 hydration retarder on the saccharose basis.
Type of
concrete
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
w/c
0.36
0.33
0.38
0.35
0.31
0.34
small
411
Each mixture was cast into the forms, after one day the samples were
removed from the forms and then varnished using curing solution. After the
hardening period of 28 days, the samples were cut to required dimensions. These
samples were then heated up to 600C in an oven to reach cracks, see Figure 1.
Figure 1:
3 Experimental methods
3.1 Basic material parameters
As fundamental physical material characteristics, bulk density b [kgm-3],
vacuum saturation moisture content wsat [kgm-3], porosity [Vol.-%] and matrix
density m [kg m-3] were determined. They were obtained using the gravimetric
method and the vacuum saturation method. The vacuum saturation moisture
content was calculated according to the equation
wsat = w
msat m0
= w ,
msat ma
(1)
where w is the water density [kg m-3], m0, msat and ma are the mass of dry
sample, water- saturated sample and mass of the immersed water - saturated
sample [kg], respectively, and is the open porosity, which is defined as the
ratio of the volume of open pores in material to its total volume. Matrix density
was calculated as
mat =
m0 ,
V (1 )
(2)
D=
m d R T
S M p p
(3)
where m the amount of water vapor diffused through the sample [kg], d the
sample thickness [m], S the specimen surface [m2], the period of time
corresponding to the transport of mass of water vapor m [s], pp the difference
between partial water vapor pressure in the air under and above specific
specimen surface [Pa], R the universal gas constant, M the molar mass of water,
T the absolute temperature [K].
On the basis of the diffusion coefficient D, the water vapor diffusion
resistance factor was determined:
Da
D
(4)
i = A t ,
(5)
where i is the cumulative water absorption [kg/m2], t is the time from the
beginning of the suction experiment. The water absorption coefficient was then
employed for the calculation of the apparent moisture diffusivity in the form [2]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
413
A ,
app
wc w0
(6)
where wc is the saturated moisture content [kgm-3] and w0 the initial moisture
content [kgm-3].
In the experimental work, the following samples were used: water vapor
diffusion coefficient 10 cylinders with the diameter 105 mm and thickness 20
mm, water sorptivity - 5 specimens 50 x 50 x 20 mm.
3.3 Sorption isotherms
The water adsorption and desorption in a porous material are based on van der
Waals forces between the surface of the porous matrix and water molecules. The
dry material mass increases after a contact with moist air because of gradual
bonding of water molecules from the air to the pore walls, in the case of
adsorption. Desorption is reversed physical phenomenon, the initial state is
capillary saturated sample. At the moment of achieving the equilibrium state
between the water vapour pressure in the moist material and in the surrounding
air this process is stopped.
The mass of samples was measured in specified periods of time until steady
state value of mass was achieved. Then, the moisture content by mass was
calculated according to the equation
um =
mw m0
m0
(7)
where mw is the mass of wet sample and m0 is the mass of dry sample [kg].
The samples were placed into the desiccators with different salt solutions to
simulate different values of relative humidity, see Table 2.
Table 2:
Without cracks
Type of
mixture
mat
wsat
With cracks
mat
[kgm-3]
[%]
[kgm-3]
wsat
[%]
BI
2423
2760
2856
162.2 16.30
BII
2388
2647
95.1
9.80 2350
2713
133.2 13.40
BBI
2240
2652
2627
182.0 18.30
BBII
2210
2578
2669
174.8 17.40
PI
1988
2809
3068
375.8 36.70
PII
1991
2779
3088
355.7 34.60
The highest bulk density exhibited the hardened HPC mixture BI. The second
highest had the BII mixture without microsilica. Nearly the same values had
mixtures denoted BBI and BBII, cement pastes had the lowest values. The bulk
density of cement mixtures with cracks was lower then of the same without
cracks. The matrix densities differed only in the range of about 5% and
increased about 10% in the case of mixtures with cracks compared to mixtures
without cracks. The highest vacuum saturated moisture content had the cement
pastes PI and PII, where the difference was within the error range, and increased
about 30 % in the case of mixtures with cracks compared to mixtures without
cracks. The most porous materials were cement pastes PI and PII, the presence of
microsilica and admixtures did not have a significant effect. The porosity
increased from 3 to 6% in the case of mixtures with cracks compared to mixtures
without cracks.
The results of measurements of water and water vapour transport parameters
of the studied hardened cement mixtures are presented in Tables 4 and 5.
The water vapour diffusion coefficient was affected by the addition of
microsilica only for cement paste PI in a remarkable way where it decreased
three to four times. However, for the mixture BII its effect was within the error
range of the measuring method. The cracks caused dramatic increase of the water
vapour diffusion coefficient in all cases.
The effect of microsilica on the liquid moisture transport parameters was
more pronounced than on water vapour transport. The water sorptivity for the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
415
basic mixture BI decreased five times due to the microsilica addition, and for the
cement paste PI by about 30%. The moisture diffusivity of mixtures with
microsilica decreased in a corresponding way. In the case of cement mixtures
with cracks the moisture diffusivity increased one order of magnitude compared
to mixtures without cracks.
Table 4:
Type
of
mixture
BI
BII
BBI
BBII
PI
PII
-6
D [10 m s ]
9725%
RH
1.06
1.10
0.287
0.288
0.136
0.357
Table 5:
Type of
mixture
BI
BII
BBI
BBII
PI
PII
2 -1
597%
RH
0.397
0.375
0.132
0.151
0.079
0.291
[-]
9725%
RH
22.5
20.9
80.3
81.4
170.7
65.4
597%
RH
59.3
61.4
175.4
152.1
295.9
81.7
With cracks
D [10-6 m2s [-]
1
]
97597525% 97% 25% 97%
RH
RH
RH
RH
1.48 1.55 15.8 15.4
2.22 1.76 10.4 13.2
1.73 1.96 13.8 12.3
2.11 2.23 11.0 10.6
2.29 3.20 10.2
7.2
2.22 1.76 10.4 13.2
Without cracks
A [kg m-2s-1/2] app [m2s-1]
3.63E-03
8.9E-10
1.80E-02
3.3E-09
1.76E-02
1.5E-08
4.04E-02
8.3E-08
2.36E-02
7.0E-09
3.52E-02
1.8E-08
With cracks
A [kg m-2s-1/2] app [m2s-1]
4.84E-02
9.2E-08
4.23E-02
1.1E-07
8.08E-02
2.5E-07
6.21E-02
2.0E-07
2.77E-01
5.5E-07
1.59E-01
2.0E-07
The effect of aggregate presence was found to be quite different for liquid
water transport than for water vapour transport. The water sorptivity was for the
basic hardened HPC mixture BI several times lower than for the cement pastes.
On the other hand, the water vapour diffusion coefficient of cement pastes was
always lower than for the basic hardened HPC mixtures. The 5-97% RH values
of the water vapour diffusion coefficient were always lower than the 97-25% RH
values, which is in accordance with the previous measurements on many other
materials including concretes and cement pastes.
Figure 2a-f shows the measured adsorption (lower curves) and desorption
(upper curves) isotherms. We can see that cement pastes had higher water
adsorption capacity than the basic hardened HPC mixtures. This is an expected
behaviour due to absence of the inert aggregates in cement pastes. The presence
of microsilica slightly increased the water adsorption capacity of basic hardened
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
0.06
BI without cracks
BI with cracks
0.04
0.02
0.02
0
0
0.2
0.4
0.6
0.8
0.2
0.4
(a)
0.8
(b)
0.02
0.02
0
0
0.2
0.4
0.6
0.8
0.2
0.4
0.6
(c)
(d)
0.14
0.12
0.12
moisture content [kg kg-1]
0.1
moisture content [kg kg-1]
0.8
0.06
0.06
0.08
0.06
0.04
PI without cracks
0.02
PI with cracks
0.1
0.08
0.06
0.04
PII without cracks
0.02
0
0
0.2
0.4
0.6
0.8
0.2
(e)
Figure 2:
0.6
0.4
0.6
0.8
(f)
Sorption isotherm of mixture (a): BI, (b): BII, (c): BBI, (d): BBII,
(e): PI and (f) PII.
Conclusions
The experimental work performed in this paper has shown that the presence of
cracks in hardened HPC mixtures dramatically changed all measured parameters.
Also the effect of both microsilica and aggregates on the moisture transport and
storage parameters was very important. The liquid water transport was found to
be affected by cracks, microsilica and aggregates in a much more significant way
than water vapour transport.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
417
Acknowledgement
This research has been supported by the Ministry of Education of Czech
Republic, under grant No. MSM: 6840770003.
References
[1]
[2]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
419
Abstract
This paper presents the results of an experimental study to evaluate the
effectiveness of using FRP wrapping for repair and seismic upgrading of square
and rectangular RC columns in buildings. The factors affecting the performance
of FRP wrapping in rectangular columns under the action of both axial and
combined axial and lateral loads are considered in this paper. These factors
include the rectangularity ratio of the column cross section, the thickness of the
FRP jacket, and the use of carbon verses glass FRP for column jacketing.
Techniques to improve the performance of strengthening rectangular columns
were also proposed and evaluated in the paper. Such techniques include rounding
the sharp edges of the columns, and transferring square into circular columns
using mortar. A total of fourteen half-scale reinforced concrete columns, divided
into two groups, are tested in this research. The first group of columns consists
of three square columns and three rectangular columns which are tested under
axial loads. The second group consists of four square columns and four
rectangular columns tested under combined axial and lateral loads.
Keywords: seismic upgrading, column strengthening, CFRP and GFRP.
Introduction
The use of FRP wrapping for repair, strengthening, and seismic upgrading of
columns has gained increasing attention in recent years [13]. FRP wrapping
was proposed for increasing the ductility of column under axial and axialWIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06041
The test program in this research aims at studying the various parameters
affecting the behaviour of reinforced concrete rectangular columns strengthened
using FRP wrapping under both axial and seismic forces. These parameters
include the rectangularity ratio of the column cross section, the thickness of the
FRP jacket, and the use of carbon verses glass FRP for column jacketing. A total
of fourteen half-scale reinforced concrete columns, divided into two groups, are
tested in this research. The first group of columns, Group I, consists of three
square columns; CS1, CS2 and CS3; and three rectangular columns; CR1, CR2
and CR3; which are tested under increasing axial loads to failure. The second
group, Group II, consists of four square columns; CS4, CS5, CS6 and CS7; and
four rectangular columns; CR4, CR5, CR6 and CR7; tested under combined
axial and lateral loads. Constant axial loads, 350 kN for square columns and 700
kN for rectangular columns, were applied to Group II columns during increasing
cyclic lateral loading up to failure. This load level is approximately 30% of the
axial load carrying capacity of the tested columns. The details of the test
specimens and program are summarized in Table 1.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Table 1:
Testing load and set-up
Group I:
Axial load
Specimen
No.
Square Columns
CS1
Control
200200 mm
CS2
CFRP
CS3
CFRP
curving
sides
CR1
Control
CR2
CFRP
CR3
CFRP
Square Columns
CS4
Control
200200 mm
CS5
GFRP
CS6
CFRP
CS7
2-layers
CFRP
CR4
Control
CR5
GFRP
CR6
CFRP
CR7
2-layers
CFRP
6/150 mm
FRP
1500
mm
412
Rectangular Col.
Designation
400200 mm
6/150 mm
612
Group II:
Lateral cyclic
loading with
constant axial
load
6/150 mm
412
1600
mm
FRP
Rectangular Col.
750
mm
400200 mm
6/150 mm
400
mm
Table 2:
612
Property
Tensile strength in fibre direction (MPa)
Elongation at breaking (%)
Tensile modulus (GPa)
Nominal laminate thickness (mm)
Fabric width (mm)
GFRP
575
2.2
26.1
0.17
300
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
CFRP
965
1.33
73
0.13
350
421
Steel cap
LVDT
LVDT &
load cell
Steel cap
a) axial load
Figure 1:
b) Failure of CR1
b) Failure of CR6
423
60
50
40
CR2
CS1
CS2
CS3
30
20
10
0
0
0.5
1.5
2.5
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
60
50
40
CR2
CS1
CS2
CS3
30
20
10
0
0
0.5
1.5
2.5
Figure 5:
60
CR1
CR2
CR3
(CR1)"
(CR2)"
50
(CR3)"
40
30
20
10
0
0
Figure 6:
0.5
1.5
2.5
60
50
CR1
(CR1)
40
(CR2)"
CR2
CR3
(CR3)"
30
20
10
0
0
0.5
1.5
2.5
Figure 7:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
425
Group I specimens are tested under increasing axial load to failure. The square
column control specimen formed inclined cracks at an average stress of
30.2 N/mm2 followed by fast progressive failure in the form of falling off the
concrete cover and buckling of longitudinal reinforcement. As for specimen
CS2, and CS3, the confinement provided by the CFRP jacket delayed the failure
to higher stresses values, 48.2 N/mm2 for CS2 and 48.9 N/mm2 for CS3. At
failure the CFRP jacket was ruptured due to hoop tensile stresses after which
buckling of steel longitudinal bars occurred. The rectangular specimens showed
similar behaviour up to failure except that the effect of confinement due to the
CFRP jacket was less pronounced. Figures 47 show the variation of axial strain
and lateral strain with axial stress for square and rectangular column specimens
respectively. Figure 8 compares the average axial stress at failure for axially
loaded specimens. FRP confinement creases the ultimate capacity of axially
loaded square columns by approximately 50%. The confinement effect was less
pronounced for rectangular columns where the increase in the ultimate capacity
was about 21%. Transferring square column to circular one slightly increase the
average failure stress by 1.5%, however the overall load capacity of the column
is increased by 5.2% from 193 kN to 203 kN due to increasing the cross section.
Average stress at failure (N/mm2)
60
48.25 N/mm2
50
48.88 N/mm2
38.875 N/mm2
40
35.625 N/mm2
30.75 N/mm2
30.2 N/mm2
30
20
10
CS1
Figure 8:
CS2
CS3
CR1
CR2
CR3
Group II specimens were tested under constant axial load (30% of the axial
load capacity) and increasing cyclic load to failure. Figure 9 shows the loaddisplacement hysteresis loops for specimen CR6 during testing. All specimens in
this group showed similar behaviour. As load increased cracking started first at
the base, and then more cracks started to develop within approximately 450 mm
from the base. With further increasing load, crack opened and concrete in
compression crushed forming plastic hinge. Figure 10 shows the envelope for
lateral load variation with lateral displacement for square specimens. The
variation of lateral load with strain measured on steel stirrups for square
specimens is displayed in Figure 11. The load-displacement envelope for
rectangular columns is shown in Figure 12. The ultimate lateral load increased
by 11%, 22%, and 44% for square column wrapped by GFRP, CFRP and two
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
200
100
0
-40
-30
-20
-10
10
20
30
40
-100
-200
-300
-400
Displacement (mm)
Figure 9:
100
CS4
CS5
CS6
CS7
50
0
-50
-40
-30
-20
-10
10
20
30
40
50
-50
-100
-150
Figure 10:
100
CS4
CS5
CS6
CS7
50
0
0
0.5
1.5
2.5
3.5
-50
-100
-150
Figure 11: Lateral load vs. lateral strain on steel stirrups for square columns.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
427
400
300
CR4
CR5
CR6
CR7
200
100
0
-40
-30
-20
-10
10
20
30
40
-100
-200
-300
-400
Figure 12: Lateral load vs. lateral displacement envelope for rectangular
columns.
350
300 kN
300
320 kN
270 kN
240 kN
250
200
130 kN
150
100
90 kN
100 kN
110 kN
50
CS4
CS5
CS6
CS7
CR4
CR5
CR6
CR7
References
[1] Lehman, D.E., Gookin, S.E., Nacamuli, A.M. & Moehle, J.P, Repair of
earthquake-damaged bridge columns. ACI Structural journal, 98(2), pp.
233-242, 2001.
[2] Mosallam, A.S. (ed.). Innovative System for Seismic Repair and
Rehabilitation of Structures Design and Application, Proc. (SRRS2),
Technomic Publishing co., Inc., 2000.
[3] Tan, K.H. (ed.). Fibre-Reinforced Polymer Reinforcement for Concrete
Structures, Proc. of the Sixth International Symposium on FRP
Reinforcement for Concrete Structures, Singapore, 8-10 July, 2003.
[4] Fam, A.Z. & Rizkalla, S.H., Confinement model for axially loaded concrete
confined by circular fiber-reinforced polymer tubes. ACI Structural journal,
98(4), pp. 451-461, 2001.
[5] Wang, Y.C. & Restrepo, J.I., Investigation of concentrically loaded
reinforced concrete columns confined with glass fibre-reinforced polymer
jackets. ACI Structural journal, 98(3), pp. 377-385, 2001.
[6] Chaallal, O. & Shahawy, M., Performance of fiber-reinforced polymerwrapped reinforcement concrete column under combined axial-flexural
loading. ACI Structural journal, 97(4), pp. 659-668, 2000.
[7] Mahmoud, Kh., Fouad, E., Ramadan, M.O. & Abd-Elalim, A., Behaviour
of axially loaded square RC column confined with sandwich FRP wraps.
Proc. International Conference: Future Vision and Challenges for Urban
Development, Paper SG96F, Cairo, 20-22 December 2004.
[8] Allam, H.M., Strengthening of square columns by a New Technique. Proc.
International Conference: Future Vision and Challenges for Urban
Development, paper SG167F, Cairo, 20-22 December 2004.
[9] Haroun, M. A. & Elsanadedy, H.M., Seismic retrofit of shear-deficient
reinforced concrete bridge columns by advanced composite jackets. Proc.
Structural Composite for Infrastructure Applications, Aswan, Egypt, 2002.
[10] Haroun, M.A., Mosallam, A.S., Feng, M.Q. & Elsanadedy, H.M.,
Experimental Investigation of Seismic Repair and Retrofit of bridge
columns by composite jackets. Journal of Reinforced Plastics and
Composites, 22(14), pp. 1243-1268, 2003.
[11] Hosny A., Shahin, H. Abdelrahman, A., & El-Afandy, T., Uniaxial tests on
rectangular columns strengthened with CFRP. Proc. Structural Composite
for Infrastructure Applications, Aswan, Egypt, 17-20 Dec., 2002.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
429
Abstract
Concrete is the most commonly used building material. It is used for buildings,
industrial structures, bridges and dams. Every day the concrete is being
improved, to achieve better characteristics, reduce price and to be environmental
acceptable. In the introduction an historical overview of concrete is given
beginning in the 1950s when the compression strength was 40 N/mm2. Today
this is the Industry standard. Concrete with compression strength greater than 40
N/mm2 is called high performance concrete (HPC). The first HPS concretes
were made in 1960. Around 1990, Reactive Powder Concrete (RPC) started to
appear. Their strengths go up to 800 N/mm2. Today, HPC is limited to bridges,
and RPC is still very rarely used. In this work we present a method for making a
RPC composite material with a compression strength of up to 170 N/mm2.
Preparation and testing of material is performed in the laboratory of the Faculty
of Civil Engineering in Zagreb. In addition to the mechanical properties, the
durability parameters are also tested. Detailed concrete mix proportions are given
in the article. The possibility that RPC could be used to construct an arch bridge
over Bakar strait (Croatia) is analyzed. The bridge would have arch span of 432
m. Due to very high compression strength of RPC the span length may be
increased, therefore reducing the total construction cost. RPC is also used to
increase the resistance to freeze-thaw cycles, increase abrasion resistance, and
reduce chloride permeability. Many large Adriatic bridges have experienced
durability problems that are primarily the result of using standard concrete. Each
of these durability enhancements provided by RPC decrease maintenance costs
and lengthens the service life of a structure, which is vital for bridges in the
Adriatic region.
Keywords: arch bridge, reactive powder concrete (RPC), construction
technology.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06042
Introduction
Cement is around 12 millions years old. Reactions between limestone and oil
shale during spontaneous combustion occurred in Israel to form natural deposits
of cement compounds. The deposits were first described by Israeli geologists
between 1960 and 1970. In 3000 years BC, the Egyptians used mud mixed with
straw to bind dried bricks. They also used gypsum mortars and mortars of lime in
the construction of the pyramids. Chinese used cementitious materials to hold
bamboo together in their boats and in the Great Wall of China. The Romans used
pozzolana cement from Pozzuoli, Italy near Mt. Vesuvius to build the Appian
Way, Roman baths, the Coliseum and Pantheon in Rome, and the Pont du Gard
aqueduct in southern France. They used lime as a cementitious material. Pliny
reported a mortar mixture of 1 part lime to 4 parts sand. Vitruvius reported a 2
parts pozzolana to 1 part lime mixture. Animal fat, milk, and blood were used as
admixtures (substances added to cement to improve properties).
Figure 1:
Maxentius basilica.
Fig. 1 shows the Maxentius basilica built in the 4th Century. These materials
were not used again until the beginning of the 19th Century. In 1824, Joseph
Aspdin of England invented Portland cement by burning finely ground chalk
with finely divided clay in a lime kiln until carbon dioxide was driven off. The
sintered product was then ground and he called it Portland cement named after
the high-quality building stones quarried at Portland, England. Portland cement
started the era of modern composition cements. In 1828, I. K. Brunel made the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
431
2.1 Introduction
This article presents the possibility of making RPC concrete with compression
strengths up to 200 N/mm2. Four different mixtures are analyzed. The first is
mixture of hybrid micro-fiber concrete; the others are composed of only one type
of fibers.
2.2 Cement
As the class of cement increases the compression strength increases. For this
mixture we have selected the Portland cement PC 55 with no mineral
ingredients.
2.3 Fine aggregate
Aggregate that is used for the making of this mixture is quartz aggregate. Two
fractions of this material are used: one with soil size of 0.1250.25 mm, other
with 0.250.5 mm, This effectively means that maximal soil size is 0.5 mm.
2.4 Steel fibers
Two different types of steel fibers are used (shorter and longer fibers). Shorter
fibers are 13 2 mm long, diameter is 0.2 0.02 mm. Minimal tensile strength is
2600 MPa. Longer fibers have curved ends; their length is 40 3 mm and
diameter 0.5 0.02 mm. Minimal tensile strength is 2600 Mpa. In both cases, the
high tensile fibers are used to achieve necessary ductility.
2.5 Silica fume
Silica fume is a pozzolanic additive, with specific area of 20 m2/g.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
2.23 g/cm3
5.6%
2.6 Superplasticizer
Superplasticizer is primary used to decrease the participation ratio of water in the
concrete mixture. Chosen superplasticizer is based on policarbon-silate. It is a
brown fluid, dissolves in water and doesnt contain chlorides.
Other parameters are:
- density
1.064 kg/dm3
- pH value
7
- alkalinity
0.31 %
- viscosity on 20C
134 mPas
Table 1:
Mixture number
M1
M2
M3
M4
Steel fibers
(kg/m3)
SF1 (40/0.5)
SF2 (13/0.2)
76
190
228
228
234
720
955
720
980
230
239
230
303
123
1112
105
945
123
1111
105
965
Superplasticizer
(kg/m3)
30
35
31
40
Water (l/m3)
190
215
190
209
24C
5%
2.41 kg/m3
140 mm
25C
5%
2.35 kg/m3
250 mm
24.5C
5%
2.36 kg/m3
190 mm
26C
5%
2.306kg/m3
220 mm
Cement (kg/m3)
Fine aggregate
(kg/m3)
Quartz sand
0.1250.25
0.250.5
(kg/m3)
Concrete
properties
temperature
pores
density
consistency
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
433
Experimental results
M1
M2
46.9
42.8
132.0
155.6
M3
M4
42.8
48.8
153.3
174.8
Figure 2:
Experimental results.
A sample
(kg/m2/s)
0
0.000613
0.00029
0.00022
0.000122
0.000089
0.0000526
0.0000314
0.0000182
0.0000133
0.0000105
0.00000888
0.00000451
0.00000242
B sample
(kg/m2/s)
0
0.000754
0.000306
0.000212
0.000122
0.0000896
0.0000518
0.0000310
0.0000181
0.0000133
0.0000107
0.00000888
0.00000451
0.00000251
C sample
(kg/m2/s)
0
0.00066
0.000291
0.000198
0.000117
0.0000864
0.0000511
0.0000302
0.0000176
0.0000131
0.0000103
0.00000864
0.00000439
0.00000241
435
shorter. The bridge arch span would be 423 m, vertical clearance is 72 m. Bridge
has 22 piers (spans are 30+20x38+30 m). Fig. 3 represents the longitudinal
section of the bridge. A cross section of the bridge is shown on fig. 4. The cross
section is constant and aerodynamic. With the exception of the foundations,
abutments, and highest piers, the whole bridge will be build from pre-cast
segments 3.8 m long. To achieve symmetric load bearing structure will be
supported from both sides. Segments that weight 718 kN are being composed
and pre-stressed in pressurising station behind the abutments and then pushed.
Computer simulation of the Bakar bridge is presented in fig 5.
Figure 3:
Figure 4:
Figure 5:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Basic design
Calculations are made for stresses during usage, strains during construction,
stresses and strains of bearing structure during construction and usage. With the
exception of dead and traffic load these calculations cover temperature load,
creep, lateral wind, and earthquake load. Table 4 shows maximal stresses for
each part of the bridge. Road slab and transverse beams are the only parts of the
structure where tensile stresses are about 15 MPa. Due to very high stresses,
passive reinforcement with ropes will be used. In transverse beams internal prestressing with addition of passive reinforcement will be used. This conception is
not very often with RPC and in future phases of the project it should be carefully
considered.
Table 4:
Part of structure
Road slab
Transverse beam
Highest pier
Arch
Maximal stresses.
Conclusion
In this article we have presented the possibility that a future arch bridge over the
Bakar strait (Croatia) could be made from RPC. Due to the very high
compression strength of RPC it can be used for large spans. RPC also has superb
durability parameters as well as a high abrasion resistance and reduced chloride
permeability. This makes RPC an ideal material for bridges on the Adriatic cost
because durability problems are primary related to the fact that
the
reinforcements protective layer is rapidly being destroyed. Winds that reach up
to 250 km/h drift large amounts of chlorides that destroy the bridge structure,
primary the arch and the columns. Each of these durability enhancements
provided by RPC decrease maintenance costs and lengthen the service life of a
structure, which is vital for bridges in the Adriatic region.
References
[1] Candrlic, V., Concrete arch bridge over Bakar straits. Proceedings for
Conference of Croatian builders, eds. V. Simovic, Cavtat, pp. 358-364,
2001.
[2] Jagar, A., High performance concrete, Faculty of civil engineering: Zagreb,
2003.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
437
[3] Mrakovcic, S., Precast arch bridges made of reactive of powder concrete,
Faculty of civil engineering: Zagreb, 2001.
[4] Edward Nawy, G., Fundamentals of high-performance concrete, John Wiley
& Sons: New York, 2001.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
439
Abstract
The elastic and inelastic seismic response of 3, 9, 17 and 25 levels of reinforced
concrete buildings (offices), located in the soft soil of Mexico City, with normal
resistance concrete (fc= 250 kg/cm2) and with high resistance concrete (fc= 700
kg/cm2) are compared. The design with normal and high resistance concrete is
made with the RDF-93 and RDF-04 codes respectively. The design results
(transversal sections dimensions, vibration periods, lateral displacements,
reinforcement steel, etc) are compared, after making a spectral modal dynamic
analysis, as well as the non-linear responses (lateral displacements, global and
local ductility demands, global distribution of plastic hinges, etc) from the
step-by-step dynamic analysis with the SCT-EW record of the 1985 earthquakes.
Introduction
441
Figure 1:
Structures description
Two kinds of structures are considered: A case (normal concrete with fc= 250
kg/cm2 and B case (high resistance concrete with fc= 700 kg/cm2). The elasticity
modulus for both cases is Ec= 14000 (fc)1/2 (221,359 kg/cm2) and Ec= 7700
(fc)1/2 + 163000 (366,723 kg/cm2), respectively. The concrete is class 1 with
volumetric weight c= 2400 kg/cm2 and = 0.2; the reinforced steel used has a
yield stress fy= 4200 kg/cm2.
Figs. 2 and 3 show the type plant and transversal cuts of 3 and 9 level
buildings; fig. 4 shows the type plant of 17 and 25 level buildings. Also, the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
800
800
800
800
800
250
800
250
The gravitational loads effects and those of second order (P-) are included in
the analysis. For A case the seismic design spectra of soft soil is considered
(Zone III) of the RDF-93 and the Zone IIIb spectra of the RDF-2004 for the B
case, considering the seismic behavior factor Q= 3 (see fig. 5).
The reinforcement steel areas of the structural elements design (beams and
columns) is made with the last mechanical elements obtained from the structural
analysis for the critical load combination; it is according to the general
requirements and ductile frames of the Concrete Norms. These designs include
the load factor and the resistance reduction factor.
Figure 2:
SECONDARY
BEAMS
VIGAS
SECUNDARIAS
(CENTIMETERS)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
150
2
800
3
800
4
800
800
800
800
800
E
VIGAS
SECUNDARIAS
SECONDARY
BEAMS
(CENTIMETERS)
astic
responses
ROOF
3.50
s7 of the3.50
structural elements of the transversal sections in B cas
is6 reduction is more important in columns, so that the resistanc
mpression3.50in the concrete has a bigger influence in the behavior o
h5igh axial loads. In the majority of B case elements, the require
areas w3.50
ere smaller; only the some external axes columns, th
4
esult similar
in both cases.
3.50
3
3.50 s of vibration
ntal period
2
3.50
res the fund
amental periods of vibration of analyzed A and B cas
1
A case structures tend to be less rigid.
7.50
STREET LEVEL
(METERS)
3.00
CIM
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
443
800
800
800
800
800
800
D
VIGAS SECUNDARIAS
SECONDARY BEAMS
(CENTIMETERS)
Figure 4:
Sa/ g
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0
Ti [seconds]
RDF-93,Q= 1
Figure 5:
RDF-04, Q= 1
RDF-93, Q=3
RDF-04, Q=3
445
MODEL
3
9
17
25
0.80
0.80
0.69
0.70
0.74
0.63
1.49
1.45
1.07
1.44
1.41
1.09
1.91
1.91
1.38
1.84
1.84
1.42
2.10
2.10
1.39
1.88
1.88
1.38
CASE
A
B
A
B
A
B
A
B
Fig. 6 has the September 19th 1985 earthquake SCT-EW accelerogram, used
later in the inelastic analysis. Fig. 7 shows the elastic and inelastic response
spectra of this record ( = 5%) and of RDF-2004, as well as the fundamental
periods of vibration location (X direction) of all the structures, A and B cases.
When the level number increases, the difference between both cases tends to be
bigger.
0.20
0.15
0.10
0.05
Sa/g
a/ g
0.00
-0.05
-0.10
-0.15
-0.20
0
10
15
20
25
30
35
40
45
50
tiempo [s]
Time (seconds)
Figure 6:
th
1.20
1.10
1.10
1.00
1.00
A case
B case
0.90
0.80
0.80
0.70
0.70
Sa/g
Sa/g
0.90
0.60
0.60
0.50
0.50
0.40
0.40
0.30
0.30
0.20
0.20
0.10
0.10
0.00
0.00
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0.0
0.5
1.0
1.5
Period (seconds)
duct=1
duct=2
Figure 7:
duct=3
duct=4
RDF-04 Q=1
2.0
2.5
3.0
3.5
4.0
Periodo [s]
Period
(seconds)
Periodo [s]
RDF-04 Q=3
duct=1
duct=2
duct=3
duct=4
RDF-04 Q=1
RDF-04 Q=3
MODEL
3
9
17
25
max [cm]
5.48
5.64
41.12
43.06
51.00
48.98
77.26
76.73
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
G
1.41
1.91
2.34
1.96
2.01
2.04
2.19
1.54
447
AXE
CASE
Elas.
Inel.
Elas.
Inel.
Elas.
Inel.
Elas.
Inel.
Elas.
Inel.
Elas.
Inel.
Elas.
Inel.
Els.
Inel.
A
3
3
A
9
C
A
17
B
A
25
B
Level
V [t]
203
164
56
52
525
331
518
329
1792
424
1757
407
3277
921
3229
910
Level
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
-8
-6
-4
-2
Figure 8:
-6
-5
-4
-3
-2
-1
B case
A case
B case
4.4 Local ductility maximum demands (L) in beams and columns and
global distribution of plastic hinges
The 3 level building does not present yield in columns. The beams in all
buildings have very similar responses, independently of the case type; the beams
maximum demands are rarely bigger in A case. The L maximum values in
beams as in columns are obtained in the 9 level building, but inside the
admissible limits from the practical point of view. Fig. 8 compares the local
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusions
Use of high resistance concretes (fc > 400 kg/cm2) in reinforced concrete
columns improves in a great deal its axial load capacity; therefore in front of
actions where compression high loads predominate its behavior is very adequate;
nevertheless, in front of less magnitude loads or even tensions its use results
almost no efficient because the resistance is given mainly by the reinforcement
steel. The most part of the energy dissipation is in the beams, no matter what
type of structure it is. The A case structures present a bigger amplitude in the
hysteretic cycles of the ratios of basal shear force-roof lateral displacement, with
bigger dissipated energy amount for inelastic deformations. The maximum
values of local ductility demands are found inside the permissible by the RDF-93
and RDF-2004. The inelastic response is considerably reduced regarding to the
elastic one, except in the 3 level building, where the elastic and inelastic
responses tend to be practically the same. In every analyzed case, reviewing the
shear forces history, the corresponding resistance is never reached; this is, there
is always a resistance reservation, which is bigger in B case buildings, so that for
this case a bigger quantity of transversal reinforcement by confinement was
required because the design conditions for transversal reinforcement must be
more strict as the fc value increases.
References
[1]
[2]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
449
Abstract
The influence of several kinds of synthetic zeolite as mineral admixtures on the
workability and performance strength of mortars cement and concrete has been
investigated. In the first series of experiments zeolites have been used to replace
10% of cement in the preparation of mortars. The second series of experiments
have regarded the preparation of high performance concrete. In order to evaluate
the influence of alkali cations, two different forms of zeolite A have been used
with one sodium (NaA or 4A) and the other with calcium (CaA or 5A) and also
to evaluate the filling effects of particles zeolites 4A and 5A with different
crystals sizes have been used. Furthermore we have used three types of cement
with a different content of clinker: CEM I 42.5R, CEM II/A-S 42.5 R and CEM
III/A-S 42.5 N. Every prepared sample has been water-cured at 20C for 2, 7 and
28 days. The results of samples (mortars and concrete) made with zeolites were
compared with those obtained with silica fume, usual fine material used in the
preparation of high performance strength concrete.
Keywords: admixtures, zeolites, mortars, concrete, compressive strength.
Introduction
Experimental
2.1 Materials
- Portland cement CEM I 32.5 R with content of clinker >95% (UNI EN 197/1).
- Portland- slag cement CEM II/A-S 42.5 R with content of clinker between 80
and 94% and content of slag ranging between 6-20%( (UNI EN 197/1).
- Blastfurnace cement CEM III /A-S 42.5 N with content of clinker between 35
and 94% and slag ranging between 6 and 65 % (UNI EN 197/1).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
451
Four different synthetic zeolites were prepared: two of which having small
particle size:
- zeolite 4A with size of crystals ranging about 1-5 m (NaAL) (Fig.1).
- zeolite 4A with size of crystals ranging about 0.2-1 m (NaAS) (Fig.2).
- zeolite 5A with size of crystals ranging about 1-5 m (CaAL) (Fig.1).
- zeolite 5A with size of crystals ranging about 0.2-1 m (CaAS) (Fig.2).
Silica Fume (SF) (trade name MAPEPLAST SF) produced and supplied by
MAPEI (Italy).
Coarse Aggregate rounded with 25 mm maximum diameter in accordance with
UNI 8520-2.
Fine Aggregate is standard sand in accordance with UNI EN 933-3.
The Superfludifiant used was acrylic polymers without formaldehyde with 30.5
solid content (trade name Dynamon SP1) produced and supplied by MAPEI
(Italy).
2.2 Preparation of zeolites
The small crystals of zeolites 4A were prepared starting from the following
initial reaction mixture: 2.37Na2O 1Al2O3 1.90 SiO2 62 H2O
The synthesis procedure was the following: sodium aluminate was added to a
solution of sodium hydroxide solution and after the homogenisation the silica
source was added. The solution was stirred for 15 min at room temperature and
then submitted to hydrothermal treatment, in static conditions, at 25C for about
7 days.
The large crystals of zeolites 4A were prepared starting from system having
the composition: 3.2 Na2O 1Al2O3 1.90 SiO2 95 H2O. The reagents admixture
was made as described above. Successively, the solution obtained has been
submitted to hydrothermal treatment at 95 C for 2 h, in stirred conditions.
After the hydrothermal treatment every samples are recovered, washed with
distilled water, dried at 110C for 12 hours and successively equilibrated at room
temperature before X-ray and microscope analysis.
The X Ray analysis confirms the presence of only zeolite A in all samples
obtained.
Through microscope analysis it is possible to evaluate the different
dimensions of crystals obtained (Fig.1, Fig.2).
The zeolites 5A were prepared by ionic exchange of zeolite 4A, 250g of 4A
have been contacted with 1liter of a 1M aqueous solution of calcium nitrate at
room temperature. This exchange procedure was applied three times. The
efficiency of exchange has been checked by EDAX (Energy Dispersive X-Ray
Analysis), the zeolite 5A obtained has a ratio of ionic content (Ca++/ (Ca+++
Na+)) of about 0.9.
2.3 Preparation of mortars
Cement mortars were prepared for determination of compressive strength of the
mixtures. Zeolites were used as direct replacement of cement on a weight to
weight basis at the levels of 10%. The sand/cement ratio was 3 and the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 1:
50
175
500
Fine
aggregate
693
Course
Aggregate
1040
Superfluidifiant
5
453
Cement
Zeolites
2 days
7 days
28 days
CEM I
-NaAS
NaAL
CaAS
CaAL
SF
-NaAS
NaAL
CaAS
CaAL
SF
-NaAS
NaAL
CaAS
CaAL
SF
18.9
17.7
21.9
26.1
25.9
20.3
22.5
20.6
28.0
26.8
30.0
21.1
15.7
15.7
20.4
20.1
22.3
13.5
29.2
21.2
30.1
34.5
33.4
29.5
32.6
28.5
34.3
35.7
37.2
28.6
23.5
20.9
26.7
30.4
32.6
22.9
34.0
25.3
31.3
36.2
39.1
36.4
42.8
39.9
37.7
42.7
42.8
40.5
42.5
23.9
30.5
38.0
40.8
39.0
CEM II/A-S
CEM III/A
Figure 2:
slum p [m m ]
250
200
CEM I
150
CEM II
100
CEM III
50
0
--
Figure 3:
SF
According to these results, most of the concrete made with zeolites CaAs and
CaAL, had a high slump and were workable. In particular, the fresh concrete
made with zeolites are more workable than that made with silica fume. This is
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
455
Figure 5:
60
50
40
30
20
10
0
10
20
30
Days
Figure 4:
70
70
60
50
40
30
20
10
0
10
20
30
Days
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 6:
70
60
50
40
30
20
10
0
10
20
30
Days
The concrete made with zeolites CEM I and both CAS and CAL exhibits
comparable values of compressive strength at 2, 7 and 28 days. They do not have
relevant differences caused by the size of the particles, probably due to the fact
that at a ratio a/c= 0.35 the pozzolanic effects of zeolites are more important than
the effects of particles packing. Also it is possible to notice at early ages a more
increment of compressive strength for samples prepared with zeolites instead of
silica fume, caused by accelerating effects of zeolites. For the concrete made
with CEM II/A-S the addition of zeolites does not improve significantly the
compressive strength at 28 days, the greater increment is observed by addition of
silica fume. It is probably due to more finesse of CEM II/A-S, that involves a
greater mechanical resistance of the samples without admixtures, in fact at 28
days we have obtained compressive strength a little advanced with the fine
additions as silica fume.
The concretes made with CEM III/A-S and zeolites CaA, exhibit an
increment of resistance with respect to concrete without admixture of about 9 %.
The greater increments are obtained at the early ages, caused by the alkali
content of zeolites that accelerate the process of activation of the fly ash.
As expected, in agreement with the results of mortars, the strength of the
concretes made with zeolite NaA, was less than all the other mixes at all ages
thus reflecting the fact that the sodium content influences negatively the
development of hardening properties of cement concrete. The increase in
strength in systems of concrete containing zeolite CaAS and CaAL can be
explained in a way very similar to the strength increase in mortars mixes. The
aggregate-matrix bond improvement induced by zeolites materials is probably
the result of a combined filler and pozzolanic effect. The filler effect leads to
reduction in porosity of transition zone and provides a dense microstructure and
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
457
thus increases the strength of the system. The pozzolanic effects help the
formation of bonds between the densely packed particles in the transition zone
through the pozzolanic reaction with the calcium hydroxide liberated during the
hydration of Portland cement to form extra binding calcium silicate hydrates
which lead to further increase in the strength.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
Ortega E.A., Cheeseman C., Knight J., Loizidu M., Properties of alkaliactivated clinoptilolite, Cement Concrete Research 30 (10), pp. 16411646, 2000.
Poon C.S., Lam L., Kou S.C., Lin Z., A study on the hydratation rate of
material zeolite blended cement pastes, Construction and building
materials, 13 (8), pp. 427-432, 1999.
Fragoulis D., Chaniotakis E., Stamatakis M.G., Zeolitic tuffs of Kimolos
Island, Aegean Sea, Greece and their industrial potential, Cement
Concrete Research, 27 (6), 889-905, 1997.
Shi C., Day R.L., Pozzolanic reaction in the presence of chemical
activators: Part II Reaction products and mechanism, Cement Concrete
Research, 30 (4), pp.607-613, 2000.
Chan S. Y.N., Ji X., Comparative study of the initial surface absorption
and chloride diffusion of high performance zeolite, silica fume and PFA
concretes, Cement Concrete Composites, 21 (4), pp. 293-300, (1999).
Taylor H.F.W., Cement Chemistry, Academic Press. London, 1992.
Naiqian F., Properties of zeolitic mineral admixtures concrete, in: S.L.
Sarkar & S. N. Gosh Eds, Mineral Admixtures in cement and concrete.
ABI Books. India pp. 393-447, 1993.
Sersale R., Frigione G., Portland-zeolite-cement for minimizing alkaliaggregate expansion, Cement Concrete Research, 17, pp.404-410, 1987.
Atkins M., Glasser F. P., Jack J.J., Zeolite P in cements: its potential for
immobilizing toxic and radioactive waste species, Waste Management, 15
(2), 127-135, 1995.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
[12]
[13]
[14]
Fu Y., Ding J., Beaudoin J. J., Zeolite-based additives for high alumina
cement products, Advanced Cement Basic Materials, 3 (1), pp. 3742,1996.
Gervais C., Ouki S. K., Performance study of cementitious systems
containing zeolite and silica fume: effects of four metal nitrates on the
setting time, strength and leaching characteristics, Journal of Hazardous
Materials 93 (2), pp.187-200, 2002.
Mazzasa F., Pozzolana and Pozzolanic Cements in P.C: Hewlett. Editors.
Leas Chemistry of Cement and concrete. Ed. Arnold, pp.241, 1995.
Su N., Fang H.-F., Chen Z H, Liu F.S, Reuse of waste catalysts from
petrochemical industries for cement substitution, Cement Concrete
Research, 30 (11), pp. 1773-1783, 2000.
Ujike I., Properties of concrete containing artificial zeolite made by
chemical conversion of fly ash in R. K. Dhir. M. J. Mccarthy Editors.
Concrete Durability and Repair Technology. Thomas Telford pp.123-131,
1999.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
459
Abstract
The typical behavioural curve of a reinforced concrete element with longitudinal
bars and stirrups under torsion comprises two distinct regions; the elastic part
until the first cracking and the part after cracking. The different character of the
response in these regions reveals the different nature of the load resisting
mechanism in each case. The present work addresses an approach that combines
two different analytical models in order to predict the entire torsional behaviour.
The prediction of the elastic part until the first cracking is achieved using a
smeared crack analysis for plain concrete in torsion, whereas for the description
of the post-cracking response the softened truss model is used. Further, the
results of an experimental investigation on the behaviour of 15 reinforced
concrete beams subjected to pure torsion are also presented in this paper. The
reported results include the behavioural curves and the values of the initial
torsional stiffness, the cracking torque moment and the ultimate torque capacity
of the beams. Analyses for the torsional behaviour of the tested beams using the
proposed approach were performed and analytical curves are produced and
compared with the experimental ones. A good agreement between predicted and
experimental results is observed.
Keywords: beams, reinforced concrete, smeared crack model, softened truss
model, torsion, tests.
Introduction
461
Experimental program
R6
RH4
RH6
RH8
Beam code
name
R4-20
R4-15
R4-10
R6-20
R6-15
R6-10
RH4-20
RH4-15
RH4-10
RH6-20
RH6-15
RH6-10
RH8-20
RH8-15
RH8-10
Longitudinal
bars
48
68
48
68
88
Stirrups
8/20 cm
8/15 cm
8/10 cm
8/20 cm
8/15 cm
8/10 cm
8/20 cm
8/15 cm
8/10 cm
8/20 cm
8/15 cm
8/10 cm
8/20 cm
8/15 cm
8/10 cm
fc (MPa)
fsp (MPa)
20.96
2.89
24.59
3.33
26.56
3.04
24.90
3.42
27.39
3.09
load cell
160
0
steel arm
tested beam
roller support
Figure 1:
Test setup.
In this study, for the prediction of the entire torsional behaviour of reinforced
concrete elements, the combination of two different theories is adopted. The
elastic till the first cracking part is described by a smeared crack analysis for
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
463
plain concrete in torsion [10] and the post-cracking part is described by the wellknown softened truss model [3, 4].
It is justified that for the elastic till the first cracking part the percentage of
steel has a minor effect on the torsional response and reinforced concrete
elements behave as plain concrete members [1]. Therefore, the analytical
smeared crack model for plain concrete in torsion proposed by Karayannis [10]
is applicable to reinforced concrete beams for the prediction of the first elastic
part till the developing of concrete cracking. This approach is using an efficient
numerical scheme for the torsional analysis of concrete that is initially based on
the elastic theory and utilizes a special numerical technique properly modified to
include the smeared cracking approach. The model is based on an analytical
technique that employs constitutive relations expressed in terms of normal stress
and crack width, for the behaviour of the crack process zones. Detailed
derivation of the equations and the solution technique of this theory can be found
in reference 10.
Table 2:
Beam
code
name
R4-20
R4-15
R4-10
R6-20
R6-15
R6-10
RH4-20
RH4-15
RH4-10
RH6-20
RH6-15
RH6-10
RH8-20
RH8-15
RH8-10
Initial stiffness
(10-3 kNcm2/rad)
exp .
exp. calc.
calc.
3771
3084
3229
3452
3557
3442
5913
6689
6186
6494
7215
6755
7744
7191
6658
3346
3624
6500
6292
6596
1.13
0.92
0.97
0.95
0.98
0.95
0.91
1.03
0.95
1.03
1.15
1.07
1.17
1.09
1.01
Cracking torque
(kNcm)
exp .
exp. calc.
calc.
217.05
201.25
200.79
240.12
259.63
264.60
322.00
364.54
418.78
378.70
355.10
372.10
345.27
330.37
368.30
204.65
235.89
337.82
379.97
343.39
1.06
0.98
0.98
1.02
1.10
1.12
0.95
1.08
1.24
1.00
0.93
0.98
1.01
0.96
1.07
Ultimate torque
(kNcm)
exp .
exp. calc.
calc.
238.49
264.87
325.40
287.28
318.38
374.22
394.76
501.27
583.42
481.14
586.91
661.63
503.69
611.95
694.97
263.26
277.52
292.30
288.75
322.81
346.65
439.99
500.89
525.01
444.21
509.88
541.61
462.20
551.14
603.72
0.91
0.95
1.11
0.99
0.99
1.08
0.90
1.00
1.11
1.08
1.15
1.22
1.09
1.11
1.15
Experimental results
2 8
8/s
400
Proposed approach
2 8
300
200
100
s = 20 cm
0
0.0
0.5
s = 15 cm
0.5
1.0
s = 10 cm
1.5 0.0
-3
0.5
1.0
1.5
600
Experimental results
2 8
500
8/s
2 8
Proposed approach
2 8
400
300
200
100
0
s = 20 cm
0.0
0.5
s = 15 cm
0.5
1.0
s = 10 cm
1.5 0.0
0.5
1.0
1.5
-3
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
900
Experimental results
2 8
800
465
700
Proposed approach
8/s
600
2 8
500
400
300
200
s = 20 cm
100
0
0.0
0.5
s = 15 cm
1.0 1.5
0.0
0.5
1.0
s = 10 cm
1.5 0.0
0.5
-3
1.0
1.5
1000
Experimental results
2 8
900
800
Proposed approach
2 8
8/s
700
600
2 8
500
400
300
200
s = 20 cm
100
0
0.0
0.5
1.0
s = 15 cm
1.5 0.0
0.5
1.0
s = 10 cm
1.5 0.0
-3
0.5
1.0
1.5
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Experimental results
2 8
900
800
2 8
700
2 8
600
2 8
8/s
Proposed approach
500
400
300
200
s = 20 cm
100
0
0.0
s = 15 cm
0.5
1.0
s = 10 cm
1.5 0.0
-3
0.5
1.0
1.5
Conclusions
The following concluding remarks are drawn from the results reported herein.
Volume of the transversal and the longitudinal reinforcement significantly
affects the torsional behaviour of the tested beams, as it was expected. Two
distinct regions can be observed in a typical experimental torque twist curve of
the tested reinforced concrete beams. The different character of the response in
these regions reveals the different nature of the load resisting mechanism in each
case. In order to describe the entire torsional behaviour of the tested beams and
based on the above observation, the combination of two different theories is
adopted. For the estimation of the elastic till the first cracking response, a
smeared crack analysis for plain concrete in torsion is used, whereas the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
467
References
[1]
Hsu, T.C., Unified Theory of Reinforced Concrete, CRC Press, Inc., Boca
Raton, Fla., 1993.
[2]
Mitchell, D. & Collins, M.P., Diagonal Compression Field Theory A
Rational Model for Structural Concrete in Pure Torsion, ACI Proceedings,
71(8), pp. 396-408, 1974
[3]
Hsu, T.C. & Mo, Y.L., Softening of Concrete in Torsional Members
Theory and Tests, ACI Proceedings, 82(3), pp. 290-303, 1985.
[4]
Hsu, T.C., Toward a Unified Nomenclature for Reinforced-Concrete
Theory, Structural Engineering, ASCE, 122(3), pp. 275-283, 1996.
[5]
Mansur, M.A., Nagataki, S., Lee, S.H. & Oosumimoto, Y., Torsional
Response of Reinforced Fibrous Concrete Beams, ACI Structural, 86(1),
pp. 36-44, 1989.
[6]
Wafa, F.F., Shihata, S.A., Ashour, S.A. & Akhtaruzzaman, A.A.,
Prestressed High-Strength Concrete Beams under Torsion, Structural
Engineering, ASCE, 121(9), pp. 1280-1286, 1995.
[7]
Rahal, K.N. & Collins, M.P., Simple Model for Predicting Torsional
Strength of Reinforced and Prestressed Concrete Sections, ACI Structural,
93(6), pp. 658-666, 1996.
[8]
Ashour, S.A., Samman, T.A. & Radain, T.A., Torsional Behavior of
Reinforced High-Strength Concrete Deep Beams, ACI Structural, 96(6),
pp. 1049-1058, 1999.
[9]
Chalioris, C.E., Study of the Behaviour and the Failure Mechanisms of
Plain and Reinforced Concrete Elements in Torsion, PhD dissertation,
Department of Civil Engineering, Democritus University of Thrace,
Xanthi, Greece, 1999.
[10] Karayannis, C.G., Smeared Crack Analysis for Plain Concrete in Torsion,
Structural Engineering, ASCE, 126(6), pp. 638-645, 2000.
[11] Karayannis, C.G., Nonlinear Analysis and Tests of Steel-fiber Concrete
Beams in Torsion, Structural Engineering and Mechanics, 9(4),
pp. 323-338, 2000.
[12] Karayannis, C.G. & Chalioris, C.E., Experimental Validation of Smeared
Analysis for Plain Concrete in Torsion, Structural Engineering, ASCE,
126(6), pp. 646-653, 2000.
[13] Karayannis, C.G. & Chalioris, C.E., Strength of Prestressed Concrete
Beams in Torsion, Structural Engineering and Mechanics, 10(2),
165-180, 2000.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
469
Abstract
The use of ferrocement, containing blended cement, to repair and enhance the
strength of brick masonry columns of old buildings and structures showing
distress has been investigated for use as a rehabilitation/strengthening material.
Intensive experimental study has been carried out on plain brick masonry
columns and brick masonry columns with single, double and triple layers of
ferrocement, with/without the use of chemical bonding agents. Use of blended
cements containing agricultural waste i.e. rice husk ash have not as yet been
investigated at all for use in ferrocement. Blended cement containing 75% OPC
and 25% rice husk ash has now been investigated for use in ferrocement.
Ferrocement containing blended cement has been observed to substantially
increase the load carrying capacity of brick masonry columns along with
decreasing the strains by about 50%. Strength improvements of up to 160% in
the load carrying capacity of columns with single layered ferrocement have been
observed whilst about 100% strength improvements have been observed in
columns with 2/3 layers of ferrocement applications. The slope of the
stress-strain curve resembles typical curves characteristic of high strengths. The
results of this study have been highly encouraging and the use of
ferrocement with blended cement is recommended for the repair and
rehabilitation/strengthening of brick masonry columns. The use of rice husk ash
for blending with ordinary Portland cement produces a stronger, impermeable
and durable material in addition to ease of in-situ application, reducing the
disposal problem of this agricultural waste and reducing the costs of repairs.
Keywords: strengthening, rehabilitation, brick masonry columns, blended
cement, old buildings, retrofit material.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06046
Introduction
Many old buildings along with few bearing historical architecture of past
civilisations were constructed by using brick masonry. Most of these old
buildings/structures have shown signs of distress/damage due to aging and action
of various offensive agencies. To maintain the safety requirements, most such
structures now require replacement/ rehabilitation/ strengthening. Strengthening
of existing columns may also be required for expansion works in the existing
buildings/structures. With the trend of conserving the heritage and specific
architecture/ faade along with avoiding environmental/disposal problems of
demolition waste, stress is laid on rehabilitation/ strengthening of the existing
elevations/structures. Resource conservation and paucity of development funds
in developing/under developed countries suggests economical options of
rehabilitation/ strengthening of existing structures. To maintain economy and
environment, use of waste materials in concrete have been investigated widely.
Lot of work has been done on use of waste brick aggregates (widely available
from old dilapidated buildings in the South Asian region) in normal/high strength
concrete by Kibriya [14] and use of agricultural waste in blended cement and
normal/high strength concrete Kibriya [58]. Ferrocement carries certain
advantages in rehabilitation/strengthening works like in-situ application, ease of
construction, use of locally available materials, no formworks, no special
skills/training required, easily moulded in any shape, very low permeability,
easily repairable, resource saving and durability vide Kibriya [9, 10], Paul and
Pama [11], ACI [12] and BOSTID [13]. The repair and strengthening of existing
structures has been a more challenging problem for civil engineers than
construction of new structures.
Use of blended cements containing agricultural waste i.e. rice husk ash have
not at all been investigated as yet for use in ferrocement. Higher content of silica
found in rice husk ashes along with low calcium oxide content is advantageous
as far as its cementing properties are concerned. Blended cements containing rice
husk ash as a partial replacement material have been found to improve strength
and durability characteristics substantially when used as cementitious material
and in high strength concretes for various purposes. Blended cement containing
75% OPC and 25% rice husk ash has now been investigated for use in
ferrocement. Selection of 25% replacement of OPC with rice husk ash has been
on the basis of excellent performance of such blended cements and concrete in
the recent studies by Kibriya [58].
2 Research methodology
This experimental investigation was carried out on nine different groups of
columns. Each group consisted of three specimen 225x225x750mm. The groups
were labeled in alphabetical order and the individual specimens were thus
numbered as 1, 2 and 3 and so on. 1st Group i.e Group A comprised of plain
brick masonry in 1:4 cement mortar and were considered as control specimen.
Group B comprised of brick masonry columns in 1:4 cement mortar with
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
471
12.5mm thick, 1:4 cement mortar plaster. Group C were brick masonry columns
in 1:4 cement mortar, simply plastered with 1:2 cement mortar. Groups D, E and
F comprised of brick masonry columns in 1:4 cement mortar with one, two and
three layers of ferrocement applied respectively. Each layer of ferrocement
applied was 15mm thick applied in two steps i.e. a layer of 7mm 1:2 cement
mortar, application of wire mesh and another 8mm layer of 1:2 cement mortar.
W/C ratio of mortar was maintained at 0.3. Blended cement containing 25% rice
husk ash and 75% ordinary Portland cement was used. Rice husk ash used for
blending was obtained by burning rice husk in an industrial furnace with
temperatures maintained around 500 to 6000C. Ashes were then cooled to 200C
and subsequently ground in a laboratory ball mill for about 120 minutes and
thereafter sieved through a #325 sieve. Ashes passing 100% through the sieve
were then used for blending with ordinary Portland cement. Sand with medium
grading was used for mortar. Upto 1% high range water reducing admixture was
used for preparing workable mortar mix. Wire mesh of expanded metal having
openings of 15mm and wire thickness of 1mm was applied with yield strength of
227 N/mm2 and ultimate strength of 379 N/mm2. 1mm thick, 2.5cm long nails
were used to bind wire mesh layers to the core. Groups G, H, and I comprised of
specimen similar to groups D, E and F with the exception that a chemical
bonding agent Sika-Latex was used in addition to nails to bind the ferrocement
layer to the core. The specimen were moist cured for 28 days at a temperature
around 200C before testing. The details are shown in Table 1.
Testing procedure
The column specimen were tested for compression using universal compression
testing machine with a 255x255mm, 5mm thick plate on top and bottom. The
observations were made for stress-strain behavior, crack pattern, cracking and
failure loads. The measurements for strain were done with the help of two
compressometers attached to the specimen. The results of the tests are tabulated
in Table 2. Permeability testing was carried out by the help of capillary rise test
according to RILEM CPC13 and Initial Surface Absorption (ISAT) tests
according to BS 1881. In addition, 50mmx50mm cubes were caste, moist cured
for 30 days and immersed in water for 30 days to observe any weight changes.
The ultimate failure and cracking load capacities of the specimen are given in
Table 2. The stress-strain curves and trend lines based on the average readings of
the specimen groups are shown in Figures 1 and 2.
4.1 Brick masonry columns (group A)
The failure loads were generally identical. The average failure load for case A
was 142kN. The crack initiation started at about 70% of the failure loads. Major
cracks were vertically oriented, however, a few horizontal cracks were also
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Details of specimen.
Group
Designation
A
B
No of
Layers
_
_
Bonding
Agent
_
_
Mortar
Ratio
_
1:4
Masonry Col
Masonry Col With
Plaster
Masonry Col with C
_
_
1:2
Plaster
Masonry Col + D
1
Nails
1:2
Ferro cement
Masonry Col +
E
2
Nails
1:2
Ferro cement
Masonry Col +
F
3
Nails
1:2
Ferro cement
Masonry Col +
G
1
Nails &
1:2
Ferro cement
S. Latex
Masonry Col +
H
2
Nails &
1:2
Ferro cement
S .Latex
Masonry Col +
I
3
Nails &
1:2
Ferro cement
S. Latex
Note: Brick Masonry in 1:4 Cement Mortar.
One layer of ferrocement mortar 15mm thick with 1:2 cement mortar.
Wire mesh with 1mm wire, 15mm mesh.
Cementitious material 75% OPC + 25% Rice Husk Ash
W/C ratio 0.3 of ferrocement.
Moist curing - 28 days
4.2 Brick masonry columns with 1:5 cement mortar plaster (group B)
The results of all specimen of group B clearly suggest that simple mortar
application can increase the failure loads by up to 35%. However, it seems to
have no effect on crack initiation and its growth as the crack growth was rapid
as in case of group A. The cracks generally started at about 70% of failure load.
The average failure load was 195kN as given in Table 2. The strains also reduced
somewhat, as compared to plain brick masonry columns as shown in Figure 1.
4.3 Brick masonry columns with 1:2 cement mortar plaster (group C)
The specimen of group C exhibited fairly good results in terms of failure loads.
The average failure load was calculated as 208kN. This clearly shows an
increase of about 46% over the failure load of the control specimen of Case A
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
473
and about 10% increase over columns plastered with 1:5 mortar ratio. The stressstrain curve is fairly straight with reduced strains as shown in Figure 2.
Table 2:
Type of Specimen
Group
Masonry Col
Masonry
With Plaster
Col
Masonry Col +
Ferro cement
(1 Layer)
Masonry Col +
Ferro cement
(2 Layers)
Masonry Col +
Ferro cement
(3 Layers)
Masonry Col +
Ferrocement(1)
Bonding Agent
Masonry Col +
Ferrocement(2)
Bonding Agent
Masonry Col +
Ferro cement(3)
Bonding Agent
D
E
F
G
H
I
Failure
Load
(kN)
138.4
141.2
146.8
192.1
198.3
196.4
211.1
208.4
204.7
340.9
394.2
367.2
299.4
268.6
279.5
310.4
301.5
270.8
354.8
396.2
377.1
280.4
359.7
260.9
281.7
320.1
318.4
Mean
Failure
Load (kN)
%
Increase
142.13
Control
195.6
37.6
208.1
46.5
367.4
158.7
282.5
98.9
294.3
107.3
376
164.8
300.3
111.5
306.7
116
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
1000
900
800
700
STRESS
600
500
400
300
200
100
0
0
0.0005
0.001
0.0015
0.002
0.0025
0.003
STRAIN
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
475
1200
1000
STRESS
800
600
400
200
0
0
0.0005
0.001
0.0015
STRAIN
D
0.002
0.0025
0.003
Linear (A)
Linear (B)
Linear (C)
Linear (D)
Linear (E)
Linear (E)
Linear (F)
Linear (G)
Linear (H)
Linear (I)
Figure 2:
4.6 Brick masonry columns with single layer of ferrocement and bonding
chemical applied (group G)
In the course of samples preparation of this group, a bonding agent, Sika-Latex
was used in addition to nailing the mesh layer. The average failure load 376kN
of this group although significantly higher i.e. 164% than the control specimen,
did not show much improvement over its corresponding group D of samples
having one layer of mesh nailed to the core, without any bonding agent.
However, the Sika-Latex bonding was effective in controlling the spalling of
outer mortar layer on failure. The failure strains were almost 10% lower than the
specimen with single layer of ferrocement without any bonding agent. The
propagation of cracks upto failure was gradual.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Discussion of results
From the summary of these results it can be clearly observed that the load
carrying capacity of ferrocement encased brick masonry columns increased
significantly. It is evident that a significant increase in failure load capacity can
be achieved using only one layer of ferrocement. Failure strains were observed to
be reduced by about 50% on application of single layer of ferrocement.
Additional layers (2 or 3) of ferrocement did not contribute much due to bond
failure between them and while bulging outwards these layers pulled the inner
mesh along and thus affecting the results. The contribution of Sika-Latex
bonding agent was not much except for controlling the spalling of mortar at
failure. The permeability of ferrocement layer is negligibly low thereby
increasing the durability of the ferrocement applied columns.
It has been observed that ferrocement containing blended cements has
performed even better than ferrocement with ordinary Portland cement on
comparing results of recent studies. Strength improvement of about 30% along
with complete watertightness in ferrocement with blended cement have been
observed. The results of this study have been highly encouraging and use of
ferrocement, with blended cement containing 25% rice husk ash, for repair and
rehabilitation/strengthening of brick masonry columns has been recommended
for practical use in the field. Use of rice husk ash for blending with ordinary
Portland cement produces a stronger and durable material in addition to ease of
in-situ application, reducing the disposal problem of this agricultural waste and
reducing the costs of repairs.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
477
Conclusions
Ferro cement having 2 and 3 layers of wire meshes are uneconomical and
did not appreciably improve the load carrying capacity as noted in this
experimental study.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 6
Damage and
fracture mechanics
481
Abstract
Open cellular lattice structures, made out of stainless steel, have been
manufactured using the micro fabrication procedure of selective laser melting. A
single geometry has been considered, namely vertical struts with cross bracing
with a cell geometry of a cubic form. Blocks of this material, with dimensions of
25 mm cube, have been tested under compression and shear loading. Three cell
sizes have been considered, namely 1.25, 2.5 and 5 mm. Uniaxial compression
and shear force versus displacement and failure characteristics of these materials
are discussed, In this way, the micro mechanical characteristics of the material
and manufacturing parameters are related to bulk properties.
Keywords: cellular, micro fabrication, selective laser melting, mechanical
properties.
Introduction
Foams have been used for a number of years as core materials in sandwich
construction for transport and general engineering systems. There is a wide
variety of foam ranging from low density, high end technology applications, e.g.
the Polymethacrylimide foam Rohacell 51WF for aerospace use [1], to high
density/low end technology applications, e.g. the aluminium foam Alporas for
civil engineering use [2]. The usual design parameters for sandwich construction
focus on elastic response (specific stiffness and strength) [3], but more
specialised behaviours include fatigue behaviour, crashworthiness and foreign
object impact performance [3, 4]. The mechanical behaviour of sandwich
structures is dependent on the mechanical properties of the foam core and
composite skin. As far as the foam core is concerned, the core can be subject to
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06047
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
483
There are a large number of possible geometries possible, which gives large
design freedom. The laser spot has a size of 50 micrometers, and any structural
feature can be built to this resolution, as long as the newly solidified material can
be built on top of the previous (solidified) layer. Figure 2 shows two cellular
structures that can be created. The geometry is based on standard RP support
geometries, but has features typical of general open cellular lattice structures.
The structure consists of vertical members with diagonal cross members only, as
currently, the process cannot build horizontal members. The powder used was
Stainless steel 316L from Sandvik [11], and the powder grain size was between
10 and 38 micrometers. The finer the powder, the more expensive and dangerous
(from combustion) it is.
Figure 2:
Figure 2 shows the cellular structures with 1.25mm and 2.5mm cell sizes. A
5mm cell size was also considered, but this proved to be problematic to make.
The struts tended to sag during build, which meant that the final quality was
unsatisfactory. In a single manufacture, six blocks were built on a steel plate
base. The manufacture time for these six blocks was about 4 hours. The time of
the process is dependent on the size and complexity of the cellular structure.
There are about 500 laser passes to create the height of blocks. The blocks were
cut from the steel bottom plate using a sharp knife, in an effort to reduce damage
to cells. An important issue is the size and quality of the struts. This has been
discussed in detail in Brocks et al. [10]. Two laser exposure times and two laser
powers were used (see Table 1).
Table 1:
Build ID
Cell size
(mm)
5
2.5
2.5
1.25
1-X-5
2L-X-2.5
2H-X-2.5
3-X-1.25
Rel. Dens.
(%)
0.8
4.1
5.4
12.5
Laser Exp
(ms)
2.5
2.5
3.0
3.0
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Las. Pwr
(A)
2.25
2.25
2.60
2.60
Figure 3:
25 mm cube specimens were crushed along the cell vertical direction in a 50kN
Instron testing machine at a displacement rate of 5mm/min. The specimens were
crushed between two steel platens and the stress and strain data was derived from
load cell and crosshead data. Engineering stress and strain measures are used.
Figure 4 gives stress v strain data for the three cell sizes, with stress being
plotted on a log scale, for clarity. Figure 5 gives the failure mode for the 2.5mm
cube cell size case. It can be seen that the progressive crush is stable and that the
major failure mode results from elastic-plastic buckling of the vertical struts.
This gives rise to a shear band forming in the complete specimen. Cells also
fail near the base, which shows the need to improve the test to ensure failure in
the bulk material. Zhou et al. [13] give an analysis of plastic buckling of struts,
and show that the critical buckling load, Pcr is given by:
Pcr =
12C
l
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(1)
485
where C is a constant relating the plastic hinge couple of the strut, and l is the
length of the strut. Hence the critical failure load should be inversely
proportional to the strut length. Figure 6 checks out this hypothesis. Equation (1)
is calibrated from the 2.5mm cell data. The effect of the change of the number of
vertical struts as a function of cell size is taken into account. The general trend is
correct, showing that compression failure for this block and loading
configuration is governed by strut length, strut cross section and parent material
property.
Figure 7 gives the effect of manufacturing parameters. It compares the
average crush stresses from the 2H tests with two 2L tests. For details of
specimens, see Table 1. It can be seen that manufacturing parameters are very
important, and it should be noted that these need to be optimised before
optimising the cell geometry.
It would be interesting to study the crush of the block in the lateral direction.
It is proposed that, crush strength would be reduced, as there are no struts in that
direction. Therefore this cellular material will be highly anisotropic.
a
b
c
Figure 4:
Stress v strain for three cell sizes (a. 1.25mm, b. 2.5mm, c. 5mm).
Figure 5:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
7.00E+06
6.00E+06
5.00E+06
4.00E+06
3.00E+06
Eqn.1
2.00E+06
1.00E+06
0.00E+00
0
10
Figure 6:
2H
Figure 7:
Foam cores in sandwich structures are subject to multi axial stresses, and so
these have to be studied [4]. A tensile Arcan test [14] has been developed, in
which combinations of tensile direct and shear stress can be applied to a block of
foam. In this paper, shear only will be described. The 25mm cube blocks are
glued onto steel plates, which are then mounted in the Arcan Rig [14]. Figure 8
gives shear stress v strain data, and Figure 9 gives the failure modes for 1.25mm
and 2.5mm cell sizes.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
487
3-2
2H
Figure 8:
Figure 9:
Comparison in failure modes for shear loading (2.5 and 1.5mm cell
sizes).
In both cases, failure initiates at the boundaries, showing the need to develop
the test further, to ensure the initiation of failure for the bulk material. In the
2.5mm cell size case, failure initiates by the buckling of the diagonal struts,
whereas in the 1.25mm cell size case, the whole block rotates, and failure is by
tensile rupture of the vertical struts. The latter failure mode gives a more abrupt
reduction in stress after failure. It would be interesting to conduct both biaxial
and hydrostatic tests, as these are important for localised loadings in sandwich
structures [4,15]. For the 2.5mm cell size, the shear failure stress is about 6MPa
which should be compared to the compression value of 10MPa. This would
indicate a failure surface typical of foams [15]. However, more unusual cellular
structures may well give more unconventional failure surfaces.
General discussion
Figure 10 compares the bulk performance of the current cellular structure with
other foams [16]. It can be seen that the structure performs well, and it should be
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 10:
Concluding remarks
489
small cell size give rise to a large number of structural features, which leads to
unrealistic numerical models for full micro mechanical modelling. Also, in the
case of foreign object impact loading, material behaviour is dynamic and highly
non linear with damage occurring. There is a need to develop homogenised
models, that characterise the effective behaviour of periodic structures [17].
Acknowledgements
The SLM machine has been funded by the EPSRC Grant number GR/598405/01
with contributions from MCP and Sandvik Osprey.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
491
Abstract
Cold-drawn steel wires have the excellent mechanical properties of strength and
toughness and are widely used in prestressed concrete structures. During
cold-drawn operations residual stresses are generated in steel wires. In this paper,
by the residual stress distribution from the finite element analysis (FEA), a
weight function method (WFM) has been used to evaluate the effective stress
intensity factors (SIFs) for cracked cold-drawn steel wires under tensile loading.
The calculation results have been compared with those obtained by the
two-dimensional (2D) FEA, which considers the residual stress redistribution in
the presence of a crack. In the present study, the effective SIFs calculated using
the WFM have shown a good agreement with those derived from the 2D FEA.
Keywords: stress intensity factor, residual stress, weight function, cold-drawn
steel wire, finite element analysis.
Introduction
In the cold-drawn procedure, the tensile strength and toughness of steel wires
increase, but a residual stress field also appears in them. After the treatment of a
further drawing with a very small area reduction or a combination of heating and
stretching the wire, the decreased residual tensile stresses still exist in the steel
wire surface [1]. It is well known that the presence of residual tensile stresses
will reduce the fatigue life of steel wires. With increasing in residual tensile
stresses, the fatigue crack growth rate increases for stress ranges close to the
fatigue limit [2]. The calculation of the SIF is one of key factors in fatigue life
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06048
app
A
C
x
app
(a)
Figure 1:
O
C
a
a
2b
D
A -A
app
(b)
(c)
Two-dimensional model
The surface crack of the cold-drawn steel wires produced by mechanical fatigue
can be assumed to be a semi-elliptical geometry (a/b<1, a is the crack depth or
minor axis of the ellipse, b the major axis of the ellipse), perpendicular to the
tensile loading direction [7], as show in fig. 1(a), 1(b), and the maximum SIF
value was obtained at the crack center C, according to the results given by
Astiz [8]. Following a well known purely linear elastic solution in the vicinity of
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
493
a 3D crack with an arbitrary curved front [9], the calculation of the SIF can be
based on 2D plane strain hypothesis on the longitudinal section of a cracked
cold-drawn steel wire under tensile loading, fig. 1(c).
As for a cracked cold-drawn steel wire under tensile loading, the effective SIF at
the crack front is decided by the combined effect of local residual stresses and
the stresses applied by external loading. According to the linear superposition
principle, the effective SIF (Keff) is the algebraic sum of the external SIF (Kapp)
due to the external applied loads and the residual SIF (Kres) due to the residual
stresses, and can be expressed as:
Keff = Kapp + Kres
(1)
3.1 Residual SIF
It is well known that, once the weight function is determined for a particular
cracked body, the SIFs can be obtained for any stress field by an appropriate
integration. So, in this section, the WFM is used to evaluate the residual SIF
(Kres).
Based on the FEA [1], the distribution of the residual stresses for colddrawn steel wires with a thermomechanical treatment is shown in fig. 2, where r
is the polar coordinate of the wire and R the radius of the wire, and the equation
used for the calculation of the residual stresses in the plane problem is as
follows:
4
5
(2)
3257.62 (r R ) + 1029.93 (r R ) .
In the edge-cracked 2D model, fig. 1c, a residual SIF can be calculated by
integrating the weight function m(x,a) and the residual stress distribution res(x)
acting on the crack length a [10], which is expressed as follows:
(3)
500
(MPa)
Figure 2:
-500
1.0
0.8
0.6
0.4
0.2
0.0
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
m( x ,a ) =
x a
G ,
a D
a
1
D
3/ 2
x
1
a
(4)
where
2
a x
a x
a x
a
x a
G , = g1 + g 2 + g3 + g 4 ,
D a
D a
D a
D
a D
5
a
a
a
a
a
g1 = 0.46 + 3.06 + 0.841 + 0.66 1 ,
D
D D
D
D
2
a
a
g 2 = 3.52 ,
D
D
2
3/ 2
a
a
a a
a
g3 = 6.17 28.22 + 34.54 14.39 1
D
D
D
D D
5
a
a
a
5.881 2.64 1 ,
D D
D
2
a
a
a
a
g 4 = 6.63 + 25.16 31.04 + 14.41
D
D
D
D
a
+ 21
D
3/ 2
a
a
a
+ 5.041 + 1.98 1 .
D D
D
y
x
P=1
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
495
Kapp = app a F ( a D )
(5)
where app is the external applied stress, D the diameter of the steel wire and F
(a/D) given by:
F( a / D ) =
a
a
a
2D
D
2D
tan
.
a
a
2D
cos
2D
app
L=10 mm
a
D/2=2.5 mm D/2=2.5 mm
Figure 4:
2.4
FEM
Tada et al.
2.1
1.8
1.5
1.2
0.0
Figure 5:
0.1
0.2
0.3
Normalised crack length, a/D
Comparison of KI
calculations.
app
0.4
497
Figure 6:
Both the WFM and the FEM described above were used to evaluate the variation
of the effective SIF as a function of the crack length. The effective SIFs
calculated from the WFM and those derived from the FEA are compared in fig. 7
for seven different crack lengths. From this figure, one can find that, for different
crack lengths under two loading levels, the maximum difference in values
between the two methods is 0.87%, which exhibits a good mutual agreement.
Comparison results between the FEA and the WFM showed that, when
calculating effective SIFs with a linear elastic analysis for cracked cold-drawn
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
240
200
WFM, app=0.4fptk
WFM, app=0.7fptk
FEA, app=0.4fptk
FEA, app=0.7fptk
160
120
80
40
0.0
Figure 7:
0.1
0.2
0.3
Normalised crack length, a/D
0.4
Conclusions
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
499
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
501
Abstract
The deformation of amorphous polymers is highly sensitive to hydrostatic
pressure and the plastic flow is non-volume preserving. Failure of these materials
is often preceded by the formation of small crack-like defects known as crazes,
bridged by the fibrils of the polymer. In this work, a population of discrete voids
is placed ahead of a crack within a boundary layer configuration to study the
effects of pressure-sensitivity and plastic dilatancy on void growth and
damage in polymeric materials. Results show that high pressure-sensitivity
significantly increases both the intensity and spatial extent of damage, which
could promote crazing. These effects can be compounded as the deviation from
the associated flow rule, | |, increases. This study offers some evidence of
pressure-sensitivity and plastic dilatancy as the major contributing parameters to
the formation and growth of microporous crazing zones in polymers.
Keywords: void growth, coalescence, pressure-sensitive yielding, plastic
dilatancy, polymer.
Introduction
2R0
X1
Figure 1:
Problem formulation
503
based on a linear combination of the mean stress and effective stress. Here, the
pressure-dependent yielding of the material is described by
3J 2 + m tan = 0
(1)
where J2 is the second invariant of the deviatoric part of the Cauchy stress tensor
, m the mean stress, the flow stress of the subsequent yield surface, and
the index for pressure-sensitivity. We assume the flow potential to take the form
= 3J 2 + m tan
(2)
where is the index for plastic dilatancy. The plastic deformation rate is given
by the non-associated flow rule
d P = P
(3)
0
E 0
=P
(4)
where N is the hardening exponent ranging from 0 to 1, and 0 the initial yield
stress under pure shear which is related to the initial tensile and compressive
yield stresses 0t and 0c, by
1
t
1 + 3 tan 0 for tension
0 =
1
0c 0t
.
c
t
0 +0
= tan 1 3
(5)
(6)
Xia and Shih [1] simplified the ductile fracture process under small-scale
yielding conditions by placing a single row of void-containing cells ahead of the
crack. In the present study, we model void growth and damage via an array of
unit-cells ahead of the crack (see Fig. 1). Each unit-cell is of dimensions D by D
and contains a discrete cylindrical void of initial radius R0. The initial void
volume fraction is given by f 0 = R02 / D 2 . A total of 23 voids are used in the
computation.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(b)
Figure 2:
Along the remote circular boundary (see Fig. 2a), the elastic asymptotic (inplane) displacement field
u1 ( R, ) = K I
1 +
E
R
(3 4 cos )cos
2
2
(7)
1 + R
(3 4 cos )sin
u 2 ( R, ) = K I
E
2
2
is applied under plane strain conditions, where R2 = X12 + X22 and = tan-1(X2/X1)
for points on the remote boundary. KI is the mode I stress intensity factor related
to the J-integral by
1 2 2
J=
KI .
(8)
E
At various stages of loading, the value of the J-integral is calculated on a number
of contours around the crack using the domain integral method. The domain
integral value was found to be in good agreement with the value given by eqn.
(8) for the prescribed amplitude KI. This consistency check assures that smallscale yielding conditions are satisfied.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
505
Figure 2 displays an example finite element mesh for the discrete void
model. By taking advantage of symmetry, only one half of the geometry needs to
be modelled. The mesh contains 4-node bilinear quadrilateral hybrid plane strain
elements. The computations are performed within the finite strain setting using
the general-purpose finite element program ABAQUS Version 6.5.1 [12].
Before proceeding to the modeling of void growth, we first examine the effects
of pressure-sensitivity on the extent of plastic deformation ahead of the crack.
Some insights can be obtained from the Dugdale strip yield model, which
assumes a long slender plastic zone at the crack tip in a nonhardening pressureinsensitive material ( = 0). The plastic zone size L under a uniaxial tensile
stress T is given as
L=
KI
8 T
(9)
For the pressure-sensitive material in eqn. (1), the tensile stress T at yielding is
T = 0/(1+1/3 tan ). Denoting L0 as the plastic zone size corresponding to =
0, i.e. T = 0, we obtain
2
1
L / L0 = 1 + tan .
(10)
3
This relation suggests an increase in the plastic zone size L with pressuresensitivity . Observe that an increase in from 0 to 20 increases L by around
25%.
A numerical study is next performed to shed some light on the extent of
plastic dissipation in a void-free pressure-sensitive polymer. The void-free
material is subjected to mode I, K-field loading in eqn. (7). An associated flow is
assumed with = . The elastic modulus of the polymer is taken to be E = 500
MPa, with a yield strength of 0 = 5 MPa, and a Poissons ratio of = 0.4. In
accordance with the Dugdale assumption, the strain hardening exponent N is set
to zero (for ideally-plastic material).
Figure 3 displays the plastic zones in the deformed mesh configuration for
three pressure-sensitive levels: = 0, 10 and 20. The plastic zones are
operationally defined by p 0.001 , where p is the accumulated plastic strain.
Observe that at the same external load level, the plastic zone spreads out as the
pressure-sensitivity level increases. For example, at J = 68 kNm-1, an increase in
from 0 to 20 raises the extent of plastic dissipation ahead of the crack by
nearly two-fold (L/L0 2). This general trend of increasing plastic dissipation
with pressure-sensitivity corroborates with the findings from eqn. (10).
In this section, an array of discrete voids is placed ahead of the crack. From
dimensional considerations, the spatial distribution of the current void volume
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(11)
The properties of the polymeric material are specified by the parameters 0/E =
0.01, = 0.4, and N = 0.1. We focus on the effects of pressure-sensitivity and
plastic dilatancy on void growth and damage ahead of the crack.
(a)
= 0
L0
(b)
= 10
L
(c)
Figure 3:
= 20
507
two voids ahead of the crack, resulting in near-tip void growth. For = 15, the
plastic zone spreads across the first four voids, leading to the formation of an
extended damage zone. This extended damage zone closely resembles the long
craze zones in polymers. One can therefore infer that pressure-sensitivity is a
major contributing parameter to the crazing phenomenon. Interestingly, we also
note that there is an increasing tendency for the voids to become oblate as they
grow larger. This suggests the effects of void shape to play an important role in
the coalescence stages of fracture.
(a)
(b)
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
=0
(b)
=15
Figure 5:
The analysis in the previous section employs an associated flow rule, i.e. = .
Experimental studies, however, have shown that the plastic volume change in
polymers does not commensurate with the predictions of the associated flow
rule, i.e. < .
Figure 6 displays the effects of on the damage distribution ahead of the
crack for = 10 and 20. One can see that an increase in the deviation from the
associated flow, | |, significantly increases both the intensity and the spatial
extent of damage. For f0 = 0.005, = 20, at J/(0D) = 0.60, an increase in | |
from 0 to 5 raises the damage intensity level by nearly 60%, and shifts the
damage process zone from X1 = 5D to X1 = 7D. The effects of increasing | |
are found to be less severe for f0 = 0.05. The non-conservative nature of the
associated flow in predicting damage has serious implications in the design of
engineering materials and structures.
Conclusion
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
509
(a)
(b)
Figure 6:
References
[1]
[2]
Xia, L. & Shih, C.F., Ductile crack growth - I. A numerical study using
computational cells with microstructurally-based length scales. Journal of
the Mechanics and Physics of Solids, 43, pp. 233-259, 1995.
Gurson, A.L., Continuum theory of ductile rupture by void nucleation and
growth: Part I - Yield criteria and flow rules for porous ductile media.
Journal of Engineering Materials and Technology, 99, pp. 2-15, 1977.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Guo, T.F. & Cheng, L., Modeling vapor pressure effects on void rupture
and crack growth resistance. Acta Materialia, 50, pp. 3487-3500, 2002.
Chew, H.B., Guo, T.F. & Cheng, L., Vapor pressure and residual stress
effects on the toughness of polymeric adhesive joints. Engineering
Fracture Mechanics, 71, pp. 2435-2448, 2004.
Chew, H.B., Guo, T.F. & Cheng, L., Vapor pressure and residual stress
effects on failure of an adhesive film. International Journal of Solids and
Structures, 42, pp. 4795-4810, 2005.
Chew, H.B., Guo, T.F. & Cheng, L., Vapor pressure and residual stress
effects on mixed mode toughness of an adhesive film. International
Journal of Fracture, 134, pp. 349-368, 2005.
Chew, H.B., Guo, T.F. & Cheng, L., Effects of pressure-sensitivity and
plastic dilatancy on void growth and interaction. International Journal of
Solids and Structures, in Press, 2006.
Tvergaard, V. & Hutchinson, J.W., Two mechanisms of ductile fracture:
void by void growth versus multiple void interaction. International
Journal of Solids and Structures, 39, pp. 3581-3597, 2002.
Kim, J., Gao, X. & Srivatsan, T.S., Modeling of crack growth in ductile
solids: a three-dimensional analysis. International Journal of Solids and
Structures, 40, pp. 7357-7374, 2003.
Chew, H.B., Guo, T.F. & Cheng, L., Vapor pressure and voiding effects
on thin film damage. Thin Solid Films, in Press, 2006.
Kambour, R.P., A review of crazing and fracture in thermoplastics.
Journal of Polymer Science, 7, pp. 1-154, 1973.
Hibbit, Karlsson & Sorensen Inc., Abaqus/Standard Users Manual,
Version 6.5.1, Vol. 1, 2005.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
511
BU Automotive
Centre of Lightweight Structures TUD-TNO,
TNO Science and Industry, Delft, The Netherlands
Abstract
In this paper, a practical orthotropic damage model is developed and tested for
composite materials during crash. The model uses the Hashins failure criteria in
which the fibre and matrix failures are described explicitly, both in tension and
compression. A linear softening degradation is proposed and a close-form
solution of the corresponding damage parameter is provided.
To reduce the mesh dependency, an embedded discontinuous element is
proposed. It is a virtual embedded element in the sense that an actual element is
divided into two zones, one elastic (undamaged) zone and one localization
(damaged) zone. Their equivalence is preserved by constraining the kinematics
and equilibrium equations. Since the damage zone is introduced into the element,
the corresponding dissipated energy due to damage is fixed, independent of the
element size.
The numerical simulations using the developed damage model show a mesh
objective result and correlate well with the energy dissipation in dynamic 3-point
bending experiments. It is concluded that an accurate material calibration is
crucial for the success of failure simulations. The traditional strain mapping
cannot be applied to failure problems in which the strain distribution is highly
non-uniform. Material parameter calibration should take into account the ratio of
the failure zone and the gauge length. A practical approximation formula is
provided.
Keywords: composite, damage model, crash simulation, embedded element.
Introduction
For plane stress situations in the continuum damage approach, the effective
is related to the total strain by the elastic stiffness tensor C0
stress
= C 0
(1)
The effective stress is mapped to the true stress by the damage operator M
= M ,
0
1 D1 0
M= 0
1 D2
0
0
0
1 D12
(2)
where D1, D2 and D12 are the damage parameters. Inserting Equation (1) into
Equation (2) yields
= M C0 = C
(3)
(1 D1 )E11
1
C = (1 D2 ) 12 E11
0
(1 D1 ) 21 E22
(1 D2 )E22
0
(1 D12 ) G
= 1 12 21
(4)
(5)
with E11, E22, G, 12 and 21 are the elasticity parameters of the undamaged
lamina. Note that when damage occurs (Di>0), the stiffness tensor is not
symmetric. To preserve a symmetric stiffness tensor, the Poissons ratio can be
related to the damage parameters [3].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
513
0 failed
e = 11 1
X
< 0 elastic
2
f
(6)
0 failed
em2 = 22 1
Y
< 0 elastic
3)
(7)
Shear mode
2
0 failed
e = 1
S
< 0 elastic
2
s
(8)
where X and Y are fiber and matrix strengths and S is the shear strength. Note
that the compression and tensile modes are treated in the same way. For notation
simplicity, the sub indexes for the tensile and compression strength parameters
are omitted.
Note that Hashins criteria have a complete decoupling of the fibre and matrix
failure modes from shear failure. Therefore, the failure criteria may overestimate
the strength of a lamina when shear is present. However, it should be mentioned
that measuring the softening part of the stress-strain curve is very difficult even
in a uniaxial test condition. A multi-axial loading surface is almost impossible to
be validated. Furthermore, the damage evolution or degradation law is, in most
cases, in an uncoupled fashion. Therefore, it is convenient to treat all failure
criteria to be independent.
It must be emphasised that although the failure modes and the degradation
law (which will be introduced in the next section) are decoupled, the complete
stress-strain relationship still contains a coupling due to the stiffness matrix C
(see Equation 3).
4 Degradation rule
Damage occurs when one of the Hashins criteria is satisfied. For simplicity, we
assume a linear degradation law (see Figure 1) after the initial strength 0 is
exceeded. Consequently, the damage parameter D can be determined by the
following relationship
= E ( D) = E0 (1 D ) = 0 h ( 0 )
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(9)
Figure 1:
It must be emphasised that Equation (9) holds for materials in the localisation
zone. Two issues related to this localisation problem should be addressed:
1) The material calibration of localisation related parameters such as the
softening parameter h and the final failure strain f. Note that the strain field
is not uniform as soon as localisation occurs. Therefore, the standard strain
mapping in uniaxial tensile examples: = u/L where L is the gauge length
and u the deformation, is not valid. The results of the experiments should be
interpreted carefully.
2) The mesh dependent problem. For a tensile bar with a uniform cross section
area A, only one element will follow the failure path. Consequently, the total
dissipation energy of the tensile bar reads
1
Wdiss = 0 f 0 l A
2
(10)
where l is the length of the failure element. Clearly, the energy dissipation is
dependent on the element size and, therefore, the simulation result is mesh
dependent.
These two issues will be discussed in the next section. First, the mesh
dependence problem is solved by introducing a localisation zone into a standard
finite element. Next, the corresponding localisation related parameters are
investigated and their relationships are outlined. Finally, formula will be
provided for practical approximation of the final failure strain and the softening
parameter.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2:
515
h + (L ) e = L = u
(11)
E0 e = Eh h = E L =
(12)
where E0 is the elastic modulus, Eh the stiffness at the damage zone and EL the
equivalent (element) elastic modulus, respectively
E h = (1 d ) E
(13)
E L = (1 D ) E
(14)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
h =
1 + d (r 1)
r=
(15)
and the local damage parameter d can be evaluated and then transferred to the
global (element) damage parameter
D=
rd
1 + d (r 1)
(16)
Note that the localisation strain h is a function of damage parameter d. This local
damage parameter can be determined by inserting Equations (13)(15) into the
linear softening law (9)
d=
(E + h ) ( 0 )
E + (E + h ) 0 (r 1)
(17)
With a known element strain , the local damage parameter can be directly
determined. This is the advantage of a linear softening assumption, which makes
the stress update simple and straightforward.
Note that the softening parameters should be carefully calibrated. The
traditional strain mapping cannot be applied to failure problems in which the
strain distribution is highly non-uniform. Material parameter calibration should
take into account the ratio of the failure zone and the gauge length.
For the final failure, d=D=1, from Equation (15), the local failure strain f and
the element failure strain have relationship
f =
r=
(18)
Suppose one uses one element to simulate a tensile bar with a gauge length L.
The material failure strain f reads
f =
Lu u
=
L
(19)
where is the length of the damage zone. Note that the material failure strain is
obtained by dividing the deformation by the length of the damage zone and not
the gauge length.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
517
Numerical simulations
Figure 3:
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 5:
test data
mat58
new model
Force [KN]
Figure 6:
10
20
30
40
50
deflection [mm]
60
70
80
90
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
519
7 Conclusions
An orthotropic damage model is developed and tested for composite materials.
The model uses the Hashins failure criteria in which the fibre and matrix
failures are described explicitly, both in tension and compression. A linear
softening degradation is proposed and a close-form solution of the corresponding
damage parameter is provided. This new model is implemented in LS-DYNA as
a user-defined material model.
To reduce the mesh dependency, an embedded discontinuous element is
proposed. It is a virtual embedded element in the sense that an actual element is
divided into two zones, one elastic (undamaged) zone and one localisation
(damaged) zone. Their equivalence is preserved by constraining the kinematics
and equilibrium equations. Since the damage zone is introduced into the element,
the corresponding dissipated energy due to damage is fixed, independent on the
element size. Therefore, mesh dependency with respect to the dissipated energy
is removed.
It is concluded that an accurate material calibration is crucial for successful
failure simulations. The traditional strain mapping cannot be applied to failure
problems in which the strain distribution is highly non-uniform. Material
parameter calibration should take into account the ratio of the failure zone and
the gauge length. A practical approximation formula is provided.
Reference
[1]
[2]
[3]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
521
Abstract
A quasistatic problem of thermoelasticity for a solid infinite cylinder with a ringshaped crack is considered. The cylinder is enclosed in a rigid thin heat
absorbing shell with a slip joint which reduces thermal impact. The external
surface of the shell is subjected to a linear heat transfer by radiation to the
surroundings. We assumed that the cylinder had initially a temperature changing
along the radius and the surrounding temperature was a function of time. The
main result is an expression obtained for the stress intensity factor depending on
time. Various particular cases are considered obtained both by the authors of this
paper and by other scientists. An analysis of the dependences of the stress
intensity factor (SIF) on time shows that with an assigned critical SIF, there are
minimum and maximum crack sizes below and above which the crack will not
grow under the given cooling conditions. For these dimensions, the maximum of
SIF does not reach the critical value. For intermediate crack sizes, growth of the
crack begins at the moment SIF reaches the critical value. Here, the crack first
grows irregularly to the size for which at the given moment of time SIF is equal
to the critical value. Extension of so called hot cracks in a solid cylinder is also
considered. These cracks arise when a cold cylinder is placed in a fusion of the
same material at the melted temperature. Further on this cylinder is rapidly taken
out of the fusion and on its surface the melting layer linked with the cylinder
surface arise. Under cooling of the cylinder, residual tensile stresses arise in the
layer which may cause growth of ring-shaped surface cracks which are
dangerous for further exploitation of this system.
Keywords: thermoelasticity, crack propagation, SIF, powdery covering.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06051
Introduction
Mathematical modeling
ij = ijT + ijP ,
u i = u iT + u iP .
(1)
The first solution, obtained with the assumption that no cracks are present,
satisfies the equations of thermoelectricity for an infinite cylinder in a
generalized plain-strain state. This solution satisfies all boundary conditions
except for those for the edges of the crack, which are loaded by a normal load
T (1 )
zz* = zz
T E (T0 )
= 2 Bi
yn J 0 ( yn ) ( Bi ky
a root of equation
n =1
yn
exp ( yn2 Fo )
2 2
n
r
2 J 1 ( yn ) yn J 0 yn
rc
+ y n2 (1 + 2k )
(2)
yJ1 ( y) = Bi ky 2 J 0 ( y) .
Here E is the elastic modulus of the cylinder material; is the Poisson ratio;
r c
is the Biot criterion;
c 1 + 0 d 2 T
1
= 0
; =
;
1 + 0d
c 1 + 0 d T
k = d/rc; d coating width.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
523
Calculations
For numeral analyses a solid cylinder of the radius rc = 5103 m made of steel 45
(its chemical composition: Fe, 0,42-0,49% C, 0,17-0,37% Si, 0,5-0,8% Mn; =
7800 kg/m3, c = 470 J/kgK, = 50 W/mK) is chosen, coating is made of bronze
101, (its chemical composition: Cu, 9-11% Sn, 0,8-1,2% P; 0 = 8700
kg/m3, c0 = 380 J/kgK, 0 = 100 W/mK); its width is d = 0,25103m. The
coefficient of heat transfer 0 = 1,5105 W/m2K. The porosity = 0,54. Using
the formula for the porous medium
= 0(1 1,5) for < 0,6
(3)
and the formula
c = c0(1 )
(4)
(5)
= 0(1 )
we obtain: = 4000 kg/m3, c = 175 J/kgK, = 19 W/mK.
As a result k = d/rc = 0,05 << 1, k = 0,0065, Bi = 5. In the case of covering
absence (d = 0) we obtain k = 0, Bi = 0rc/ = 15.
Figures 1 and 2 show the relations between time Fo = at/ rc2 (a heat
diffusing) and axial strains zz* depending on the radius r/rc for the cases of the
cylinder with the covering (Figure 1) and without it (Figure 2). As we can see
from the graph on the figures powdery coverings essentially reduce
thermoelastic stress in cylinder.
pzz = Tzz ,
u pz = 0 ,
z = 0,
(6)
z = 0,
(7)
z = 0,
(8)
r = rc,
0 < z < ,
(9)
rz = 0
p
rz
= 0,
p
r
u =0
r = rc,
0<z<
(10)
According to the problem (6) (10) the calculation of stress-strain state of the
cylinder with a ring-shaped crack is fulfilled. This solution allowed to us
calculate stress intensity factor (SIF) as a criterion of crack propagation [1, 2].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
525
elasticity ij , u i
(, ) =
(11)
Bn ( n ) = b n ( n ) +
rd
2
t
g(t ) cos n dt
2
rc J 0 ( n ) 0
rc
(12)
pzz = 2F(r ) ,
z = 0,
(13)
u pz
z = 0,
(14)
= ,
r
, is the constant normal
rc
displacement along the crack which is determined from the balance condition of
the cylinder in the plane of the crack.
rd
rc g (t )dt =
0
rc
P
F(x )xdx
4 rd
(15)
g ( ) =
1 P
211
a
(
)
+
yF(yrc )
dy
yF
yr
arcsin
c
2
4rc
y
2 1 yF(yrc )dy 4 K1 ( ) sh
1
ch
+
4
K1 ( )
sh
(
)
g
d
I ( ) ch ch d ,
2
0
0 1
0 < < , = rd rc .
(16)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
K I = 2
rd
r c g (r d )
(17)
K *I =
(1 )
(T0 )
2K I
1/ 2
rc T E
on time for Fo for different ring-shaped crack sizes = rd/rc in the case of rigid
slip joint where d = 0 is shown. In Figure4 the same for d = 2,5104m. In case
of flexible joint SIF is 12 % larger for the same sizes of the cylinder and the
shell.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
527
Figure 4: Dependence SIF on time Fo fir different crack sizes (with covering).
References
[1] Zhornik A.I. Nonsteady problem of the theory of elasticity for a solid
infinite cylinder with a penny-shaped crack, enclosed in a thin shell,
International Applied Mechanics, USA 1993, July P. 47 52.
[2] Zhornik A.I., Zhornik V.A., Kashitsyn L.P. Nonsteady problem of
thermoelectricity for a solid infinite cylinder will a ring-shaped crack,
enclosed in a thin shell, Advanced Computational Methods in Heat Transfer
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
529
Abstract
The inside insulation has been investigated in the past by means of scientific
institutions as a necessary method in connection with the thermal improvement
of worth-preserving facades of architectonic heritage. In the case of fair faced
brickwork it is established on homogenous areas of the wall, for instance as a
variant of capillary active systems or as a classical version with a
humid-dependent vapor barrier. With regard to the insulation of special details,
e.g. embedded wooden beam ends within areas of outside walls, the opinion of
experts differs considerably. Both the total ventilation of the beam ends and
complete air tightness are proposed. In line with the new European directive on
the energy efficiency of buildings, innovative ideas should also be developed in
the field of building physics in order to realize the restrictions without damage to
building envelope parts. The main point of this paper is the investigation of the
hygrothermal situation of embedded beam ends in inside insulated outside walls,
dependent on time and climate boundary conditions. The topics are treated by
means of in situ measurements in two test houses in the eastern part of Germany
under condition of use and by experiments in connection with an erected test
stand. A comprehensive numerical simulation of the coupled heat- and mass
transfer supports the measurements to reduce the experimental work. In principle
there are several options to avoid the effects of the hygrothermal deficiencies of
the structure caused or intensified by inside insulation. Hygrothermal effects are
quantified and solutions are found for workmanship without damage.
Keywords:
wooden beam ends, worth-preserving facades, heating energy
saving, hygrothermal performance, damages to structures.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06052
Introduction
In case of worth preserving facades there are some conflicts: for an example on
the one hand the changing climate conditions challenge the mankind to develop
political strategies and to implement measures that reduce greenhouse gas
emissions. Plans of European Member States call for 20% ...40% decrease by
2020. Another key objective, which is linked to the environmental protection, is
to make European states less dependent on oil and gas. Current forecasts indicate
that without measures this dependency will jump to approximately 70% by 2030.
A directive of the European Parliament and of the Council on the energy
performance of buildings was published and discussed in the past. The updated
regulation came into force in Dec.16th 2002 [1]. The directive should be realized
in the member states of the European Union up to 2006. About 41% of the total
final energy demand of Europe is used in the residential and tertiary sectors.
Space heating is by far the largest energy end-use of households in EU Member
States, namely 57%. Therefore the European directive covers four main
elements:
-Establishment of a general framework of a common methodology for
calculation the integrated energy performance of buildings.
-Application of minimum standards on the energy performance to new buildings
and certain existing buildings when they are renovated.
-Certification schemes for new and existing buildings on the basis of the above
standards and public display of energy performance certificates and
recommended indoor temperatures and other relevant climatic factors in public
buildings and buildings frequented by the public.
-Specific inspection and assessment of boilers and heating and cooling
installations.
About 32% of the current stock of the 150 million residential dwellings in the
15 old EU Member States was built prior to 1945, about 40 % between 1945 and
1974 and 28% since 1974. An upgrade of thermal insulation regulations and
improved efficiency for installed equipment for existing dwellings, bringing
them close to current buildings codes, would help to realize an important saving
potential, making it a very desirable and in most cases a cost-effective option.
So the directive is explicit valid for existing buildings too. In the face of the
low turnover rate of buildings with a lifetime from 50 to more than 100 years it
is clear that the existing stock of buildings contains the largest potential for
improving energy performance in short and medium periods of time.
On the other hand due to an architectonic heritage the use of a thermal
insulation composite system or another outside insulation is not possible. An
inside insulation may be an alternative if the hygrothermal and acoustic
behaviour is considered. But in regard with the inside insulation of wooden beam
ends embedded within areas of outside walls, the opinion of experts differ
considerably. For this reason the investigation is to be done both by means of
numerical simulation of the coupled heat- and mass transfer included the effects
of ventilation and a lot of in situ measurements in test houses under condition of
use and by experiments in connection with an installed test stand in a laboratory.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
531
Owing to the requirements of the new European directive and to the demand for
a comfortable room-climate the thermal insulation of outside walls of existing
buildings should be increased. A good solution for homogenous areas in case of
worth- preserving facades is the so-called capillary-active inside insulation or a
classical version with a humid-dependent vapor retarder [2]. The results of the
numerical simulation of the coupled heat and moisture transfer, represented in
this paper are carried out by means of the program Delphin [3], validated also
in international projects.
Fig. 1 represents the moderate moisture distribution by means of a capillary
active calciumsilicate inside insulation without use of a classic vapor barrier.
This method guarantees the drying out of moisture caused for instance by driving
rain at the facades. Owing to the capillary diffusivity function of the material
calciumsilicate with a maximum of 10-5m2/s the condensation at the cold site of
the inside insulation is limited in wintertime and due to the deleted vapor barrier
a sufficient drying out process takes place during summer period.
Figure 1:
100
300
12
10
15
y in mm
500
20
20
700
15
10
900
1,100
1,300
100
Figure 2:
200
300
x in mm
400
500
700
20+
18 to 20
16 to 18
14 to 16
12 to 14
10 to 12
8 to 10
6 to 8
4 to 6
2 to 4
0 to 2
100
3
300
8
500
y in mm
600
700
10
15
900
5
10
15
1,100
1,300
100
200
300
400
500
600
700
x in mm
Figure 3:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
533
Figure 4:
Figure 5:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 6:
Experimental investigation
535
Figure 7:
Course of the relative humidity of the air gape in the front side of
the beam end H4.
Figure 8:
The effect of the additional heating supply through the heating channel to the
temperature in front of the beam ends demonstrates fig. 8. After the heating flow
to H4 is mobilized (Jan.4th.2002) the differences between the temperatures at the
beam ends H4 and H3 are compensated.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 9:
Conclusions
Figure 10:
537
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1]
[2]
[3]
[4]
[5]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
539
Abstract
The repair of cracked timber beams on site frequently involves crack injection
with adhesives followed by the application of the chosen strengthening system,
involving bonded or bolted metal plates or rods or bonded pultruded GFRP or
CFRP profiles. Although strengthening design will not generally take into
account the contribution of injecting adhesives in the crack volume region it is
felt that this contribution to overall strengthening of the component or structure
may be of some significance. A test programme was therefore developed at
LNEC to assess the strength and stiffness of timber beams which were treated
with epoxy adhesives after they had developed failure in a 4 point bending test. It
is recognised that partially broken beams may have very narrow, almost
invisible, fissures, which are difficult to impregnate. As well as the forgoing
partially separated broken beams may present extremely irregular split ends,
which are difficult to bring back into perfect alignment and contact. These two
factors may reduce the effectiveness of the strength restoration and were
considered in the experimental study. However, the specific beam failure
appearance is not likely to give a clear indication of how efficient the
consolidation turns out to be. Test results show that epoxy adhesive injection of
cracked timber beams may lead to a not insignificant, but somewhat variable,
strength recovery (0.45 to 1.11) and to a significant stiffness recovery (0.74 to
0.99) of broken beams. Having these results in mind, although on site repair of
broken timber beams should not rely on epoxy adhesives injection only, its
contribution to the stiffness of strengthened beams is significant and one should
make use of this advantage.
Keywords: cracked timber beams, structural adhesives, consolidation, strength
and stiffness recovery.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06053
Introduction
Old timber structural members may present failure due to biological damage,
which is responsible for timber strength reduction or cross section loss, or due to
unacceptable stresses resulting from inadequate design or from structural
accidental or intentional modifications. Furthermore older structures intended for
a particular purpose are currently being subjected to types of use for which the
original design parameters were not intended.
Although the replacement of damaged members by new timber or other
materials may seem the easiest solution, on site repair may be a better cost option
in cases where only a small timber length is damaged, or when that member is
part of a complex structural system and its removal may jeopardise decorative
ceilings and/or limit the occupancy of the building, Even the replacement of one
carrier beam may with traditional and well-proven carpentry techniques involve
the whole or partial removal of a roof structure with its corresponding time and
cost delay. In fact removal of sound surrounding components often puts hitherto
undamaged and adequate parts of the structure at risk.
The repair of cracked timber beams on site frequently involves crack injection
with adhesives followed by the application of the chosen strengthening system,
namely involving bonded or bolted metal plates or rods or bonded glass fibre
profiles. It is interesting to note that in the early days of injection technology
with timber (borrowed from Concrete repair practice in the late 1950s early
1960s) attempts were made to restore overstressed classical truss rafter systems
with inappropriate low viscosity 2 part epoxy so-called injection systems (Ref
Aberdeen Music Hall) - the lack of success resulted in a secondary repair method
using bolted U beam sections.
Current strengthening design will not generally take into account the
contribution of injecting adhesives in the cracked area, so-called consolidation. It
is felt that this contribution might be significant. There was also the question as
to whether it would be possible to identify in advance some typical failure
configurations likely to benefit in practice from the so-called consolidation
technique.
A test programme was therefore developed at LNEC with the intention of
assessing the strength and stiffness of timber beams which were consolidated
with epoxy adhesives after they had developed failure in 4 point bending tests.
References to studies involving timber crack injection could not be found in
the literature, although they have some similarity with glued timber delamination
repair. For the specific case of crack consolidation, with complex irregular
surfaces of the adherend it is necessary to find adhesives with suitable rheology.
Thick gap-filling adhesive which usually contain fillers are not deemed to be
suitable, whereas very fluid adhesives will penetrate more easily but will not be
capable of filling small voids left inside the timber member, reducing bonding
efficiency at those points. This was observed in previous trials by the authors [1],
where a poor crack consolidation (and corresponding low strength recovery) was
obtained by the combined use of a thick gap-filling adhesive followed by
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
541
injection with a very fluid one. To avoid this problem, a thixotropic adhesive was
used in the current experimental programme.
Another peculiarity of this kind of repair lies in the fact that preparation of
clean bonding surfaces is not possible. Wood surface oxidation and any possible
air borne or liquid contamination will have taken place since the time between
fracture and repair intervention could be days or even years depending upon
discovery of the fracture and application of the chosen remedial solution.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 1:
Table 1 presents individual test results in the first (initial sound state) and in the
second (consolidated) round of tests, concerning the global modulus of elasticity
and bending strength.
In annex A are presented some examples of the beams after the second round
of bending tests. The fibre separation lines on both faces and bottom edge are
enhanced on the photos (fine lines for old cracks before consolidation, thick
lines for cracks developed on the consolidated beams).
Load-deformation diagrams obtained for each beam, in both rounds of
bending tests (initial state and consolidated beam), are presented in the annex B.
The average recovery of strength due to consolidation with the epoxy
adhesive was 0.72 (ranging from 0.45 to 1.11). In most cases, the initial part of
the load-displacement test diagrams obtained before and after consolidation were
very similar. Therefore, similar global modulus of elasticity was obtained 0.90
efficiency was obtained with individual beams (ranging from 0.74 to 0.99).
This seems to indicate that consolidation was effective till a certain point
under moderate loads although it was not enough to recover the performance of
the solid timber beam for stress levels closer to the initial bending strength.
This is confirmed by the fact that bending failure in the consolidated beams
generally affected the same material, that is to say that it occurred in the same
areas where the initial beam had previously failed.
There is not a clear correspondence between crack pattern and efficiency of
the repair: apparently similar cracks have conducted to distinct results within the
range of strengthening values obtained.
Despite the small number of beams in each group, there is no clear distinction
between groups 1-10 and A-F. In both cases, the application of the epoxy
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
543
adhesive took place several months after failure, thus simulating the application
of adhesive to old timber surfaces.
Table 1:
Beam
Econs/Eini
Fini
Fcons
Fcons/Fini
11921
9684
0.81
29.88
14.00
0.47
16193
15504
0.96
56.99
63.13
1.11
17371
14680
0.85
45.74
20.77
0.45
12479
12033
0.96
43.93
33.18
0.75
12504
10660
0.85
31.49
20.93
0.66
13378
12843
0.96
41.10
42.48
1.03
11706
(1102)
32.68
6.11
15261
15105
0.99
40.42
33.84
0.84
14431
12162
0.84
44.39
24.26
0.55
10
15683
14989
0.96
50.40
27.49
0.55
14169
10519
0.74
59.05
43.54
0.74
13364
11668
0.87
44.54
36.68
0.82
14183
13913
0.98
51.66
27.49
0.53
Mean (1-10)
Eini
0.91
0.71
12636
11400
0.90
48.39
23.42
0.48
13565
12345
0.91
58.67
55.30
0.94
Conclusions
It is recognised that partially broken beams may have very narrow, almost
invisible, fissures, which are difficult to impregnate. Together with the fact that
partially separate broken beams may present extremely irregular split ends,
which are difficult to bring back into perfect contact. These two factors may
reduce the effectiveness of the consolidation and were considered in the
experimental study. However, the specific beam failure appearance is not likely
to give a clear indication of how efficient the consolidation turns out to be.
Nevertheless, test results show that epoxy adhesive injection of cracked
timber beams may lead to a not negligible, but somehow variable, strength
recovery (0.45 to 1.11) and to a significant stiffness recovery (0.74 to 0.99) of
broken beams.
Having these results in mind, although on site repair of broken timber beams
should not rely on epoxy adhesives injection only, its contribution to the stiffness
of strengthened beams is significant and should not be ignored.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Annexes
Annex A
Beam 1
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Beam 3
Beam 6
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Beam E
545
Annex B
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
547
Acknowledgement
This work was carried out in the scope of the European Research Project
CRAFT-1999-71216 LICONS Low Intrusion Conservation Systems for
Timber Structures (www.licons.org), which had a much wider scope.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1] COLORETIM CT 98-9548, COmposites LOcal REinforcement for TIMber
structures, Final Report Consolidating tests: bending tests on timber
beams strengthened with FRP.
[2] Norma Portuguesa NP 4305: Madeira serrada de Pinheiro bravo para
estruturas. Classificao visual Visual strength grading of maritime
pine.
[3] European Standard EN 408: Timber structures Structural timber and
glued laminated timber Determination of some physical and mechanical
properties. Draft revision, November 1999.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
549
Abstract
Commercial Al-Li alloys have strength and weight advantages over non-Al-Li
alloys. The fracture behavior of these alloys is unusual and has limited their use.
The fracture mode, described as delamination, is intergranular, along the broad
grain boundaries parallel to the rolling plane of the plate. Microtexture analyses
have shown that delaminations occur along boundaries with greater than 30
misorientation. However, it was observed that relatively few of the high angle
boundaries exhibited this behavior. Some grains of the retained deformation
texture show high internal misorientation, which is a measure of stored strain
energy. Delamination tends to occur between these grains and adjacent,
recrystallized grains. Nanoindentation studies indicate a higher hardness for the
high internal misorientation grains. These results suggest that the delamination
could be reduced by processing the alloys to minimize grain-to-grain property
disparities.
Keywords: microtexture, nanoindentation, Al-Li, delamination.
Introduction
551
Side
Groove
Figure 1:
Side
Groove
In earlier work [14, 15], microtexture methods were used to examine the
misorientation angles of delaminating boundaries, the possible role of special
boundaries, and similarities between boundary planes and precipitate planes.
These studies showed that delamination occurs at high angle boundaries in
excess of 30 misorientation. No conclusive correlation was found with grain
boundary types, crack path and precipitate distributions [15]. However, Kalu and
Wagner [14] showed that delaminations in alloy 2090 tended to occur adjacent to
Experimental
Fractured 2090 compact tension samples were sectioned normal to the fracture
surface and crack direction in a region containing several delaminations.
Sectioning was performed with a slow speed diamond saw. Fracture surfaces
were protected by painting with an acetone soluble stop-off lacquer (Microstop),
mounted in 1.25 diameter epoxy mounts and metallographically polished with a
specimen load of ~ 4lbs./sample. The polishing procedure involved the following
abrasives and times: 600 grit SiC until flat, 1200 grit SiC for 15 minutes, 3 m
Diamond paste with Struers Blue lubricant for 30 minutes on Buehler Texmet
2000 cloth, 1 m Diamond paste with Struers Blue lubricant for 30 minutes on
Buehler Texmet 2000 cloth. Final polish used a 1:1:2 mixture of Stuers colloidal
silica (OP-S), soap (Liquinox) and water. Final polishing was for 1 hour on
Buehler Mastertex cloth. Samples were then freed from the epoxy and cleaned in
acetone and alcohol with ultrasonic agitation. This method provided a flat, fairly
strain-free surface, which was well suited for collection of microtexture data
along the delamination cracks. The surface quality was comparable to that
obtainable from electropolishing methods, but with a much flatter surface. A flat
surface is important since even the minor undulations common to electropolished
samples interfere with the microtexture data collection at the required 70 of tilt,
especially in vicinity of cracks. Grains on either side of a delamination were
studied via microtexture.
A JEOL 6400 scanning electron microscope (SEM) equipped with a
microtexture system was used to study microtexture and the distribution of
nanoindents.
Microtexture data were collected in large sets of ~ 40,000 solved diffraction
patterns per set, with a spatial separation of 0.5 m. Analyses of the data
included grain boundary misorientation; internal misorientation; discrete texture
from points of the data set; identification of grains by proximity to specific,
characteristic orientations (or texture components); and internal misorientation.
Internal misorientation is a measure of the spread of orientations bound by a high
angle boundary; i.e. it is essentially a quantification of the amount of
substructure. Since the recrystallization process removes substructure, a low IM
is an indication of a recrystallized condition. The data were plotted as colorcoded maps, where appropriate. Data were collected from around delamination
cracks and from undeformed material. A simple method of indicating the grain
orientation will be used in this paper, namely showing the distribution of indexed
points on a discrete pole figure. Discrete pole figures associated with
microtexture maps can be used to depict the grain orientations.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
553
High
IM
Figure 2:
GOSS
Figure 3:
Cu
555
shows an IM of near 5, which is about twice as high as the Goss grain across the
crack. Large internal misorientation was taken as an indication of high stored
strain energy, such as would be expected for a grain of the deformation texture.
Low internal misorientation was an indication of recrystallization subsequent to
rolling, which is likely to have developed during solution heat treatment of the
sheet prior to artificial aging.
For sheet or plate to be given a T8 aging treatment, a stretch of about 6 8%
is applied subsequent to solution heat treatment and prior to aging.
Recrystallized grains (without substructure) are then given a significant
dislocation density which aids in precipitate nucleation. The increased density of
precipitates increases the strength considerably. However, the grains with well
developed substructure should exhibit precipitation on subgrain boundaries, and
should develop an even higher dislocation density due to dislocation
multiplication effects during deformation. These grains may develop greater
hardness during aging. A nanoindentation study was used to investigate this
possibility. Figure 4 shows a series of nanoindents traversing a delamination
crack. The grain on the right side of the crack, which was near the Brass
orientation, showed a hardness increase of about 20% greater than the grain
across the crack. This was despite the indent being at the very edge of the crack,
which might be expected to result in a low hardness value.
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusions
References
[1]
[2]
[3]
[4]
Sanders, Jr., T.H., & Starke, Jr., E.A., (eds). Al-Li Alloys I, Proceedings of
the First International Aluminum-Lithium Conference, Stone Mountain,
Georgia, 1980.
Sanders, Jr., T.H. & Starke, Jr., E.A., The effect of slip distribution on the
monotonic and cyclic ductility of Al-Li binary alloy. Acta Metall., 30, p.
927, 1982.
Csontos, A.A. & Starke, Jr., E.A., The effect of inhomogeneous plastic
deformation on the ductility and fracture behavior of age hardenable
aluminum alloys. International Journal of Plasticity, 21, pp. 1097-1118,
2005.
Howe, J.M. & Vasudevan, A.K., Structure and deformation behavior of T
sub 1 precipitate plates in an Al-2Li-1Cu alloy. Metall. Trans. A, 19A, p.
2911, 1988.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
557
Crooks, R., Wang, Z., Levit, V.I. & Shenoy, R.N., Microtexture,
microstructure and plastic anisotropy of AA2195. Materials Science and
Engineering A, A257(1), pp. 145-152, 1998.
Jata, K.V. & Vasudevan, A.K., Effect of fabrication and microstructure on
the fracture initiation and growth toughness of Al-Li-Cu alloys. Materials
Science and Engineering A, A241, pp. 104-113, 1998.
Suresh, S., Vasudevan, A.K., Tosten, M. & Howell, P.R., Acta Metall.,
35, p. 25, 1987.
Vasudevan, A.K. & Doherty, R.D., Acta Metall., 35, p. 1193, 1987.
Vasudevan, A.K., Ludwiczak, E.A., Baumann, S.F., Howell, P.R.,
Dougherty, R.D. & Kersker, M.M., Mater. Sci. Technol., 2, p. 1205, 1986.
Lynch, S.P., Materials Science and Engineering A, A136, p. 25, 1991.
Lynch, S.P., Muddle, B.C. & Pasang, T., Ductile-Brittle Fracture
Transitions in 8090 All-Li Alloys. Acta mater, 49, pp. 2863-2874, 2001.
Adams, B.L., Orientation Imaging Microscopy: Emerging and Future
Applications. Ultramicroscopy, 67, pp. 11-17, 1997.
Schwartz, A.J., Kumar, M. & Adams, B.L., Electron Backscatter
Diffraction in Materials Science, Kluwer Academic/Plenum, New York,
2000.
Kalu, P.N. & Wagner, J.A., A Microtexture Investigation of the Fracture
Behavior of Al-Li Alloy, 2090-T81. The 3rd Light Weight Alloys for
Aerospace Applications, ed. E.W. Lee, TMS, Warrendale, PA, pp. 157167, 1995.
Hales, S.J. & Crooks, R., Recrystallized and unrecrystallized Al-Li alloys
for elevated temperature service. Proc. 3rd International Conf. on
Recrystallization and Related Phenomena, ed. T.R. McNelley, MIAS, pp.
511-518, 1997.
Oliver, W.C. & Pharr, G.M., An Improved Technique for Determining
Hardness and Elastic Modulus Using Load and Displacement Sensing
Indentation Experiments. J. Mater. Res., 7(6), pp. 1564-1583, 1992.
Torralba, J.M., Velasco, F., Costa, C.E., Vergara, I. & Caceres, D.,
Mechanical behaviour of the interphase between matrix and reinforcement
of Al 2014 matrix composites reinforced with (Ni3Al)p. Composites: Part
A, 33, pp. 491-496, 2002.
Liu, C., Qin, S., Zhang, G. & Naka, M., Micromechanical properties of
high fracture performance SiCp-6061/6061 Al composite. Materials
Science and Engineering A, A332, pp. 203-209, 2002.
Ogura, T., Hirosawa, S. & Sato, T., Quantitative characterization of
precipitate free zones in Al-Zn-Mg(-Ag) alloys by microchemical analysis
and nanoindentation measurement. Science and Technology of Advanced
Materials, 5, pp. 491-496, 2004.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
559
Abstract
The aim of this work has been the study of an electrochemical treatment that
allows the prevention and halting of chloride corrosion processes in passive
reinforced concrete. This treatment is based on a desalination system that
removes free chloride in concrete simultaneously increasing the pH around
reinforcement steel until its repassivation.
With this purpose a concrete has been manufactured with mixing proportions
close to that used in civil engineering. The chloride attack has been carried out at
room temperature, and the chloride removal process has been started at several
concrete ages. Chloride ions have been determined before and after the attack, as
well as the corrosion potential, in order to establish comparison between the
grade of corrosion of the reinforcement steel before and after treatment,
according to exposition time and initial corrosion situation. Relationships
between electrochemical parameters have been determined.
The studied procedure is a non-destructive and non-intrusive method that
does not change either the strength of the concrete or its external appearance.
The procedure uses non-pollutant materials, so no environmental damage is
caused.
Keywords: electrochemical chloride extraction, steel-reinforced concrete,
repassivation.
Introduction
26 cm
11 cm
Figure 1:
561
11 YY-specimens
11 XX-specimens
1 control specimen
chloride attack (7d)
11 ZZ-specimens
1 control specimen
1 control specimen
1 control specimen
1 control specimen
1 control specimen
ECE
(7d; 3 specimens)
ECE
(7d; 3 specimens)
ECE
(7d; 3 specimens)
ECE
(21d; 3 specimens)
ECE
(21d; 3 specimens)
ECE
(21d; 3 specimens)
ECE
(42d; 3 specimens)
ECE
(42d; 3 specimens)
ECE
(42d; 3 specimens)
Figure 2:
Concrete
Cement CEM I 42,5R
Calcareous Sand 0/5
Calcareous gravel 5/12
Calcareous gravel 12/18
Water
w/c
Compressive strength 28d
Steel bars AEH-400N
Diameter
Modulus of elasticity
Breaking stress
346 kg/m3
1289 kg/m3
521 kg/m3
739 kg/m3
225 L/m3
0.65
45 MPa
3 mm
4100 kp/cm2
4500 kp/cm2
Figure 3:
563
X-specimen set
(7 day attack)
-110
Y-specimen set
(56 day attack)
-110
Z-specimen set
(90 day attack)
-110
-283
-524
-621
157
376
464
X-specimen
set
(7 day attack)
28.3
Y-specimen
set (56 day
attack)
13.3
Z-specimen
set (90 day
attack)
17.1
5.84
12.6
10.9
6.73
9.38
8.90
28.3
13.3
17.1
40.0
38.6
34.8
60.2
66.7
61.6
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
70
X-specimen set
62,1
60
Y-specimen set
52,38
50
Z-specimen set
51,5
45,4
40,46
40
32,18
30
28,3
23,87
20,26
20
17,43
16,99
11,27
10
0
0
Week
Figure 4:
565
0,35
0,32
0,30
0,28
0,25
0,22
0,20
0,15
0,12
0,10
0,05
0,00
0,1
0,08
Exterior
0,02
0,02
0,05
0,04
Interior
Week
0,60
0,51
0,50
0,44
0,40
0,33
0,29
0,30
0,20
0,10
0,00
0,19
0,17
Exterior
0,12
0,02
0,02
0,07
Interior
Week
0,60
0,57
0,50
0,6
0,48
0,40
0,35
0,34
0,30
0,21
0,20
0,17
Exterior
0,10
0,00
0,09
0,02
0,02
Interior
Week
Figure 5:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusions
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
567
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 7
Adhesion and adhesives
571
Abstract
The characterization of materials or the study of resistant response associated
with new constructive solutions often convey progressive adjustment of models
of analysis starting from the results obtained by means of test campaigns. The
reliability of the conclusions will obviously be conditioned by the amplitude of
the experimental development which, as a result, should contain a sufficiently
representative sampling of cases. In the situations where significantly multiple
parameters intervene, specimens that involve possible values and combinations
should be examined, which in practice remarkably increases the number of
specimens to be treated. The case being presented deals with the evaluation of
the effectiveness of certain fixed joints carried out by gluing metallic bars into
timber pieces by means of adhesives of different formulations. The set of
calculus patterns, including the extensive variability of the associated
magnitudes, has been treated in a parametrized form. The geometry of the
studied pieces, as well as the sustenance conditions and the load, are
automatically generated by means of software specially designed for those
purposes. The resulting information is structured in an orderly way to facilitate
its reading and also its manipulation if needed. The files of the process obtained
in this way can optionally incorporate the meshing sequences, as well as the
commands of the calculus, analysis and postprocess. The adopted strategy
considerably reduces the time assigned to the phases of the geometrical
definition and resolution, and enormously simplifies the interpretation of the
results.
Keywords: numerical analysis, parametrized design, glued joints.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06056
Introduction
This work is placed in an investigation project directed to the study of the joints
behaviour in timber pieces by means of steel bars glued-in with an adhesive. The
approach causes the development of two parallel complementary lines of
advance: the first one is of experimental nature and the second one is based on
the use of the numerical analysis technique.
On the one hand it is pretended to use the results obtained in the experiments
for calibrating gradually the theoretical models in such a way that they, in their
turn, could be applied for extrapolating not specifically experimented situations
or facilitating the adoption of design decisions in certain practical cases. On the
other hand the numerical analysis can be useful for the selection of those
mechanisms of connection that prove to be more operative in advance, reducing
the temporary and economic costs associated with the creation of the specimens.
Figure 1:
Table 1:
d
10
10 12
12 14
573
60
90
120
150
180
60
90
120
150
180
60
90
120
150
180
Figure 2:
Li
Lm
Li=L
3.L 500
60
90
120
150
140
60
90
120
150
140
60
90
120
150
140
180
270
360
450
500
180
270
360
450
500
180
270
360
450
500
Le
150
150
150
Lb
210
240
270
300
330
210
240
270
300
330
210
240
270
300
330
Reference
a=6.d
Units: mm
48
60
72
1a
1b
1c
1d
1e
2a
2b
2c
2d
2e
3a
3b
3c
3d
3e
The attached diagram shows the proposed combinations for the 9 geometrical
parameters to deal with, as well as the type of the timber and the type of the
applied adhesives. This set appears to be sufficiently full for establishing
relevant conclusions in relation with the incidence of each magnitude and even
facilitating formulations that will permit to define the necessary anchorage
lengths.
The numerical models are calculated with the help of the application of Ansys
Multiphysics, Swanson Analysis Systems Inc., initially assuming an elastic and
lineal behaviour, noticeably correct in accordance with the experimentally
obtained stress-deformation diagrams for load values not surpassing the 75-80
per cent of the limit number.
The volume of the pieces, the possible parametrization of its geometrical and
mechanical values lead us to the development of a specific software that
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
575
principle, ample divisions are managed to tackle the first assessment of the
problem. Later comes the restructuring of the configuration on the basis of a
radial scheme that serves to a better estimation of the stress distribution in
different materials, especially in the areas close to the contact surfaces (fig. 3).
Figure 3:
Figure 4:
At the beginning of the process the coordinates (x, y) of the 105 points, that
define the base plane, are stored in matrices. In the area nearest to the centre of
the section the mentioned values are deduced by means of a polar system of
reference, while in the exterior panels a Cartesian type of analysis is proposed. It
should be paid attention to the fact that the parametrization of the process of
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 5:
577
At the end of this phase the volumes that are situated between the zones z1
and z2 (additional band of 2 mm) relative to the bottom of the bar and the
adhesive are eliminated. This elimination is decided after the real specimens are
analyzed with a very reduced efficiency of the adhesive in this area, as a
consequence of the filling process itself. In figure 4 the resulting slot caused by
the elimination commands is estimated (VDELE, NV1, NV2, NINC: delete
volumes from NV1 to NV2 in steps of NINC).
To finish the preprocess of the model, the obtained volumes are related to the
involved materials, a task in which controlling the numeration of the first items
proves to be fundamental. The selection is realized by means of ranks, a reason
for which in each section at first the central steel elements are coded, then the
relatives to the band of the adhesive and finally the constituent elements of the
timber.
VSEL,Type,Item,,VMIN,VMAX (values of the item range),VINC (increment)
VATT,MAT,REAL,TYPE (parameters to be assigned)
The meshing only affects the final volumes (VMESH,ALL). With a previous
discretization, a subdivision of each edge in two halves (ESIZE,2) is considered
correct. The application permits to refine the mesh in determined areas, but
advanced studies indicate that the resultant model is sufficiently approximate
without further modifications of the set.
In the module of the calculus the transversal displacements coerce the symmetry
planes (used for reducing the real specimen to its eighth part). So the external
load is applied to the extreme surface of the steel bar, and the phase of analysis
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 6:
Figure 7:
The diagrams show curved lines with maximal numbers focused on the ends
of the anchorage area which are much greater than the average value traditionally
used in the references. Both peaks turn to be similar for reduced longitudes,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
579
increasing noticeably the exterior one when the union is realized with more
profundity.
Figure 8:
Conclusions
The means in parallel to experimental technique and numerical analysis, for the
study of the resistant behaviour of timber-steel joints as the described ones,
supposes a sufficiently complete and illustrative strategy. The comparison of the
results allows to calibrate gradually the virtual models in such away that they
prove to be useful for verifying practical cases that are not specifically subjected
to tests, or for predicting possible situations of risk. Likewise, the mentioned
idealizations are useful for assessing constructive solutions at the very beginning,
thus reducing the relative costs of production and of the load of possible
prototypes. In this generic frame the computerized configuration of the models
remarkably simplifies the generation and postprocess tasks appealing to the
parametrization of its geometry.
Acknowledgements
This research is sponsored by the Ministry of Science and Technology through
research project title uniones metlicas encoladas con adhesivos en barras de
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
581
Abstract
It is admitted that a solid body is obtained by means of extraction from an
infinite medium, and at the same time its state changes. To describe this
phenomenon there a linear theory of elasticity of isotropic materials of the
second order was used, generalized for the case of availability of the initial
stressed state described by tensors of various ranks. They depend upon the
physical nature of the body, the form of its boundary and position of points for
their determination. The boundary effect existence of surface tension and
surface energy is described. The analysis of known test data regarding values of
these quantities, wave dispersion in resilient media, calculus results on the basis
of the model offered and one-dimensional theory of the crystalline lattice of
elastic energy density for a number materials allowed values of elastic modulus
and initial stresses to be determined. The created model of medium is used for
the description of the stressed state of bodies adhered together, when their
adhesion is considered to be one of the kinds of contact interaction. The calculus
results of adhesion energy are confronted with the known data.
Keywords: theory of the elasticity, materials of the second order, the initial
stressed state, and surface energy, adhesion energy.
Introduction
It is known that particles of solid body - atoms, ions, interact with the aid of the
potential forces [1]. Therefore a change in the position at least of one particle of
a body affects the position and the mutual influence of all remaining particles.
The formation of the new section of free surface, for example, with the division,
is accompanied by the separation of the sets of particles. On the one side, the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06057
Interaction of the atoms of real body is described with the aid of the
potential of pair wise interaction. It is assumed that for the continuous
simulator interaction of infinitely small particles is also potential, paired. The
potential of interaction is proportional to their volumes dV and dV1
( = const - material density), that have at the moment of the time t of
G
G
G G
position R and R 1 . Constant of proportionality = R 1, R (further
2.
3.
)
) (
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
583
G
G
d = R 1 R dV1 dV .
(1)
G
(
(
) (G
))
(d ) = w dV = R1 R + ( u ) R1 R dV1 dV =
G G T
G
= R1 R ( u )dV1 dV .
( (
(2)
))
G
G
G
Potential gradient F = R 1 R is written in the form
( )
G G
G
G G
GT
F R1 R = F (a ) + F
G
u =0
G
u .
(3)
G
In this case for the relative displacement u is considered valid the idea in the
G
form Taylor series according to the external degrees of vector a [7]
G
u =
k ! [rG(k )u ]G
k =1
G
... a k .
(4)
a =0
w=
P (m )T ... ( rG (m )u ) .
m =1
where:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(5)
0(m )
n =1
G G Gm
[F (a )a ]dV1 ,
C (m , n ) =
(6)
(7)
G G
G
1
a n aG F a m dV1 .
n! m!
(8)
G
It follows from idea (5) that the infinite sequence rG (m )u of the gradients of
displacements is the characteristic of the strained state in the infinitely small
G
environment dV of point r . Its elements - these are the generalized
displacements, determined in dV . By the generalized forces, which work on
them, appear the stress tensors, determined by equalities (6) - (8). They compose
{ }
(
(
))
G
n (P ( ) (P ( ) ...))
2
)))
( (
))
G
(n (P ( ) ...)) = ( )
3
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
G
r V ,
(9)
G
r S ,
(10)
G
r S .
(11)
585
that possess surface densities (k ) , k = 0 ,1, ... . With k = 0 this is the density of
the classical surface forces, which accomplish work on the true displacements of
material points. With k > 0 - this is of density over forces [8]. System (9) - (11)
must be augmented by the differential form of the law of conservation of
momentum of pulse and by initial conditions for pour on displacements and their
speeds.
G G
2.4 Version of the potential = R1 , R determination
It can be realized for the isotropic medium. In this case it suffices to examine
one-dimensional situation. Equation (9) taking into account (6) - (8) which
G
f = 0 takes the form
1 2 u1
c2 t 2
2 u1
x12
+ b12
4 u1
x14
+ b2 4
6 u1
x16
+ ...
(12)
, D j = D1b j 12( j 1) ,
j = 1, 2 , ... ,
c=
(13)
................................. ................
The search for the solution in the form u1 = e i ( t kx1 ) , i = 1 , gives the
possibility to obtain dispersion law dependence (k ) [1], in the form of the
power series
2
c
= k 2 b12 k 4 + b2 4 k 6 ... .
(14)
The same law can be built theoretically, with the use of ideas of solid state
physics [1], or on the basis of data of experiments [9]. With the comparison it is
G
possible to determine values b j . Function = (a ), a = a can be
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
p e
, p =
p =1
a
bp
(15)
Substitution (15) in (7), (8), and then in (13) reduces to the infinite system of
linear algebraic equations for the unknowns p . Its solution determines
potential (15) of material, for which are obtained the values b j .
g kk g ll +
(16)
1
ui , j + u j ,i and Z ijk = ui , jk the component of the classical strain
2
tensor and second gradient of displacements; , the Lame coefficients;
g ij =
587
0 =
Wp
k
,b =
3W p
2
4k A1
, k=
2 +
, 2 A1 = 2 + , b 2 = b1 2 + b2 2 . (17)
the estimations of additional elastic constants are made for the number of
materials b and 0 . The results of calculations are represented in table 1.
Table 1:
Element
W p Dg / m 2
1
2
3
4
5
6
7
Mg
Al
Si
Ni
Cu
Ag
Au
0,73
1,04
1,24
1,44
1,12
0,95
1,23
b 1011 (m )
0 Dg / m 2
2,67
4,65
7,54
2,49
1,89
1,98
1,29
1,28
1,92
3,21
2,50
1,57
1,28
1,46
The use of a dispersion law [1] for determining the parameter b leads to the
results, commensurate with those given.
3.3 Adhesion of two solid bodies
By adhesion of two solid deformed bodies B (1) and B (2 ) in the model of
medium is understood this form of their contact, when the rectilinear material
fiber, normal to the surface of contact S a and which intersects it, in the process
of the deformation of the united of body B = B (1) B (2) preserves not only its
integrity (as in the case of classical, solid contact), but also smoothness of the
distribution of its deformations. This idea gives the possibility to solve the
problem about the stress-strained state of the body, which are in a state of
adhesion as the combined task about the state of piecewise-uniform body
B = B (1) B (2 ) with the condition of the absence of external actions on it.
Adhesion energy F of bodies B (1) and B (2 ) is determined by loss per the unit
surface area, along which occurred the adhesion, their free energy [12]
F = W p(1) + W p(2 ) W p(1,2 ) .
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(18)
the thickness, for much the great significance of the parameters b ( j ) , is obtained
the formula
F =
W p( 1 )W p( 2 ) ( k ( 1 ) + k ( 2 ) )2
W p( 1 ) k ( 2 )2 + W p( 2 ) k ( 1 )2
, k( j) =
( j)
,
1 ( j )
(19)
( j ) Poisson ratio. The results of the comparison of the results of theory with
the data of experiment are represented in table 2.
Table 2:
Metal
(Dg
glass Dg m 2
aluminum (Al)
0,214
0,150
silver (Ag)
0,181
0,120
copper (Cu)
0,210
0,110
Conclusion
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
589
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 8
Optimal design
593
Abstract
The main purpose of this study is to present results of the design and damage
analysis of industrial vessel shell microstructures made of fibre reinforced plastic
laminates and subject to mechanical and aggressive chemical loads due to
sulphuric and hydrochloric acids and high temperatures. Methods to obtain
optimal performance microstructure are use of balanced and symmetric laminate
layering with optimal fibre strengthening directions and volume fractions.
Optimal selection of chemically resistant fibres and matrix is essential for
endurance. Use of thick enough walls is cost-effective for vessel bottoms. Two
vessel winding manufacturing technologies are compared for optimal selection.
Keywords: composite materials and structures, optimal design, stress corrosion
cracking.
Introduction
Composites used for structural engineering are commonly formed from five
commonest material groups, metals, ceramics, glasses, elastomers and polymers.
Design rules and codes are based continuum models. Another trend is to use
more microscopic modelling. Useful information of composite design is
presented by Agarwal and Broutman [1] and Barbero [2]. Chemical resistances
are studied by Aveston and Sillwood [3] and, Kawada and Srivastava [4]. In
aggressive enough loadings all material alternatives and combinations suffer
accumulation of damage. Typical damage events are stress corrosion wearing of
outer surface, fracture of fibres and degradation of matrix. The main purpose of
this study is to present results of design and damage study analysis of industrial
vessel walls.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06058
2.1 General goals of designing thin shelled structures for process industry
In metallurgical chemical industry concentrated sulphuric and hydrochloric acids
are processed at about boiling temperature in thin shelled large vessels. The
focus is now on shells made inorganic composites, Figure 1.
Biopolymers
Natural
fibres
Polymers,
elastomers
Ceramics,
glasses, inorganics
Old
Metals T=10y
A B New
T=0.1y
FRP,inorganic
FRP
biobased
MMC
Rm,new
Composites
Metal -metal-interphases
a)
Rm,old
b)
Figure 1:
Background theory
L Q11
= Q
T 12
LT 0
Q12
Q22
0
0 L
0 T
Q66 LT
(1)
c 2 s2
2sc x
L
= s 2 c 2 2sc
y
T
LT k sc sc c 2 s 2 xy
k
{ L } k = [T]{ x } k
595
(2)
)
(c
(3)
4
+ s4
Q16 Q16, k = (Q11 Q12 2Q66 )sc 3 (Q22 Q12 2Q66 )s3c
Q26 Q21, k = (Q11 Q12 2Q66 )s3c (Q22 Q12 2Q66 )sc 3
The global midplane strain induces global stresses at layer k with angle to xaxis
x
Q11
y = Q12
xy
k Q16
Q12
Q22
Q26
Q16 x 0
Q26 y 0 ,
Q66 xy 0
{ x } k = [Q ]{ x 0 }
(4)
Nx
Ny =
N xy
x
hk -1 y dz =
k =1
xy
k
k =n h
k
k=n h
k
k =1
k -1
k dz
(5)
= Q = Q ( 0 + z ) = Q 0 + zQ
In component vector form the force resultant vector is
Nx
N y = Aij
N xy
[ ]
x0
x
0
y + Bij y
0
xy
xy
[ ]
(6)
[ Aij ] = Qij
k =1
k=n
( hk hk-1 ) ,
(7)
k =1
t1
1=+45 h0 = -6
k=2
x
k=3
t2
2=-45
h1=--3 h2 = 0
t3
3=-45
h3=3
k = 4 = n t4
4=+45
z
T
y
Figure 2:
h4 = 6
L
Principle of a symmetric and balanced composite +45/-45/-45/+45.
The ABD matrix relates the midplane strains and plate curvatures to resultant
force vector N and moment vector M. The ABD matrices are extensional
stiffness matrix A, coupling stiffness matrix B and bending stiffness matrix D.
N A B 0
A B 0
=
M B D
B D
A 0
0
B
(8)
In formulating the design goals and constraints the symmetry condition is the
stiffest one. These rules can be utilised to obtain simple and well functioning
microstructures.
3.1.3 Symmetric laminate structure as one optimisation goal
The useful result of using symmetry goal is that the B matrix is made zero. Using
the case study data
[Q ]k=1,+45deg = [Q ]k=4,+45deg
(10)
N A 0 0
M = 0 D
A 0 0
0 D
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
A 0
(11)
597
(12)
Q16 ' sin( ) = s = (Q11 Q12 2Q66 )sc 3 + (Q22 Q12 2Q66 )s3c = Q16
(13)
N x A11
N y = A12
N xy A16
A12
A22
A26
A16 x 0 A11
A26 y0 A12
A66 xy0 0
A12
A22
0
, T = t
0 x0
0 y0
A66 xy0
(14)
(15)
A12 x 0
N x
x
1
pr 1
= pr
0 , = t = t
N
A22 y
t 0.5
0.5
y
y
(16)
N xy = A66 xy 0
Q12 x 0
Q22 k y 0
(17)
1
A66
a12 N x
1 A22
=
a 22 N y det A A12
,
A12 N x
A11 N y
(18)
Q11
Q12
Q12 a11
Q22 k a12
a12 1
pr
a 22 0.5
(19)
2
c2
T,k = s
LT,k sc sc
2sc x,k
2sc y,k
c 2 s 2 xy = 0
(20)
d
5
b)
2a
hackles
4
c)
a)
Figure 3:
d)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
599
1
N
Rm =
.
Rm = 10 a t b 100 t 0163
1
N
Rm0
(21)
tp
, Rm [ MPa ], t[ years]
(22)
300
Bending Rm= 250MPa,
new samples
tested in air.
Bending Rm=150MPa,
after 2 years in liquid acid
200
Ultimate
strength Tensile
(MPa) Rm=
100
Tensile test
Log(Rm)-Log(T)
prediction model for samples
held in gas 5 years
Rm= 69 MPa
k
190MPa
new
Tensile
Rm=100MPa
2 years in liquid
Tensile
Rm= 80MPa
2 years in gas
0
0
Figure 4:
da
= AK I n = 9.55 10 11 K I 3
dt
units: V
[ ],
m
s
t[sec], [ MPa]
(23)
1
Y ' = Y 2
KI = aY' a , n 3 ,
a is applied stress, static or dynamic, a is crack length at time t, Y is a
geometrical factor for crack. For a crack of length 2a in a plate, Y = 1 , Y =
Since the life of fibres is decisive for the whole composite, the fibre life can
be regarded as the life time of the composite under full loading. Initial crack
size is ai and final aIC. Stress corrosion life is from initial to final crack size
dt =
da
AK I n
da
A aY' a
tC
dt = t C =
AY ' n a n
0
C
1
1 2 ( 2 n ) (24)
a
1 n
ai
For short time testing, the applied stress is raised up to tensile strength while the
crack starts at initial size. Stress intensities at initial and final crack sizes are
K Ii = a Y ' a i , K IC = a Y ' a C K IC = max Y ' a i , max = Rm (25)
Analytical optimisation
Ef = 70000
f = 0.22
Em = 3400
m = 0.38
s=
2L
,e
d
L =
e 1
e + 2s
Ef
Em
fu ( Im) = 0.012
T =
e 1
e+2
Ef
2(1 + f )
Em
Gm =
2(1 + m )
Gf =
v=
Gf
Gm
G =
, z =1
v 1
v+z
(26)
(27)
601
0
1
LT TL
LT TL
11
11
(28)
ET
= LT E L
22 , 21 = 12
0
22
1
E2
E1
LT TL 1 LT TL
12
G12 12
0
0
Next the transformed Q k for layer k are calculated. The two optimising
conditions of symmetry and balancing have been applied
For k= 1 to 4, calculation of global stresses in lamina k
(30)
s = sin( k ) , c = cos( k )
{ x } k = [Q ]{ x 0 }
(31)
Next k
5. For k = 1 to 4 loop for calculating lamina k stresses and strengths
Calculate A matrix
4
[ Aij ] = t
(( ) ( ) ( ) + ( Q ) )
(32)
Q12 a11
Q22 k a12
(33)
x,k
=
y,k
Q11
Q12
ij 4
a12 1
pr
a22 0.5
(34)
Strength in layer k
Strength in L direction is related to volume fractions of fibres
EL
V
= p fL VfT = 1p VfL
ET
VfT
, VfL + VfT = Vf 0
(35)
(36)
(37)
(39)
(40)
Fk ( k ) = L L T + T + LT 1
LU k LU k TU k TU k LTU k
(41)
Next k
5. Next k,
for each lamina stresses and strengths and failure criteria
4. Next Ith, Angle (Ith) is varied
3. Next Iss, s = 5000, 20, 1, aspect ratio is decreased if
environmental attack by stress corrosion cuts fibres
2. Next IVf, Vf = 0.14 to 0.46 is lower at inner and higher for support layer
1. Next Itt, wall thickness
Ef
Em
Figure 5:
2L
2r=d
mL
603
goal is to constrain product of failure criterions. The third goal is to obtain stress
corrosion life of fibres sufficiently large by fibre and matrix material selections.
A. At new materials, interfacial matrix strength is i=83MPa, LT strengths are
LUk=320, TUk=125. When long fibres s=5000, failure F=0.073. Strains
x=0.0015, y=0.0028 are close to allowed strain 0.0015.
B. Next matrix degradation is assumed from i =83 to 40 causing F rise to 0.1
C. If cracking decreases aspect ratio to s=20, with lowing of matrix strength to
i=40, then F increases further but not much to 0.11.
D. If cracking of fibres is maximal, then aspect ratio is minimal, s =1. If also
matrix is degraded to i=40, then LUk=40 and TUk=40. Failure criteria are large
predicting full failure F >4. Also strains are large x=0.004, y=0.01.
E. Stress corrosion cracking life at pressure 1 MPa gave hoop stress 45. At 1N
H2SO4 acid load reasonable threshold stress intensity is KIi=5 for typical E-glass
giving 0.02 years. Somewhat more chemically resistant E-glass with KIi=1 may
give 0.1 years. Most optimal is to use such fibres which endure these acids well.
Microscopy characterisation
2
3
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Conclusions
Acknowledgements
The assistance in SEM analyses by Mr. Markku Levomki, MSc, is gratefully
acknowledged. This research is supported by EU Asia Link Project (Contract
Reference no.: ASI/B7-301/98/679-023)
References
[1]
[2]
[3]
[4]
605
Abstract
In this paper we deal with the topology and standard optimization of unbraced
steel frames with rigid beam-to-column connections. The optimization has been
performed by the Mixed-Integer Non-linear Programming (MINLP) approach.
The MINLP performs a discrete topology and standard dimension optimization,
while continuous parameters are simultaneously calculated inside the continuous
space. As the discrete/continuous optimization problem of steel frames is
non-convex and highly non-linear, the Modified Outer-Approximation/EqualityRelaxation (OA/ER) algorithm has been used for the optimization. Two practical
examples with the results of the optimization are shown at the end of the paper.
Introduction
Steel frames
Steel frames, see Figure 1, are proposed to be analyzed under the combined
effects of the self-weight of frame members, uniformly distributed variable load,
concentrated variable load on each storey and an initial frame imperfection.
Second-order elastic structural analysis is performed by considering a
geometric nonlinearity due to P- and P- effects. While the P- effect is
associated with the influence of the axial force on the beam-column member
flexure, the P- effect denotes the influence of axial force acting through the
relative sideways displacements of the element ends. In this paper, both effects
are included in the nonlinear stiffness matrix of the individual frame member by
usage of stability function approach. Chen and Lui [5] have comprehensively
described this approach and have presented stability functions sii and sij. These
stability functions are different for compressive and tensile axial forces. Since
they give the indeterminate numerical solution when axial force is zero,
simplified expressions for stability functions S1 and S2, introduced by Kim et al.
[6], are rather used. The shear deformation effect is neglected considering the
fact that only slender structural members are subjected to buckling for which
shear deformation is insignificant.
Design/dimensioning of steel frames is performed in accordance with
Eurocodes 3 for the conditions of both the ultimate limit and serviceability limit
states. When the ultimate limit state of beam-column members is considered, the
elements were checked for bending moment, vertical shear, shear buckling,
interaction between bending, shear and axial force and interaction between
bending and axial compression/buckling. The ultimate moment capacity is
calculated by the elastic method. Since the second-order elastic global analysis is
used, the in-plane buckling lengths of compression members are calculated
considering the non-sway mode.
Considering the serviceability limit state, the vertical deflections of beams in
the individual storey were calculated by the elastic method. The total deflections
subjected to overall load max and the deflections subjected to the variable
imposed load 2 are calculated to be smaller than the limited maximum values:
span/250 and span/300, respectively. The horizontal deflections are also checked
for the individual storey and for the structure as a whole. Both types of
horizontal deflections are checked for the recommended limits: the relative
horizontal deflection of each storey should be smaller than each storey
height/300 and the horizontal deflection of the top of the frame must be smaller
than an overall frame height/500.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
607
Lc
q
F
H
Lc
Lb
Lb
Lb
Figure 1:
s.t.
(MINLP-G)
x X = {x R : xLO x xUP}
m
y Y ={0,1}
where x is a vector of continuous variables specified in the compact set X and y
is a vector of discrete, mostly binary 0-1 variables. Functions f(x), h(x) and g(x)
are nonlinear functions involved in the objective function z, equality and
inequality constraints, respectively. Finally, By+Cx b represents a subset of
mixed linear equality/inequality constraints.
The above general MINLP model formulation has been adapted for the
optimization of mechanical superstructures (MINLP-SMS). The resulted
formulation that is more specific, particularly in variables and constraints. It can
also be used for the modelling of steel frames. It is given in the following form:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
h( x ) = 0
g(x ) 0
A( x ) a
Ey e
(MINLP-SMS)
Dy + R( x ) r
e
( )
Py + S (d ) s
Ky e + L d cn k
st
x X = {x R : xLO x xUP}
m
y Y ={0,1}
In the model formulation, included are continuous variables x={d, p} and
discrete binary variables y={ye, yst}. Continuous variables are partitioned into
design variables d={dcn, dst} and into performance (nondesign) variables p,
where subvectors dcn and dst stand for continuous and standard dimensions,
respectively. Subvectors of binary variables ye and yst denote the potential
existence of structural elements inside the superstructure (the topology
determination) and the potential selection of standard dimension alternatives,
respectively.
The mass or economical objective function z involves fixed mass or cost
charges in the linear term cTy and dimension dependent mass or costs in the term
f(x).
Parameter nonlinear and linear constraints h(x)=0, g(x) 0 and A(x) a
represent the rigorous system of the design, loading, stress, deflection, stability,
etc. constraints known from the structural analysis.
Integer linear constraints Ey e are proposed to describe relations between
binary variables.
Mixed linear constraints Dye+R(x) r restore interconnection relations
between currently selected or existing structural elements (corresponding ye = 1)
and cancel relations for currently disappearing or nonexistent elements
(corresponding ye = 0).
Mixed linear constraints Kye+L(dcn) k are proposed to define the continuous
design variables for each existing structural element. The space is defined only
when the corresponding structure element exists (ye =1), otherwise it is empty.
Mixed linear constraints Py+S(dst) s define standard design variables dst.
Each standard dimension dst is determined as a scalar product between its vector
of standard dimension constants q and its vector of binary variables yst. Only one
discrete value can be selected for each standard dimension since:
d st = qi yist
iI
st
yi = 1
iI
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
609
Optimization
Examples
5.0 m
F = 10.0 kN
q = 50.0 kN/m
q = 50.0 kN/m
5.0 m
F = 10.0 kN
15.0 m
5.0 m
F = 10.0 kN
5.0 m
5.0 m
5.0 m
15.0 m
Figure 2:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
611
Variable imposed loads s = 1.60 kN/m2 (snow) and w = 0.137 kN/m2 (wind)
are defined as the uniformly distributed surface load in the model input data.
Both, the horizontal concentrated load at the top of the columns (wind) and the
vertical uniformly distributed line load on the beams (snow and wind) are
calculated considering the intermediate distance between the portal frames.
Internal forces are calculated by the elastic first-order analysis for non-sway
frame mode. The dimensioning of steel members is performed in accordance
with Eurocode 3 for the conditions of both ultimate limit state (ULS) and
serviceability limit state (SLS).
5.0 m
5.0 m
HEA 180
HEA 180
HEA 340
HEA 180
5.0 m
15.0 m
5.0 m
HEA 260
HEA 340
HEA 260
5.0 m
HEA 260
HEA 180
HEA 260
HEA 260
HEA 180
HEA 260
HEA 260
HEA 180
HEA 260
HEA 260
HEA 180
HEA 260
HEA 180
HEA 260
5.0 m
15.0 m
Figure 3:
6.5 m
14
6m
.0 m
40
2.8
20.0 m
0.50 m
HEA 120
HEA 260
HEA 260
6.5 m
HEA 500
HEA 500
6.0 m
Figure 4:
10 2.0 m
20.0 m
Figure 5:
Conclusions
This paper presents the topology and standard dimension optimization of steel
frames with rigid beam-to-column connections. The optimization has been
performed by the Mixed-Integer Non-linear Programming (MINLP) approach.
The MINLP was found to be very successful optimization technique for solving
the frame structures.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
613
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
615
Abstract
This paper presents the optimization of metal-plate-connected plane timber
trusses with special emphasis on joint flexibility. The optimization is performed
by the non-linear programming approach. Since various truss design parameters
such as type of truss configuration, span/depth ratio, number and type of
diagonal and vertical members and type of joint connections simultaneously
affect each other, all of these parameters are proposed to be considered
simultaneously in a single mathematical model. The optimization model for cost
optimization of timber trusses was thus developed. The economic objective
function was defined to minimize the structures self-manufacturing costs,
subjected to the design, stress and deflection (in)equality constraints. The finite
element equations were as the equality constraints defined for the calculation of
the internal forces and the deflections of the structure. The stiffness matrix of the
structure was composed by considering the fictiously decreased cross-section
areas of all the flexible connected elements. Constraints for the dimensioning of
the timber members were determined in accordance with Eurocode 5 in order to
satisfy the requirements of both the ultimate and the serviceability limit states.
The cross-section dimensions and the number of fasteners were defined as
independent optimization variables. A numerical example demonstrates the
applicability of the presented approach.
Introduction
span/depth ratio,
617
and deflection (in)equality constraints. The finite element equations were as the
equality constraints defined for the calculation of the internal forces and the
deflections of the structure.
The design constraints for the timber trusses were determined in accordance with
Eurocode 5 [12] in order to satisfy the requirements of both the ultimate (ULS)
and the serviceability limit state (SLS). Considering the ULS, the truss members
were checked for the tensional as well as the compressive/buckling resistance.
The required number of fasteners was also calculated for each joint. At the SLS
the vertical deflections of the truss girders were checked.
Since the bracing members (diagonals and verticals) are flexibly connected,
their stiffness decreases. In finite element analysis we consider the joint
flexibility in such a way that cross-sectional areas Am of all bracing members are
replaced by a fictiously decreased cross-section area Am* [4]:
A*m =
Am
E m ,mean Am
1
1
1+
+
K k
Lm
K
ser k m ,2
ser m ,1
(1)
where km,1 and km,2 represent the numbers of fasteners at both ends of the
considered m-th bracing element and Kser denotes the fasteners slip modulus,
taken for different types of fasteners from Table 7.1 of Eurocode 5. Em,mean stands
for the mean value of the modulus of elasticity.
(NLP)
upper chord
i,
li
vertical
A
i,
dia
l
na
go
hi
X
lower chord
Figure 1:
bi
A i = bihi
cost = ct bi hi li + c fm + c fl 2 k m + c fm + c fl k j
i =1
m =1
j =1
(2)
where cost represents the self-manufacturing (material and labour) costs of the
structure; ct denotes the price of the manufactured and embedded timber material
per m3; the sum of the products between widths bi, heights hi and lengths li of i,
iI, timber members represents the volume of the truss in m3 (see Figure 1); cfm
is the material cost of one fastener together with the adjoining steel plates, while
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
619
cfl denotes the manual labour costs for handling, assembling, drilling and bolting,
defined per one fastener. Considering that the required number of fasteners is
equal for both ends of each intermediate member, the total number of fasteners in
the m-th member is 2km, where km = km,1 = km,2. The last term of the objective
function represents the sum of all fasteners required in joints of the chord
members. Variable kj stands for the number of fasteners of the j-th joint, jJ. It is
evaluated considering the resultant force on account of the axial forces of all
intermediate elements connected to joint j. Since the dimensions of steel plates
depend directly on the number of calculated fasteners, the costs of steel plates are
included in the values cfm and cfl.
The input data of the optimization model is the truss geometry (coordinates
of joints), the supporting and loading conditions, the diameter of the considered
fasteners, the thickness of the metal plates, as well as the material characteristics
of all the used components (timber, fasteners, plates).
The cross-section dimensions bi and hi of i, iI, truss timber members and
the number of fasteners km and kj are defined as independent optimization
variables.
The finite element equations for the calculation of internal forces and
deflections of the structure are defined as equality constraints. The stiffness
matrix of the structure is composed by considering the fictiously decreased
cross-section areas of all the intermediate timber elements (diagonals and
verticals) in accordance with Eq. (1). The ULS and SLS design conditions,
described in Section 2 are defined as inequality constraints.
Numerical example
Fi
i
F = qL
h0
L = 30 m
Figure 2:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Truss B
Truss C
[EUR]
2984.33
3575.14
3322.59
[kg]
1291.29
1486.19
1356.99
16/13
18/15
17/21
21/17
17/20
21/17
181
248
244
44.82
79.85
82.62
59.88
95.79a
Total costs
Timber mass
It should be noted, that the max. displacements of truss B is not equal to L/300 = 100 mm due to the
rounding of cross sections before the reanalysis. After the first optimization phase (continuous
optimization), the SLS conditions were active and the displacement was equal to 100 mm.
621
different truss alternatives, see Table 2. The influence of the height of the truss
on the final results is presented by the ratio B/A, which represents the ratio
between the obtained costs (masses) of the truss B (h = 250 cm) and A (h=400
cm). The possible error, caused by neglecting the flexibility of the joints in the
timber trusses, is exposed through the ratio B/C which represents the ratio of the
obtained costs (masses) between truss B (considering joint flexibility) and C
(neglecting joint flexibility).
Table 2:
costs ratio
B/A
B/C
B/A
B/C
1.198
1.076
1.151
1.095
Conclusions
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
623
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
625
Abstract
The optimisation of structures and materials is a justifiably popular engineering
topic. Contemporary research is concentrated, among others, into cost
minimisation, structural efficiency and intelligence, in compliance with
environmental and social preservation. As a counterpart this paper puts the
accent on the time dependent aspect of constructions, such as the life cycle cost,
the possibility to make (non-) structural changes and recycling or reuse of
building material. In search of an optimisation of this dynamic efficiency of
constructions, a design strategy has been developed at the Vrije Universiteit
Brussels (Departments of MeMC and ARCH). This strategy is presented here. It
considers the temporal character of constructions from the first sketches
onwards.
Keywords: adaptability, reuse, design strategy, construction kits, generating
system, temporal/temporary.
Introduction
In society buildings have been - and still are - designed in terms of end states.
The moment the first sketches are drawn, the constructions finality is planned or
denied. Because of their static nature, which they acquired ab initio, most
buildings are not suited to meet the demands of a quickly changing society. As a
result many building components end up as waste or are brought back in
circulation by means of expensive and consuming industrial processes. Changing
functions, quickly evolving living and building trends, the amount of new
materials and improved techniques are some mutations the built environment
has to go through and has to provide appropriate answers to. Although humans
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06061
627
Figure 1:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 2:
To make effective use of the proposed system, the set of basic forms should
be provided with basic dimensions. In order to achieve optimal flexibility and
combination, the basic elements should have the same dimensions. If differences
are unavoidable HendrickxVanwalleghem proposed to solve the problem
using the rules of either halving or doubling. Both are the result of an easy
mathematical manipulation and create a geometrical series. Halving is easy if
one uses flexible elements: they can be folded. Starting with a square with side
x one finds: x, 2x, 4x, 8x, 16x
The fractal model in Fig. 2 can be projected on all materials and all scales and
thus can define the basic elements for each material type. Grouping ALL possible
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
629
Figure 3:
Figure 4:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
631
The awareness of the limited life time of our heritage is ever increasing; slums
are demolished, old train stations are replaced by prestigious ones, offices are
refurbished and monuments are carefully renovated or the object of restoration
Still professional developers and real estate owners pay little attention to the
temporary character of a construction: even during the study and the drawing
phase this aspect is often forgotten or even simply ignored. The point is that if
you want to face changing uses during the life time of the construction, static
solutions will make transformations extremely difficult if not impossible. It
could happen that some structural elements still perform in a satisfactory way,
but the owner will often prefer to demolish and start over. This causes a lot of
debris. Consequently, the actual society is missing a dynamic design strategy,
allowing transformations and adaptations during the life cycle of a construction.
The Hendrickx Vanwalleghem design strategy takes these characteristics of
temporality and adaptability of constructions into account, from the first sketches
on. It allows every construction part to transform into another configuration by
adding, deleting and transforming basic elements of the same system and
combining it with elements coming from other. One of the consequences is that
so-called dry connections are used: bolts and nuts, screws, click-systems
[6, 7].
Figure 5:
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 7:
633
development of (multi-) modular systems, but with the additional property that
they can be used with different design scales.
Furthermore, it is not the module which is standardised but the
(dimensional) modifiable basic elements from which it is composed. The latter is
the key difference with modular construction systems.
Figure 8:
5M
8M
3M
4M
2M
2M
Multi-modular grid.
Figure 9:
Generated compatible
design grids.
Conclusion
References
[1] De Wilde W.P., Conceptual design of lightweight structures: the role of
morphological indicators and the structural index. Proc. of 3rd International
Conf. on High Performance Structures and Materials 2006, ed. Wessex
Institute of Technology, United Kingdom, 2006.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
635
Abstract
This paper presents the study of a facade. One of the aims of the designing team is
the minimisation of the weight of the facade. Due to architectural constraints, the
geometry of the facade is fixed to a highly statically indeterminate truss. Therefore,
the minimisation of the weight of the facade is limited to a section optimisation.
The objective function is the indicator of volume. Since the number of members
is of the order of 1000, exhaustive search methods are impractical. We used a
simple iterative process to find optimal sections. In the first iteration all sections
are equal. This allows the forces to flow through the facade as if the facade has a
constant stiffness. In the next iterations sections are adapted to the forces that arise
from the previous iteration. This method quickly converges to an optimal section
layout. The results of this method are corroborated by a genetic algorithm. We find
that the truss facade with optimal sections consumes less material than an arch
with push rods, that transfer the forces to the arch. Furthermore, the influence of
buckling can be ignored.
Keywords: morphological indicators, optimization, genetic algorithm, statically
indeterminate, truss.
1 Introduction
This paper describes a methodology to optimize sections of a (highly) indeterminate truss. It is especially applicable for fixed geometries, where only sections
need to be optimized. The method is iterative and consists of updating the stiffness of the members according to forces in that member in the previous iteration.
The main advantage of this process is that it consumes considerable less time than
an ordinary optimization algorithm (genetic algorithm). Typically it will take the
process on the order of 10 iterations before convergence occurs. This means that
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06062
637
ume of an isomorphic structure with unit span, with at least one section of each
element dimensioned on its unit allowable stress, subjected to a load system with
E
compares the displacement
unit resultant. The Displacement Indicator = L
of different structural systems. It is the maximum displacement of an isomorphic
structure with unit span in a material with unit Youngs Modulus, with at least
one section of each element dimensioned on its unit allowable stress, subjected
to a load system with unit resultant. The analytical expressions of both W and
have been established in [3] and [2] for trusses, beams, arches, cables, cable stayed
structures, masts and frames subjected to a limited number of (simple) load cases
and support conditions. For statically determinate structures these Morphological
L
if instabilities, self
Indicator are only function of the geometrical slenderness H
weight and second order effect are neglected. Efficiency curves with respect to
minimum volume of used material can be established (Figure 3).
2 Method
The first iteration consists of making all sections equal, thus making all modules
equally stiff. The system is then solved, using a simple finite element program.
The solution can be interpreted as follows: it is a set of sections that is needed
to transfer the loads to the supports, if the stiffness is homogeneously distributed
over the framework (Figure 4). In reality the stiffness will not be homogeneous,
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
since member sizes vary. Therefore an update of the framework stiffness will yield
a more accurate flow of forces. This is done in the second and every successive
iteration, until convergence of the objective function W .
3 Results
The process converges rapidly (Figure 5(a)). We observe that after 1 iteration the
value of W is 1.38. After 10 iterations this value becomes 1.27. The optimum
is reached after 31 iterations (W = 1.26) (Figure 5(b)). The objective function
decrease between the first and final (best) iteration is 10%. This decrease becomes
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
639
1% between iteration 10 and the final one. This means that early on, the process
gives a very good estimation of the optimum. Of course this method has to be
compared to other optimization techniques. Using the genetic algorithm toolbox
of MatLab, a slightly better optimal solution of W = 1.20 is reached. However, it
takes on the order of 104 more time to reach this optimum. This has consequences
on the design process: because one finite element analysis takes about 2 seconds,
the answer of the iterative process is known after 1 minute, whereas the GA yields
its result after 5 hours. The advantage of having a slightly better solution does
not weigh up against the enormous loss in calculation time, especially during the
conceptual design stage.
The results are obtained by incorporating buckling. Optimal solutions vary
slightly when buckling is ignored. There are two reasons for this: member lengths
2
are small and sections are buckling efficient, with form factor ( H
= 8 with
I
section, H height and I moment of inertia) low (circular tubes).
The alternative geometry (Figure 6) yields a W -value of 1.07.
The previous solution obtained by the update iteration is compared to an arch
with push rods. According to [3] the value of W of this structural system is 1.13.
This means that the (alternative) truss facade (will transfer the vertical loads more
efficiently than an arch with push rods. The inclusion of horizontal (wind) loads
will only widen the gap between the two structural systems, because the arch is
inefficient at transferring this type of loads. It is not surprising that the truss solutions exhibit an arch shape of its most compressed members and a suspension cable
shape of its most tensioned members.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
641
truss with the previous sections. Within the limits of the conceptual design phase,
the results are accurate enough to guide the designer in her/his choices.
With regard to the application, this method yields member sections that make
the facade very light. The truss solution is even better than the arch with push rods,
wind loads excluded.
A setback of this method is that some boundary conditions (e.g. maximum displacement) are difficult to incorporate. This is due to the method itself. The process
can be used for two situations:
1. fixed geometry: only the member sections are to be optimized;
2. truss members used as grid: the truss works as base structure and members
that are not useful tend to have sections that converge to zero (Figure 7). This
type of optimization echoes the topology optimization method developed by
Bendsoe [1]. The problem will be to find the optimal grid.
References
[1] M.P. Bendsoe. Optimization of Structural Topology, Shape and Material.
Springer, New York, 1995.
[2] P. Latteur. Optimisation et predimensionnement des treillis, arcs, poutres et
cables sur base dindicateurs morphologiques. VUB, Brussels, 2004.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
643
Abstract
In this paper the optimization of the single-storey industrial building steel
structure is presented. The structure consists of the main portal frames, which are
mutually connected with the purlins. All structural elements are proposed to be
built up of standard hot rolled I sections. The structural optimization is
performed by the Mixed-Integer Non-linear programming approach, MINLP.
The MINLP performs a discrete topology and standard dimension optimization,
while continuous parameters are simultaneously calculated inside the continuous
space. Since the discrete/continuous optimization problem of steel frames is
non-convex and highly non-linear, the Modified Outer-Approximation/EqualityRelaxation (OA/ER) algorithm has been used for the optimization. Alongside the
optimal structure mass, the optimal topology (an optimal number of main portal
frames and purlins) as well as all standard cross-section sizes have been
obtained. The paper includes the theoretical basis and a practical example with
the results of the optimization.
Introduction
Single-storey frame structures are extensively used for industrial, leisure and
commercial buildings. In order to obtain efficient frame designs, researchers
have introduced various optimization techniques, appropriate either for the
continuous or the discrete optimization. OBrien and Dixon [1] have proposed a
linear programming approach for the optimal design of pitched roof frames.
Guerlement et al. [2] have introduced a practical method for single-storey steel
structures, based on a discrete minimum weight design and Eurocode 3 [3]
design constraints. Recently, Saka [4] has considered an optimum design of
pitched roof steel frames with haunched rafters by using a genetic algorithm.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06063
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
645
for the conditions of both ultimate limit state (ULS) and serviceability limit state
(SLS).
Lf
Lf
Lf
Lf
Lf
Lf
Figure 1:
LL
s.t.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(MINLP-G)
x X = {x R : xLO x xUP}
m
y Y ={0,1}
where x is a vector of continuous variables specified in the compact set X and y
is a vector of discrete, mostly binary 0-1 variables. Functions f(x), h(x) and g(x)
are nonlinear functions involved in the objective function z, equality and
inequality constraints, respectively. Finally, By+Cx b represents a subset of
mixed linear equality/inequality constraints.
The above general MINLP model formulation has been adapted for the
synthesis of mechanical superstructures (MINLP-SMS). The resulted
formulation that is more specific, particularly in variables and constraints. It can
be used also for the modelling the steel frames. It is given in the following form:
min z = c T y + f ( x )
s.t.
h( x ) = 0
g(x ) 0
A( x ) a
Ey e
Dy + R( x ) r
(MINLP-SMS)
( )
Py + S (d ) s
Ky e + L d cn k
st
x X = {x R : xLO x xUP}
m
y Y ={0,1}
The MINLP model formulation for mechanical superstructures is proposed to be
described as follows:
Included are continuous variables x={d, p} and discrete binary variables
y={ye, yst}. Continuous variables are partitioned into design variables d={dcn,
dst} and into performance (nondesign) variables p, where subvectors dcn and
dst stand for continuous and standard dimensions, respectively. Subvectors
of binary variables ye and yst denote the potential existence of structural
elements inside the superstructure (the topology determination) and the
potential selection of standard dimension alternatives, respectively.
The mass or economical objective function z involves fixed mass or cost
charges in the linear term cT y and dimension dependant mass or costs in the
term f(x).
Parameter nonlinear and linear constraints h(x)=0, g(x) 0 and A(x) a
represent the rigorous system of the design, loading, stress, deflection,
stability, etc. constraints known from the structural analysis.
Integer linear constraints Ey e are proposed to describe relations between
binary variables.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
d st = qi yist
iI
iI
647
(1)
M el , Rd =
Wel f y
M0
(2)
(3)
where MSd is the design bending moment, Mel,Rd is the elastic resistance moment,
fy is the yield strength of structural steel, Wel is the elastic section modulus and
M0 is the partial safety coefficient.
Resistance to axial force of the beams and columns:
N Sd N pl ,Rd
N pl , Rd =
A fy
M0
(4)
(5)
where NSd is the design axial force, Npl,Rd is the compression resistance, A is the
cross-section area, M0 is the partial safety coefficient.
Compression/buckling resistance of columns:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
N Sd N b, Rd
N b, Rd =
A fy
(7)
M1
VSd V pl , Rd
V pl ,Rd = Av
fy
3 M0
(9)
where VSd is the design shear force, Vpl,Rd is the design shear resistance and Av is
shear area.
Interaction between axial force and bending moment:
N sd
M sd
+
1 .0
N pl , Rd M el , Rd
(10)
A f y / M1
k LT M sd
LT Wel f y / M 1
1.0
(11)
L
200
max,Lf
2, L
L
250
2,Lf
Lf
200
Lf
250
(12)
(13)
where max.L is the vertical deflection of beams and max.Lf is the vertical
deflection of purlins for the terminated state, 2,L is the vertical deflection of
beams and 2,Lf is the vertical deflection of purlins for the variable load.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
649
H
150
(14)
The optimization
The example
8.0 m
(b): for serviceability limit states: 1.00g + 0.90(s + w + F + F). While internal
forces were calculated by the elastic first-order analysis the deformation of frame
members were calculated by the force method.
The portal frame superstructure was generated in which all possible
structures were embended by 30 portal alternatives, 20 purlin alternatives and
different standard size variation. The superstructure also comprised 24 different
standard hot rolled European wide flange I beams, i.e. HEA sections (from HEA
100 to HEA 1000) for each column, beam and purlin separately. The material
used was steel S 355.
The optimization was performed by the MINLP optimization approach. The
task of the optimization was to find the optimal structure mass, the optimal
topology (the optimal number of portal frames and purlins) and optimal standard
sizes.
The optimization was carried out by a user-friendly version of the MINLP
computer package MIPSYN, the successor of PROSYN [6] and TOP [9]. As
an interface for mathematical modelling and data inputs/outputs GAMS (General
Algebraic Modelling System), a high level language, was used [10]. The
Modified OA/ER algorithm and the two-phased optimization were applied,
where GAMS/CONOPT2 (Generalized reduced-gradient method) [11] was
used to solve NLP subproblems and GAMS/Cplex 7.0 (Branch and Bound)
[12] was used to solve MILP master problems.
The optimization model contained 130 (in)equality constraints, 183
continuous and 122 binary variables. The final optimal solution of 150,87 tons
was obtained in the 12th main MINLP iteration.
The optimal result represents the mentioned optimal structure mass of 150,87
tons, the obtained optimal topology of 12 portal frames an 20 purlins (see Figure
2) and the calculated optimal standard sizes of columns, beams and purlins (see
Figure 3).
m
.54
u4
m
1
0
.
1
50
30.0 m
Figure 2:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
651
0.50 m
HEA 140
8.0 m
HEA 900
HEA 900
7.5 m
HEA 800
HEA 800
18 1.67 m
30.0 m
Figure 3:
Conclusions
The paper presents the topology and standard dimension optimization of the
single-storey industrial building steel structure. The optimization is proposed to
be performed by the Mixed-Integer Non-linear Programming (MINLP)
approach. The MINLP was found to be a successful optimization technique for
solving this type of structures.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
E.J. O'Brien, A.S. Dixon, Optimal plastic design of pitched roof frames
for multiple loading, Comput. Struct. 64, 737-740, 1997.
G. Gurlement, R. Targowski, W. Gutkowski, J. Zawidzka and J.
Zawidzki, Discrete minimum weight design of steel structures using EC3
code, Struct. Multidisc. Optim. 22, 322-327, 2001.
Eurocode 3, Design of steel structures, European Committee for
Standardization, 1992.
M.P. Saka, Optimum design of pitched roof steel frames with haunched
rafters by genetic algorithm, Comput. Struct. 81, 1967-1978, 2003.
S. Hernndez, A.N. Fontn, J.C. Perezzn, P. Loscos, Design
optimization of steel portal frames, Adv. Eng. Software. 36, 626-633,
2005.
Kravanja, Z. and Grossmann, I.E., New Developments and Capabilities in
PROSYN - An Automated Topology and Parameter Process Synthesizer,
Computers chem. Eng., 18, 1097-1114, 1994.
Kravanja, S., Kravanja, Z. and Bedenik, B.S., The MINLP optimization
approach to structural synthesis. Part I: A general view on simultaneous
topology and parameter optimization, Int. J. Numer. Methods Eng. 43,
263-292, 1998.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
[9]
[10]
[11]
[12]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
653
Abstract
In modern structural control the application of discrete modal sensor arrays is a
commonly used technique to obtain the modal state vectors. In this paper, a genetic
algorithm is used to find the optimal positions to place modal sensor arrays on simple structures such as beams, plates and shells. The performance criterion is taken
as the steady state observability Grammian of the system and includes spillover
prevention as well. The performance of optimally placed modal sensors in the linear range is discussed. The variation in the performance of these modal sensors in
the nonlinear range is investigated.
Keywords: geometrical nonlinearity, modal sensors, Lyapunov equation, genetic
algorithm.
1 Introduction
There has been consistent research on smart materials and structures for the last
two decades. Geometrically linear theories and numerical methods have been
developed by many authors, e.g. Crawley and de Luis [1], Tzou and Tseng [2].
Considerably less work can be found in the area of geometrically nonlinear modelling of smart structures. Structural nonlinearity has been taken into account for
interlaminar stress analysis by Icardi and Di Sciuva [3], for large deflection shape
control in Yi et al. [4], Mukherjee and Chaudari [5] and Lentzen and Schmidt [6].
Active buckling control and post-buckling analysis has been done by Krishna and
Mei [7] and Chandrashekhara and Bhatia [8].
Piezothermoelastic analysis including nonlinearity is discussed in Tzou et al. [9]
and Pai et al. [10].
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06064
2 Structural model
The FE analysis of the structural response is performed using the theory of composite laminated shells given by Schmidt and Reddy [16]. The geometrically nonlinear strain displacement relations are based on the Reissner-Mindlin hypothesis and are valid for small strains and moderate rotations of the midsurface normals [16]
2
1
0
= +3 + 3 2
1
3 = 3 +3 3
33 = 0
with
= +
1
2
1 0 0
,
2
0
0
0
0
1
1
1
1
v | + v | b b + b v + b v ,
1 1 1
1
1
2
=
b b v v b v| b v | ,
2
1 0
1 1 1
0
1
1 0
1
+ v + v
3 =
and 3 = v v | .
2
2
The following abbreviations were used:
0
1 0
0
0
v | + v | b v 3 ,
=
2
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(1)
=v | b v 3
655
=v 3, +b v .
and
The internal and external virtual work are evaluated in a total Lagrangian fashion. The second Piola-Kirchhoff stress and Green-Lagrange strain are chosen to
express the mechanical part of the internal virtual work. Consequently, the electric
variables are defined as referring to the initial undeformed configuration [17].
3 Performance criterion
The performance criterion to optimise the modal sensor positions requires modal
voltages. The linear equations of motion of a laminated composite structure with
embedded piezoelectric layers can be written as
q
Fe
M 0
q
Kqq Kq
(2)
=
+
Kq K
Qe
0 0
0
where the above matrices are found in Lentzen et al. [17]. There, M is the mass
matrix, Kqq is the linear elastic stiffness matrix, Kq = Kq are the linear electromechanical coupling matrices and K is the piezoelectric capacitance matrix. The
electromechanical response of the structure is denoted by {q, }, where q and
are the systems generalised displacements and sensor voltages, respectively. The
externally applied forces (Fe ) and charges (Qe ) are denoted by the right subscript e.
After static condensation of the electric variables, equation (2) can be written in
modal form as
R
BjL VL
(3)
pj + j2 pj =
L=1
L =
CjL pj .
(4)
j=1
Here the induced voltage of the Lth sensor and the control voltage at the Lth actuator are denoted by L and VL , respectively. The sensing constant of the Lth sensor
due to unit modal displacement of the j th mode is denoted by CjL and the j th modal
actuation constant due to unit applied voltage of the Lth actuator is denoted by BjL .
In order to address the observability problem one should convert the standard
FE equations into state space equations as follows [15]:
x = [A]{x} + [B]{u}
(5)
y = [C]{x}
(6)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
p1
1 p1
. . . pn
Ai =
A = diag(Ai ),
B11
.
B = ..
Bn1
0
0
1
1 C11
..
and C =
.
1
n Cn1
V1
.
u = ..
,
VR
1
.
y = ..
n pn
T
i
0
0
i
(7)
B1R
..
.
0
..
.
BnR
..
.
0
1
1 C1S
..
.
n1 CnS
A linear time invariant system (A, B, C), with s outputs is completely observable,
if any of the following conditions are satisfied [19].
1. The (sn n) observability matrix [O] has rank n, where
T
[O] = [C] [C][A] [C][A]2 [C][A]n1
(8)
2. The observability Grammian [Wo ] is full-rank
T
e[A]t [C][C]T e[A] t dt
[Wo ] =
(9)
(10)
The observability test based on the rank is binary in nature (i.e. it tells us whether
the system is observable or not). Additionally the extent of observability is
required. Therefore, in the present work the objective function proposed by Hac
and Liu [18] is used, given by:
2(no +nr )
2no
2(no +nr )
2no
2n
r
2n
o
J=
j
(j )
j
(j ) , (11)
j=1
j=1
j=2no +1
j=2no +1
where no and nr are the number of observed modes and residual modes, respectively. For sensor optimal places, j is the eigenvalue of the steady state observability Grammian (i.e. [Wo ]). Note that in the objective function (i.e. Eq. 11) a
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
657
product term is included to account for least controlled modes. If any of the modes
is least controlled then the total objective function value goes to zero.
4 Genetic algorithm
Genetic algorithms (GAs) are random search techniques based on the mechanics
of natural selection and genetics. Genetic algorithms are used to explore the global
extremum of the given linear or nonlinear function. Although randomised, genetic
algorithms can efficiently explore the new generation with better fitness.
The GA is used to maximise J (i.e. Eq. (11)) for a given number of sensors.
4.1 Algorithm
1. Create a random initial population of sensors.
2. Evaluate each member of the current population by computing its fitness
value (i.e. J as given in Eq. (11) ), and select parents based on their fitness
value.
3. Children are produced by mating a randomly selected pair of parents at a
randomly selected site, known as crossover and by making random changes
to a single parent, known as mutation.
4. Replace the current population with the children from the new generation.
5. Repeat the algorithm for a prescribed number of generations.
4.2 Modal sensors
The principle of discrete modal sensor arrays is depicted in Figure 1. By choosing
the gains i in a particular way, the modal sensor will respond only to the mode j.
The gains are obtained by solving the orthogonal system of equations
[G]{j } = {ej },
(12)
where [G] = gki , which is the modal voltage of the k th sensor due to the unit
modal displacement of mode i, {j } are the linear gains of all sensors to sense
mode j and {ej } is the unit vector in the direction j.
Modal sensor signal
Linear combiner
Sensors
Structure
5 Numerical results
5.1 Clamped plate
A plate, with the dimensions [1101101] mm as shown in Figure 2, is taken as a
numerical example. The plate consists of aluminium and is meshed with a [11x11]
grid. The material properties of the aluminium and PVDF layers are displayed in
Table 1. The position of the four patches has to be optimised in order to observe the
first four modes. The GA parameters used in the analysis are population size (30),
crossover probability (50%), mutation probability (10%) and number of generations (40). After application of GA, the optimal sensor positions for the plate are
obtained as depicted in Figure 2. The first four natural frequencies are calculated
as 728 Hz, 1486 Hz, 1486 Hz and 2193 Hz.
Table 1: Aluminium and piezo material properties.
Aluminium
PVDF
E [Gpa]
[]
70
0.3
2
0.3
[kg/m3 ]
d31 [m/V ]
2700
2800
2.2 1010
d33 [m/V ]
1.062 1011
111
111
121
85
37
41
82
86
38
42
11
110 mm
110 mm
110 mm
81
121
11
110 mm
In order to examine the effectiveness of the modal sensors, the transient modal
sensor signals have to be investigated. One of the popular explicit time integration
techniques (i.e. central difference method) is used to integrate the equations (2)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
659
in the time domain. As an initial displacement field, the superposition of the first
four modal displacements is prescribed, with the respective modal amplitudes of
0.01, 0.007, 0.007 and 0.003 resulting in a mid-point deflection of 0.3616 mm. Figure 3 shows an arbitrary configuration of sensors that is considered in the present
work for comparison purpose. Figures 4-7 compare the results obtained with the
optimised and the arbitrary configuration. It can be observed in Figures 4-7 that the
modal sensor signals for the optimal sensor configuration are correct and are barely
aliased with those of higher modes. In case of the arbitrary sensor configuration the
modal sensor signals are strongly aliased with those of higher modes. Figures 8-9
show the modal signals for both optimal sensor configuration and arbitrary sensor
configuration in the nonlinear case. It can be concluded that the induced membrane
stresses which are not considered in the linear and modal analysis are the main
cause for the failure of modal sensor arrays in the nonlinear range of deformations.
2
x 10
x 10
optimal configuration
arbitrary configuration
optimal configuration
arbitrary configuration
2
x 10
2
x 10
optimal configuration
time[sec]
time[sec]
optimal configuration
arbitrary configuration
arbitrary configuration
time[sec]
time[sec]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
optimal configuration
arbitrary configuration
optimal configuration
arbitrary configuration
time[sec]
time[sec]
Figure 9: Nonlinear
signal.
second
mode
6 Conclusions
In the present work, GA is used to find optimal placement for modal sensors. Transient analysis is performed with linear and nonlinear FE, based on first-order shear
deformation moderate rotation theory. By numerical example it is shown that the
principle of modal sensor arrays yields good results for the genetically optimally
placed sensor patches in the range of small displacements. In the geometrically
nonlinear case, it is found that the induced membrane stresses are the prime cause
for the failure of modal sensor arrays.
References
[1] E. F. Crawley and J. de Luis, Use of piezoelectric actuators as elements of
intelligent structures. AIAA Journal, 25, 1373-1385, 1987.
[2] H. S. Tzou and C. I. Tseng, Distributed piezoelectric sensor/actuator design
for dynamic measurement/control of distributed parameter systems: a piezoelectric finite element approach. J. Sound Vib., 138, 17-34, 1990.
[3] U. Icardi and M. Di Sciuva, Large-deflection and stress analysis of multilayered plates with induced-strain actuators. Smart Mater. Struct., 5, 140-164,
1996.
[4] S. Yi, S. F. Ling and M. Ying, Large deformation finite element analyses
of composite structures integrated with piezoelectric sensors and actuators.
Finite Elements in Analysis and Design, 35, 1-15, 2000.
[5] A. Mukherjee and A. S. Chaudari, Piezolaminated beams with large deformations. Int. J. of Solids and Structures , 39, 4567-4582, 2002.
[6] S. Lentzen and R. Schmidt, Nonlinear finite element modelling of composite
structures with integrated piezoelectric layers. High Performance Structures
and Materials II, WIT Press, Southampton-Boston, 67-76, 2004.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
661
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
663
Abstract
In this paper the physical and mathematical models of deformations
(displacements and strains) and stress in the cold process of trapezoidal thread
rolling is presented. The process is considered as a geometrical and physical
non-linear, initial as well as boundary value problem. The phenomena on a
typical incremental step were described using a step-by-step incremental
procedure, with an updated Lagrangian formulation. The state of strains was
described by Green-Lagranges tensor, while the state of stress by the second
symmetrical Pioli-Kirchhoffs tensor. The object was treated as an elastic (in the
reversible zone) and visco-plastic body (in non-reversible zone) with mixed
hardening. The variational equation of motion in three dimensions for this case
was proposed. Then, the finite elements methods (FEM) and dynamic explicit
method (DEM) were used to obtain the solution. The application developed for
the method of finite elements ANSYS 8.1 provides a complex time analysis of
displacement, strains and stresses occurring in the object. The boundary
conditions for a displacement increment determined in model investigations were
used. Examples of calculations of influence of a friction coefficient on the states
deformation and stress were presented.
Introduction
One of the most widespread machine elements is the thread. Above 60% parts of
modern machines, devices and mechanism has threaded holes, whose
performance by screw-tap in high-plastic steels, some non-ferrous metals and
their alloy, pose difficult problems in technological aspects. The difficulty at
threading hard materials characterized by a large ductility and a high elastic
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06065
Figure 1:
Figure 2:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 3:
665
t
t
t y = t F2 ( t y) t y +
tt F1[] ( VP ) tt F1[] t t ( VP ) ( VP )
eq + t
t
t F3 ( t eq ) t eq ,
t st
tt (eqVP )
= 2 t e ij(VP) t e ij(VP) /3 ,
where t (VP)
eq
(VP)
t
eq
2 t e ij(VP) t e ij(VP) /3
(1)
are
t ij
1
[ t D (E) kl tt A]
t ~ ** t ijkl t
1 t S
(2)
t (E)
t ij = t C ijkl t kl
~ ~
(E)
t
tt Sij*[ tt Sij tt Cijkl
t kl t A],
(3)
where:
t ~** t ~* t (TE) t ~
t S = t Sij t Cijmn t Smn
(4)
t~
t Sij
t ~ t (E) t ~
t Sij t C ijkl t Sij
(5)
2t 2 t~ t
t y ( t C+ t E T )
stress
2t
3
t y
t ij
tensor,
y (VP)
eq
t (VP) t
t eq
(6)
are
the
components
of
tensor
(E)
strain tensor, t Cijkl
are the components of elastic constitutive tensor.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
667
i , t u i )= t J ()
t J ( t u i , t u
(7)
where tu i , tu i , tu i are the ith increment components of the displacement,
velocity and acceleration vectors, respectively. While using the conditions of
stationary of functional t J and a finite element method, we can write an
equation of motion in the form:
[ tt M]{tr} +[ tt CT ]{tr} + ([tt K T ] +[tK T ]){tr} ={tR} +{tF} +{tt F} +{tt R} (8)
where mass matrix [ tt M ] , damping matrix [ tt CT ] , stiffness matrix [ tt K T ] and
force vector {tt FT } are known at time t. However, increment stiffness matrix
[ tK T ] , external incremental load vector {t R} , internal incremental forces
3 DEM solution
Assuming that an increase of temporary step t is very small, it is possible to
execute a linearization of equation (8) and using the incremental decomposition
we obtain:
[M ]{t tt r} + [CT ]{t tt r} + [K T ]{t tt r} = {tt tt FT } +{t tt Q}.
(9)
Then using the central difference method (DEM), in which it is assumed that:
{t r} =
1
1
({t + t r} {t t r}) , {t r} = 2 ({t + t r} 2{t r} + {t t r}) (10)
2 t
t
(11)
where:
~
1
1
[CT ],
[M ] = 2 [ M ] +
2t
t
~
{t tt Q T } = {tt tt FT } +{t tt Q} [K T ]{t tt r} +
+
t t
2{ r} {
t 2
r}
[M ] +
(12)
t t
r}
[CT ].
2t
The integration method requires that the time step t is smaller than critical
value tkr, which can be calculated from the mass and stiffness properties of the
complete element assemblage: t t kr = TN / , where TN is the smallest
period of the finite element assemblage with n degrees of freedom.
Model investigation
The model investigation was conducted in order to settle the course deformation
layer top sample executed from the plastic material, as well as with the aim to
qualify boundary conditions for displacements indispensable to numeric analysis
of the trapezoidal tread rolling process. To model investigations were applied to
the samples in a rectangle shape with the following dimensions:
a)
b)
Figure 4:
The stamp (a) and the meshed sample before deformation (b).
Two samples were joined by sides with a plot mesh, and were closed in a metal
form. Then the samples were subjected to the deformation by a perpendicular shift
of rectilinear motion in the model stamp of an outline of trapezoidal thread rolling.
The view of the deformed mesh of complete elements is presented in Fig. 5.
a)
Figure 5:
b)
The mesh after deformation for =0,2 (a) and =0,39 (b).
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
669
Figure 6:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
2
ELEMENTS
DISPLACEMENT
APR 13 2005
21:59:14
Y
Z
APR 13 2005
21:59:14
STEP=1
SUB =100
TIME=1
DMX =15.6
4
NODAL SOLUTION
NODAL SOLUTION
APR 13 2005
21:59:14
STEP=1
SUB =100
TIME=1
EPTOINT (AVG)
DMX =15.6
SMN =.00995
MX =3.616
SMX
MN X
Z
Z
.00995
.911405
1.813
2.714
3.616
APR 13 2005
21:59:14
STEP=1
SUB =100
TIME=1
SINT
(AVG)
DMX =15.6
SMN =.517E+08
MX
SMX =.178E+10
MN
X
.517E+08
.485E+09
.918E+09
.135E+10
.178E+10
Figure 7:
DISPLACEMENT
APR 19 2005
22:49:06
STEP=1
SUB =100
TIME=1
DMX =16.85
2
DISPLACEMENT
STEP=1
SUB =100
TIME=1
DMX =16.85
node 158
APR 19 2005
22:49:06
node 160
node 127
Figure 8:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Figure 9:
Figure 10:
Figure 11:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
671
Figure 12:
Conclusions
The paper presents a numerical analysis of the process for thread rolling. The
process is considered as a geometrical, physical non-linear initial and boundary
value problem. The phenomena on a typical incremental step were described
using a step-by-step incremental procedure, with an updated Lagrangian
formulation. The variational nonlinear equations of the objects motion for a
typical incremental step were derived from the stationary condition for these
functionals. These equations were unravelled using the spatial digitization by a
finite elements method. The adequate algorithm to solve the equations of the
motion assuming Rayleighs proportional dissipation was developed for this
method. The application developed for the method of finite elements ANSYS 8.1
provides a complex time analysis of displacement conditions, strains and stresses
occurring in the object. The mathematical models of the process, the algorithms
for the solution of discrete equations of motion and the application in the
ANSYS system developed in the present work could be used to improve the
design process for the thread rolling process.
References
[1]
[2]
[3]
673
Abstract
Limit analysis with rigid block models is a tool successfully used in recent years
for the assessment of ancient masonry structural elements and small buildings. In
this paper, the interface yield functions for three-dimensional models are defined
at interpolation points, instead of using the generalised stresses approach. This
approach leads to very simple expressions for the yield functions and flow rules
and therefore, renders the mathematical programming problem easier to solve than
the generalised stresses formulation. The solution for the limit analysis problem is
obtained using the load-path following approach. The validation of the present
proposal shows good agreement compared with non-linear finite element results.
Keywords: limit analysis, rigid block assemblages, non-associated flow, numerical
integration.
1 Introduction
A valuable model for the structural assessment of ancient masonry structures is
the limit analysis of rigid block assemblages interacting through no-tension and
frictional interfaces. The reason for this affirmation lies in the fact that masonry
has low tensile strength and quasi-brittle failure; therefore, at collapse, the cracks
render the structure as a set of rigid blocks rocking and sliding between them.
In the rigid blocks modelling strategy, the degrees of freedom are related to
the blocks, and the stress and strain variables are related to the interfaces. Therefore, blocks can be regarded as extended nodes and the interfaces as structural
elements. Previous works in this subject have used mostly a generalised stress
approach for the interfaces [14]. This means that the generalised stresses are,
for instance, the normal and shear forces and the bending and torsion moments.
Besides, Livesley[5, 6], using a different approach, verified the contact at points
located at the interface corners.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06066
(1)
u 1 = 0
(2)
= 0
TQ
Fv C
Fc +
0
(3)
0
T = 0
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(4)
(5)
(6)
Fi
Fi
675
Fj
s 1k
nk
s 2k
o
Fj
Block j
Block i
V1 =
V2 =
np np
1 dA
.
= J
1lm l m
(7)
l=1 m=1
np np
2 dA
.
= J
2lm l m
l=1 m=1
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(8)
N=
M1 =
lm l m
(9)
l=1 m=1
x2 dA
.
= J
lm x2m l m
(10)
l=1 m=1
np np
.
= J
np np
M2 =
T=
np np
dA
x1 dA
.
= J
lm x1l l m
(11)
l=1 m=1
np np
.
(1 x2 + 2 x1 )dA= J
(12)
l=1 m=1
In the approach proposed in this paper, it is not necessary to calculate the stress
resultants over the interfaces, but it can be done by means of eqns. (712). The
yield function characterisation is performed by limiting the normal and shear stresses
at each integration point. The normal stress limits are: from bellow, the effective
compressive stress and from top the zero value due to the no-tension hypothesis.
Therefore, eqns. (13) and (14), represent the yield functions for the normal stress.
Here, fce f is the effective compressive stress, which takes into account the effect of
transverse cracking and the fact that masonry presents quasi-brittle failure, while
the model features perfect plastic behaviour. The shear stresses are limited by the
cohesionless Coulomb criterion, as already mentioned. This criterion conducts to a
quadratic function as illustrated in Figure 2 by the circle of radius , where is
the friction coefficient. Nevertheless, in order to simplify the solution to the mathematical programming problem, a piecewise linear approximation is proposed and
illustrated also in Figure 2 by an inscribed octagon. The resulting yield functions
are expressed by eqns. (15) and (16). It is observed that, due to the absolute value
operands, these two expressions represent the eight linear functions illustrated in
Figure 2.
c
fce f
(13)
s1s4 |1 | + ( 2 1)|2 | + 0
s5s8 ( 2 1)|1 | + |2 | + 0
(14)
(15)
(16)
There are, therefore, ten linear yield functions for each integration point. The
yield functions for all the integration points at every interface in a model can be
cast in matrix form, in such a way that eqn. (4) takes the form of eqn. (17). Here, N
is the matrix resulting from the assemblage of all the yield functions in the model.
0
N Q
(17)
677
Coulomb criterion
Approximation
s1
s2
t
s14
0
1 /|1 |
( 2 1)2 /|2 |
1
0
2 /|2 |
s58
( 2 1)1 /|1 |
x1 , x2 and xn , respectively. The flow directions for the yield modes are in Table 1.
For the compression and tension yield modes the flow consists on a normal relative
displacement rate in the negative and positive directions of the xn axis, respectively.
The flow directions for the sliding failure modes are obtained from the normality
rule but neglecting the relative displacement rate along the xn axis, due to the
hypothesis of zero dilatancy. The rows of Table 1 are used to assemble the columns
of matrix N0 , in eqn. (18), which gives the generalised strains resulting from the
flow rule condition. Here the vector q gathers the generalised strains for all the
integration points in a model.
q = N0
(18)
The displacement rates for a generic block i are the translation displacement
rates at the block centroid uti and the angular displacement rates uri . These vectors are referred to the global coordinates system. Eqn. (19) gives the generalised
strains at the integration point p of the interface k in terms of the displacement
rates at the blocks i and j; see Figure 1. Here, Tkg is the matrix that transforms a
vector from the global coordinates system to the local interface k system; c pk , ci
and c j are the position vectors of the point p, and the centroids of blocks i and j,
respectively. Eqn. (19) can be used to assemble the compatibility set of equations
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
(19)
(20)
l2
x1
l2
l1
l1
679
analysis solution indicates that the centre of twisting is located at the interface
centroid, therefore, the maximum sliding and normal displacement rates are at the
interface corners. As the blocks are rigid, separation will occur at the inner points,
and contact will be possible only at the interface corners. Under this conditions, it
is straightforward to show that eqn. (21) proportionates the value of the failure torsion moment, where c is the distance from the interface centroid to any corner; see
eqn. (22) and Figure 3. Michalowski and Gawecki [9] arrived to an analogous conclusion for a circular interface where c is replaced by the circle radius in eqn. (21).
It is possible to arrive to the same result always that the dilatancy coefficient is
positive, even if it does not equal the friction coefficient.
T = c(N)
c = l12 + l22
(21)
(22)
If the dilatancy coefficient is negative, the normal displacement rates are again
proportional to the sliding displacement rates. Nevertheless, no separation but penetration exists between blocks, contact is only possible at the very interface centroid and, therefore, the torsion moment strength is zero (contact on a single point).
If the dilatancy coefficient is zero, contact points or areas are not determined
by the failure mechanism and neither, the normal stress distribution. This means
that for every possible normal stress distribution, there exists a torsion moment
strength. It is evident that the values of these strengths lie between zero and that
given by eqn. (21). The most fiscally meaningful case consists on a uniform normal stress distribution over the interface. For this case, eqn. (23) gives the torsion
moment strength, where the torsion constant cT is given by eqn. (24) [4]. It is
observed that eqns. (21) and (23) are the same except for the torsion constant definition.
T = cT (N)
l2
l2
1
l2 + c
l1 + c
+ 2 ln
c + 1 ln
cT =
3
2l2
l1
2l1
l2
(23)
(24)
The cT /c ratio can be calculated for the whole range of interface aspect ratios,
and the observation is made that it has a small variation between 0.50 and 0.54.
Therefore, a first but marginal conclusion is that the torsion constant can be approximated as half the distance from the centroid to the corner. A more important observation is that, for zero dilatancy, the torsion strength of the interface can take values between zero and that given by eqn. (21), but the strength for a uniform normal
stress distribution is about half the way between the former values. Therefore, it
is of fundamental importance, in limit analysis problems with zero dilatancy, to
obtain solutions with even normal stress distributions over the interfaces.
Orduna and Lourenco [4, 10] showed that for this type of problem it is also
important to take into account the loading history. Here, a solution procedure, akin
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Theoretically minimum
Theoretically maximum
0.427
0.553
FEM
Load-path following
0.479
0.465
to the load-path following one is used, which agrees with the integration points
approach. The main idea behind this procedure is to keep the normal stresses distribution as uniform as possible, while the variable loads are applied by small
increments.
5 Validation
The validation of this type of three-dimensional models is not an easy task due
to the lack of experimental or analytical results to compare with. Therefore, the
example presented here is a comparisons against a finite element (FEM) non-linear
model analysed with the package DIANA [11].
Figure 4(a) shows the model of a masonry hollow pile. The pile is built of
dry masonry blocks with dimensions 0.20.20.4 m. The pile dimensions are
0.60.81.2 m. The material volumetric weight is 20 kN/m3 and the friction
coefficient is 0.7. The permanent loads are the self-weight of the blocks. The variable loads are proportional to the blocks weight, but horizontally applied in the
direction of the larger base side (X direction). The compressive effective stress,
according to the load-path following procedure, is steadily increasing.
Figure 4(b) shows the failure mechanism obtained by the FEM analysis. The
same failure mechanism is obtained by limit analysis and Figures 4(c), (d) show it
from two different viewpoints for a better understanding. A range of ultimate load
factors is possible for this mechanism and for infinite compressive strength. It is
noted that the ultimate load factor is defined as the ratio between the variable loads
causing failure on the structure, and their nominal values, in this case numerically
equal to the self-weight of the blocks. Table 2 presents the ultimate load factor
obtained from different approaches. If the reaction on the overturning blocks set
is concentrated on the interface A only, Figure 4(d), with zero stresses at interface
B, the ultimate load factor would be 0.427, the minimum possible for this mechanism. If there are non-zero normal and shear contact forces on interface B, the
last one opposing to the upper block overturning, the maximum possible ultimate
load factor for the mechanism shown equals 0.553. The ultimate load factor calculated with the load-path following procedure is 3% lower than the FEM value. It
is possible to verify that the load-path following procedure agrees very well with
the FEM results.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
681
1.2
0.8
0.6
(a)
(b)
joint A
joint B
(c)
(d)
Figure 4: Masonry pile; (a) model; failure mechanisms for infinite compressive
strength: (b) FEM failure mechanism; (c) and (d) different views of the
limit analysis failure mechanism.
6 Conclusions
A limit analysis formulation using numerical integration techniques at the interfaces has been proposed for rigid block assemblages. The importance of even normal stress distributions over the interfaces was demonstrated. Comparisons, not
presented here due to space limitations, show that good agreement is obtained, at
interface level, between the integration points approach and the yield functions
obtained by constant normal stress distributions over regular shaped parts of the
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgment
This work was supported by project PROMEP/103.5/04/1322 funded by the Secretary of Public Education (SEP) of Mexico.
References
[1] Begg, D. & Fishwick, R., Numerical analysis of rigid block structures including sliding. Computer Methods in Structural Masonry 3, eds. J. Middleton &
G. Pande, Portugal, pp. 177183, 1995.
[2] Baggio, C. & Trovalusci, P., Limit analysis for no-tension and frictional
three-dimensional discrete systems. Mech Struct Mach, 26(3), pp. 287304,
1998.
[3] Ferris, M. & Tin-Loi, F., Limit analysis of frictional block assemblies as a
mathematical program with complementarity constraints. Int J Mech Sci, 43,
pp. 209224, 2001.
[4] Orduna, A. & Lourenco, P., Three-dimensional limit analysis of rigid blocks
assemblages. Part I: Torsion failure on frictional interfaces and limit analysis
formulation. Int J Solids Structures, 2005. Accepted for publication.
[5] Livesley, R., Limit analysis of structures formed from rigid blocks. Int J Num
Meth Engrg, 12, pp. 18531871, 1978.
[6] Livesley, R.K., A computational model for the limit analysis of threedimensional masonry structures. Meccanica, 27, pp. 161172, 1992.
[7] Orduna, A. & Lourenco, P., Cap model for limit analysis and strengthening
of masonry structures. J Struct Eng, 129(10), pp. 13671375, 2003.
[8] Bathe, K.J., Finite Element Procedures in Engineering Analysis. PrenticeHall, 1982.
[9] Michalowski, R.L. & Gawecki, A., Limit torque for a frictional joint. Mech
Struct Mach, 24(4), pp. 499512, 1996.
[10] Orduna, A. & Lourenco, P.B., Three-dimensional limit analysis of rigid
blocks assemblages. Part II: Load-path following solution procedure and validation. Int J Solids Structures, 2005. Accepted for publication.
[11] TNO Building and Construction Research, Delft, The Netherlands, DIANA
Users Manual Release 7.2, 1999.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Section 9
Reliability of structures
685
Abstract
We propose the finite element simulation technique to model the process of
diffusional creep and stress relaxation that occurs in Cu-damascene interconnects
of integrated circuit devices in the processing stage. On the length scale of the
interconnect lines (microns), the stress-induced mass flow constitutes the major
mechanism of inelastic deformation. The mass flow problem is coupled to the
stress analysis through vacancy flux and equilibrium vacancy concentration,
allowing independently for the concentration profile and evolution of the stresses
and strains in an iterative process to be solved. We decompose the total
displacement field into the elastic part and the inelastic mass flow contribution.
Performing the stress analysis in the configuration with accumulated inelastic
displacements, we ensure that the shape of the interconnect line is compatible
with external geometrical constraints throughout the simulation. This approach
has been implemented in the software package that seamlessly integrates the
problem-oriented code with the commercially available finite element program
MSC.Marc. We apply the technique to model the Coble creep phenomenon by
introducing the nanoscale grain boundary region having the thickness of the
order of several layers of atoms. As an illustration, the problem of stress
relaxation in a single grain subjected to prescribed displacements and tractions is
examined.
Keywords: copper interconnects, diffusional creep, grain boundary, nanoscale
deformation, finite elements.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06067
Introduction
Typical
Cu grain
Grain
boundary
Ta liner
Low-k
dielectric
m
Figure 1:
687
example, by Kobrinsky et al. [1]. The deformation may be attributed to the CuCu grain boundary and Cu-Ta interfacial sliding, as well as to the material build
up due to diffusional creep. The experimental background for this assumption
was provided by Gross et al. [2], who developed an AFM method to measure
out-of-plane deformation resulting from thermal cycling and applied this
technique to observe the Cu-polyimide interconnect structure.
In this paper we develop the numerical method to predict the diffusion-based
deformation that occurs in the process of interconnects fabrication. To describe
this phenomenon adequately, the mathematical formulation is needed that
accounts for the coupled nature of mechanical variables and concentration of
species such as Cu atoms, vacancies and impurities. The decoupling of creep
problem is achieved by its decomposition into the linear elasticity and mass flow
subproblems, which are solved in an iterational process. Rzepka et al. [3] applied
similar technique to a 3-D model of interconnect. We propose a different
approach by utilizing the thermodynamical coupling equations and the concept
of grain boundary region of finite thickness. A general discussion of latticebased thermodynamics with respect to diffusional creep in interconnects was
presented by Garikipati et al. [4]. Although our treatment is considerably
simplified, it nevertheless gives adequate description of the creep/stress
relaxation behaviour, and at the same time can be easily adapted to use with
existing commercial finite element packages. We also devote special attention to
the modelling of grain boundary regions, which are expected to be the major
path of vacancy diffusion at the temperature ranges used in interconnects
processing.
(1)
(2)
v0
where Qm , gb = Qm
( < 1) . We are not aware of a precise method to measure the activation energy
for vacancy motion through the grain boundary and will treat it as an adjustable
parameter that may be affected by impurities, grain boundary roughness,
crystallographic orientation and whether the opposing interface has similar
diffusivity.
Neglecting the local vacancy relaxation strain (Hirth and Lothe [8]) the mass
or atomic flux jA is opposite to the vacancy flux jv :
jA = jv .
(3)
The atomic flux field can be treated as the diffusive flow velocity field in a body
that is statically fixed. The gradient of the mass flow velocity jv defines the
cr = 1 2 jv + jv
).
(4)
The creep strains defined above are caused by diffusion mass flow due to stress
field gradients. Assuming that total strain consists of elastic and creep
components, the elastic part of the strain can be related to the total stress by
Hookes law
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
= C : ( cr )
689
(5)
: C : ( cr )dV = P
(8)
where P is the increment of the externally applied load. This equation is the
basis for the finite element formulation of mechanical subproblem.
symmetry
Grain
boundary
symmetry
Figure 2:
V 11dV
relaxation function
(9)
= 0 exp ( t ) .
In this equation, is the Maxwell model relaxation time defined through spring
stiffness and dashpot viscosity as = . We use the concept of
relaxation time to characterize the time period of relaxation of average stress in
the model, and to scale the time t in the graphs throughout the rest of this
section.
Table 1:
Parameter
Grain width d
Burgers vector b
Grain boundary region thickness
Atomic volume
Melting temperature Tm
Value
100 nm
0.25 nm
1 nm
1.18 10-29 m3
Youngs modulus E
Poissons ratio
128 Gpa
0.33
286.8 Mpa
2 10-5 m2/s
113.3 kJ/mole
5e-15 m3/s
67.98 kJ/mole
83.7 kJ/mole
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
1356 K
unloaded
loaded
Displacement y, nm
691
T = 0.3Tm
T = 0.5Tm
T = 0.6Tm
t=0
1.000
T = 0.7Tm
T = 0.8Tm
T = 0.9Tm
t=
loaded at t = 0
0.100
0.010
0.001
-50
-40
-30
-20
-10
10
20
30
40
50
x, nm
Figure 3:
grain. Note that relaxation time is different for different temperatures, for
example 1 s for T = 0.5Tm and 0.001 s for T = 0.7Tm .
1.4
610-4
T=0.3Tm
1.2
T=0.7Tm
610-3
11/0
0.03
0.09
0.8
0.3
0.6
0.6
0.4
y
x
0.2
0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
y/d
Figure 4:
1.0
t=
0.9
0.10
0.8
<11>/0
0.7
0.20
0.30
0.40
0.6
0.5
0.4
0.80
0.90
1.10
0.3
0.35
1.00
0.95
0.2
0.1
t = 0.01
0.0
0.0
Figure 5:
0.1
0.2
0.50
0.65
0.80
t = 0.1
0.3
0.4
0.5
t /
0.6
0.7
0.8
0.9
1.0
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
693
DL 1 +
=
(10)
2
dDL
kTd
where is the shear strain rate, d is the grain width, e is the effective
thickness of grain boundary (on the order of magnitude of Burgers vector b ) and
is the applied stress. Note that the diffusivities DL and Dgb in eqn (10) are
based on the energy of both formation and motion of vacancies. They are
different from the vacancy motion only diffusivities Dv , L and Dv , gb used in
our finite element formulation which explicitly includes the vacancy formation
energy in eqn (2) for stress dependence of vacancy concentration. We obtain the
theoretical estimate of relaxation time by computing the dashpot viscosity from
eqn (10) as = . The shear strain rate in eqn (10) accounts for the
contribution of both Nabarro-Herring and Coble mechanisms to creep rate,
which is consistent with the numerical modeling approach of sections 2 and 3.
Figure 6 shows good agreement of theoretical and numerical predictions for
relaxation time. The largest difference is in the temperature range 600-900 K.
1.0E+09
Maxwell model
Relaxation time , s
1.0E+07
FEA
1.0E+05
1.0E+03
1.0E+01
1.0E-01
1.0E-03
1.0E-05
400
500
600
700
800
900
1000
1100
1200
1300
Temperature T, K
Figure 6:
Conclusions
The stress-induced mass flow constitutes the major mechanism of creep in Cudamascene interconnects during thermal processing of integrated circuit devices.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Acknowledgements
This work was supported by the National Science Foundation, Division of
Manufacturing and Industrial Innovation, under Grant No. DMI-0300216. The
donation of the Cu-damascene sample by IBM is gratefully acknowledged.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
695
Abstract
Assessing the safety of a structure through inspection has always been part of the
practice of the structural engineer. Information obtained through these
inspections can be both objective and subjective. One can update the reliability
of an existing structure based on objective and subjective information through
the fuzzified Bayes theorem.
This paper cites the algorithm for computing the posterior probability of
failure based on visual inspection of existing structures by incorporating fuzzy
set theory into the Bayes theorem.
The failure design probability of a structural steel frame is updated.
Results indicate that the preciseness of the membership function associated
with the input data has little or no influence on posterior failure probability.
Keywords: reliability, Bayes theorem, fuzzy sets, existing structures, steel frame.
Introduction
In engineering practice one often has to deal with qualitative and vague
evaluations, commonly known as fuzzy. Such subjective fuzzy information is
to be found in the results of inspections of structures (such as bridges, frames and
so on) and are accompanied by those of a deterministic and probabilistic kind.
In particular, when we consider the question of evaluation of the safety level
of an existing structure, it appears evident that we cannot avoid taking into
account the quality of workmanship, the state of conservation of the elements
making up the structure and so on. Estimation of these qualities can be expressed
subjectively through variations of a linguistic nature. The values of these
variables, as for example the quality of the workmanship is good, or the state
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06068
697
those proposed in [3] and [7] (triangular and piecewise linear in form) in our
calculations. The two solutions obtained are then compared.
The Bayes theorem provides a method that allows inclusion of new information
in a priori probabilistic evaluations, thus producing a new probability value for
the occurrence of an event (Benjamin and Cornell [9]).
When there is no fuzzification and we take into account the theorem of total
probability, the Bayes theorem is defined by the relation:
P (B J | A) =
P ( A | B J ) P (B J )
m
(1)
P (A|Bk ) P (Bk )
k =1
In Eqn (1) BJ is the Jth unknown random parameter having a known a priori
distribution P(BJ ), whose probability is to be updated; A is the random sample
representing the parameter to be inspected; m is the number of mutually
exclusive and totally exhaustive events. P(BJ | A ) is commonly known as the
posterior probability of BJ after statistical event A has taken place.
Equation (1) is thus valid if event A can be expressed objectively, which is
not the case of a fuzzy event. Therefore the above formulation of the Bayes
theorem cannot be used with information of the type: the crack is small or the
state of bolt conservation is poor, and so on, which are vague and imprecise and
cannot by defined with any certainty. To express such fuzzy information,
specific linguistic variables are introduced. In particular, function (x) is
introduced to express the probability of fuzzy event .
Taking this into account, we find that the conditional probability of fuzzy
event can be expressed as:
~
P A | BJ =
x
~
A
(x ) f x| B J (x ) dx
(2)
where fx|BJ (x) is the conditional probability function that predicts event BJ for
support x. Equation (2) is valid if the support is continuous while if the support is
discrete, the equation is written:
~
P A | BJ = A~ (x ) P x | B J (x )
x
(3)
Briefly stated, the posterior probability of event BJ, which takes into account
the observation of fuzzy event , is obtained by modifying Eqn (1) in accordance
with Eqns (2) and (3).
The fuzzy Bayes theorem ([10], [11]) can therefore be expressed by the
following relations. Where the support is continuous, taking into account Eqn
(2), Eqn (1) is written:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
A~ ( x ) f x | B J ( x ) dx P (B J )
P BJ | A = x
m
A~ ( x ) f x | Bk (x ) dx P (B k )
k=1 x
(4)
where the support is discrete, taking into account Eqn (3), Eqn (1) is written:
~
x A ( x ) P x| B J P (B J )
~
P BJ | A =
m
A~ (x ) P x| Bk P (B k )
k=1 x
(5)
3 Examples of application
The example under consideration (see [7]) concerns the updating of the design
failure probability of a steel frame following an inspection. It is assumed that
such a probability is equal to Pf = 10-5, and that this represents a mean reliability
value that could be calculated by considering the construction of a large number
of similar structures. Numerous components (both structural and non-structural)
are inspected. Since the conditions of these components influence the overall
failure probability of the structure, they assume the role of probability
parameters. It is supposed that during the inspection the following parameters are
considered: connections, foundations, alignment, columns, beams, braces, bolts
and paint.
3.1 Statistical parameters
We consider event BJ which may correspond to event B1 (= the structural
component is safe) or to event B2 (= the structural component is near failure).
The conditional probability function of random event BJ, owing to the
support of quality x, supposedly assumes the following quadratic forms:
N 6
g ( X | B1 ) = i
(0.1) 2 (x 10 ) 2 + 1.0
6
(6a)
N 6
g( X | B2 ) = i
(0.1) 2 x 2 + 1.0
6
(6b)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
699
0.14
P(X|Safe)
1
0.12
0.10
3
4
5
6
0.08
0.06
0.04
0.02
0
Figure 1:
4
6
Support of Quality, x
10
0.14
0.12
0.10
P(X|Failure)
6
5
0.08
4
0.06
0.04
0.02
1
0
Figure 2:
4
6
Support of Quality, x
10
P ( x) = 0.30 x + 1.1
0.10 x + 0.5
if
x 1
if 1 x 2
if 2 x 3
if 3 x 5
if
x5
0
0.20 x 0.4
0.40 x 1.0
N ( x) =
0.40 x + 3.0
0.20 x + 1.6
if
if
if
if
if
if
0 x2
2 x3
3 x 5
5 x7
7 x8
8 x 10
(8b)
if
if
if
if
if
x5
5 x7
7 x8
8 x9
9 x 10
(8c)
G ( x) =
0
0.10 x 0.5
0.30 x 1.9
0.40 x 1.7
0.10 x
(8a)
701
1.00
0.80
Normal
Good
Grade of Membership
Poor
0.60
0.40
0.20
0.00
Figure 3:
4
6
Support of Quality, x
10
1.00
Grade of Membership
0.80
Normal
Poor
Good
0.60
0.40
0.20
0.00
Figure 4:
4
6
Support of Quality, x
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
10
P( x ) =
0.20 x
0.20 x + 2.0
N ( x ) =
0.25 x 1.5
0
G ( x ) =
if
x4
if
x>4
if
x5
if
x>5
if
x6
if
x>6
(9a)
(9b)
(9c)
(10)
k =1
P ( A1 |
P( B1J ) = 2
~
BJ ) P ( BJ )
(11)
P ( A 1| Bk ) P ( Bk )
k =1
~
P ( A 2 | BJ ) P ( B1J )
2
~
P( A 2 | Bk ) P( Bk )
(12)
k =1
703
P( BJi ) =
~
P( A i | BJ ) P( BJi 1 )
2
~
i 1
P( A i | Bk ) P( Bk )
(13)
k =1
Equation (13) above is immediately explicable by using Eqn (2) when the
support is continuous and Eqn (3) when it is discrete.
Table 1:
CASE N
Parameter
(1)
(2)
Posterior failure probability for the type 1 (col. 5) and type 2 (col.
6) membership functions adopted.
Connection
Foundation
Alignment
Columns
Beams
Bracing
Bolts
Painting
Connection
Foundation
Alignment
Columns
Beams
Bracing
Bolts
Painting
Connection
Foundation
Alignment
Columns
Beams
Bracing
Bolts
Painting
Index Ni
Quality Ai
Pif ( 1 )
Pif ( 2 )
1
1
1
2
4
4
4
6
1
1
1
2
4
4
4
6
1
1
1
2
4
4
4
6
Normal
Poor
Normal
Poor
Normal
Poor
Poor
Normal
Good
Normal
Poor
Normal
Good
Normal
Poor
Good
Good
Good
Good
Normal
Good
Normal
Good
Normal
1.00E-5
3.16E-5
3.16E-5
6.94E-5
6.94E-5
9.52E-5
1.31E-4
1.31E-4
3.17E-6
3.17E-6
1.00E-5
1.00E-5
7.29E-6
7.29E-6
1.00E-5
1.00E-5
3.17E-6
1.00E-6
3.17E-7
3.17E-7
2.32E-7
2.32E-7
1.69E-7
1.69E-7
1.00E-5
3.11E-5
3.11E-5
6.76E-5
6.76E-5
9.24E-5
1.26E-5
1.26E-5
3.22E-6
3.22E-6
1.00E-5
1.00E-5
7.32E-6
7.32E-6
1.00E-5
1.00E-5
3.22E-6
1.04E-6
3.34E-7
3.34E-7
2.44E-7
2.44E-7
1.79E-6
1.79E-6
(3)
(4)
(5)
(6)
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
Zadeh L.A., Fuzzy sets. Information and Control, Vol. 8, pp. 338-353,
1965.
Kaufmann A., Gupta M.M., Fuzzy mathematical models in engineering
and management science. North-Holland, Amsterdam, 1988.
Blockley D.I., Predicting the likelihood of structural accidents. Proc. Inst
Civ. Eng., 59(2), pp. 659-668, 1975.
Blockley D.I., Analysis of structural failures. Proc. Inst Civ. Eng., 62(1),
pp. 51-74, 1977.
Brown C.B., Yao T.P., Fuzzy sets and structural engineering. Journal of
Structural Engineering, ASCE, 109(5), pp. 1211-1225, 1983.
Itoh S., Itagaki H., Application of Fuzzy-Bayesian Analysis to Structural
Reliability. Proc. of ICOSSAR '89, ASCE, N.Y., pp. 1171-1174, 1989.
Chou K.C., Yuan J., Fuzzy-Bayesian approach to reliability of existing
structures. Journal of Structural Engineering, ASCE, 119(11), pp. 32763290, 1993.
Wu H.C., Bayesian system reliability assessment under fuzzy
environments. Reliability Engineering & System Safety, 83, pp. 227-286,
2003.
Benjamin J.R., Cornell C.A., Probability, statistics and decision for civil
engineers. Mc Graw Hill, New York, N.Y, 1970.
Asai K., Negoita C.V., Introduction to fuzzy set theory. Ohm Press,
Tokio,1978.
Kandel A., Fuzzy mathematical techniques with applications. Addison
Wesley, Reading, Massachusset, 1986.
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
705
Abstract
3D wall panels are used in the construction of exterior and interior bearing and
non-load bearing walls and floors of building of all types of construction. This
system consists of a welded wire space frame integrated with a polystyrene
insulation core and two layers of concrete on both sides. In this paper, attention
is focused on the experimental measurements of the seismic response of 3D wall
panels surrounded by a steel bending frame. The approach of quasi-static cyclic
loading is employed using horizontal actuators to the combined system. The
vertical, lateral and horizontal displacements are measured by LVDT equipment.
The failure mechanism of 3D wall panels is described in detail. The evaluation
of strength and stiffness degradation of the whole system is presented based on
the envelope force-displacement curve of actual specimens under cyclic loads.
The results of the current study are shown in the form of ductility factors,
hysteresis loops and load-displacement envelope curves. The comparison
between the ductility of sole steel frames, 3D shear walls and the combined
system as the main theme of the current research is presented. Finally, this work
clarifies the benefits of using 3D wall panels as a strengthening method for
existing steel frame buildings and confirms the feasibility resistance of such
combined systems.
Keywords: 3D-panels, combined system, cyclic loading.
Introduction
3D wall panels are used in construction of exterior and interior bearing and nonload bearing walls and floors of building of all types of construction. This system
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
doi:10.2495/HPSM06069
Figure 1:
3D Sandwich panel.
Salmon and Einea [1] presents the results of full-scale test of prototype
sandwich panel under transverse loading in a vertical position. Nijhawan [2]
measured experimentally the interface shear force and designed the shear
connectors. Eiena et al. [3] used the plastic composite diagonal elements to
implement in sandwich panel as shear connector for increasing the thermal
insulation of this system.
Through this study the behavior of 3D panels in combination to steel moment
frame was investigated, the fracture mechanism of concrete wythes and the
adequacy of steel bars designed based on ACI 318-95 and procedure of PCI
design handbook [4].
Theoretical study
max
y
Experimental program
Four 3D walls are provided to be combined with portal steel frames. These
specimens are considered to represent the critical and structural elements with a
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
707
Figure 2:
Specimen
No.
Wall
Dimensions
(cm)
Specific
Gravity
(kg/m3)
Max. app
lied force
(tone)
035
036
037
038
120*64
120*64
120*64
120*64
2330
2310
2290
2280
87
91
82
85
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
Cube
compressive
strength
(bars)
387
404
364
378
Core No.
Core Dimensions
(Diameter*length)
Specific
gravity
(kg/m3)
Max.
applied
force (kg)
1
2
3
4
5
107.6*54.4
100.4*54.4
99.5*54.4
108.3*54.4
103.3*54.4
2150
2170
2230
2100
2250
5350
5100
5250
5400
5200
Table 3:
Specimen
No.
1
2
Specimen
Dimensions
Tensile strength.
Specific
gravity
(kg/m3)
Max.
applied
force (kg)
Slump
(cm)
2300
2320
3700
3500
8
8
15*30
15*30
Core
Compressive
strength
(kg/cm2)
230.2
219.4
225.9
232.3
223.7
Cylinder
tensile
strength
(kg/cm2)
52
50
709
cracking state, yielding state and ultimate state. The applied displacement is
started from 0.5 mm to 5 mm in 10 cycles. In the second phase of loading, the
increment of displacement is increased to 1 mm and after 19 cycles in
displacement of 14 mm the displacement increased to 2 mm and after 23 cycle
the increment increased to 4 mm and in 30 cycle system was failed.
Linear transducer of types LVDT is used to measure and monitor the
horizontal displacements at top, mid height and bottom of specimens. The
measured values of applied load and displacement are recorded by a computer
data logger capable of measurement to sensitivity ranges of 0.1 N, 0.001 mm,
respectively.
40
30
20
Displacement (mm)
10
0
0
10
15
20
25
30
-10
-20
-30
-40
-50
Cycle Number
Figure 4:
Figure 5:
Cycle No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Reaction Forces in
Both sides
(kN)
back
forth
Panel Stiffness in
both sides (kN/m)
back
forth
15.144
27.891
39.463
50.134
60.301
69.308
77.705
85.582
92.177
104.830
116.267
126.511
135.015
143.335
153.12
162.463
169.196
176.249
187.332
194.126
194.889
186.294
172.539
150.83
125.381
114.359
104.619
87.384
55884
45237
40847
37871
35397
33387
31713
30042
28468
26106
24235
22688
21520
20516
19825
19214
18586
17996
16895
15783
14006
11800
9790
7075
5070
4080
3332
2546
136762
83804
63561
50860
43414
38511
34540
31671
29432
26847
247.76
23284
21971
21158
20509
19686
18892
18228
17144
15992
14351
12252
8505
6179
4916
3697
2994
2582
16.808
29.113
42.745
53.966
63.828
72.346
79.857
86.559
93.353
108.191
121.580
134.373
146.174
156.906
166.95
176.630
184.905
192.553
206.064
214.369
217.773
213.666
174.111
152.937
143.06
123.793
116.588
112.512
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
711
80
200
60
150
50
0
-60
-40
-20
-50
20
40
40
100
20
0
-80
-60
-40
-20
20
40
60
80
-20
-100
-40
-150
-60
-200
-80
-250
Displacement (mm)
Displacement (mm)
Discussion
Figure 7:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
200
150
100
50
0
0
10
20
30
40
Displacement (mm)
Figure 8:
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
50
Table 5:
Specimen
without gap
Wall panels
713
max
035
21.32
2.44
8.73
036
22.07
2.58
8.55
Conclusions
The current work describes the cooperation between 3D panels as infill wall and
traditional steel frames. The following conclusions remarks are raised based on
previous work by the first author [5]:
- For all specimens, the plastic hinge is formed at the extreme fiber of the wall
section and at the vicinity at the base, above the foundation. The distribution and
propagation of cracks show that 3D sandwich panels with limited height,
behaves in shear performance. The observed horizontal crack at the base prior to
ultimate state may be due to sliding at the vertical reinforcement. This caused
considerable reduction in the strength, stiffness and energy dissipation of the
specimens.
- Externally reinforcing 3D wall panels, which basically behaves in shear
manner, enhances more ductility in performance design approach and increases
substantially load carrying capacity of system.
Finally, this study clarifies benefits of using lightweight prefabricated panels
as a strengthening method for existing of steel frame building and confirms
feasibility resistance of such compound system.
Acknowledgement
The authors gratefully acknowledge the financial support from Pre-fabricated
sandwich Panel firm, SAP company in Tehran for present work
References
[1]
[2]
[3]
[4]
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
WIT Transactions on The Built Environment, Vol 85, 2006 WIT Press
www.witpress.com, ISSN 1743-3509 (on-line)
715
Author Index
Abdel-Mooty M. A. N. ............ 419
Aiello R.................................... 449
Akita K. ........................... 351, 359
Al-Jabri K. S. ..................... 93, 381
Alnuaimi A. S. ........................... 93
Aoyama E. ............................... 169
Arasteh A................................. 269
Artemev A. .............................. 103
Avalle M. ................................. 249
Avila J. A................................. 439
Belingardi G. ........................... 201
Bhandari Y............................... 331
Bitar M. A................................ 419
Bossuyt S. ................................ 239
Boufas S................................... 311
Bouquerel J. ............................. 259
Brooks W................................. 481
Caliez M. ................................. 211
ern R.................................... 409
Chalioris C. E. ......................... 459
Cheng L. .................................. 501
Chew H. B. .............................. 501
Chiandussi G............................ 201
Cizmar D.................................. 429
Cooreman S. ............................ 239
Couroneau N.............................. 73
Crea F. ..................................... 449
Crooks R. ................................. 549
Cruz H...................................... 539
Custdio J. ............................... 539
Cuypers H. ................................. 21
Darling T.................................. 323
Dascotte E................................ 225
De Bolster E............................... 21
De Temmerman N. .................... 41
De Wilde W. P...............................
................... 3, 13, 21, 31, 625, 635
Debacker W. ............................ 625
Degrieck J. ............................... 259
Dns G. .......................... 301, 311
717
High Performance
Structures and Materials
III
Edited by: W. P. DE WILDE, Vrije
Universiteit Brussel, Belgium,
C. A. BREBBIA, Wessex Institute of
Technology, UK
Most high performance structures require
the development of a generation of new
materials, which can more easily resist a
range of external stimuli or react in a nonconventional
manner.
Featuring
authoritative papers from the Third
International Conference on High
Performance Structures and Materials, this
book addresses issues involving advanced
types of structures, particularly those based
on new concepts or new types of materials.
Particular emphasis is placed on intelligent
smart structures as well as the application
of computational methods to model, control
and manage these structures and materials.
The book covers topics such as: Damage
and Fracture Mechanics; Composite
Materials and Structures; Optimal Design;
Adhesion and Adhesives; Natural Fibre
Composites; Failure Criteria of FRP; Nonlinear Behaviour of FRP Structures; Material
Characterization; High Performance
Materials; High Performance Concretes;
Aerospace Structures; Reliability of
Structures; Ceramics in Engineering.
WIT Transactions on The Built Environment
Volume 85
ISBN: 1-84564-162-0 2006 apx 350pp
apx 129.00/US$233.00/193.50
Find us at
http://www.witpress.com
Save 10% when you order from our encrypted
ordering service on the web using your credit
card.
Computational Mechanics
for Heritage Structures
B. LEF;HERIS, Technical University of
Crete, Greece, M. E. STAVROULAKI,
Technical University of Crete, Greece, A.
C. SAPOUNAKI, , Greece,
G. E. STAVROULAKIS, University of
Ioannina, Greece
For thousands of years people have built
great structures utilizing the mechanics of
beams, arches, columns and their
interactions. It is now possible to use new
computer technology to examine the great
structural achievements of the past and to
WIT eLibrary
Home of the Transactions of the Wessex
Institute, the WIT electronic-library provides
the international scientific community with
immediate and permanent access to
individual papers presented at WIT
conferences. Visitors to the WIT eLibrary
can freely browse and search abstracts of
all papers in the collection before
progressing to download their full text.
Visit the WIT eLibrary at
http://library.witpress.com
Computational Methods in
Materials
Characterisation
Editors: A.A. MAMMOLI, University of
New Mexico, USA and C.A. BREBBIA,
Wessex Institute of Technology, UK
Papers from the first international
conference on this subject. Topics covered
include:
Parameter Identification;
Thermomechanical Behaviour; Damage
Mechanisms; Foams; Polymers; and
Interface Phenomena.
Series: High Performance Structures and
Materials, Vol 6
ISBN: 1-85312-988-7 2003 368pp
121.00/US$193.00/181.50
Evolutionary Algorithms
and Intelligent Tools in
Engineering Optimization
Edited by: W. ANNICCHIARICO,
Universidad Central de Venezuela,
Venezuela, J. PERIAUX, Pole Scientifique
Dassault Aviation, France, M.
CERROLAZA, Universidad Central de
Venezuela, Venezuela, G. WINTER,
Universidad de la Palmas de Gran
Canaria, Spain
Evolutionary Algorithms are very powerful
techniques used to find solutions to realworld search and optimization problems.
They are artificial intelligence techniques
which mimic nature according to Darwins
principal of the survival of the fittest in
order to explore and explode large search
space and find near-global optima.
In this book, a large spectrum of innovative
evolutionary and intelligence methods are
Impact Loading of
Lightweight Structures
Introduction to Motion
Based Design
Deployable Structures
Analysis and Design
C.J. GANTES, National Technical
University of Athens, Greece
Accessible to practicing structural engineers
and graduate students with no previous
knowledge of the field, this title formulates
and solves the complex engineering design
problems with which deployable structures
are associated. It also presents the issue of
design of snap-through type deployable
structures in an organized way which will be
of interest to more experienced readers.
Series: High Performance Structures and
Materials, Vol 2
ISBN: 1-85312-660-8 2001 384pp
132.00/US$198.00/198.00
WIT Press
Ashurst Lodge, Ashurst, Southampton,
SO40 7AA, UK.
Tel: 44 (0) 238 029 3223
Fax: 44 (0) 238 029 2853
E-Mail: marketing@witpress.com