Beruflich Dokumente
Kultur Dokumente
OF A PRECAST CONCRETE
HOLLOW-CORE FLOOR SUPER-ASSEMBLAGE
A THESIS
SUBMITTED IN PARTIAL FULFILLMENT
OF
MASTER OF ENGINEERING
AT THE
UNIVERSITY OF CANTERBURY
BY
CAMERON MACPHERSON
UNIVERSITY OF CANTERBURY
CHRISTCHURCH, NEW ZEALAND
2005
TABLE OF CONTENTS
Abstract...vi
Acknowledgements... vii
1 INTRODUCTION...1-1
1.1 Introduction..1-1
1.2 Previous Relevant Research 1-3
1.3 Findings from Matthews (2004)...1-4
1.3.1 Origin of the Test Specimen Super-Assemblage. 1-4
1.3.2 Loading System... 1-5
1.3.3 Major Findings: Beam-to-Floor Slab connection 1-6
1.3.4 Major Findings: Lateral Frame-to-Floor connection... 1-7
1.3.5 Analytical Findings.. 1-8
1.4 Findings from Lindsay (2004) 1-9
1.4.1 Hollow-core Connection Details. 1-9
1.4.2 Major Conclusions and Recommendations... 1-11
1.5 Outline of this Thesis. 1-13
1.6 What is new and significant in this Thesis?...1-15
1.7 References..1-16
Page i
2.8 Strength and Failure Mode Prediction Models.. 2-28
2.8.1 Moment Capacity...2-28
2.8.2 Low Cycle Fatigue. 2-29
2.9 Discussion of Results. 2-30
2.10 Conclusions2-32
2.11 References..2-32
Page ii
4 FORENSIC ANALYSIS OF FAILURE AND BEHAVIOUR MODES.. 4-1
Section Summary. 4-1
4.1 Introduction..4-2
4.2 Torsional Behaviour of the Transverse Support Beams.. 4-4
4.2.1 Determination of the Torsional Demand. 4-9
4.2.2 Transverse Beam Torsional Capacity 4-12
4.3 Displacement Incompatibility 4-14
4.3.1 Identified Issues and Previous Findings 4-14
4.3.2 Analytical Modelling of Displacement Incompatibility 4-17
4.3.3 Scope for Future Advancements 4-19
4.4 Lateral Load Capacity of the System.4-20
4.4.1 Bowstring Effect 4-21
4.4.2 General System Strength... 4-26
4.4.3 Localised Strengths 4-28
4.5 Beam Elongation... 4-31
4.6 Low Cycle Fatigue of the Reinforcement.. 4-33
4.7 Comparisons with Previous Research... 4-34
4.7.1 Performance of the Seated Connection..4-35
4.7.2 Performance of the Perimeter Frame to Hollow-Core Connection... 4-37
4.7.3 Overall Performance Comparisons 4-38
4.8 Conclusions4-40
4.9 References..4-41
Page iii
APPENDIX A SUB-ASSEMBLAGE EXPERIMENTS: CONSTRUCTION & RESULTS...A-1
A.1 Construction Details and Photographs A-1
A.2 Material Tests..A-5
A.3 Testing Photographic Log... A-1
A.3.1 Experiment Set-up and Instrumentation. A-6
A.3.2 Specimen 1.. A-7
A.3.3 Specimen 2.. A-8
A.4 Loading Protocol.......................A-10
A.5 Low Cycle Fatigue Prediction...A-11
A.6 References. A-12
Page iv
D.1.3 Column Contribution.. D-2
D.2 Super-Assemblage StiffnessD-4
D.3 Torsion Test D-5
D.4 General Super-Assemblage Performance... D-6
D.4.1 Central Column Joint Reinforcing Bar Slip D-6
D.4.2 Hollow-core floor unit pull-off... D-8
D.4.3 Beam Column Joint Rotation Contributions.. D-10
D.4.4 Reinforcement Strain Information.D-12
D.5 West Transverse Beam Torsion D-18
D.5.1 Experimental Results D-18
D.5.2 Calculation of Torsional Demand. D-19
D.5.3 Calculation of Torsional Capacity D-20
D.6 Displacement Incompatibility... D-23
D.6.1 Experimental Results D-23
D.6.2 Analytical Modelling D-28
D.7 Beam Elongation...D-31
D.8 Low Cycle Fatigue Fracture of Reinforcement D-34
D.9 Observed Bowstring Effect... D-36
D.10 Super-Assemblage Capacity Mechanism Calculations D-38
D.10.1 Phase 1 and 3: Longitudinal ... D-38
D.10.2 Phase 2: Transverse.. D-40
D.11 References. D-46
Page v
ABSTRACT
Recent earthquakes and laboratory research has demonstrated the seismic vulnerability of
buildings constructed with precast prestressed concrete hollow-core floor systems. However,
further investigations have shown that with simple detailing enhancements, significant
improvement in the seismic performance of hollow-core floor systems can be expected. The
New Zealand Concrete Structures Standard provides two acceptable solutions for the
connection of hollow-core floor units to concrete supporting beams (Amendments to NZS
3101, 2004). The second of these acceptable solutions is currently untested, and the present
experimental investigation aims at validating this solution as well as several other new
detailing enhancements. Two dimensional sub-assemblage component experiments are
conducted utilising a newly developed, more realistic loading configuration. Based on the
sub-assemblage research findings, a concrete frame super-assemblage is constructed
incorporating a reinforced connection that rigidly ties the floor into the supporting beam and
an articulated topping slab cast onto a timber infill solution that bridges between the hollow-
core units and parallel longitudinal frame beams. The full-scale structure is cyclically tested
in both the longitudinal and transverse directions to inter-storey drifts of 5%. Visual and
instrumental observations from the experiment are presented and discussed. A forensic
analysis of the experimental observations is performed. Torsional hinging behaviour that is
observed due to a strong floor-to-beam connection and weak support beam is explained and
experimentally observed displacements are compared with computational modelling. The
lateral load capacity of the super-assemblage is predicted and it is seen that it is necessary to
incorporate all sources of strength enhancement. Finally, the results from this present
investigation are compared and contrasted with similar previous research investigations and
recommendations for the forthcoming revision of the New Zealand Concrete Standard, NZS
3101, are made. The experimental observations demonstrate that with appropriate detailing
enhancements there can be significant seismic behaviour improvements whereby a relatively
fragile hollow-core flooring system is transformed into a robust frame-floor system where all
the damage is transferred into the plastic hinge zones of the supporting moment resisting
frame.
Page vi
ACKNOWLEDGEMENTS
This research was undertaken in the Department of Civil Engineering, at the University of
Canterbury under the supervision of Professor John Mander and Professor Des Bull. I would
like to thank John and Des for their support, encouragement, expertise and guidance
throughout this research and my university studies.
This project would simply not have been possible without the support of many
companies and organizations. Financial support has been made for this research by a grant
from the EQC Research Foundation. Products and support were provided by the following
organisations: University of Canterbury, Firth Industries Ltd, Stresscrete, Pacific Steel and
Fletcher Reinforcing, Construction Techniques and McDowels Concrete Accessories. All of
this financial and in-kind support is gratefully acknowledged.
Jeff Matthews and Renee Lindsay, from whom this research has continued on from,
deserve enormous amounts of thanks. Their friendliness, knowledge and experience from
their studies have helped me considerably. Jeff and Renees high standard of work has also set
high qualities for me to aspire to.
I would like to thank John Trowsdale for his support throughout my research, and
acknowledge his work on the sub-assemblage part of this project. Johns practical expertise
and friendliness were invaluable in completing the experiments. I would also like to thank my
fellow postgraduate students for their friendship and understanding of postgraduate life.
Full credit must go to all of the technical staff of the Civil Engineering Department. I
am thankful for everyones helpfulness and guidance. I would like to give special thanks to
Russel McConchie, Nigel Dixon, Tim Perigo and Richard Newton for the amount of effort
they have put into my research and for all of their support and friendliness that made my
experimental work so enjoyable. I would also like to thank Dr Stefano Pampanin for his
guidance, expertise and encouragement. I have learnt a lot from Stefano throughout my
postgraduate studies.
Finally, I would like to thank my friends and family for all the love and support and
most especially, Ryoko for her guidance, patience and support throughout my time at
university.
Page vii
1. Section 1 Introduction
1.1 INTRODUCTION
Recently, precast concrete construction has become the conventional form of construction in
New Zealand. The ease of construction, rapid erection times and the appeal of the off-site
construction and factory-like controls available in the precast, prestressed industry are reasons
for its popularity. Since the 1980s, one of the most common construction types in New
Zealand has been the use of precast prestressed hollow-core floor units seated on reinforced
concrete moment resisting frames. However, the proliferation of precast concrete construction
has preceded research in many areas and there are concerns over the seismic performance of
these hollow-core floor systems in New Zealand. The possibility for the loss of seating to
occur, lack of composite bond strength with the topping concrete, cracking of the webs of the
hollow-core units and the general lack of structural redundancy have been highlighted as main
factors that could lead to poor performance and collapse of several precast concrete buildings.
Over the last few years, research at the Universities of Auckland and Canterbury has sought to
investigate the seismic performance of these precast systems and the level of seismic
Recently, research was conducted to fully assess the adequacy of a typical New
Zealand designed and built concrete structure with hollow-core floors (Matthews, 2004;
Zealand multi-storey concrete frame building with hollow-core floor units was constructed.
investigating lateral load resisting systems, and the effects of floor diaphragms have often
been overlooked by the simplification of the rigid diaphragm assumption. The results from the
Page 1-1
Matthews (2004) experiment showed that precast concrete structures in New Zealand may not
perform as well as would be hoped and the collapse of floors could be expected. The overall
collapse of the floor units during the experiment showed similar traits to some failures
observed after the Northridge (1994) earthquake (Norton et al, 1994), and the main concern
became the detailing used for the connections of the hollow-core floors to the concrete frame.
The investigation into the seismic vulnerability of existing buildings provided several
for future construction, the retrofit of existing structures with hollow-core floors, as well as
further investigations into the problems. The issues surrounding the use of prestressed floors
that have stemmed from the Matthews (2004) experiment, previous research and earthquake
events, have been recognised in New Zealand Concrete Design Standards. Specific
requirements are in place prescribing minimum seat widths and different connection solutions
that will address the shortcomings highlighted with traditional construction methods
response to the Matthews (2004) testing a repair was carried out on the concrete super-
assemblage and second phase testing was undertaken by Lindsay (2004) and Lindsay et al
(2004). This experiment investigated improved connection details for the hollow-core floor
systems for new construction that have been included in proposed amendments to the New
This thesis follows the research completed by Matthews (2004) and Lindsay (2004).
This chapter initially briefly discusses the previous sub-assemblage and super-assemblage
research related to hollow-core floor systems. A summary of the Matthews (2004) and
Lindsay (2004) experiments is provided, including the origins of the test specimen, the
development of the loading system, and a review of the main test findings. The main sections
Page 1-2
1.2 PREVIOUS RELEVANT RESEARCH
The performance of hollow-core floor systems has been examined in many experimental
Herlihy et al (1995), Oliver (1998) and Herlihy and Park (2000) have all examined the seat
requirements of hollow-core floor units due to beam elongation for a variety of end
connection details. Further research by Bull and Matthews (2003) and Liew (2004) have
similarly tested various connection details, but introduced the relative rotation between the
floor slab and supporting beam as the main cause of damage. A more in-depth summary of
There have been many super-assemblage research programs relevant to this project
involving concrete frame structures with and without floor slabs. Zerbe and Durrani (1989,
1990), Fenwick and Megget (1993), Restrepo (1993), Fenwick et al (1995) and Lau (2002)
have conducted tests examining beam elongation and the contribution of floor slabs to the
overall performance and strength of lateral load resisting elements. However, due to the
respective test set-ups used, beam elongation has been either promoted or restrained and the
results were not realistic. Despite sub-assemblage research previously investigating hollow-
core seating connections, prior to the Matthews (2004) investigation, a full-scale experiment
examining the seismic adequacy of hollow-core floor systems had not been undertaken. The
need for these tests is due to three-dimensional effects, such as beam elongation, torsion and
displacement incompatibility between the perimeter concrete frames and flooring, which all
displacement incompatibility has been acknowledged as a source of damage (Lau et al, 2002
and Priestley et al, 1999), but little quantitative research has been done to explore the issue.
Page 1-3
1.3 FINDINGS FROM MATTHEWS (2004)
experimental program was initiated to examine the seismic adequacy of such systems in the
New Zealand building stock. A two bay by one bay concrete moment resisting frame
performed relatively well whereas the hollow-core flooring system behaved poorly and
collapsed at moderate levels of seismic intensity. The following subsections briefly describe
the origin and details of the super-assemblage including the seated and lateral connections of
the hollow-core floor units, the development of the self-equilibrating load frame and the main
concrete moment resisting frame building built in New Zealand over the last two decades.
Figure 1-1 shows a view of a representative structure from which test specimen was derived
from. The corner section was chosen to examine the effects of earthquake actions in different
directions - longitudinally as well as orthogonal to the span of the flooring system. The bay
widths were 6.1m long and the 300 series hollow-core flooring spans over two bays past an
intermediate column and is seated on the two end beams. Although this type of construction
whereby the double span flooring is present is not the most common form, it was chosen to
investigate a worst-case scenario. Buildings of this type have been constructed in Wellington,
which is a high seismicity region of New Zealand, whereas the use of double span hollow-
core is less frequent in Auckland and other regions of New Zealand. For these reasons, the
descriptions of the findings from the Matthews (2004) test will indicate, the serious
Page 1-4
inadequacies of the hollow-core flooring were centred on the supporting beam-to-slab
Figure 1-1 Selection portion representing the test super-assemblage (Matthews, 2004)
An integral part of the Matthews (2004) experimental program was the development of the
loading frames. As a strong wall or strong floor was unavailable to provide reaction support
for loading, a self-equilibrating load frame was constructed to load the structural super-
assemblage. The loading frames were designed and implemented such that under lateral
loading, the building deforms in a realistic manner and that the advent of beam elongation is
neither restrained nor promoted. The self-equilibrating primary loading frame is shown in
Figure 1-2. By controlling the actuators within the loading frame, equal and opposite diagonal
loads are applied to the tops and bottoms of the columns ensuring that the system is in
equilibrium (no axial forces are induced in the beams) and that axial forces cancel within the
system.
To ensure that the columns displace parallel to each other as they would in a real
building, a secondary load frame (SLF) was used. The frame, as shown in Figure 1-2, acts to
keep the columns parallel to each other but also allows beam elongation to occur through a
Page 1-5
reduction of the initial angle of the frame. The design of the loading frames and test specimen
allows the super-assemblage to be tested in both longitudinal and transverse directions. For
longitudinal loading, additional rams are attached at the base of the back columns to maintain
the same column inclinations and ensure that any artificial torsion in the transverse beams
Ac B
tua tor
tor tua
D Ac
E Ac
tor tu a
tua tor
Ac A
Primary
Loading frame Primary
Loading frames
Figure 1-2 Self equilibrating and secondary loading frames used by Matthews (2004)
In compliance with the overlying investigation into the seismic abilities of existing buildings,
the test super-assemblage included the floor slab connection with the supporting beam
predominantly used in New Zealand construction practice. The seating connection featured a
300 series hollow-core unit seated on a mortar bed on either a 20mm or 40mm seating. The
cores at the end of the floor units were plugged with common plastic end bungs to prevent
concrete from entering the cores. A 75mm thick topping slab was cast on top of the floor units
and conventional Grade 300 starter bar reinforcement was used together with the cold drawn
mesh in the topping slab. Initial damage to the supporting beam-to-floor slab connection
during testing occurred earlier than expected and the mode of failure was different to what
was previously assumed. Originally it was thought that loss of seating occurred due to the
floor units sliding off the seat of the supporting beams. However, it became apparent that the
Page 1-6
relative rotation between the flooring and beam was the major source of damage. The mortar
bed possessed sufficient rigidity to ensure that the floor units could not slide, and the negative
moments induced by the relative rotation caused diagonal compressive struts and orthogonal
tensile stresses, led to splitting of the unreinforced webs of the hollow-core. Positive moments
created large crack openings, and under cyclic loading the connection deteriorated
considerably, ultimately causing the entire support at both ends of the structure to be lost. A
photograph illustrating the overall failure of the flooring is shown in Figure 1-3, and a
diagram of the supporting beam floor slab connection is shown in Figure 1-4(a).
Figure 1-4(b) shows the connection used for the Matthews (2004) experiment between the
perimeter moment resisting concrete frame and the adjoining parallel hollow-core unit. In
addition to the damage incurred by the end connection, the first hollow-core unit adjacent to
the perimeter frame showed signs of distress at an early stage of testing. The construction
connection inadvertently rigidly tied the floor into the perimeter beams, and under lateral
displacements, the flooring was forced to displace in the double curvature pattern of the frame
beams. Hollow-core flooring sags by nature and this undesirable deformation caused internal
cracking of the webs of the hollow-core unit and the eventual failure of the lower half of the
first unit.
The displacement incompatibility coupled with the tension forces in the floor
diaphragm caused the cold drawn mesh reinforcement within the topping slab to fracture and
a large tear to form within the topping concrete along the interface between the first and
second hollow-core units spanning parallel to the perimeter frame. This caused the centre
column of the front frame to displace outwards, transverse to the plane of loading. The
broader ramifications of this can be seen if this outward column movement was applied to
Page 1-7
several floors of a multi-storey structure during an earthquake. Premature buckling of the
interior intermediate columns could result in the potential collapse of the structure. This also
identifies the necessity to tie the intermediate column into the floor diaphragm, and this was
Figure 1-3 Overall failure of the floor units in the Matthews (2004) experiment
analytical work was undertaken to complement the observed results. A capacity mechanism
analysis was used to calculate the overall strength capacity of the super-assemblage. A
method was proposed to determine the contribution of the floor diaphragm to the overall
(2004) to predict beam elongation. The method uses rigid body kinematics and a rainflow
counting technique to predict growth of plastic hinges under increasing rotations. For both
beam elongation and the strength capacity prediction models, the comparisons with the
Page 1-8
1.4 FINDINGS FROM LINDSAY (2004)
A continuation of the work done by Matthews (2004) was conducted by Lindsay (2004). This
investigation aimed to verify recommended details for the improved performance of future
investigated past construction techniques (1985-2000) and the resulting problems with
seismic performance. Lindsay (2004) repaired and reconstructed the Matthews (2004) full-
scale super-assemblage specimen and then retested it under a more rigorous loading regime.
Modifications were made to improve the performance of the hollow-core floor system for
both the lateral and end seating connections between the frame and hollow-core units, as well
as more adequately tying the central column into the floor diaphragm.
As shown in Figure 1-4(c), Lindsay (2004) employed a beam to floor connection detail that
used a low-friction bearing strip and a compressible backing board, which replaced the plastic
end plugs in the cores at the end of the units. The seating connection detail used during the
Matthews (2004) test could be likened to a semi-rigid type of connection. The poor
behaviour led to further testing of a pinned type of connection by Lindsay (2004), whereby
the low friction bearing strip allows the floor units to slide horizontally relative to the
supporting beam, and the compressible backing board to compress reducing large
compression stresses entering the floor unit. The overall performance was satisfactory, and
significant improvements were seen relative to the Matthews (2004) support condition. The
connection was able to sustain sizeable relative rotations with overall structural inter-storey
drifts reaching +5%. The compressible backing board did not compress as expected which
indicates beam elongation effects were causing pull-off of the flooring units off the seating
Page 1-9
away from the supporting beam. The low friction bearing strips also did not perform as
originally expected, whereupon the rough side of the bearing strip slid on the concrete seat
and the low friction side adhered to the floor unit. Significant spalling of the concrete ledge
(seat) of the supporting beams later in the experiment reduced the seat width to minimal
75mm concrete
topping
(c) Lindsay (2004) floor-slab connection (d) Lindsay (2004) lateral connection
Figure 1-4 Hollow-core connection details used by Matthews (2004) and Lindsay (2004)
To overcome the issues of the displacement incompatibility that occurs between the perimeter
frame and the floor units, a topping slab was cast onto a timber infill that bridged between the
hollow-core units and perimeter frame beams. Figure 1-4(d) shows a diagram of the
connection. The timber infill was proposed subsequent to the poor behaviour witnessed in the
Matthews (2004) experiment to allow more desirable one-way hollow-core action. The thin
slab was designed to provide a flexible interface and accommodate the displacement
Page 1-10
incompatibility between the floor and perimeter beams. Overall, the infill detail showed a
marked improvement (when compared to Matthews (2004). However the ductile mesh used to
reinforce the topping slab fractured and a major longitudinal tear formed along the interface
between the infill and hollow-core, which interfered with the transfer of inertia and structural
interaction forces from the floor to the moment resisting frame. The presence of the drag bar
reinforcement tying the centre column into the floor plate ensured that there was no outward
translation of the centre column similar to what was observed during the Matthews (2004
experiment.
The improved detailing utilised in terms of the seated and lateral hollow-core connection
details and adequate tying of the central column into the floor, meant that the super-
assemblage was able to sustain inter-storey drifts up to +5%, until the low cycle fatigue of the
beam longitudinal reinforcement occurred. Although the experiment was success in many
areas, it also uncovered other issues that the experiment was unable to address. Some of the
detailing alterations made from the Matthews (2004) experiment still required further
enhancement and the following recommendations for professional practice and improving
1. Cold drawn or ductile mesh should not be used and conventional reinforcement should
be used in its place for diaphragm topping reinforcement over precast hollow-core
units.
2. In the presence of an infill slab offsetting the flooring from adjacent frame beams, the
starter bars used should extend over the infill, across the infill-to-hollow-core floor
interface and extend above the first floor slab unit. This will increase the ductility at the
Page 1-11
3. Hollow-core units should not be seated on potential plastic hinge zones of supporting
beams. This should be overcome by having an infill slab over the highly deformable
hinge region, and hooked longitudinal reinforcement within the supporting beam to
concentrate large deformations away from the seated hollow-core units and close to the
column face.
4. A second generation low friction bearing strip should be employed for future use,
whereby the floor slab slides horizontally on the strip relative to the supporting beam,
and the strip adheres to the concrete seating. A strip featuring longer teeth on the
rough side will help to minimise spalling by evening out the surface roughness of the
seat concrete and help to ensure that the bearing strip will stay on the seat as designed.
In addition, transverse reinforcement should be used within the concrete seating of the
supporting beams.
Similarly, several recommendations were made by Lindsay (2004) for future research. These
1. Due to the large amount of time and resources required to undertake large full-scale
performed, similar to those performed by Bull and Matthews (2003) and Liew (2004).
Despite the limitations of this method of testing, the comparable performance and
assemblage tests are of great use in order to easily and relatively quickly assess the
adequacy of many other types of seated connections. Such tests could allow the
generation of a database of results and allow for a wide range of details to be tested
including all the different types of precast elements used in New Zealand construction.
The effects of beam elongation should also be incorporated into the testing procedure.
Page 1-12
2. Of a more urgent nature, the second of the two details included in the 2003 Amendment
to the New Zealand Concrete Structures Standard (NZS 3101, 2004) needs to be tested
existing buildings, a retrofit connection solution needs to be developed and tested such
Other questions relating to hollow-core seated connection details were raised including; what
is the effect of negative seating (a bridged connection detail when the hollow-core units are
cut too short)? What is the effect of different beam geometries? What are the effects of
placing additional shear reinforcement in the cores of the units? Recommendations were also
made for further analytical research. The torsion of the transverse beams needs to be modelled
to gain an idea of when full activation of the starter bars occurs. Also analytical modelling of
This thesis comprises of three main sections. Following this introductory section, Section 2
research at a sub-assemblage level is more closely reviewed and the experimental set-up is
described. The results are presented and from which the conclusions and recommendations
are drawn.
Section 3 covers the construction of the moment resisting frame and the modifications
made from the original Matthews frame (2004), as well as the final super-assemblage
specimen details and hollow-core connection details. The experimental loading regime and
instrumentation are outlined. The second part of this section provides visual and instrumental
Page 1-13
observations from the testing, hysteretic performance of the structure and a discussion of the
major results.
frame and the first floor slab unit is undertaken and comparisons between the analytical
results and experimental observations are made. The lateral system capacity is calculated from
the individual components of the system and compared with the experimental results. The
bowstring effect of floor slabs is also examined. The beam elongation results from the
between this experiment, Matthews (2004) and Lindsay (2004) in terms of the floor seating
connection, the lateral connection between the frame and initial hollow-core unit, and the
The final section of this thesis provides a summary and major conclusions from the
research presented. Recommendations are made for professional practice and future research.
Considerable material has also been included at the end of this thesis in the various
appendices. This material supplements the information provided in the main sections of the
thesis.
investigation and features photographic logs of the construction and testing as well as
Page 1-14
Appendix D offers additional instrumental results from the testing and calculations to
The investigation presented in this thesis experimentally examines the seismic capabilities of
a reinforced hollow-core support connection detail that has been proposed in amendments to
the New Zealand Concrete Design Standard (NZS 3101, 2004). To this point, the specific
detail has not been tested, and this research describes the performance of the connection detail
specimen has been constructed and incorporates the new detailing enhancements that seek to
address the shortcomings of hollow-core floor systems that have been highlighted from
previous research.
incompatibility between the floor and frame elements of a structure. Non-linear modelling has
been undertaken and the results compared with actual values obtained from the experiment.
The model has then been extended and examples of the possible applications of the model are
discussed. The bowstring effect in floor slabs is examined and a method is introduced to allow
for the influence it has on the lateral load capacity of the super-assemblage.
The overall findings from this research produce a greater level of understanding of
hollow-core floor systems and the three-dimensional interaction between floor slabs and
reinforced concrete frames. Important recommendations are made for both professional
design practice and construction as well as providing ideas for future research. According to
the results of the hollow-core connection detailing, direct recommendations are made for the
Page 1-15
1.7 REFERENCES
Bull D.K, Matthews J.G, 2003, Proof of Concept Tests for Hollow-core Floor Unit
Fenwick R.C, and Megget L.M, 1993, Elongation and load deflection characteristics of
reinforced concrete members containing plastic hinges, Bulletin of NZNSEE, Vol 26,
Fenwick R.C, Davidson B.J, and McBride A, 1995, The influence of slabs on elongation in
Herlihy M.D, Park R and Bull D.K, 1995, Precast Concrete Floor Unit Support and
Herlihy M.D, and Park R 2000, Precast concrete floor support and diaphragm action,
Lau D.B.N, Fenwick R.C and Davidson B.J, 2002, Seismic performance of R/C perimeter
frames with slabs containing prestressed units, Proceedings for NZSEE conference,
Lindsay R.A, 2004, Experiments on the seismic performance of hollow-core floor systems in
Lindsay R.A, Mander J.B, and Bull D.K, 2004a, Preliminary results from experiments on
Page 1-16
Lindsay R.A, Mander J.B, and Bull D.K, 2004b, Experiments of the Seismic Performance of
Liew H.Y, 2004, Performance of Hollow-core Floor Seating Connection Details, Masters
New Zealand.
Matthews J.G, 2004, Hollow-core floor slab performance following a severe earthquake, PhD
New Zealand.
Matthews J.G, Bull D.K, and Mander J.B, 2003a, Background to the Testing of a Precast
Matthews J.G, Bull D.K, and Mander J.B, 2003b, Preliminary Experimental Results and
Seismic Performance Implications of Precast Floor Systems with Detailing and Load
Matthews J.G, Mander J.B, Bull D.K, 2004, Prediction of beam elongation in structural
concrete members using a rainflow method, 2004 NZSEE Conference, Rotorua, NZ,
Meija-McMaster J.C and Park R, 1994, Tests on Special Reinforcement for the End Support
of Hollow-core Slabs, PCI Journal, Vol 39, No. 5, pp. 90-105, September-October.
NZS3101, 1995, Concrete Structures Standard, NZS3101, Parts 1 & 2, Standards New
NZS3101, 2004, Amendment No. 3 to 1995 Standard (NZS3101), Standards New Zealand,
Page 1-17
Norton J.A, King A.B, Bull D.K, Chapman H.E, McVerry G.H, Larkin T.J and Spring K.C,
OGrady C.R, 2004, Letters to the Editor, Journal of the New Zealand Structural Engineering
Oliver S.J, 1998, The Performance of Concrete Topped Precast Hollow-core Flooring
Systems Reinforced with and without Dramix Steel Fibres under Simulated Seismic
Christchurch, Canterbury.
Priestley M.J, Sritharan S, Conley J.R, Pampanin S, 1999, Preliminary results and
conclusions from the PRESSS five-storey precast concrete test building, PCI Journal,
Restrepo J.I, Park R and Buchanan A, 1993, Seismic behaviour of connections between
Zerbe H.E, and Durrani A.J, 1989, Seismic Response of Connections in Two-bay R/C Frame
2829-2844.
Zerbe H.E, and Durrani A.J, 1990, Seismic Response of Connections in Two-bay R/C Frame
Subassemblies with a floor slab, ACI Structural Journal, Vol. 87, July-August, pp 406-
415.
Page 1-18
2. Section 2 Sub-Assemblage Experimental Investigation
SECTION SUMMARY
seismic loads. The support connections are reinforced and rigidly tie the floor into the
supporting beam, the first of which has been proposed in Amendment No. 3 to the New
Zealand Concrete Standard, (NZS 3101, 2004). A newly developed test set-up is described
which incorporates both the relative rotation between the supporting beam and hollow-core
floor, and tension in the floor, to represent the elongation of a parallel moment resisting
frame, as the sources of damage. The sub-assemblage specimens were tested up to equivalent
inter-storey drifts of 5.0% and visual and instrumental observations from the experiments are
outlined. A simplified analytical study is conducted to better understand the performance and
observed failure modes. Results show promising behaviour, and after testing in a concurrent
revision of the New Zealand Concrete Standard, NZS 3101, are made.
Page 2-1
2.1 INTRODUCTION
Recent earthquake events and research undertaken has raised serious concerns of the seismic
the University of Canterbury have showed that a hollow-core floor diaphragm supported by a
ductile moment resisting concrete frame structure performed inadequately under simulated
seismic loads. Further experimental research conducted by Lindsay (2004) and Lindsay et al
(2004) demonstrated that improved detailing significantly improved the performance of these
hollow-core floor systems. Amendments to the New Zealand Concrete Design Standard (NZS
3101, 2004) stipulate two new acceptable solutions. The two connections could be
characterised as either pinned or rigid. The former pinned connection detail was tested
by Lindsay (2004), but the latter remains untested, and is the subject of this research.
In this section, the seismic performance of two hollow-core floor unit-to-support beam
dimensional sub-assemblage component test format. The loading set-up was developed from
previous sub-assemblage tests (Bull and Matthews, 2004 and Liew, 2004) whereby tensile
elongation effects were imposed on the floor-to-beam connection. This was considered
necessary to simulate the effect of beam growth that takes place in plastic hinge zones as a
result of cyclic loading effects. The two specimens feature rigidly reinforced connection
details, one of which is the detail proposed in Amendment No. 3 to NZS 3101:1995 (2004)
This section firstly reviews research relevant to these experiments, both at sub-
assemblage and super-assemblage formats and the progression made from this research to the
current specifications detailed in the New Zealand Concrete Design Standard. The details of
the test specimens and experimental set-up are provided and discussed. The second part of
Page 2-2
2.2 PREVIOUS RESEARCH FINDINGS AND NEW ZEALAND GUIDELINES
There have been several research programs at a sub-assemblage level that have investigated
Oliver (1998) and Herlihy and Park (2000) examined seat requirements of hollow-core units
due to beam elongation. In each set of experiments, various end connection details were
investigated under either a monotonic or a cyclic horizontal load, to simulate the loss of
seating (pull-off) effects that occur due to beam elongation. The experiments were primarily
undertaken to investigate whether the cast in-situ topping slab and starter bar reinforcement as
well as different reinforced core configurations are sufficient to stop the floor units from
collapsing when the hollow-core unit was pulled off the seat of the supporting beam.
between the hollow-core floor units and supporting beams, designed to carry the weight of the
floor units, performed well when a loss of seating occurred. The use plain round bars also
showed that a ductile connection could be achieved due to an early loss of bond strength
enabling large plastic elongations to be sustained. A limitation of the use of the beam-to-floor
connections used within structures expected to experience large horizontal movements was
floor connection details used in New Zealand construction. The testing showed that the
hairpin type of connection (reinforcement hooked at both ends connecting the floor and
supporting beam at the bottom of the hollow-core cells) performed adequately but concluded
that there was insufficient composite bond strength between the cast-in-place topping and the
hollow-core units to maintain a ductile connection when there was no reinforcement crossing
the topping to hollow-core interface. The authors raised concern over the performance of the
connections within structures under large horizontal movements and proposed an alternative
connection detail, the paperclip using plain round bars (two layers of reinforcement in the
Page 2-3
shape of a paperclip between the hollow cores and supporting beam) that could be used to
resist these large displacements. Further research by Herlihy and Park (2000) included a
monotonic downward loading to induce tension in the topping and to observe the amount of
conclusions were found to suggest that deformed bars traditionally placed in topping slabs are
insufficient to resist the effects of beam elongation that occurs in ductile frames. Oliver
(1998) concluded that the paperclip tie reinforced connection is not suitable in situations
where beam elongation is critical, such as the corner regions of a ductile concrete moment
resisting frame structure. The use of steel reinforced fibre concrete (SFRC) was also
investigated, and it was shown that the increased tensile capacity of the concrete due to the
under the tension induced in the floor system from beam elongation effects. The failure
modes witnessed from these tests were different to the damage observed in the aftermath of
the Northridge Earthquake, when a loss of seating failure of hollow-core occurred (Norton et
al, 1994) and photographs showing the different failure modes are shown in Figure 2-1. The
relative rotation between the hollow-core floor slab units and the supporting beams was not
examined and therefore, failure due only to the induced tension is unrealistic.
construction prior to 2001. The connection detail consisted of hollow-core units seated on a
mortar bed on a 20mm or 40mm concrete seating. This mortar bed did not allow the floor
units to slide horizontally relative to the supporting beam as originally assumed. The cyclic
relative rotations between the floor and supporting beams, due to lateral loading of the super-
Page 2-4
assemblage, resulted in diagonal compression stresses from the seating to the topping slab and
orthogonal tensile stresses within the unreinforced webs of the hollow-core units. Due to the
cyclic loading, these tensile stresses led to cracking and ultimately, the loss of seating failure
of the hollow-core units (Figure 2-1(c)). The experimental results showed that the relative
rotation between the flooring and supporting beam was the major contributing factor to
failure.
(a) Meadows Apartment Building, Northridge (b) Herlihy and Park (2000) failure mode.
Earthquake 1994 (Norton et al, 1994).
(c) Matthews (2004) failure mode. (d) Liew (2004) Specimen 2 failure mode
Figure 2-1 Comparison of the failure modes observed during previous experiments.
One of the actions taken following the inadequate behaviour of the hollow-core floor system
observed during the Matthews (2004) experiment was the formation of a Technical Advisory
Group on precast floors (TAG). TAG sought to address the identified problems associated
with hollow-core floor systems and proposed new hollow-core support connection solutions
Page 2-5
Bull and Matthews (2003) conducted four proof-of-concept sub-assemblage tests that were
commissioned by Precast NZ. The testing regime introduced the relative rotation between the
floor slab and support beam as the chief failure mode by way of a cyclic vertical load at the
end of the 6m long hollow-core units. A diagram of the test set-up is shown in Figure 2-2(d)
and the specific details for the test specimens are included in Figure 2-2.
(a) Control Specimen (200 and 300 series) (b) Flexible seating detail
Figure 2-2 Seating connection details and the test configuration used by
Bull and Matthews (2003)
Two of the specimens comprised of traditional connection details, one with a 300 series
hollow-core unit and the other with a 200 series hollow-core unit, denoted as the control
specimens. The traditional connection details (for the 200 and 300 series hollow-core units)
comprised of the floor unit seated on a mortar bed, core end plugs to prevent concrete from
entering the cores, and conventional continuity starter bar reinforcement in the topping slab.
The floor-to-beam connection for the 300 series hollow-core unit was identical to the detail
used in the Matthews full-scale super-assemblage experiment (Matthews, 2004). The other
two specimens were newer connection details, recommended by TAG, and one of which
effectively created a pinned connection with a compressible backing board and a low
Page 2-6
friction bearing strip, and the other was a rigid paperclip connection, tying the beam and
floor slab together with reinforced cores. The traditional methods behaved poorly with web
cracking observed at low inter-storey drifts and full delamination of the hollow-core units
from the topping slab. There was also significant displacement of the floor units down the
face of the supporting beams at higher drifts. The damage patterns were similar between the
200 series and 300 series hollow-core specimens as well as what was observed in the full-
scale three-dimensional Matthews (2004) tests (Figure 2-1(c)). The two newer methods
behaved well and were able to sustain high drifts without a loss of seating occurring.
As a result of the previous research and the widespread use of precast floors in
construction practice, specific details governing the use of precast concrete systems have been
implemented within New Zealand design standards. Amendment No. 3 to NZS 3101:1995
(2004) provides two acceptable solutions for connecting hollow-core floors into supporting
beams, as indicated by Figure 2-3(a) and (b). The first solution is a pinned, simple type of
connection, which features the hollow-core unit seated on 75mm seat with a low friction-
bearing strip, and has a compressible backing board between the end face of the hollow-core
unit and supporting beam. This detail was experimentally tested and validated, firstly by Bull
and Matthews (2004) and then in a full-scale super-assemblage experiment (Lindsay, 2004).
Significantly improved behaviour was observed (as compared to Matthews (2004)), and the
beam-floor connection was able to withstand super-assemblage inter-storey drifts to +5%. The
second of these acceptable solutions has currently not been tested experimentally. This
type of connection with reinforcing tying the hollow-core units into the supporting beam. The
rigid reinforced detail requires more effort to construct (as compared with the simple
connection detail tested by Lindsay (2004)) with the need to pre-cut cores and place extra
reinforcement.
Page 2-7
(a) Hollow-core with compressible backing on low friction bearing strips.
Page 2-8
However, the rigid connection offers redundancy by being tied into the supporting beams, and
because of the chance for hollow-core units to arrive on site short (due to shrinkage, creep or
construction tolerances), this type of connection construction can be rendered more suitable.
Liew (2004) continued the research from Bull and Matthews (2003) and investigated
the relative rotation characteristics of hollow-core seated connections. Two new connection
details were investigated as well as a retrofit solution, which are all shown in Figure 2-4. The
first detail included a 70mm seating, and each core (4 cores) of the hollow-core unit was
reinforced with Grade 500 R16 reinforcement and Grade 500 D12 starter bars (where the
R and D denote plain round and deformed reinforcement respectively). The second
specimen was identical to the first specimen, except that there was no seating. This negative
seating case was to investigate whether units that arrive short to site could be propped and
connected to the supporting beam with a strongly reinforced bridging connection. The third
specimen used the same traditional detail previously tested by Matthews (2004) and Bull and
Matthews (2003) but featured an angle section catcher to act as a retrofit measure.
The findings from Liews (2004) experiments highlighted that over-reinforced beam-
a drift of 1%, and despite suffering significant damage, Specimen 1 was able to function
without collapse up to 4% drifts. The detail however performed poorly in comparison to the
paperclip detail tested under the Bull and Matthews (2004) research program (Figure 2-2(c)),
which utilised two D12 (Grade 300) paperclips instead of the four R16 (Grade 500)
paperclips. The zero seating specimen performed inadequately and failed under the expected
2% drift design demand. Although the moment capacity of the connection was suitable, the
lack of shear resistance from the zero bearing proved to be totally inadequate. The retrofit
detail tested did not perform as well as the author would have hoped, with the angle bracket
effectively catching the unit, but at the same time restraining the rotation of the floor unit.
Page 2-9
Hurricane "Ductile Mesh"
HD12 starter bars
612 100
D24
43
R60 Plastic end plug
4 R16 Paperclips
65
H/C unit seated on mortar pad
70
4 R16 Paperclips
65
P la stic e nd p lug
Liew (2004) concluded that it would be interesting to repeat the experiments including the
effects of beam elongation. In the early sub-assemblage research performed, only horizontal
pull-off loads were used. The sub-assemblage experiments by Bull and Matthews (2003) and
Liew (2004) included cyclic vertical loading inducing relative rotations between the hollow-
core floor unit and the supporting beams. To this point, no sub-assemblage experimental
Page 2-10
program has investigated the effect of both horizontal tension in the floor and relative rotation
3101:1995 (2004), as shown in Figure 2-3(b), currently has not been experimentally tested.
For this reason, this connection detail between the supporting beam and hollow-core is the
The two test specimens are described in the following subsection and can be seen in Figure
2-5(a) and (b). Full design and construction details of the specimens are provided in
Appendix A.
2.3.1 Specimen 1
The seated connection detail used for this experiment is the second detail proposed by
Amendment No. 3 to NZS 3101:1995. The 6m long 300 series hollow-core unit was seated on
a 75mm wide seat and a low friction bearing strip. The connection features two of the four
hollow cores reinforced and filled with concrete. Starter bars consisted of D12 (Grade 300)
reinforcement spaced at 300mm centres. Two cores of the four hollow cores were reinforced
with hooked R16 (Grade 300) bars placed close to the bottom of the cores. The topping slab
consisted of D12 (Grade 300) reinforcement, which was lapped with the starter bars, and
placed at 300mm centres in both directions. It is also worth noting that backing boards were
used in place of traditionally used end plugs, to prevent concrete entering the un-filled cores.
This is to aid the beam-to-slab interface becoming the critical zone, concentrating the
rotational demand at this interface and not into the hollow-core unit.
Page 2-11
D12 Starter Bars (Grade 300)
2.3.2 Specimen 2
Specimen 2 was based principally on Specimen 1. However the Grade 300E reinforcement
was replaced with Grade 500E deformed reinforcement to represent present construction
practice in New Zealand. The hollow-core unit is similarly seated on a 75mm seat and a low
friction bearing strip, and features two reinforced cores. Starter bar reinforcement consisted of
D12 bars placed at 300mm centres above each of the cores. The R16 reinforcement used for
Specimen 1 in two of the four cores has been replaced by 2-D12 bars for Specimen 2. The
reason for this was to provide equal top and bottom reinforcement (4-D12) and hence equal
Page 2-12
positive and negative connection moment capacities. Transverse reinforcement was used
within the reinforced cores, with the placement of one leg of R10 reinforcement at 150mm
spacings, tied between the starter bar and bottom reinforcement. This was to investigate
to overcome situations when overly high shear demands are present, which might be
experienced with long spans, bridges or other cases. The reinforcement passing the beam-to-
slab interface (4-D12 reinforcement top and bottom) has been de-bonded over 150mm with
plastic tubing to reduce the risk of reinforcement fracturing and to support the critical section
occurring at the beam-slab interface. In addition to this, a thin plastic strip was placed on the
top of the starter bars beneath the concrete surface to both act as a crack initiator and further
The inelastic rotations experienced by plastic hinges of concrete frame structures under
seismic loading cause the phenomenon of beam elongation, whereby the beam grows in
length. Beam elongation has been quantified through many research programs over the last
decade such as Fenwick et al (1999). Typical values for beam elongation vary from between
2% and 5% of the beam depth, and these estimated values have been used to explain
permanent deformations of frame structures as well as a way of assessing precast floor seating
widths.
Matthews (2004) and Matthews et al (2004) presented a new theory for predicting
elongation for structural concrete members. The theorem is based on simple rigid body
kinematics and utilises a rainflow counting method to calculate the amount of elongation due
to inelastic bending of a plastic hinge in a frame system. Elongation occurs only if plastic
Page 2-13
deformations exceed those of previous peaks. In its essence, the theorem can be given by the
expression:
Gi T u ecr (2.1)
in which Gi = elongation of the ith hinge; T = rotation of the hinge; and ecr = eccentricity
between the centre of gravity of the concrete (c.g.c) for the beam and the centroid of the
The elongation of a frame (Gel) with ith plastic hinges can be separated into the elastic
T e
n
G el G eel G pel
p T p T y cr (2.2)
i 1
the frame.
The growth of the beam lengths occurs as tensile strains within the reinforcement are
higher than the accompanying compression strains within the concrete. Under reversing
loading, compression yielding does not fully close the flexural cracks caused by the
reinforcement which has previously yielded in tension. Figure 2-6(a) illustrates beam
elongation by showing the relationship between the hysteretic behaviour and recoverable
the loading, with respect to the plastic rotations and the yield drift and ecr values are a
function of the structural and geometrical properties of the structure. The ecr eccentricity value
for example is a function of the reinforcement and the presence of topping reinforcement and
prestressing strands of adjoining floor slabs. Figure 2-6(b) and (c) show the derivation of the
ecr values for a typical concrete frame. For a more in-depth explanation of the development of
Page 2-14
2.5.1 Sub-Assemblage Elongation
With the beam elongation prediction model proposed by Matthews et al (2004) and further
verified by Lindsay (2004), there became a method to introduce beam elongation effects into
tension forces in the floor through a displacement controlled loading, which was modelled on
the beam elongation prediction model. The drift displacement history was controlled by the
vertical actuator in a prescribed manner, and as the theorem predicts beam elongation in terms
of rotation (of a plastic hinge or frame) under lateral loading, the use of Equation 2.2 allowed
beam elongation results to be calculated for each inter-storey drift increment. After
accounting for geometry (as the horizontal actuator rotates), an input file was developed
listing the target ram displacements for both the vertical and horizontal rams at each
increment of drift.
frame used during the Matthews (2004) project as well as the full-scale tests associated with
this current research. Beam elongation was calculated as one half of the total frame
elongation, or the pull-off displacements which one end support of the hollow-core flooring
units would experience. The amount of beam elongation calculated was based on an exterior
moment resisting frame and it is accepted that there would be a difference between this value
and that experienced by a floor unit that is seated away from the frame due to out-of-plane
bending and torsion of the transverse supporting beams. Also, a yield drift rotation, T y , of
0.5% was assumed and used for the generation of the beam elongation data for these
experiments. It is later verified that although this was a priori assumption, it was a valid
approximation.
Page 2-15
Load
3 5
Deflection
'el T+p
Unloading
Tp- 5
recovery
4
Unloading
recovery
3
Elastic Elastic
growth growth
2 1
T y Ty DriftT
(a) Detailed diagram of beam elongation showing the hysteretic relationship and recoverable
elastic and non-recoverable plastic components.
T
Elongation due to
positive moment
Instantaneous centre
of rotation (I.C.R)
(b) Exterior plastic hinge lever arms for reinforced concrete beams
I.C.R
Elongation due to T
negative moment
Prestressing
strands
c.g.s
ecr |0.275D D
jD
ecr |0.225D c.g.c
Elongation due to
positive moment
Instantaneous centre
of rotation (I.C.R)
Page 2-16
2.5 EXPERIMENTAL SET-UP
The construction of the test specimens, albeit sub-assembly components of a real structure,
was conducted in accordance with customary site practice. Figures 2.7(a) and (b) shows a few
stages of the construction process. Initially, the support beams were cast to half height using
typical reinforcement for a concrete frame beams. The 6m long 300 series hollow-core units
were then lowered into place on top of a low friction bearing strip. The top flanges of two of
the hollow-core voids were pre-cut 75mm wide and approximately 800mm long to allow for
the placement of the connection reinforcement. The connection and topping reinforcement
was put in place together with cast in beams to tie down the support beam, and the top half
Tensile tests were carried out on samples of the reinforcement used in the two tests.
For Specimen 1, the average yield strengths were 311 MPa for the D12 reinforcement and 325
MPa for the R16 reinforcement. For Specimen 2, the average yield strength for the D12
(Grade 500) reinforcement was 570 MPa. Standard structural ready mix concrete was used
with a specified 30 MPa 28-day strength, 19mm maximum aggregate size and a 100mm
slump. Test day compression cylinder test strengths were 41 MPa and 35 MPa for the topping
concrete of Specimens 1 and 2 respectively, and 45 MPa for the supporting beams. Further
The configuration of instruments used in the tests is illustrated in Figure 2-7(c). Strain
gauges were attached to the starter bar and hairpin bar reinforcement to measure the level
and amount of propagation of inelastic behaviour into the topping slab and hollow-core unit.
Linear potentiometers were used to measure the relative movement of the supporting beam
and floor slab as well as the horizontal pull-off and vertical dropping of the hollow-core unit
relative to the supporting beam. Inclinometers were used to attain an accurate level of the drift
Page 2-17
imposed on the floor. Purpose-built data acquisition and control programs were employed to
track the experimental data and automatically control the two actuators loading the specimen.
A new loading rig and set-up was developed to test the two specimens. This loading
set-up, through the use of dual displacement controlled actuators, loads the specimen in such a
way that both the relative rotation between the floor and supporting beam and induced tension
in the floor are integrated. A schematic of the set-up is shown in Figure 2-7(d). In the same
way as the Bull and Matthews (2003), and Liew (2004) sub-assemblage tests were
undertaken, a vertical actuator was located at the end of the 6m long hollow-core unit. This is
able to provide negative and positive moments at the connection of the supporting beam with
the hollow-core unit. In addition, a 35kN concrete block is placed at mid-span to generate
extra shear force at the connection, which assures an equivalent shear force distribution to the
full-scale prototype structure. The development from the previously conducted sub-
assemblage tests (outlined in Section 2.2) is the coupling of the motions of the vertical and
horizontal actuators. The horizontal actuator induces a net tension force in the floor slab and
complete reversing cycles to inter-storey drift amplitudes of 0.5%, 1%, 2%, 3%, 4%,
and 5%. The inter-storey drift used throughout this section refers to the rotation (as a
percentage) of the hollow-core floor unit with respect to the supporting beam. Positive drift
denotes tension on the bottom of the hollow-core unit and negative drifts imply tensions
within the starter bars and topping concrete of the floor unit. It should also be noted that from
analysis of the test data, the target drifts and displacements were not always achieved due to
the rotation and sliding of the supporting beam relative to the ground. Throughout the results
and analysis presented in the following sections, this discrepancy has been accounted for.
Page 2-18
(a) Photograph showing the pre-cutting of the (b) Supporting beam reinforcing and
hollow core. connection detail (Specimen 1).
Inclinometers (spaced
along the floor unit)
Vertical
(drift controlling) Reaction Frame
Actuator
Reaction Frame
35kN Concrete
Weight
Page 2-19
2.6 EXPERIMENTAL RESULTS: SPECIMEN 1
Figure 2-8 presents photographs of damage observed for the testing of Specimen 1 and further
results are displayed in Appendix A. A crack opened at the floor unit-to-supporting beam
interface during the 0.5% cycle and this widened to around 5-6mm during the 1% cycle
peaks. During the 2% cycles, this continuity crack had opened to around 10mm on the
tension side and 5-6mm on the compression side, and following the second cycle to 2%, the
residual opening at this interface was 9-10mm, which signified the amount of beam
elongation pulling the floor unit off the seat of the supporting beam. To this point, the only
other signs of damage were hairline cracks that opened up at 2% within the topping concrete.
However, on the subsequent cycle these cracks closed and did not reopen throughout the
remaining testing. The 3% and 4% cycles were characterised by the continuing widening of
the beam-floor opening, and at the conclusion of the test after a second 4% cycle, the
The performance of this specimen is easier to describe by discussing what did not
occur during the test. The large crack opening at the beam-floor interface showed that the
rotational demand imposed on the floor unit was nil, and no damage of the floor unit was
observed. The bearing strip performed its intended purpose of allowing the unit to slide
horizontally relative to the supporting beam while staying in place on the concrete seat. The
improved performance of the bearing strip also meant that no spalling of the concrete seat of
Page 2-20
(a) Cracking of topping at 2%. (b) Beam-to-floor interface crack at the end of
testing.
Figure 2-8 Photographs of the testing of Specimen 1.
A collection of the important instrumental results is shown in Figure 2-9 for Specimen 1. The
overall hysteretic performance of the specimen is shown by the moment rotation relationship
in Figure 2-9(a). The hysteresis shows good energy dissipation of the beam-to-floor
4%. The maximum positive and negative moment capacities occurred at the first cycle to
0.5% respectively, and the following cycles to 4% showed a constant level of moment
strength. It is noticeable for the negative drift cycles, that there were reductions in strength
between the first and second cycle. The theoretical predictions show good agreement, and the
derivation of the nominal and overstrength theoretical values are outlined in Section 2.8. The
overstrength was determined from material testing for the reinforcement, and was taken to be
the ratio of the ultimate tensile stress to the yield tensile stress. Behaviour from the strain
gauged reinforcing is presented in Figures 2-9(b) through (e). The notation of the north core
and south core refers to the northern most and southern most reinforced cores, or the second
and fourth cores. Each of the four graphs for the starter and hairpin show similar trends.
Page 2-21
100
Moment
(kNm)
80
60
40
20
0
-5 -4 -3 -2 -1 0 1 2 3 4 Drift (%) 5
-20
-40
-60 Experiment
Nominal Capacity
-80 125% Overstrength Capacity
-100
1 2 3
1 & 4: The drift displacement peak (time)
2 & 5: denotes is a count of the half cycles.
4 5 6
3 & 6: Appendix A provides a list of the
Yield: displacement peaks for each count
Hairpin bar
45 45
40 40
35 35
30 30
strain
strain
25 25
20 20
15 15
10 10
5 5
0 0
0 10 20 30 40 0 10 20 30 40
-5 -5
Drift displacem ent peak (tim e) Drift displacem ent peak (tim e)
(b) North core starter bar strains. (c) South core starter bar strains.
50 45
45 40
40 35
35
30
30
strain
strain
25
25
20
20
15 15
10 10
5 5
0 0
-5 0 10 20 30 40 0 10 20 30 40
Drift displacem ent peak (tim e) -5
Drift displacem ent peak (tim e)
(d) North core hairpin bar strains. (e) South core hairpin bar strains.
Page 2-22
The strain measured by the strain gauges at the critical section at the beam-floor interface
showed distinctly higher strains. This matches the experimental observations whereby the
damage and cracking was concentrated at the beam-to-floor interface, and there was minimal
amounts of cracking into the topping slab over the hollow-core unit. This indicates that the
design objective of concentrating the rotational demand at the critical section was achieved.
For the starter bar reinforcement, yielding occurred early either at 0.5% or during the first
cycle to -2% at the floor to beam interface. Whereas yield penetration reached the topping
strain gauges 150mm away during the subsequent 2% cycles. Yield penetration however did
not progress back into the beam until the final stages of the test. The same trend is shown for
the hairpin reinforcement, where the plastic strains were highest at the critical section.
A collection of photographs from the testing of Specimen 2 are presented in Figure 2-10 and
extended results are shown in Appendix A. Very little damage occurred during the 0.5%
cycles where there was only cracking at the sides at the end of the floor unit of the beam-slab
interface and minor cracking within the topping slab. At 1% the continuity crack along the
interface between the floor slab and supporting beam had opened to around 4mm, at +2% it
had closed to be a hairline crack and at 2% it was 8-10mm wide. During the second cycle to
+2%, spalling of the seat concrete of the supporting beam was observed which progressively
worsened throughout the remainder of testing. At the second cycle to 3%, the continuity
crack had widened to be 20mm at both the top and bottom of the hollow-core unit (Figure
2-10(b)). On the first cycle to +4%, a loud noise and accompanying drop off in load-carrying
capacity was noticed at +3.8% suggesting reinforcement fracture of one of the bottom D12
bars. On the subsequent second cycle to 4% there were similar fractures at +2.84% and then
at 0.9%, -2.2% and -3.3%. Following the 4% cycles, the spalling had extended over the
Page 2-23
whole width of bearing of the floor unit. The bearing strip had become clearly visible as can
be seen in Figure 2-10(c), and the seat width had been reduced to approximately 20mm. There
was also evidence of the floor unit dropping down the face of the supporting beam. Prior to
the 4% cycles the damage was moderate and repairable. However, following the first cycle
to +4% the damage could be categorised as major and a similarly damaged building could
only be entered under extreme caution. At +5% the bearing strip had split, was sliding
horizontally off the seat with the hollow-core unit, and the unit had dropped by around 10mm
down the face of the supporting beam. The beam-floor crack had opened to widths over
35mm. At +4.8% on the second cycle to +5% the north starter bar fractured and the seated
(a) Topping and beam-floor interface cracking (b) Beam-floor interface crack at
at 2%. +3% drift.
(c) Extent of seat concrete spalling. (d) Failure after a loss of seating (at 4.78% on
the second cycle to 5%.
Figure 2-10 Photographs of the testing of Specimen 2
Page 2-24
Figure 2-11 displays the key results derived from the instrumentation from the testing of
20kN concrete weight was added at the midspan for upward loading (unloading from a
negative drift peak), and both midspan weights taken off for downward loading (unloading
from a positive drift peak to a negative drift peak). The increased bending moment gradient
heightened the shear force at the support imparting higher shear demands on the beam-floor
connection. In this case the compatibility with the full-scale testing was forsaken in order to
The instances of the reinforcement fracture are shown on Figure 2-10(a) by the sudden
decrease of load carrying capacity. The overall hysteresis shows good performance up to the
first cycle to +4%, but after the first reinforcement fracture (at 3.83% on the first cycle to
+4%), there was a significant reduction in stiffness and load carrying capacity of the floor-to-
beam connection. There was a distinct reduction of peak moments between the first and
second cycles to +4%, largely due to reinforcement fracture at +2.8%, and on the second
cycle to 4%, a series of three reinforcing fractures reduced the load carrying capacity
appreciably. The capacity of the floor-to-beam connection during the 5% cycles was
minimal as a result of the prior reinforcement fractures, and overall failure occurred when the
final starter bar fractured, under a positive drift, and the floor unit lost its seating at +4.78%
+4%, and the maximum negative moment capacity occurred at 0.5%. This shows that for the
positive drift cycles there was a good degree of strain hardening of the bottom reinforcement,
which would have occurred prior to the reinforcement fracture. The theoretical predictions
overlain on the moment rotation relationship show good agreement with the experimental
results.
Page 2-25
100
Moment
(kNm)
80
60
First
40
Fracture
20
Drift (%)
0
-6 -5 -4 -3 -2 -1 -20 0 1 2 3 4 5 6
-40
Fracture
Fracture
-60 Experiment
Nominal Capacity
-80 125% Overstrength Capacity
-100
1 2 3
1 & 4: The drift displacement peak (time)
6
2 & 5: denotes is a count of the half cycles.
4 5
3 & 6:
Appendix A provides a list of the
Yield:
displacement peaks for each count
Hairpin bar
45 45
40 40
35 35
30
30
25
strain
25
strain
20
20
15
15
10
10
5
0 5
-5 0 10 20 30 40 0
0 10 20 30 40
-10 -5
Drift displacem ent peak (tim e) Drift displacem ent peak (tim e)
(b) North core starter bar strains. (c) South core starter bar strains.
40 45
35 40
30 35
30
25
25
strain
strain
20
20
15
15
10
10
5
5
0 0
0 10 20 30 40 0 10 20 30 40
-5 -5
Drift displacem ent peak (tim e) Drift displacem ent peak (tim e)
(d) North core hairpin bar strains. (e) South core hairpin bar strains.
Page 2-26
A comparison between the moment rotation behaviour for Specimen 1 and Specimen 2 shows
the relative strengths of the two reinforced connection details. For Specimen 2, the higher
positive moment capacity of the floor-to-beam connection and extra mid-span loads under
upward loading would have imparted larger compression reaction forces on the seat, and the
Figures 2-11(b) and (c) present the strain behaviour of the starter bars, and similarly
2-11(d) and (e) the hairpin reinforcement for the floor-to-beam connection. The behaviour is
similar to that observed for Specimen 1, where the starter and hairpin reinforcement show
similar trends. Pre-yield compression strains were experienced during the early stages of the
test, however, following tensile yielding, the strains never returned below this yield strain
value and the reinforcement hysteretic behaviour would have been characterised by a distinct
shift into positive tensile strains. Yielding occurred early for the starter bars during the first
cycle to -1%, whereas yield penetration reached the topping concrete strain gauges (150mm
from the end of the beam-floor interface) during the 2% cycles and 150mm back into the
beam later during the 3% cycles. Plastic strains either side of the critical section of the
To investigate the effects of vertical acceleration during seismic excitation, the testing
procedure could have been modified. Vertical accelerations have important implications, and
are accommodated for in design standards. Upward accelerations reduce the gravity loads on
the flooring, causing a reduction of the bending moment gradient, and extending the demand
on the topping reinforcement. The use of the additional concrete weight could have been used
oppositely (extra midspan loads for downward loading, and no loads for upward loading).
However, the strain relationship exhibited by the starter bar reinforcement shows that
maximum demand is placed on the reinforcement at the interface, and the strain reduces
further into the topping. The extent of which the strain diminishes away from the supporting
Page 2-27
beam over a greater length was not fully investigated, but it is assumed that either the
continuous starter bars, or starter bars lapped with the same type of diaphragm reinforcement,
To assess the strength of the reinforced floor-to supporting beam connections for both
specimens, simplified moment equations have been used in order to calculate the positive and
negative moment capacities. The structural actions imposed on the connection are shown in
Figure 2-3(c) and the moment capacities were calculated from the common reinforced
concrete relationship:
Mn = Asfy(d-d) (2.3)
in which As = the cross sectional area of tension reinforcement, fy = yield stress of tension
reinforcement, d = effective depth to the centre of the tensile reinforcement taken from the
outermost compression fibre, and d = depth from the outermost compression fibre to the
Actual values for the yield stress and reinforcment and the measured lever arm
between the top and bottom reinforcement were used in the calculations. The overstrength
moment capacity of the two reinforced floor-to-beam connections was also calculated as
125% of the nominal (yield strength measured from material tests) strength. Table 2-1
The moment capacities have been plotted overlaying the experimental moment
rotation behaviour and can be seen in Figures 2-9(a) and 2-11(a). It can be seen that there is
good agreement with the experimental data for both of the test specimens. The moment
rotation relationship for Specimen 1 is illustrated in Figure 2-9(a) where it can be seen that
Page 2-28
the maximum positive and negative moments were experienced during the first cycle to
0.5%. Despite this, the predictions for Specimen 1 showed that the experimental values lay
within the 125% overstrength and nominal strength bounds. A similar quality of prediction is
evident for Specimen 2, as shown in Figure 2-11(a). Although the first cycle to 0.5%
produced the highest negative moment, there was some degree of overstrength or strain
hardening that was predicted by the calculated values. From these results, it is evident that the
When assessing the ultimate deformation capacity of a structure, low cycle fatigue of
(1994). Further investigation by Dutta and Mander (2001) produced a low cycle fatigue
theory that enables the plastic rotation life capacity of reinforcement to be calculated. The
D L p 1 L p
Tp 0.16 1 2 N f
0.5
(2.4)
D' D 2 L
in which Tp = plastic drift; L = lever arm of the cantilever column; D = overall member depth;
D = distance between outer layers of steel, 2Nf = the number of reversals to the appearance of
the first fatigue crack with which Nf is the effective number of constant amplitude cycles for a
Page 2-29
variable amplitude displacement history and Lp = equivalent plastic hinge length, or the length
Miners well-known cycle counting method can be utilised to calculate the equivalent
2
T
Nf i T
(2.5)
ref
where T = drift for the ith cycle of loading and Tref = reference drift for an equivalent constant
amplitude loading history. A more complete description of this theory and the calculations are
present in Appendix A.
calculated. Using Equation 2.4 and an equivalent plastic hinge length of the un-bonded length
of reinforcment (150mm), a plastic drift Tp of 0.03 radians was found. Together with an
elastic yield rotation of 0.7% that was derived from the experimental observations, the
rotational capacity of the reinforcement before fracture is 0.037 radians, or 3.7% as calculated
by Equation 2.4. This compares well to the observed first fracture that occurred at +3.83%
drift.
The visual and instrumental observations show that both specimens behaved well. Specimen 1
was able to sustain drifts of 4% with no major damage to the supporting beam and hollow-
core floor slab. Specimen 2 performed well up to the low cycle fatigue fracture of the
reinforcement at +3.83% drift, but was still able to withstand further 4% and 5% cycles
until a loss of seating failure occurred at +4.78% on the second cycle to +5%. The relative
rotation damage was concentrated at the critical section at the interface between the
Page 2-30
supporting beam and floor interface for both specimens, and there was very little damage to
the hollow-core units and topping concrete away from this critical section. However, there
were still some shortcomings observed during the testing of Specimen 2. Large amounts of
seat spalling was observed, which minimised the bearing area for the floor unit. During the
final stages of testing, the amount of deterioration of the concrete seat meant that the hollow-
core floor unit was sliding down the face of the supporting beam. The low friction bearing
strip performed well for Specimen 1, but inadequately for Specimen 2, whereby the bearing
strip slid horizontally with the hollow-core floor unit relative to the supporting beam. The un-
bonding of the reinforcement passing the critical section for Specimen 2 also showed an
The maximum moments sustained by Specimen 2 were 70% greater than the
reinforced connections with Grade 300 and 500E reinforcement. The larger moment capacity
would have caused larger vertical reactions to be resisted by the seating concrete of the
supporting beam. This can help to explain why the bearing strip performed as intended for
Specimen 1, but not for Specimen 2, where the bearing strip slid horizontally with the hollow-
core unit relative to the seat of the supporting beam. In turn, the poor performance of the
bearing strip would have accelerated the spalling of supporting beam concrete seat. It is
difficult to infer at what point a floor-to-beam connection becomes over-reinforced, but the
relative damage between the two specimens indicates that the Grade 500E reinforced
connection may have been too strong. Also, the plain round R16 bars (used for Specimen 1)
appear to be more suitable in high ductility connections under high drift demands than the
Page 2-31
2.10 CONCLUSIONS
From the experimental investigation presented in this chapter, the following conclusions may
be drawn:
NZS 3101:1995: a rigid, reinforced floor-to-beam connection, behaved well, and was
2. A new testing procedure that introduced the realistic coupling of relative rotation of the
hollow-core and supporting beam and the net tension in the floor slab from beam
successfully.
3. Simple analytical methods were proposed to assess the strength capacity of the hollow-
core to supporting beam connections. Similarly low cycle fatigue theory was used to
predict the rotation limit leading to reinforcement fracture. Both analyses provided
4. Evidence of over-reinforcement was observed with the use of Grade 500E reinforcement
with increased amounts of damage being observed for Specimen 2. The presence of plain
round reinforcement was more capable of resisting high ductility demands than the
2.11 REFERENCES
Bull D.K, Matthews J.G, 2003, Proof of Concept Tests for Hollow-core Floor Unit
Dutta A and Mander J.B, 2001, Energy Based Methodology for Ductile Design of Concrete
the New Zealand Structural Engineering Society (SESOC), Vol 12, No. 1, pp 35-40.
Herlihy M.D, Park R and Bull D.K, 1995, Precast Concrete Floor Unit Support and
Herlihy M.D and Park R 2000, Precast concrete floor support and diaphragm action,
Lindsay R.A, 2004, Experiments on the seismic performance of hollow-core floor systems in
Lindsay R.A, Mander J.B, and Bull D.K, 2004a, Preliminary results from experiments on
Lindsay R.A, Mander J.B, and Bull D.K, 2004b, Experiments of the Seismic Performance of
Liew H.Y, 2004, Performance of Hollow-core Floor Seating Connection Details, Masters
New Zealand.
Mander J.B, Panthaki F.D, Kasalanati A, 1994, Low Cycle Fatigue Behaviour of Reinforcing
Steel, Journal of Materials in Civil Engineering, ASCE, Vol 6(4), November, pp 453-
468.
Page 2-33
Matthews J.G, 2004, Hollow-core floor slab performance following a severe earthquake, PhD
New Zealand.
Matthews J.G, Mander J.B, Bull D.K, 2004, Prediction of beam elongation in structural
concrete members using a rainflow method, 2004 NZSEE Conference, Rotorua, NZ,
Meija-McMaster J.C and Park R, 1994, Tests on Special Reinforcement for the End Support
of Hollow-core Slabs, PCI Journal, Vol 39, No. 5, pp. 90-105, September-October.
NZS3101, 1995, Concrete Structures Standard, NZS3101, Parts 1 & 2, Standards New
NZS3101, 2004, Amendment No. 3 to 1995 Standard (NZS3101), Standards New Zealand,
Norton J.A, King A.B, Bull D.K, Chapman H.E, McVerry G.H, Larkin T.J and Spring K.C,
Oliver S.J, 1998, The Performance of Concrete Topped Precast Hollow-core Flooring
Systems Reinforced with and without Dramix Steel Fibres under Simulated Seismic
Christchurch, Canterbury.
Zealand.
Page 2-34
3. Section 3 Super-Assemblage Experimental Investigation
SECTION SUMMARY
Recent earthquakes and laboratory research has demonstrated the seismic vulnerability of
buildings constructed with precast prestressed concrete hollow-core floor systems. However,
follow-up research has shown that with simple detailing enhancements, significant
improvement in the seismic performance of hollow-core floor systems can be expected. The
present experimental research aims at validating several new detailing enhancements. Based
following details: (i) a reinforced connection that rigidly ties the floor into the supporting
beam, (ii) an articulated topping slab cast onto a timber infill solution that bridges between
the hollow-core units and parallel longitudinal frame beams; (iii) a specially detailed plastic
hinge zone in the beams that support the floor, that concentrates damage away from the
hollow-core units; (iv) Grade 500E reinforcing steel used in the main frame elements; and (v)
mild steel deformed bars in the concrete topping in lieu of the customary welded wire mesh.
The full-scale structure is cyclically tested in both the longitudinal and transverse directions
to inter-storey drifts of 5%. Visual and instrumental observations from the experiment are
presented and discussed. The experimental observations demonstrate that with appropriate
detailing it is possible to avoid undesirable failure modes induced in the hollow-core flooring
and to shift all damage into the plastic hinge zones of the supporting moment resisting frame.
Recommendations for the forthcoming revision of the New Zealand Concrete Structures
Page 3-1
3.1 INTRODUCTION
Research over the last decade at the Universities of Auckland and Canterbury has sought to
determine the seismic adequacy of precast buildings with hollow-core flooring systems in
New Zealand and to provide better solutions to preserve life safety. One of the major concerns
identified from both overseas earthquake events and experimental investigations was the
connection of these precast hollow-core units to the surrounding lateral load resisting system.
Under earthquake loading that induces significant storey lateral displacements, the potential
for the hollow-core units to lose their seating, leading to a possible partial or full collapse of
the floor, focused research into the connection of the floor units with the surrounding
perimeter beam. The collapse of the hollow-core units during testing by Matthews (2004) and
Matthews et al (2003) identified serious flaws for existing precast concrete frame structures
with hollow-core flooring structural systems. A continuation of this research (Lindsay, 2004
and Lindsay et al, 2004) demonstrated that enhanced reinforcement details could be
implemented into new structures that should result in improved behaviour in a seismic event.
(Matthews, 2004, and Matthews et al 2003, 2004) and on new structures (Lindsay, 2004, and
Lindsay et al 2004).
The popularity and widespread use of precast concrete is recognised in New Zealand
design standards where there is specific reference to precast concrete flooring support
conditions. Amendment No. 3 to the current New Zealand Concrete Structures Standard NZS
3101:1995 provides two specific acceptable solution details for the connection of hollow-
core floor units to reinforced concrete frame supporting beams. The second solution specifies
a reinforced connection that rigidly ties the floor into the supporting beam, but to this point
herein presents an experimental investigation of the effectiveness of this solution and its
Page 3-2
adequacy for inclusion into the upcoming revised New Zealand Concrete Code and use in
Following a summary of the findings from previous research, this section initially
provides an overview into the experimental set-up. The construction of the test specimen and
the modifications made to the concrete moment resisting frame super-assemblage from the
previous test programs are discussed. The specific details of this series of experiments are
then provided including the hollow-core seating details, the lateral connection between the
hollow-core unit and the parallel perimeter beams as well as the diaphragm reinforcement.
The set-up of the experiment, including the material properties of the specimen, the
instrumentation and the displacement drift controlled loading regime used are then discussed.
The second part of this chapter provides the visual and instrumental observations of the
testing. The hysteretic behaviour and strengths in terms of load displacement relationships of
the system are shown. The results and overall effectiveness of the design changes and new
experimental program was initiated to examine the seismic adequacy of such systems in the
New Zealand building stock (Matthews, 2004; Matthews et al 2003, 2004). A two bay by one
bay concrete moment resisting frame was constructed to represent a one-floor corner segment
of a typical New Zealand 10-20 storey precast concrete frame building. General results from
that program showed that although the concrete frame performed relatively well, the semi-
rigidly supported hollow-core flooring system, behaved poorly and collapsed at a relatively
low level of seismic intensity. The main source of damage occurred due to the relative
rotation between the supporting beams and floor and the expected sliding action of the floor
Page 3-3
on the supports provided by the beams did not occur. A snapping action at the hollow-core-to-
support beam connection and in particular within the webs of the hollow-core led to the
eventual failure of the floor. Recommendations for future connection details were made to
ensure that any inelasticity and relative rotation was centred at the interface between the floor
and supporting beam thereby protecting the hollow-core units. The connection between the
floor and the frame spanning parallel to the hollow-core units also performed poorly as a
result of the displacement incompatibility between the flooring, which is designed to sag and
the perimeter beams, which deform in double curvature. The undesirable deformation
demands placed on the hollow-core caused internal web cracking, fracturing of the cold
drawn mesh that reinforces the diaphragm and ultimately, the failure of the lower half of the
first hollow-core unit next to the two bay perimeter beams. Analytical work provided a more
accurate means of determining the lateral strength capacity of the structure accounting for the
influence of the floor slab contributing to the strength of the frame. A theory was developed
to predict beam elongation (Matthews et al, 2004). A rainflow counting theory was
advanced and validated against the experimental results. This provided a suitable method to
In response to the poor behaviour of the hollow-core flooring during the Matthews
(2004) experiment, a Technical Advisory Group on precast floors (TAG) was formed and new
tests were carried out by Bull and Matthews (2003) under sponsorship from Precast NZ Inc.
One of the new beam to floor connection details, which featured a low-friction bearing strip
with a compressible backing board replacing the plastic bungs in the end cores of the units,
performed well and was consequently recommended for firstly full-scale testing, and
Page 3-4
Lindsay (2004) repaired and reconstructed Matthews (2004) full-scale super-
assemblage specimen and then retested it. Whereas the Matthews (2004) experiment was
looked towards developing solutions for improved performance for future precast concrete
construction practice. The seating detail included a low friction bearing strip and
compressible backing board, and was implemented along with a connection isolating the first
hollow-core unit from the perimeter beams by way of a timber infill solution. A more
adequate connection was used which tied the central column into the floor plate. These
withstood up to inter-storey drifts up to +5%. Despite the good performance, the bearing strip
was observed to slide on the seat (the ledge cast along the beams that support the precast
hollow-core floor), the compressible backing board did not compress and hollow-core units
seated on or near the highly deformed plastic hinge zones of the supporting beams showed
cracking across the corners of the hollow-core units. The concrete of the seat spalled
considerably, and although the hollow-core was relatively undamaged, the units were
beginning to slide off the seat and down the face of the supporting beam. The infill detail
behaved well and proved to be a suitable flexible link isolating the beams and floor. However
premature fracture of the diaphragm ductile mesh produced a longitudinal tear within the
topping concrete along the interface between the timber infill and the first hollow-core unit.
assemblage component level. Meija-McMaster and Park (1994), Herlihy et al (1995), Oliver
(1998) and Herlihy and Park (2000) tested various reinforced hollow-core to beam
connections. Tests were primarily featured pull-off tension loads, to determine the adequacy
of reinforcing to be able to carry the weight of the floor if a loss of seating occurs. Among the
results, the tests showed that plain round reinforcing could provide suitably ductile behaviour.
Page 3-5
By comparing the aforementioned sub-assemblage component research with the
failure modes witnessed during the Matthews (2004) experiment, it was realised that the
earlier work was fundamentally flawed in that the rotational component of deformation was
not included in the sub-assemblage experiments. Subsequently, testing by Bull and Matthews
(2003) and Liew (2004) included a cyclic vertical force into the testing set-up to create cyclic
relative rotations between the supporting beam and precast floor slab. The experiment
towards performance. By forming a strong, rigid connection between the floor slab and
supporting beam, the relative rotational demand concentrated the damage at the relatively
weaker regions in the hollow-core floor slab away from the supporting beam. The brittle
nature of hollow-core meant that failure occurred at low imposed drift demands. It is therefore
important that there is a consideration of the influence of reinforcing the connection between
The seated connection detail used for this experiment is the second acceptable solution
detail given in Amendment No. 3 to NZS 3101:1995. The connection features two of the four
cores within the hollow-core units reinforced and filled with concrete, and is
diagrammatically shown in Figure 3-1(c). Deformed 12mm Grade 300 (D12) reinforcing is
used at 300mm centres for the starter bars, which is lapped with the same steel as the
diaphragm reinforcement. In the two reinforced cores, R16 Grade 300 bars (the R denotes
plain round bars) were placed close to the bottom of the cores, with both ends hooked to
Page 3-6
Reaction Pinned steel
frame tie beams
Unit 4
750x750
columns 300 series Unit 3
Transverse hollowcore
beams units Unit 2
475x750
N Unit 1 transverse beams
W E
750 wide 400x750
timber infill Plan longitudinal beams
S
6100 6100 6100
3500
Hollowcore Units
Timber Infill
Transverse Beam
Hooked Bar
(in addition to the regular reinforcing) Longitudinal Beam
Supporting Beam
Low friction
bearing strip
Hollowcore Units
200mm*25mm Timber boards
Perimeter beam
Page 3-7
To prevent concrete entering the two non-filled cores of each hollow-core unit, a backing
board was used in place of the more conventional end plug. This was used to ensure that the
rotation of the floor units relative to the beam occurs at the critical section at the beam-to-
floor interface and that the relatively brittle hollow-core unit did not experience high
rotational demands. The backing board substitutes for the conventional use of end plugs,
which create a concrete stub part way into the cores, which, under relative rotations, causes
prying and splitting forces within the un-reinforced hollow-core webs. The hollow-core unit
was seated on a low friction bearing strip and the seat widths are 50mm and 75mm at the east
and west ends of the test structure respectively as shown in Figure 3-1(c) and (d). The code
Amendment prescribes a 75mm seating, even so, it was decided to investigate the effect of a
shorter seat width for cases when hollow-core units arrive on site shortened, due to shrinkage
The infill detail used in this experiment stems from the recommendations made by Lindsay
(2004) and is shown in Figure 3-1(e). A 75mm concrete topping is cast on a 750mm long by
25mm thick timber plank infill running between the first hollow-core unit and the longitudinal
beams. The starter bars from the perimeter beams extend 600mm into the topping above the
first hollow-core unit. The longer starter bars and use of conventional reinforcement instead
of cold drawn or ductile mesh was used as a means to increase the ductility capacity at the
infill-hollow-core interface and mitigate the risk of fracture from occurring, as observed
previously (Matthews, 2004; Matthews et al 2003, 2004; Lindsay, 2004; Lindsay et al, 2004).
For this experiment, Grade 300 D12 bars (the D designates bars with deformations) have
been used at 300mm centres in both directions. The use of D12 bars in the topping slab is
beneficial because they are the same as the starter bars used, and all lapping is between the
Page 3-8
same type of reinforcement. The current standards as proposed in the amendment to NZS
3101:1995, state that starter bars should extend to the larger of either 20% of the hollow-core
span, or the development length (ld) plus an additional 400mm, which in some cases can be a
considerable length. In this case, the starter bars and diaphragm bars are the same cross
sectional area and spacing, meaning effectively full-length starter bars with no curtailment.
The starter and lapped bars were strain gauged to investigate the optimal length that starter
bars should be. Also, two XD20 drag bars (where the X denotes Grade 500 bars), which
were successfully implemented into the testing by Lindsay (2004), were cast through the
topping over the flooring and the central column to tie back the column to the floor.
The full-scale super-assembly specimen, shown in Figure 3-1(a), was a two-bay by one-bay
structure designed as a lower storey corner section of a multi-storey precast concrete moment
resisting frame building. The pretensioned flooring system runs parallel to the longitudinal
beams (east-west), past the central column, and are seated on the transverse beams. The frame
dimensions were identical to the original Matthews (2004) supper-assemblage to maintain the
same loading set-up and to be able compare results. The columns were 750mm square in
section, spaced at 6.1m centres with an inter-storey height of 3.5m. The longitudinal beams
were 750mm deep by 400mm wide, while the transverse seat beams were 750mm deep by
475mm wide. The super-assemblage was constructed in a similar fashion as it would be done
on a construction site. A photograph of the frame during construction can be seen in Figure 3-
at the structural details of existing buildings, whereas the second experiment (Lindsay, 2004)
along with the present investigation are aimed at providing new solutions for future
construction. With this in mind, it was decided to use Grade 500E seismic reinforcement
Page 3-9
throughout the frame, which is currently the most commonly used grade of steel in New
Zealand construction practice. The concrete frame was designed according to NZS
3101:1995, and is in keeping with the original Matthews frame strength and hierarchy of
strength. Following weak beam-strong column capacity design principles, the beam plastic
hinges were designed to the same strength as the original cases, however, using Grade 500E
steel. The beams required 6-XD25 bars, the exterior columns required 12-XD25 bars and the
Figure 3-2 Photograph showing the precast concrete frame after it has been erected.
During previous testing, the hollow-core units seated on the potential plastic hinge zones of
the supporting beams suffered damage due to the high deformation occurring in these zones.
To avoid such deformation demand on the floor units, a hooked bar was placed within the
beam to force the hinge to occur close to the column face as shown in Figure 3-1(b). Another
negative feature revealed in the previous experiments was the spalling of the seat concrete. To
overcome this, transverse reinforcement was placed within the seat to tie the seat concrete
back into the beam, as shown in Figure 3-1(c). A second generation of low friction bearing
strip was used which featured longer teeth to both increase the friction resistance on the
underside of the strip, which is in contact with the ledge of the seat, and to negate the effects
Page 3-10
3.3.4 Material Properties
New Zealand manufactured Grade 500E seismic reinforcement was used throughout the
which used Grade 300 seismic reinforcement. Table 3-1 presents a summary of the tensile test
information while a full set of stress strain curves are presented in Appendix C.
Column & Beam Transverse Steel YR12 530 714 190 - 0.0005 0.0152
Support Connection R16 336 469 186 4.48 0.0031 0.0200
Starter Bars and Diaphragm D12 307 447 181 3.65 0.0029 0.0218
Drag Bars YD20 580 724 246 3.82 0.0012 0.0156
Ready mixed concrete with a specified target compression strength of 30MPa, a maximum
aggregate size of 19mm and a slump of 120mm was used to cast the precast specimen
components. The topping slab was a 30MPa pump mix with a maximum aggregate size of
13mm. Table 3-2 shows the 7-day, 28-day and test day cylinder compressive strengths
Page 3-11
3.4 EXPERIMENTAL SET-UP
The testing of the super-assemblage was completed in four Phases: (i) longitudinal loading;
(ii) transverse loading, (iii) re-loading in the longitudinal direction, and (iv) re-loading in the
transverse direction. For Phase 1 and 3 loading, the set-up shown in Figure 3-3(a) illustrates
the location of the secondary load frames and primary load frames, which were attached to the
south frame of the super-assemblage. In each bay there were two actuators in opposing
diagonal directions: one that is displacement controlled, the other force controlled. Actuators
attached to the bottom of the north columns were displacement controlled and used to
maintain the same inclinations as the south columns and thereby minimise the torsion in the
transverse beams. Figure 3-3(c) shows an elevation of the loading setup for the south centre
column. For the previous experiments by Matthews (2004) and Lindsay (2004) this was the
setup for the transverse loading only, but this setup remained for the entire duration of testing
for this experiment. These actuators, by applying equal and opposite forces were able to keep
the centre column at a suitable transverse (north south) inclination during the longitudinal
(Phase 1 and 3) loading again minimising the torsion of the longitudinal beams. During the
transverse Phase 2 loading, these actuators ensured that the central column was at the same
inclination as the corner columns. The configuration shown in Figure 3-3(b) shows the
transverse set-up that was attached to the east and west frames (as shown) to allow loading in
3.4.1 Instrumentation
Due to the size of this test specimen, an extensive array of instrumentation was required to
acquire all the experimental data. A custom designed program and data acquisition system
developed at the University of Canterbury were used to record and save the data from all of
the various types of instruments at each increment of load. Similarly, a control program was
Page 3-12
developed, in house, to actively control each of the actuators to load the structure in a
displacement-controlled manner. For a full account of the control system and algorithms
Diagrams showing the instrumentation can be seen in Figure 3-4. The load was
monitored by load cells that were attached to each of the eight hydraulic actuators used during
the experiment. Similarly, rotary potentiometers were able to monitor the actuator
displacements. Each leg of the secondary load frames was strain gauged in a full-bridge
configuration allowing the load to be monitored and the redistribution of load through the
system to be followed.
structure through inclinometers, linear potentiometers and displacement transducers. The drift
measure the inter-storey drift of the structure or inclination of the columns. Sonic
displacement transducers were used at the base of the columns to accurately measure the
column sliding movement. Linear potentiometers of different sizes were placed at numerous
positions of the building to measure any large displacements, such as the flexural, shear and
elongation deformations of the plastic hinges, movement of the hollow-core units and relative
movement of the timber infill. The floor diaphragm was also extensively instrumented by a
grid pattern of Demec points, where the distance measured between the points allow strain
patterns of the entire floor to be generated. Vertical levelling was undertaken at discrete points
on the floor to obtain a contour map of the vertical displacements of the diaphragm as well as
provided an absolute measure of the vertical displacement of the whole structure. Finally,
starter bar and hairpin reinforcement for the support connections, starter bars passing over
the infill and the bars tying the central column into the floor were also been strain gauged to
monitor the level of load or inelasticity that the reinforcement were subjected to.
Page 3-13
(a) Load frame and hydraulic actuator set-up for Phase I and 3 in-plane.
Secondary loading frame
Ac
tua
tor
D
E
ator
tu
Ac
Self equilibrating loading frame
(b) West end elevation of the load frame and hydraulic actuator set-up for Phase 2 out-of-plane
testing.
Hydraulic actuator (Ram H)
Reaction frame
Reaction frame
Linear bearings
(c) Side Elevation view of the central column load frame and loading set-up.
Page 3-14
Strain Gauge Location Strain Gauge Locations
300 600 600 600
Timber Infill
3
T 1 2 4 5
B 1 2
Hollowcore Units 50mm pots
30mm pots measuring the
horizontal pull-off and vertical Potentiometers measuring the
drop of the floor units relative relative vertical displacement
to the supporting beam between the floor diaphragm
Low Friction bearing strip and perimeter beams
on the 50 or 75mm seat
(a) Hollowcore support potentiometers and (b) Timber infill potentiometers (16 sets) and
strain gauge instrumentation. strain gauge instrumentation.
Side elevation
(d) Location of beam plastic hinge and beam column joint potentiometers.
Page 3-15
3.4.2 Loading Protocol
Experimental loading protocols at the University of Canterbury have commonly been based
on cyclic loading at increasing ductilities (Park, 1989). For the Matthews (2004) experimental
investigation into the seismic performance of existing structures a new loading pattern was
developed. The aim of the displacement loading regime was that it would be indicative of
what a similar structure would experience during a real earthquake event. The loading
protocol was based on studies into probable drift demands of a range of precast concrete
frame buildings subjected to time history analyses under a suite of different earthquake
records.
For the Lindsay (2004) experiment, the experiment was centred on investigating the
conservative assessment of the seismic capacity was sought and hence a higher displacement
demand was necessary. This experiment also investigates the performance of a new solution,
and is utilising the same loading protocol as what was used during the Lindsay (2004) tests
offering a direct comparison of the results. The experiment consists of four phases comprising
of: (i) longitudinal loading of two completely reversed cycles of 0.5%, 1% and 2%; (ii)
transverse loading of two completely reversed cycles of 0.5%, 1%, 2% and 3%; (iii)
of 3%, 4% and 5%; and (iv) transverse re-loading of one complete reversed cycle of 3%,
and then two completely reversed cycles of drift amplitudes of 4% and 5%. The Matthews
(2004) or Lindsay (2004) experiments did not conduct Phase 4. Therefore, for comparative
purposes, only the first three phases of results are reported herein.
Page 3-16
3.5 EXPERIMENTAL OBSERVATIONS
Phase 1 longitudinal loading included two cycles at increasing drift amplitudes to 2%. Some
photographs of the key damage are shown in Figure 3-5. Throughout the descriptions of the
test results, a positive inter-storey drift refers to when the tops of the columns move in an east
direction and the bottoms of the columns move west. From the early stages of testing,
diagonal torsional cracks appeared at the ends of the transverse beams and continued to
extend and widen throughout this phase of testing. Figure 3-5(b) shows early stages of torsion
with X type cracking. Cracks, propagating diagonally outwards from the longitudinal beams
across the infill slab first appeared at +0.25% and continued to extend into the topping
concrete of the first hollow-core unit. Figure 3-5(c) shows the tendency of the crack pattern to
arch towards the south central column. At +1%, a longitudinal crack was observed between
the first and second hollow-core unit, and after the 2% cycles this had extended to be around
6m long. Damage in the plastic hinge zones was largely restricted to one major crack at the
beam to column interface, and another significant crack around 300mm from the column.
Some instances of spalling of seat cover concrete were evident at +1% and after the 2%
cycles, the spalling had not extended but had worsened in a few areas as can be seen in Figure
3-5(d). A single continuity crack formed along the beam to floor interfaces, as shown in
Figure 3-5(a), hairline soffit cracks were first noticed underneath the a filled core at the east
end of Unit 1 and then a web crack was observed at +2% propagating at 45 degrees from the
seat to the topping. Beam elongation was illustrated both with the residual crack openings and
by the sliding of the floor units off the supporting beams (which could be seen by the
exposure of unpainted soffit of the hollow-core floor units). Results showed that the bearing
strips worked as expected. The residual drift after the 2% cycles was around 1.1% and the
Page 3-17
(a) Continuity crack forming at the east end at (b) Torsional cracking of the North West
+2%. Transverse Beam at 1%
(c) Cracking over the infill and floor of the (d) Removed spall concrete at east end and
west bay (looking towards south centre exposure of the bearing strip after Phase I
column) after Phase I testing. testing.
Figure 3-5 Damage to the test super-assemblage during and after Phase I longitudinal
loading.
Page 3-18
3.5.2 Phase 2: Transverse Loading
A selection of photographs showing the behaviour of the test specimen under Phase 2
transverse loading is shown in Figure 3-6 where the convention for postive drift is when the
tops of the columns move north and the bottoms of the columns move south. The early stages
of testing exhibited little new damage, and behaviour was characterised by the pre-existing
cracks (caused by Phase 1 loading) opening wider. In a similar fashion to the longitudinal
beams, the rotation experienced by the transverse beams was concentrated at or near the
column face rather than being distributed over a conventional plastic hinge zone length.
Figure 3-6(a) and (b) show the damage in these zones, where crack widths were
approximately 15mm at +3% drift. There was little new damage to the floor and topping slab
in general, although elongation of the transverse frames was clearly apparent with large
openings between the column and topping slab at the top of the beams under negative
moments. At +3% drift, the opening was around 15mm at the top of the transverse beams at
the north ends and the topping slab was pulling away from the south corner columns by
approximately 25mm at 3%. Vertical deformations of the floor also became apparent with
the floors dropping away from the column, which could also be seen from the shear
deformation along of the large cracks at the ends of the transverse beams. The large
deformations occurring at the northern end of the east transverse beam, due to shear and
flexure, resulted in a crack to propagate from the seat through the northernmost filled core of
Unit 4 and into the topping slab, as shown in Figure 3-6(c). This damage did not worsen
during the rest of the testing. After the completion of the Phase 2 transverse loading to 3%
the residual drifts were approximately 1.6% and the structure had suffered moderate damage
Page 3-19
(a) Damage to the south end of (b) Damage to the north end (c) Hollowcore web cracking
the west transverse beam at of the west transverse beam at and damage at the north east
+3%. +3%. corner after Phase 2 testing.
Figure 3-6 Notable damage from Phase 2 transverse testing.
Figure 3-7 shows some key photographs of the Phase 3 longitudinal re-loading of the test
specimen. During the initial stages of loading below 2% no additional significant damage
occurred. At +3% drift, several of the infill slab cracks extended over the hollow-core units
and small vertical displacements were first noticed (~1mm) at these cracks. A 1m long soffit
crack, 1-2mm wide, appeared at the west end of Unit 1 underneath one of the filled cores.
However no more damage to the hollow-core or beam seats of the supporting beams was
witnessed throughout the remaining testing. Upon loading towards 3% drifts, crushing of the
top concrete of the south centre column occurred at 2.4%, which can be seen in Figure 3-
7(d) and was matched by a small load drop off. It was also apparent to see the torsional nature
of the transverse beams. Whilst the south frame and columns were inclined the transverse
beams appeared to remain near vertical, acting to minimise the relative rotation imposed on
the hollow-core-to-beam seating connections. Cracking at the ends of the transverse beams
Page 3-20
showed between 2mm and 5mm of outwards movement and evidence of torsional hinging. It
was expected that the transverse beams would rotate rigidly with the corner columns with
slight inclination changes towards the midspan due to the eccentric loading of the flooring. At
4% drifts, significant amounts of concrete had become loose and fallen from the plastic
hinge zones at the ends of the south, east and west beams, and several of the reinforcing bars
became visible at the bottom of the southwest hinge and top of the southeast hinge of the
south beams. Although the structure was still in a stable condition, the damage in some areas
became irreparable and major components would need to be replaced for further structural
use. The phenomenon of beam elongation was again illustrated on several instances. Figure 3-
7(c) shows the gap opening of around 25mm wide at the southwest column, and Figure 3-7(e)
demonstrates by way of the unpainted floor, the amount of sliding of the floor units off the
supporting beams.
Prior to the first cycle to 4% drift, buckling of the compression bars at the bottom of
the southwest beam was observed, as illustrated in Figure 3-7(f) and at the top of the
southeast beam. On the accompanying cycles with an opposite bending moment, the bars did
not straighten completely in tension. During the final cycle of loading to 5% drift, at 1.14%
and then at 0.36% unloading from 5%, the inside and outside top reinforcing bars
respectively at the southeast beam hinge, fractured as shown in Figures 3-7(g) and (h).
Although the load-carrying capacity of the structural system was jeopardised, collapse
prevention was maintained; all but two reinforcing bars ensured that the frame remained
stable and the lack of damage to the hollow-core units and seating support mitigated any
Page 3-21
(a) Super-assemblage specimen at +5% inter- (b) Damage to southeast plastic hinge at +5%
storey drift. of the south frame.
Direction of
movement
(c) Top view: crack opening (d) Top view: (e) Evidence of elongation and the
(~25mm) at between the southwest concrete crushing at floor units sliding on the bearings.
column and floor topping slab. south centre (seen as the unpainted soffit, Unit 1
column. west end)
(f) Longitudinal beam bar buckling (g) Longitudinal (h) Longitudinal beam bar fracture
(southwest beam bottom, south beam bar fracture at 0.36% (southeast beam top
frame). at 1.14% inside bar, south frame).
(southeast beam top
outside bar)
Page 3-22
3.6 HYSTERETIC PERFORMANCE AND SYSTEM STRENGTHS
Figure 3-8 presents the lateral base shear force displacement (drift) hysteretic performance of
the super-assemblage for each of the three loading phases. Figure 3-8(a) shows the behaviour
for Phase I loading, where a maximum positive and negative base shear capacities of 1460kN
and 1380kN was observed. These maxima both occurred on the first cycle to 2%, although
it must be noted that there was only a small reduction in base shear capacity during the second
cycle.
The hysteresis loop for Phase 1 loading exhibits a small reduction of stiffness at low
inter-storey drifts. This could be explained by the opening and closing of cracks and the
general slop associated with the pins within the load frames and actuators. Phase 2
conversely shows no premature deterioration of strength or stiffness and that the longitudinal
Phase 1 had little effect on the structural behaviour. The maximum positive and maximum
negative base shear forces were 755kN and 801kN respectively which both occurred on the
first cycle to 3%. There was no loss is base shear capacity during the second 3% cycle. The
non-symmetrical layout of the super-assemblage, with the tie frame at the north end can help
to explain the differences between the positive and negative base shear capacity values. Under
negative drift cycles, the south beam starter bars that pass over the infill tying the beam into
the floor are activated under negative bending moments. A crack extending along this beam-
infill interface opened up during testing to negative drifts, and was 8mm wide near the
columns and 1mm wide at midspan of the south beams. Under positive drifts there are no
starter bars along the northern edge of the floor to be activated, and hence a slightly smaller
Page 3-23
2000
Base Shear
Force (kN)
1500
W E
1000
500
Inter-Storey Drift (%)
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
-500
-1000 W E
-1500
-2000
1000
Base Shear
Force (kN)
S N 800
600
400
200
Inter-Storey Drift (%)
0
-4 -3 -2 -1 0 1 2 3 4
-200
-400
S N
-600 Pre Torsion
-800 Post Torsion
-1000
2000
Base Shear
Force (kN)
W E 1500
1000
500
Inter-Storey Drift (%)
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-500
W E
-1000
Phase 3
-1500 Phase 1
-2000
Page 3-24
The hysteretic performance of the super-assemblage during Phase 3 longitudinal loading is
shown in Figure 3-8(c). The maximum positive base shear was 1583kN and the maximum
negative base shear force was recorded at 1536kN which both occurred during the first cycle
to 3%. Noticeable reductions in the base shear capacity were observed between the first and
capacity. Sudden drop-offs in load due to crushing of the topping concrete either side of the
south central column can be seen in Figure 3-8(c) firstly at 2.4% and at +3.5%. The
overlying hysteretic behaviour from Phase 1 shows that the damage from the transverse Phase
2 loading did not have a detrimental effect, and the maximum positive and negative base
shear forces were observed during Phase 3 testing. Subsequent longitudinal testing to high
drifts would, due to the longitudinal beam reinforcing bars fracturing (on the second cycle,
unloading from 5%), result in a reduction in the base shear capacity of the structure.
Perhaps the most significant feature of the present experiment was the extent of torsional
twisting evident, under longitudinal loading, of the two transverse floor-supporting beams.
Five inclinometers were installed on the western transverse beam to gauge the amount of
torsion experienced. Figure 3-9(a) presents the torsion, in terms of percentage drift as a
function of time for Phase 1 and Phase 3 loading, and the true extent of the relative torsional
twist that occurred in the plastic hinge zones at the ends of the transverse beams is illustrated
in Figure 3-9(b). It can be seen that during early stages of testing, the rotation of the beam
followed that of the corner columns - rigid rotation which would be expected. In the latter
stages of Phase 1 the beam rotation was similar to the columns for positive drifts but less for
negative drifts. This can be explained by the eccentric loading of the flooring units rotating
Page 3-25
the beam in a positive direction, and torsional moments generated by the moment resisting
degrades. During testing, cracking was observed on the transverse beam hinges at early stages
and at -1%, X type torsion cracks were evident. The torsion behaviour for Phase 3 loading,
seen in Figure 3-9(a) showed interesting results and that the transverse Phase 3 loading had
had a marked effect. The results illustrate that the beams remained essentially upright
throughout the test. There is an apparent tendency towards positive drifts accounting for the
eccentric floor support. For example, at the first cycle to +4% the beam was at an average
+1% inclination, and the first cycle to 4%, the beam remained at around +0.1%. All five of
the inclinometers show similar behaviour. This indicates that the torsion occurred at the ends
of the beams in the plastic hinge zones. Similar behaviour and damage was observed at the
eastern end of the super-assemblage. Appendix D shows the location of the five west beam
inclinometers.
The presence of torsional hinges had important effects on the behaviour of the
structure. The first issue is that the torsion could reduce the strength of the flexural hinges
placing increased demand on the reinforcement and concrete. This could possibly jeopardise
the integrity of the structure. However, no major detrimental behaviour was observed. The
positive implications of the torsional hinging was that the relative rotation between the
flooring units and supporting beam, which is the primary cause of damage to the supporting
beam-to-floor connection, was effectively small and can help explain the improved
performance of the seated connection and lack of damage as compared to the floor tested by
Matthews (2004).
The torsional behaviour is quite different to what was observed during the previous
Page 3-26
expected that in a similar way to the previous tests, the torsion would be distributed over the
length of the beam rather than at torsional hinges at the beam ends. The current experiment
1.8% of the earlier experiments) passing into the column, resulting in a reduced torsional
stiffness. However, the most likely cause of the torsional hinging is the effect of the
reinforcement connecting the supporting beam and hollow-core units. The rigid floor-to-beam
connection acted to keep the beam more upright, imposing an opposing torsional force as the
column and beam rotates. The floor-to-beam connection detail used in the Lindsay (2004)
experiment was in effect a pinned connection, and this lack of fixity of the connection detail
would mean that the additional torsion applied to the supporting beam would not be present.
6 Phase 3
5
Torsion Rotation (%)
4 Phase 1 loading
3 loading
2
1
0
-1
-2
-3
SW Column Inclination Inclinometer 1
-4 Inclinometer 2 Inclinometer 3
-5 Inclinometer 4 Inclinometer 5
-6
4 Phase 1 loading
3
2 loading
1
0
-1
-2
-3
-4
-5
-6
-7
-8
(b) Torsional twist of the south hinge of the west transverse beam
(radians/metre in plastic hinge zone)
Figure 3-9 Torsion rotational behaviour of the western transverse beam.
Page 3-27
Large rotation at the South-West
end of the beam Column
Eccentric weight of
floor units
(c) Damage and evidence of (d) Torsion of the east (e) Torsion of the east
torsional hinging (northwest transverse beam at +5%. transverse beam at 5%.
connection).
Figure 3-10 Torsional behaviour and damage of the transverse beams.
Page 3-28
3.7.2 Hollow-core-to-Supporting Beam Connection
Beam torsion played a significant role in the performance of the super-assemblage. While
torsion individually may be considered a sign of adverse behaviour, it had a beneficial effect
on mitigating the damage within the hollow-core-to-supporting beam connection. Figure 3-9
shows the difference between the rotation of the columns and the west transverse supporting
beam. The low rotations measured by the inclinometers indicated that the transverse beams
were remaining close to vertical, and the hollow-core floor slabs horizontal. Therefore, the
relative rotations between the floor slab and supporting beam was minimal.
The connection detail performed admirably and the super-assemblage structure was
able to sustain inter-storey drifts up to 5% with damage restricted to the support beam alone
and not the floor. At the conclusion of testing, there was minor diagonal web cracking at the
eastern end, both at the south and north sides of Units 1 and 4 respectively. A camera that
could be placed into four of the hollow cores showed that there was no observable internal
cracking of those particular cores. Soffit cracking was observed under several of the filled
reinforced cores at both ends of the structure. However these cracks appear to have had little
effect on the performance. The single crack that formed along the support beam-to-floor
interfaces on the topping concrete and lack of any other cracking in the adjacent floor
confirmed the objective to concentrate the relative rotation on this plane and restrict the
rotational demand on the floor units. The bearing strip was also adequate and allowed the
floor units to slide relative to the supporting beam whilst staying on the seat. At the 4%
peaks, around 16mm of pull off was recorded and could be seen underneath the structure,
while the bearing strip on the seat remained in place. The improved friction resistance and
better grip associated with the newer generation of bearing strip seems to have improved the
performance of the bearing strip. The minimal amount of seat spalling was also a positive
feature of the experiment. It is difficult to pinpoint the exact reasons why there was little
Page 3-29
spalling, but aside from the torsion of the supporting beams reducing the amount of rotation
experienced by the supporting beams, the bearing strips evened out the surface irregularities
of the seat concrete and the presence of transverse reinforcing tying the seat concrete back
The performance of the seated connection detail showed very similar performance
characteristics to the same connection detail tested under two dimensional sub-assemblage
conditions outlined in Section 2. The close similarity of observed performance between the
dimensional tests that are much easier to undertake in terms of time, cost and resources.
It must also be noted that there was no discernible difference in the behaviour between
the 50mm and 75mm seating ends of the structure. Minor spalling amounted for a loss of the
50mm seat, however this was confined to a small area of around 300mm wide. Although good
behaviour of both ends shows that the connection detail can be utilised over smaller than code
specified seat widths (75mm), short seat widths should not be specified. The experiment was
undertaken in a controlled situation where the seating details were constructed as designed
and the seat widths were measured. Specification of a minimum of 75mm seat accounts for
the likely maximum drifts while maintaining a minimum bearing width for the construction
tolerances and shrinkage and creep of the hollow-core units that occur, and specification
should adhere to minimum seat widths of 75mm or more. Furthermore, the laboratory
investigation was not a real time dynamic test such that neither vertical accelerations nor
higher mode effects were applied. This further underlines the need to maintain 75mm bearing
widths.
To overcome some of the deficiencies observed during the Matthews (2004) and Lindsay
Page 3-30
supporting beams, to promote column face hinging (see Figure 3-1), and seat transverse
reinforcing were implemented. The minor damage that occurred within the hollow-core units
at the southern corners of the structure point to the added hooked bar performing as designed
restricting the plastic hinge to the column face. The hinging itself was confined to a small
area, approximately within half the beam depth from the column face for all of the plastic
hinges within the structure, so it is difficult to say that the hooked bar was solely responsible.
However, the unit seated on the northeastern corner plastic hinge did experience web
cracking, which can be seen in Figure 3-6(c), which further supports the contention that it is
preferable to seat hollow-core floor units away from areas of high deformation within
supporting beams, and isolation infill-type details should be used over plastic hinge zones.
The presence of the second generation bearing strip possibly contributed to a reduction of
stress concentrations along the seat, thereby reducing the spalling (as compared to Lindsay
(2004)).
The inclusion of the timber infill connection isolating the longitudinal beams from the floor
system performed moderately well during Lindsays (2004) experiment. However, that
experiment demonstrated the disadvantage of using relatively brittle welded wire mesh. In the
present experiment the provision for longer starter bars and use of conventional reinforcement
(instead of ductile mesh) proved to be successful. The crack pattern extended through the first
floor unit and across the second floor unit indicating that the longer starter bars were more
suitable and able to distribute the loads over a larger area. The use of conventional
reinforcement within the diaphragm also proved to be successful. Longer starter bars
throughout the topping concrete ensured that no major cracks appeared and therefore the
Page 3-31
3.7.5 Displacement Incompatibility
Figure 3-11 presents the displacement incompatibilities between the perimeter longitudinal
beams and the first hollow-core floor unit at peak drift amplitudes under Phase 1 and 3
loading. The displacements are plotted along the length of the structure from the west end to
the east end and were derived from levelling carried out on the floor. The results shown
indicate the amount of relative vertical displacements between the floor units (Unit 1) and
Figure 3-11(a) shows the vertical displacements for Phase 1 loading. It can be seen
that there was a relatively symmetrical relationship between the positive and negative drifts.
The maximum incompatibility was approximately -10.5mm at +2% drift on the west side of
the south centre column meaning that the floor had dropped relative to the beam. Figure
3-11(c) illustrates a similar trend but distinct shift in the negative direction indicates how
much the floor dropped following transverse loading. The damage incurred by the transverse
beam hinges was associated with a shearing displacement that lowered the beam and flooring
together. The effect of the transverse loading is also shown in Figure 3-11(b) where the 2%
cycle peak incompatibilities are compared for both loading phases. The crossover point that
occurred close to the centre of the structure (6100mm) shows the floors had dropped
approximately 8mm. The maximum incompatibility evident shows a 33mm drop of the floor
relative to the beam at +4% drift. This corresponds to a 4.4% slope over the infill in a north
south sense. Due to the lack of observed damage or tearing of the topping concrete, it can be
deduced that such slopes could be accommodated by the infill topping concrete and starter
bars. Further displacement incompatibility results from the experiment, and accompanying
Page 3-32
10
Position along fram e (m m )
5
0
Displacement (mm)
-10
-15
-20 Phase 1: 1%
Phase 1: -1%
-25 Phase 1: 2%
-30 Phase 1: -2%
-35
(a) Relative vertical displacement incompatibilities during Phase 1 loading (displacement of the
floor with respect to the longitudinal beams).
10
5
Pos ition along fram e (m m )
0
Displacement (mm)
-10
-15
-20
Phase 3: 2%
-25 Phase 3: -2%
Phase 1: 2%
-30
Phase 1: -2%
-35
(b) Comparison of the 2% cycle displacement incompatibilities between Phase 1 and Phase 3.
10
5
Position along fram e (m m )
0
Displacement (mm)
-15
Phase 3: 2%
-20 Phase 3: -2%
Phase 3: 3%
-25 Phase 3: -3%
Phase 3: 4%
-30
Phase 3: -4%
-35
(c) Relative vertical displacement incompatibilities during Phase 3 loading (displacement of the
floor with respect to the longitudinal beams).
Figure 3-11 Vertical displacement incompatibilities between the floor and longitudinal
beams during Phase 3 loading. (Note a negative displacement means that the floor has
dropped relative to the beams)
Page 3-33
3.8 CONCLUSIONS
From the experimental investigation that has been presented in this section, the following
NZS3101:1995 and included the Amendment No. 3 rigid connection hollow-core seating
longitudinal reinforcing bars in the beam fractured due to low cycle fatigue. The floor
system sustained only minor damage, which in no way impaired life safety. The onset of
occurred at +1% and irreparable damage to the plastic hinges zones of the south beams
drifts of 5%, where damage to the hollow-core units and seated support was minimal.
Previously, the details investigated herein were untested, and for that reason Amendment
No. 3 to the current New Zealand Structures Standard NZS 3101:1995 limited inter-
storey drifts of this class of construction to 1.2%. In light of the good performance, this
3. The articulated timber infill slab connecting the lateral load resisting system longitudinal
beams with the floor slab performed well with no major cracking and tearing
experienced. The use of longer starter bars and conventional reinforcement in place of
the cold drawn and ductile mesh ensured that any displacement incompatibility was
accommodated and diaphragm action was maintained. The performance also verifies the
Page 3-34
4. Three-dimensional effects of torsional hinging have arisen throughout the course of this
investigation. A rigid beam-to-floor connection acted to keep the floor horizontal and
beams upright forcing torsion of the transverse beams to occur over a limited zone in the
plastic hinges at the ends of the transverse beams under high drift rotations of the super-
assemblage. More research and analysis of torsion within precast concrete structures is
required.
5. Precast concrete moment resisting frames designed and constructed with Grade 500E
Plastic hinge zones appear to be less than half the depth of the beam long, and most
rotation is centred at large cracks at the beam-column interface. More research needs to
be carried out utilising various reinforcement ratios and investigating the buckling and
concrete frames.
3.9 REFERENCES
Bull D.K, Matthews J.G, 2003, Proof of Concept Tests for Hollow-core Floor Unit
Herlihy M.D, Park R and Bull D.K, 1995, Precast Concrete Floor Unit Support and
Herlihy M.D and Park R 2000, Precast concrete floor support and diaphragm action,
Page 3-35
Liew H.Y, 2004, Performance of Hollow-core Floor Seating Connection Details, Masters
New Zealand.
Lindsay R.A, 2004, Experiments on the seismic performance of hollow-core floor systems in
Lindsay R.A, Mander J.B, and Bull D.K, 2004a, Preliminary results from experiments on
Lindsay R.A, Mander J.B, and Bull D.K, 2004b, Experiments of the Seismic Performance of
Matthews J.G, 2004, Hollow-core floor slab performance following a severe earthquake, PhD
New Zealand.
Matthews J.G, Bull D.K, and Mander J.B, 2003a, Background to the Testing of a Precast
Matthews J.G, Bull D.K, and Mander J.B, 2003b, Preliminary Experimental Results and
Seismic Performance Implications of Precast Floor Systems with Detailing and Load
Matthews J.G, Mander J.B, Bull D.K, 2004, Prediction of beam elongation in structural
concrete members using a rainflow method, 2004 NZSEE Conference, Rotorua, NZ,
Page 3-36
Meija-McMaster J.C and Park R, 1994, Tests on Special Reinforcement for the End Support
of Hollow-core Slabs, PCI Journal, Vol 39, No. 5, pp. 90-105, September-October.
NZS3101, 1995, Concrete Structures Standard, NZS3101, Parts 1 & 2, Standards New
NZS3101, 2004, Amendment No. 3 to 1995 Standard (NZS3101), Standards New Zealand,
Oliver S.J, 1998, The Performance of Concrete Topped Precast Hollow-core Flooring
Systems Reinforced with and without Dramix Steel Fibres under Simulated Seismic
Christchurch, Canterbury.
laboratory testing, Bulletin of the New Zealand National Society for Earthquake
Page 3-37
4. Section 4 Forensic Analysis of Failure and Behaviour Modes
SECTION SUMMARY
Present experimental research has sought to verify new detailing requirements for the use of
hollow-core flooring in precast concrete structures. A full-scale two bay-by-one bay concrete
displacements in both the longitudinal and transverse directions. This chapter provides a
behaviour that is observed due to a strong hollow-core floor connection with a weak
potential of modelling for future research and construction are provided. The overall base
shear capacity of the structure is predicted and is shown to agree well with experimental
compressive forces are transferred into beams running parallel and adjacent to the prestressed
concrete floor units. Finally, the present research is compared and contrasted with previous
hollow-core flooring system is transformed into a robust frame-floor system where all the
Page 4-1
4.1 INTRODUCTION
Over the previous two decades, prestressed concrete hollow-core flooring systems have
become one of the preferred forms of precast construction in New Zealand. This research
described herein seeks to explain certain key experimental observations made in a lateral
Unit 4
750x750
columns 300 series Unit 3
Transverse hollowcore
beams units Unit 2
475x750
N Unit 1 transverse beams
W E
750 wide 400x750
timber infill Plan longitudinal beams
S
6100 6100 6100
3500
(b) View of the floor of the super-assemblage. (c) Load frame set-up.
Figure 4-1 Super-assemblage test specimen details.
Page 4-2
The experiment is the third super-assemblage examination on precast concrete structures. The
first of these was conducted by Matthews (2004) on existing buildings with pre-2001
construction details. Matthewss work discovered that existing hollow-core flooring details
performed unsatisfactorily and can be expected to behave poorly in a design basis earthquake
Amendment No. 3 was fast-tracked into the New Zealand Concrete Design Standard (NZS
3101, 2004). In Amendment No. 3, two acceptable solutions are given for detailing hollow-
core-to-supporting beam connections. The first of these was tested by Lindsay (2004). This
current research has investigated the viability of the second acceptable solution. Where past
research has focused on lateral load resisting systems overlooking the effect of floor
diaphragms, the interaction between the two became a principal focus for these investigations.
One significant finding in the current strong floor-to-weak support beam experiment
was the extent of the plastic torsional response of the supporting beams. This section seeks to
provide a forensic analysis of these experimental results. Several other failure mode aspects of
displacement incompatibility between the hollow-core flooring and the adjoining parallel
displacement incompatibility is summarised and the experimental results are compared with
results from computer modelling. New design solutions to combat the problems associated
with displacement incompatibility and reducing floor slab damage and the potential benefits
theoretical pushover analysis. Comparisons of the theoretical and experimental lateral force-
deformation capacity results are provided and reasons for the discrepancies are explained.
Page 4-3
Additional results obtained from the experiment are compared with existing theoretical
relationships where the beam elongation values are evaluated against the rainflow analysis
experiment is compared and contrasted with respect to the preceding two investigations
(Matthews, 2004 and Lindsay, 2004) as a means to complete the overall investigation into the
The behaviour of the transverse beams from observations of the experiment demonstrated
interesting torsional behaviour. Figure 4-2 presents the torsional behaviour of the west beam
observed during the experiment. Figure 4-2(c) shows the amount of torsional twist that
occurred within the south hinge of the west transverse beam. The rotations were measured
from inclinometers spaced evenly along the west transverse beam, which can be seen in
Appendix D. The results show that torsional hinges formed at the ends of the beams and that
the middle of the beam remained vertical and the flooring horizontal for most of the testing.
The comparison between Phase 1 and Phase 3 shows that there was a large increase in the
torsional twisting at the ends of the transverse beam following Phase 2 transverse loading.
During Phase 2, the ends of the transverse beams formed flexural plastic hinges, and as a
result of the damage caused due to the high flexure and shear actions, the torsional stiffness of
Phase 1 and 3 longitudinal loading and compares the observations with calculated torsional
demands and capacities. The torsional behaviour of the transverse beam was not measured or
analysed during Phase 2 transverse loading, but it is explained that the damage sustained by
the transverse beams during transverse loading has a significant impact on the torsional
Page 4-4
South-West Large rotation at the South-West
Column end of the beam Column
4
Phase 1 Loading Phase 3 Loading
3
2
1
0
-1
-2
-3
SW Column Inclination Inclinometer 1
-4 Inclinometer 2 Inclinometer 3
-5 Inclinometer 4 Inclinometer 5
-6
(b) West beam inclination during Phase 1 and 3 loading.
6
5
Angle of Twist (rad/m)
4
3 Phase 1 Loading Phase 3 Loading
2
1
0
-1
-2
-3
-4
-5
-6
-7
-8
(c) Torsional twist of the south hinge of the west transverse beam
(radians/metre in plastic hinge zone)
Figure 4-2 Observed torsional behaviour of the west transverse beam
Page 4-5
Figure 4-3(a) shows an idealised diagram of the hypothesised relationship between the
torsional strength (torsion moment) and torsional twist. In a similar way to the calculation of
the shear capacity of a reinforced concrete beam section, the torsional strength comprises of
strength contributions from the concrete, Tc, and the reinforcement, Ts. The ultimate torsional
moment strength, Tu, is a sum of the concrete and reinforcement contributions and is shown as
Torsion
Strength Monotonic torsion
capacity: Tu = Tc + Ts
Ultimate Torsion
Strength, Tu = T c + Ts
Tc
GKc r Cyclic torsion capacity:
Tu = Ts
Torsional Demand
Cracking Torsion = Tc
Torsion
Twist
Phase 3 response: soft post-
Phase 1 response: initial stiff pre-
cracking response due to damage
cracking response and small torsional
during Phase 2 transverse loading,
twist rotations
large torsional twist rotations
O = load path dependent parameter (O<1) that depends on the degree of previous loading and
flexure-shear-torsion interaction
(a) Schematic relationship between the torsion force and torsional twisting
South-West
Column
Torsional
Capacity of the
beams resisting
North-West the torsion at the
Column beam ends
Page 4-6
The derivation of the torsional capacity of reinforced concrete sections is complex, and it is
difficult when considering interaction with flexure and shear actions and pre- and post-
cracking behaviour. Significant work has been undertaken by Hsu (1993) and Mitchell and
Collins (1991), particularly with prestressed structures, and Park and Paulay (1975) for
reinforced concrete structures. These sources provide examples for the calculation of torsional
demands and capacities for concrete structural elements and several aspects of this work have
been reflected in the New Zealand Concrete Structures Standard, NZS 3101: 1995.
For compatibility torsion, which is the case in this experiment, a reduction of the
torsional moment can occur due to the redistribution of internal forces when cracking first
occurs. NZS 3101:1995 provides an equation for calculating the cracking torsional moment
for a section, and is the same as the formulation given by Mitchell and Collins (1991). The
f c' Acp2
Tc I (4.1)
3 p cp
where Tc = design torsional moment force at section at the ultimate limit state; or cracking
torsion; I = strength reduction factor; fc = concrete compressive strength; Acp = area enclosed
by the outside perimeter of the concrete cross-section; and pcp = outside perimeter of the
reinforcement ties and diagonal concrete compression struts form, and the contribution to the
torsional resistance from the reinforcement can be calculated. Park and Paulay (1975) provide
transverse and longitudinal reinforcement, which is shown as Equation 4.2. The equivalent
design expression for the nominal torsional strength from NZS 3101:1995 relates to the
Page 4-7
amount of transverse reinforcing and stems from the work presented by Park and Paulay
(1975).
At f y Al f ly
Ts 2 A0 (4.2)
s P0
where Ao = area enclosed by the connecting lines between the centres of the longitudinal
Prior to torsional cracking occurring or the torsional strength given by the concrete
being exceeded, the response is stiff and defined by the pre-cracking stiffness, GK. After
cracking occurs, the torsional stiffness is significantly reduced. As is shown in Figure 4-3, the
effect of cyclic loading has a large effect on the torsional stiffness whereby the concrete
torsional strength contribution diminishes, and the ultimate torsional strength becomes Ts, the
torsional capacity of the reinforcement. To calculate the torsional twists corresponding to the
cracking and ultimate torsion moments, relationships from Mitchell and Collins (1991) and
Park and Paulay (1975) for the torsional stiffness were used. For the initial stiffness, the
section is represented by an equivalent thin walled tube or hollow section with the same
overall dimensions and is a function of the section geometry. The post-cracking torsion
stiffness is governed by the deformation of the reinforcing. The pre-cracking stiffness was
calculated using:
4 A02 t
GK G (4.3)
po
where GK = the pre-cracking torsional stiffness; G = the shear modulus; A0 = area enclosed by
the centreline of the thin walled tube with thickness, t; and po = perimeter of the area. The
Page 4-8
3 Acp
t (4.4)
4 p cp
Where Acp = area enclosed by the outside perimeter of the concrete cross section, pcp =
To find the post-cracking stiffness, GKcr, a ratio of the post-cracking stiffness to pre-
cracking stiffness was calculated using methods included in Park and Paulay (1975). This
method uses the aspect ratio of the section, and a modified reinforcement ratio expression.
For the present experiment, the postulated relationship between torsional strength and
torsional twisting incorporates a distinct reduction of the torsional strength due to flexure,
shear and cyclic interaction. The reduced torsion capacity represents the damage that occurred
from Phase 2 transverse loading, whereby high shear forces and bending moments were
present and flexural plastic hinging occurred. The concrete contribution to the overall
torsional strength is rapidly diminished, and the torsion demands are sustained by the
the intersection of the torsional demand and idealised torsion-twist capacity values, an
estimation of the expected torsional twists can be found. The following sub-sections describe
the calculation of the torsional strength and stiffness parameters as well as the torsional
demands such that the relationship (Figure 4-3(a)) can be quantitatively examined.
The torsion due to eccentrically supported flooring systems on beams can be easily calculated.
However, the added torsional demand that can be caused by a moment resisting reinforced
connection between floors and supporting beams, under relative rotations between the slabs
and beams, is less well understood. Concern has been raised in New Zealand as to the
compressible backing board and low-friction bearing strip on an external supporting beam
Page 4-9
(OGrady, 2004). This connection detail was experimentally tested in the Lindsay (2004)
experiment, and only minor effects of torsion were observed. The degree of twist did not raise
any issues over the performance of the floor. Elliot et al (1998) discussed the effects of
eccentrically loaded external beams for both the isolation type floor-to-beam connection
If the beam is tied into the floor slab through the normal arrangement of continuity tie bars
concreted into some of the opened cores in the floor slab, the torsional stress is virtually
eliminated. There are two reasons for this: (1) a reaction force R generated in the floor plate
will prevent the top of the beam from experiencing inward deflections and (2) the eccentricity
of the load is reduced because of an extended bearing at the end of the cast-in-place infill.
This statement is fundamentally flawed when lateral displacement conditions are considered.
The relative rotations between the floor and supporting beam induce moments in the floor to
beam connection and these moments generate significant torsion demands at the ends of the
beams. The assumption that the floor remains rigidly connected to the supporting beam, and
rotates with the beam is false. Previous experiments (Matthews, 2004 and Lindsay, 2004) as
well as the current experiment, have shown that under lateral displacements when the
supporting beam rotates, the floor slab stays predominantly horizontal. This relative rotation
between the floor and supporting beam induces a force couple within the reinforcement
connecting the floor and beams. Figure 4-4 shows the falsely claimed behaviour and the
observed behaviour of the reinforced beam-floor connections. The induced moments caused
by the relative rotations impose torsion on the beam, as shown in Figure 4-3(b) and Figure
4-4(b). Therefore, the statement made by Elliot et al (1998) that torsionally stresses are
Page 4-10
(a) Erroneous claimed behaviour of the (b) Observed behaviour of the reinforced
reinforced seated floor slab connection seated floor slab connection
(Elliot et al 1998)
Figure 4-4 The torsional demand from reinforced hollow-core to floor slab connections
In addition to the torsion moment generated by the reinforced floor-beam connection, the
effects of beam elongation also cause torsional demands. The growth of the perimeter frame
beams, parallel to the floor span, exert tension forces into the flooring, which acts to pull the
floor slabs off the seating of the supporting beams. This means that the reaction force, R,
generated in the floor plate is not present, and the tension force from beam elongation further
The torsional demands calculated for this experiment were derived from the results
from two-dimensional sub-assemblage testing (discussed in Section 2). The moment capacity
of the floor-to-beam connection (for one hollow-core unit) was observed from the
experimental results and good comparisons were found between the calculated positive and
negative moment capacity values. This moment capacity was used to find the torsional
moment demand imposed on the transverse beams, as is shown in Figure 4-4(b). Accounting
for the moment generated by the connection between the supporting beam and four hollow-
core units, an applied torsional moment at the beam ends of 79kNm was calculated (same for
both positive and negative moments). For the compressible backing board connection
(Lindsay, 2004), the eccentric loading of the hollow-core flooring on the concrete seat of the
supporting beams does result in torsion under static conditions. Under lateral loading, the
moment induced by the section under relative rotations was minimal as it was observed that
the compressible backing board did not actually compress and a force couple could not be
Page 4-11
achieved. A torsional moment demand from the eccentric floor loading of 20kNm at each end
of the transverse beams was calculated, although it must be noted that this is a lower bound
For Phase 1 and 3, the amount of bending moment and shear force acting at the ends of the
transverse beams would have been small, as determined from the static loading of the floor
slabs. Therefore, the following calculations of the torsional capacity of the concrete transverse
beam do not include flexure and shear actions, as they are sufficiently small to negate. Under
different circumstances, for example a 45 degree earthquake whereby the longitudinal and
transverse directions are simultaneously loaded, then the torsion would have to be considered
Using the equations and methods presented, the torsional strength and stiffness values
for the transverse beams were calculated. The cracking torsion moment was calculated to be
77kNm and the torsion contribution from the reinforcement was found to be 200kNm. Using
the stiffness formulation presented in Equations 4.3 and 4.4, the pre-cracking stiffness was
95500kNm2 and from methods presented in Park and Paulay (1975), the post-cracking
stiffness was found to be 4% of the pre-cracked stiffness (3800kNm2). Using these calculated
torsional strength and stiffness values, the torsion-twist relationship was generated
(Figure 4-5). This offers a quantitative form of the idealised torsion-twist relationship shown
in Figure 4-3(a).
The representation of the torsional demand and capacity shown in Figure 4-5, can be
used to explain the amount of torsional twisting that occurred within the plastic hinge regions
of the west transverse beams (Figure 4-2(b)). The intersection between the demand for the
current experiment and capacity curves for Phase 1 indicates that there would be a torsional
Page 4-12
rotation of around 0.0064 radians (0.64%). This is approximately true for the experimental
discrepancies seen during the later stages of Phase 1, where at the first and second cycles to
1%, the difference between the rotation of the column and the supporting beam adjacent to
the plastic hinges was around 0.7%. At the first and second cycle peaks to -2% drift, the
torsional twist at the ends of the west transverse beam was around 1%. The reduced stiffness
for Phase 3 longitudinal re-loading indicates that significantly higher torsional twisting
occurred. The intersection point between the capacity and demand for Phase 3 shows that
torsional drifts of around 0.021 radians (2.1%) could be expected. During the later stages of
the experiment, the ends of the beams were experiencing rotations of over 4%. The calculated
value of 2.1% underestimates the amount of torsional twisting of the torsional hinges.
However, the suggested representation still provides a relatively good approximation of the
300
250
Torsion Moment (kNm)
200
150
100 Phase I
Phase III
50 Current Experiment Demand
Lindsay (2004) Experiment Demand
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Torsional Twist (rads/m)
confidently predict torsional performance. Uncertainties arise due to the difference between
various methods in calculating the torsional capacity of section as well as the assessment of
the torsional demand on the supporting beams. However, the calculations provided a
reasonably accurate insight into the amount of torsional rotations of the supporting beam
Page 4-13
plastic hinge zones that can occur. However, the fact that the simple calculations produce
estimates of torsional rotation that are less than what was shown by the actual experimental
results, means that torsion of floor supporting beams needs to be considered seriously.
The occurrence of vertical displacement differences between flooring systems and lateral load
resisting frame systems has been accepted as a source of damage for floor diaphragms.
However, little analytical and experimental work has been done to assess the incompatibility.
The displacement incompatibility arises due to the different deformation patterns of the
flooring and frame elements, and is schematically shown in Figure 4-6. Under lateral loading,
the frame beams deform in double curvature, whereas flooring systems, that span parallel to
the frame, are most commonly designed to act as though simply supported. In cases where the
flooring is adequately tied into the beams and is forced to deform in an undesirable double
curvature pattern, the floor slab is prone to damage and effective load paths and transfer of
Displacement Displacement
incompatibility incompatibility
between the frame between the frame
and h/c unit and h/c unit
Hollow-core unit
sags and deforms Frame deforms in
in single curvature double curvature
Laus (2002) experiments observed damage between the perimeter beams and floor and
concluded that the vertical movement of the floor relative to the beams has the potential to
cause failure of the slab. Displacement incompatibility was evident due to damage that
Page 4-14
occurred during the hollow-core super-assemblage experiment conducted by Matthews
(2004). A strong connection between the first hollow-core unit and adjoining perimeter beams
(Figure 4-7(a)) forced the floor to deform in double curvature and these deformations could
not be sustained by the webs of the hollow-core unit adjacent to the perimeter beams.
Ultimately the localised over-stressing resulted in the collapse of the bottom half of the first
hollow-core unit beside the main frame. The adverse deformations also resulted in fracturing
of the cold drawn mesh reinforcing the topping slab and caused an unzipping tear
longitudinally between the first and second floor units which can be seen in Figure 4-7(c).
This tear was first noticed when cracking was observed adjacent to the centre column at
0.75% inter-storey drift, and by 1.95% all of the cold drawn mesh crossing this crack had
fractured.
750mm
First timber
Second Hollow-core First Hollow-core Unit
infill
Hollow-core Unit
Unit Perimeter beam
Perimeter beam
Column face
Column face
(a) Lateral connection used in the Matthews (b) Lateral connection used in the Lindsay
(2004) experiment (2004) experiment
(c) Tearing damage observed during (d) Infill floor damage witnessed during the
Matthews (2004) testing Lindsay (2004) experiment
Figure 4-7 Lateral perimeter beam floor unit connection details and damage.
Page 4-15
The Lindsay (2004) experiment investigated future construction details addressing the
problems observed during the Matthews (2004) test. The lateral connection between the
perimeter beams and initial hollow-core floor unit consisted of a topping slab cast over a
timber plank infill 750mm wide, as shown in Figure 4-7(b). This was intended to be a flexible
link isolating the two structural systems to accommodate the vertical displacement
incompatibilities while still being able to transfer inertia forces from the floor to the frame.
Figure 4-7(d) shows the damage incurred by the infill slab during the testing. The solution
was able to satisfactorily withstand the displacement incompatibility and damage to the
flooring was minimised. However the premature termination of the starter bars at the interface
between the infill and first hollow-core unit created a zone of weakness where tearing was
observed and the ductile mesh reinforcing the topping slab fractured.
The testing of precast seismic structural systems in the PRESSS research program (Priestley
et al 1999) included connection solutions designed to overcome the problems associated with
Figure 4-8, connected the double tee flooring to the perimeter frame at discrete points on the
lower levels of the structure. The X plate connectors were designed by capacity design
principals to act as an energy dissipating device, and limit the amount of inertia forces
Page 4-16
4.3.2 Analytical modelling of displacement incompatibility
The analysis involved the construction of an analytical model in the inelastic time history
dimensional model was set up to model the longitudinal lateral loading of the front frame, and
consisted of members representing the frame and floor elements and springs connecting the
floor to the frame. Different spring members were used to model the level of fixity of the end
support conditions, and along the frame between the frame and flooring. The model was
calibrated by assessing the base shear capacity of the system and comparing it to the
experimental values.
Figure 4-9 shows the displacement incompatibility between the longitudinal perimeter
beams and the adjacent parallel hollow-core unit from the experiment and modelling for both
the Lindsay (2004) and current experiments. The lateral spring properties modelling the infill
connection were kept the same, but the end spring properties were altered to model the pinned
type seating connection for Lindsay (2004) and the rigid connection in the present
experiment. The results show promising correlation between the experiment and analytical
model. The overall deformation trends are matched although the model slightly
parameters of the model allowed other areas of interest to be investigated. Sensitivity analyses
were performed examining the effect of: different end support conditions, different lateral
connection details, different bay lengths and numbers of bays and different flooring types.
The effects of the different parameters was seen by examining both the displacement
incompatibility patterns as well as the overall system base shear capacities, and the results
Page 4-17
12
Displacement Incompatibility 10
8
6
4 Position along fram e (m m )
2
(mm)
0
-2 0 2000 4000 6000 8000 10000 12000
-4
-6 Experiment 1%
Model 1%
-8 Experiment 2%
-10 Model 2%
-12
8
6
4
2 Position along fram e (m m )
(mm)
0
-2 0 2000 4000 6000 8000 10000 12000
-4
-6 Experiment 1%
Model 1%
-8 Experiment 2%
-10 Model 2%
-12
(b) Observed and modelled displacement incompatibility from the current experiment
Figure 4-9 Observed and modelled displacement incompatibility
From the sensitivity analyses, it was seen that the end connection detail, through the stiffness
of the end spring, had a large influence on system strength capacity and displacement
incompatibility values. The base shear capacity for the system was 25% higher for the rigid
connection than the pinned and semi-pinned connections. This is matched to a certain extent
by the higher ultimate base shear capacity of the current experimental structure than the
equivalent Lindsay (2004) base shear capacity. For the rigid end conditions, the maximum
displacement incompatibility occurred on both sides adjacent to the central column, as the
floor unit deformed in single curvature due to the fixity at the ends. For the pinned end
Page 4-18
conditions, which allowed the floor to simply sag, displacement incompatibility peaks were
seen close to the supports as well as either side of the central column, and the displacement
incompatibility values were generally higher than for the fixed end case.
The computational modelling conducted has more recently been extended by work
completed at the University of Canterbury through several research projects. Taylor (2004)
used results from sensitivity analyses of the controlling parameters, to find the points of
minimum displacement incompatibility. For each of the influencing parameters, a list of the
made to be able to form the basis for the location of connectors. This work preceded work by
McKenzie (2004), who investigated the importance of higher mode inertia forces and if
energy dissipating discrete connectors were able to reduce the demand imposed on lateral load
resisting systems. Jensen (2004) analysed the diaphragm action due to the effects of different
locations and numbers of floor-frame connectors. More favourable conditions were seen for
several connectors as the inertial forces could be distributed over a wider area, however the
development of post-tensioned rocking frame and wall solutions involved with damage
particular, research has focused into isolating the floor and lateral load resisting elements to
limit damage, while still maintaining a connection capable of transferring inertia and
structural interaction forces from the floor to the lateral load resisting system. Work has been
done testing various mechanical connectors and devices that can join floor and frame
Page 4-19
members, and one such connector is shown in Figure 4-8. Connectors that possess energy
dissipation qualities, which are able to limit the inertia forces being transferred to a perimeter
lateral load resisting frame, which in turn restricts damage, have previously been developed
and tested. Naito et al (2004) modelled connectors for double tee floors and found that
desirable manner. The welded X plate type connectors used in the PRESSS program, as
shown in Figure 4-8, use shear forces to produce a flexural response, which increases the
Despite this previous research, similar work investigating the viability of systems
involving these connectors has been minimal in New Zealand and research should look to
develop systems tailored for New Zealand construction practice. The timber infill solution has
in other building situations, the timber infill may be undesirable, and a flat-soffit option with
flooring butted up against the perimeter frame may be required. In these cases, discrete
connectors that allow vertical displacement but are fixed horizontally transferring inertia
This section discusses the calculation of the lateral load capacity of the super-assemblage and
the comparisons of the predicted strengths with the actual behaviour. The lateral load capacity
of the system was found by incorporating all sources of strength enhancement, including the
starter bar and beam-to-floor slab reinforcement, diaphragm reinforcement and the influence
of the floor slab through the bowstring effect. Full calculations of the lateral load capacity
Page 4-20
4.4.1 Bowstring Effect
floor system provides restraint due to the growth of the concrete frames. This restraint causes
large compression forces in the frame beams, which significantly increases the moment
capacity of the frame. Tensile forces are induced in the flooring, which are transferred to the
frame beams in a compression arching manner, which explains the name the bowstring
effect, where the tension in the floor acts as the string, and the compression arch represents
the bow. Fenwick et al (1999) demonstrated that the presence of floor slabs parallel to
perimeter frame beams can cause a significant increase in the negative moment capacity of
the beams. The bowstring effect is further explained by Bull (2003) and was observed during
experiments conducted by Matthews (2004) and Lindsay (2004). The bowstring effect was
also graphically evident during the present experiment. Cracking within the topping slab over
the timber infill and hollow-core floor units near the east and west ends of the structure,
arched inwards towards the south centre column. The crack distribution followed the same
end
connection
eccentricity (e) reinforcement
2e
Page 4-21
Figure 4-10 shows the location of the compression arch and the centroid of the tension tie in
the floor. The depth of the compression arch, the eccentricity, is the distance between the
centroid of the tension tie in the floor, and the centroid of the compression force at the inside
edge of the south perimeter beams. The tie force is a combination of the capacity of the
reinforcement at the hollow-core to support beam connection as well as friction of the floor.
This resultant tie force acts at a distance of the eccentricity and is a combination of the
attributing tension forces from the end connection reinforcement over a width of two times
the eccentricity. The tie force can be found from the following:
2eAs f y
T 2PWle (4.5)
s
where e = midspan eccentricity of the tie force with respect to the centroid of the compression
strut (2e is the width that contributes to the combined tie force); T = combined tie force;
l = overall span of arch (the length of the structure for this case); As = end connection
reinforcement area (starter bars and hairpin bars); fy = end connection reinforcement yield
strength; s = end connection reinforcement spacing; P = coefficient of friction for the bearing
By equilibrium, the tension force in the floor slab is transferred to the perimeter beams
by diagonal compression struts. These diagonal struts follow the observed crack pattern in the
arching pattern demonstrated in Figure 4-10. The eccentricity can be found by considering the
diaphragm reinforcement orthogonal to the tie force. The moment generated by the force
couple between the floor slab and perimeter beams can be equated with the moment generated
by the distributed load given by diaphragm reinforcement orthogonal to the tie force, w. The
compression arch is equivalent to the moment from the uniformly distributed load given by
the diaphragm reinforcement, which is in the shape of a parabola (as is the case for simply
supported beams under distributed loads). The mechanism can be expressed as follows when
considering equilibrium:
Page 4-22
wl 2 Pl
Te (4.6)
8 4
Where the first and second parts of (the right hand side of) Equation 4.6 respectively represent
the uniform distribution of slab topping reinforcement along the entire floor span, and the
drag bars anchored into the central column. Equation 4.6 can be expanded to give:
Asd f yd l 2 Ast f yt l
Te (4.7)
8s sd 4
where Asd = area of one bar of the diaphragm reinforcement; fyd = yield strength of the
diaphragm reinforcement; ssd = spacing of the diaphragm reinforcement; and Ast = area of one
This formulation provides an admissible mechanism from which the eccentricity can
be found (by combining Equations 4.5 and 4.6), and therefore the combined tie force and by
equilibrium, the compression axial force induced in the south perimeter beams can be
calculated. This solution provides an upper bound method as it assumes that the reinforcement
at the end supports of the hollow-core floor slabs, and diaphragm reinforcement orthogonal to
Using these equations, and the measured material properties of the reinforcement, an
eccentricity of 2.5m was found. From strain gauge information of the drag bars tying the
interior column into the floor plate, it was observed that the reinforcement was yielding, and
the inclusion for this reinforcement in the mechanism analysis (Equation 4.6) was verified
(see Appendix D for strain information for the drag bar reinforcement). The calculated value
of the depth of the compression arch or eccentricity lies over the middle of the second hollow-
core unit. The location of the calculated tie force and depth of the compression arch matches
well with the cracking pattern witnessed throughout testing as is shown by 4-11(a) whereby a
schematic of the bowstring effect is superimposed on a photo of the floor. During testing, a
longitudinal crack over 7m long formed at the interface between the first and second hollow-
Page 4-23
core units, close to the calculated centroid of the tie force, which indicates that the calculated
eccentricity for the mechanism is supported by the observed behaviour. Using the calculated
eccentricity value, a combined tension force distributed over 5m, of 1254kN was found. By
equilibrium, this tension force is resisted by a compression force of the same magnitude,
which, in this case is experienced by the central column (at the top of the arch). Figure
4-11(b) shows the internal and external actions for the plastic hinges adjoining the south
centre column. The compression force, N, which is transferred through the topping slab into
the perimeter beams, has a significant effect the strength of the plastic hinges of the southeast
N N
Eccentricity = e
2 * Eccentricity = 2e
T
Centroid of the tension tie in the floor
diaphragm
Page 4-24
As it is displayed in Figure 4-11(b) the negative moment capacity of the plastic hinges
framing into the south centre column are increased by the axial compression force which is
shown to act at the level of the topping slab. The moment contribution from both of the plastic
hinges framing into the interior column can be found by the following:
yield strength of tension reinforcement; d = the distance from the extreme compression fibre
to the centroid of the tension reinforcement; d = the distance from the extreme compression
fibre to the centre of the compression block; and N = the axial compression force due to the
bowstring effect.
Using the moment capacity Equation 4.8, the total moment from both of the southeast
and southwest beams hinges (joining the south centre column) experienced by the column is
1859kNm. This is compared to 1004kNm when the bowstring effect in the floor slabs is not
included.
The compression axial forces due to the bowstring effect in this case have been
represented at the height of the topping slab although it is acknowledged that the distribution
of the compression axial forces is more complex than what has been described. The inclusion
of the bowstring effect has been done by way of a simplistic mechanism approach, although it
is acknowledged that it is an upper bound method. Due to this, a significant increase of the
strengths of the perimeter beam plastic hinges was noted due to the compression axial force.
The following sub-sections include the bowstring effect into calculations of the total super-
assemblage lateral load capacity and compare these predictions with the experimental results.
Page 4-25
4.4.2 General System Strength
An important part of finding the capacity of the structure is assessing the negative moment
capacity enhancement provided by tension reinforcement within the topping slab. Matthews
(2004) provided a method of calculating the activated slab width and accompanying strength
from the starter bar reinforcement. This method was based on the experimental findings from
the Matthews (2004) experiment and is a function of the specific structure. It provided an
accurate means of calculating the lateral load capacity of the system. Comparisons were made
with the equivalent methods prescribed in New Zealand Concrete Structures Standard (NZS
3101:1995) and the American Concrete Code (ACI 318-02) and it was found that these
methods significantly underestimated the amount of strength contribution from the activated
tension flange. Undervaluing the amount of moment capacity enhancement creates the
important that the slab and diaphragm reinforcement is accounted for correctly.
starter bars according to increasing lateral displacement. Two distinct points are required to
define the width of slab activation: the onset of yielding, and full plasticity. These points were
found by examining the crack widths crossing the interface where starter bar or diaphragm
reinforcement is present. The torsional stiffness of the beam is important in determining the
rate of activation of the starter bar reinforcement. For beams of relatively low torsional
stiffness, the starter bars in the middle of the beam will be activated much later from those at
the end. In this experiment, it was observed that the beams formed torsional hinges at the
ends, and can be classified as having low torsional stiffness. However, a sizeable crack
formed in the topping concrete at the beam-to-floor interface and it was determined that the
initial yield point and point of full plastification were similar to that assumed by Matthews
(2004).
Page 4-26
For this experiment, the slab width at the first yield point was found to be the width of the
timber infill (0.75m), which incorporates three starter bars or longitudinal diaphragm bars,
and full plasticity was reached at 2% drift with an activated slab width of 3.05m,
approximately half the width of the bay. This effective slab width is the same as what was
used by Matthews (2004) and Lindsay (2004) and Figure 4-12 shows the contributing slab
widths.
Activated tension flange Activated tension flange
JD
The overall strength of the system observed from the experiment is shown in
Figure 4-13 for both the longitudinal and transverse loading phases. The theoretical pushover
curves are superimposed to assess the validity of the theoretical calculations. This predicted
strength capacity incorporates the nominal and overstrength strengths of the beams, starter bar
contributions during the progression from yielding to full plasticity, the effect of the end
support beam-to-floor slab connection reinforcement and the axial compression forces
induced by the bowstring effect. As it can be seen by examining Figure 4-13(a) for both Phase
1 and 3 loading, the correlation is satisfactory. The overstrength capacity strengths (assumed
as 125% and 140% of the nominal strengths) have been plotted as upper bound values and
actual strength of the system lies between the two overstrength predictions. It can be seen that
the later cycles are well predicted; the 3% and 4% cycle peaks or maximum strengths are
slightly underestimated by the nominal theoretical values but well matched by the 40%
overstrength prediction.
Page 4-27
The base shear capacity for the transverse loading phase is shown by Figure 4-13(b).
The accompanying theoretical pushover curve shows good agreement with the experimental
results. For positive drifts, the experimental results experimental values fall in between the
upper bound overstrength capacity values and the lower bound nominal strengths. The
negative drift predictions conversely, are higher than the actual results. This could signify that
there was less starter bar reinforcement activated than what was assumed during calculating
the pushover curve. The experimental values graphed are for before and after a torsion test
that was conducted mid-way through Phase 2 loading. It is discussed further in Appendix D,
but it can be seen that the torsion test undertaken did not cause a reduction in the load
It should also be noted that a yield inter-storey drift of 0.64% was used for the yield
strengths. As the hysteretic behaviour indicates, this theoretical yield drift compares well with
the observed experimental yielding point and elastic stiffness. A more in depth assessment of
the super-assemblage stiffness as well as details of the calculation of the yield drift are located
in Appendix D.
The individual or combined plastic hinge strengths are shown by the moment rotation
relationships in Figure 4-14. The plastic hinges of the south frame for longitudinal loading are
shown in Figure 4-13(a) for both the southeast and southwest hinges, and the combination of
the centre hinges framing into the centre column. The general agreement is good with the later
cycle peaks lying between the nominal and overstrength estimates. The comparison between
the observed experimental behaviour theoretical predictions for the central column is good,
which indicates that the method, albeit an upper bound approach, used to quantify the added
Page 4-28
1800
1600
Base Shear
Force (kN)
1400
1200
1000
800
600
400
200
Inter-Storey Drift (%)
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-200
-400
-600
-800
Experiment Phase 3
-1000
Experiment Phase 1
-1200
Nominal Prediction
-1400 25% Overstrength Prediction
-1600 40% Overstrength Prediction
-1800
(a) Phase 1 and 3 Longitudinal loading: Experimental base shear capacity and theoretical
pushover curve.
1200
Base Shear
Force (kN)
900
600
300
-300
Experiment (post-torsion)
-1200
(b) Phase 2 Transverse loading: Experimental base shear capacity and theoretical pushover
curve.
Figure 4-13 Base shear capacity: experimental and theoretical relationships.
Page 4-29
1200 3000 1200
Moment (kNm)
Moment (kNm)
Moment (kNm)
1000 2500 1000
Southwest hinge, south beams South centre hinges Southeast hinge, south beams
(a) Phase 1 Longitudinal loading: plastic hinge strengths
1200 1200
1000
Moment
1000
Moment
(kNm)
(kNm)
800 800
600 600
400 400
200 200
Rotation (%) Rotation (%)
0 0
-4 -3 -2 -1 -200 0 1 2 3 4 -4 -3 -2 -1 -200 0 1 2 3 4
-400 -400
Pre-Torsion Pre-Torsion
-600 -600
Post-Torsion Post-Torsion
-800 Nominal Strength -800
Nominal Strength
-1000 25% Overstrength -1000 25% Overstrength
-1200 -1200
North hinge, West transverse beam North hinge, East transverse beam
1200 1200
1000
Moment
1000
Moment
(kNm)
(kNm)
800 800
600 600
400 400
200 200
Rotation (%) 0
Rotation (%) 0
-4 -3 -2 -1 -200 0 1 2 3 4 -4 -3 -2 -1 -200 0 1 2 3 4
-400 -400
Pre-Torsion Pre-Torsion
-600 -600 Post-Torsion
Post-Torsion
-800 -800 Nominal Strength
Nominal Strength
-1000 -1000 25% Overstrength
25% Overstrength
-1200 -1200
South hinge, West transverse beam South hinge, East transverse beam
(b) Phase 2 Transverse loading: Individual hinge strengths.
Figure 4-14 Individual hinge strengths: experimental and theoretical pushover relationships.
Experimental and theoretical comparisons for the transverse loading are provided in Figure
4-14(b). It is shown that the theoretical strength model represents the actual behaviour well.
The positive moment capacities for each of the hinges are matched well by the overstrength
Page 4-30
capacities for the southeast and southwest hinges of the transverse beams. The nominal
moment capacity values provide a more suitable match in this case. This overestimation could
mean that the amount of activated slab contributing is less than what was specified; half of the
One of the significant outcomes from the Matthews (2004) experiment was the development
of a rational mechanics based theory to predict beam elongation. A rainflow counting method
was been employed to account for the growth of plastic hinges in a frame system under
inelastic rotations. The theory is briefly explained in Section 2 of this thesis, but further
information is provided by Matthews (2004). The calculations made for the theoretical
Figure 4-15(a) and (b) displays the observed and theoretical beam elongation results
for Phase 1 and 3 longitudinal loading. The elongation values represented are for the total
frame and a consideration of the pull-off demands imposed on the floor seating connections
require the values shown to be halved. As it is illustrated, the observed and predicted results
agree well, especially during Phase 1. The theoretical elongations slightly underestimate the
actual values up until the first cycle to 4% during Phase 3, after which the experimental
values level off and even decrease. The elongation predictions of the final loading cycles to
5% are not reached due to instrumental errors, but despite this, the agreement is satisfactory
up until the 4% cycles. After Phase 1 loading, the permanent elongation was 32mm, from
which the elongation continued to grow during Phase 3 re-loading and at the end of testing,
Page 4-31
Figure 4-15(c) shows the frame beam elongation results obtained from the Phase 2
transverse loading. The predicted values display elongation for the outside transverse frames
and are displayed by Figure 4-14(d). The beam elongation theory once again predicts the
actual behaviour well. The actual results are marginally lower than the predicted values,
however the difference is small. At +4% the predicted beam elongation was 30mm and the
recorded experiment elongations were 29mm and 30mm for the west and east bays
respectively.
120 120
Elongation
Elongation
T heor y
(mm)
(mm)
100 100
80 80
60 60
40 40
Phase 1 20
20
Phase 3
0 0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Inter-storey Drift (%) Inter-storey Drift (%)
(a) Phase 1 and 3 longitudinal experimental (b) Phase 1 and 3 longitudinal theoretical beam
beam elongation elongation
35 35
Elongation
Elongation
T heor y
(mm)
(mm)
30 30
25 25
20 20
15 15
10 10
East Bay
5 5
West Bay
0 0
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
Inter-storey Drift (%) Inter-storey Drift (%)
(c) Phase 2 transverse experimental beam (d) Phase 2 transverse theoretical beam
elongation elongation
Figure 4-15 Experimental and theoretical beam elongation results.
Page 4-32
4.6 LOW CYCLE FATIGUE OF THE REINFORCEMENT
The fracture of the reinforcement observed during the later stages of the experiment can be
predicted by low cycle fatigue theory. The fracture of the longitudinal reinforcement during
the Lindsay (2004) experiment was able to be predicted to a satisfactory level of accuracy.
Low cycle fatigue is also addressed in Section 2 for the sub-assemblage tests. The low cycle
fatigue theory enables the plastic rotation life capacity of the reinforcement to be calculated
D L p 1 L p
2 N f
Tp 0.16 1
0.5
(4.9)
D' D 2 L
where Tp = plastic drift; L = lever arm of the cantilever column; D = overall member depth;
D = distance between outer layers of steel; 2Nf = the number of reversals to the appearance of
the first fatigue crack with which Nf is the effective number of constant amplitude cycles for a
variable amplitude displacement history; and Lp = equivalent plastic hinge length, which is
given by:
2
T
Nf i T
(4.11)
ref
where T = drift for the ith cycle of loading and Tref = reference drift for an equivalent constant
amplitude loading history. Using a reference drift of 5% an applied cyclic demand of Nf = 4.5
cycles is calculated for the experiment. This signifies that 4.5 equivalent cycles of 5% drift
caused the low cycle fatigue fracture of the reinforcement. By re-arranging Equation 4.9, the
Page 4-33
theoretical equivalent number of equi-amplitude cycles can be found. Using a reference drift
of 0.05 radians, a yield drift of 0.0064 radians and a equivalent plastic hinge length of
511mm, an equivalent number of 3.6 cycles to 5% was calculated This underestimates the
first fracture, which occurred after 4.5 cycles. Given that conventional wisdom considers that
any result that falls within 25% of the predicted value is satisfactory, it is evident that the
experiment (NF-Expt = 4.5) and theory (NF-Theory = 3.6) match to a satisfactory level.
The research presented in this thesis report follows work that has identified the seismic
inadequacies of precast hollow-core floor systems and investigated solutions to address the
identified problems. With the results and understanding gained from each of the three super-
nuances of hollow-core behaviour have become clearer. There has been a progression of
design solutions and correspondingly better performance. This experiment, as it was aimed to
do, has answered many questions and proven the recommendations given by previous
researchers. This sub-section briefly assesses aspects of this experiment in relation to the
investigations by Matthews (2004) and Lindsay (2004) as a way to conclude this particular
Figure 4-16 shows the lateral load demand imparted on the structure during each of
the three super-assemblage experiments as well as the timing of significant failure events. As
it can be seen, the cyclic demand for the Lindsay (2004) and current investigations of new
construction details, was more onerous than the original Matthews (2004) experiment, which
examined of existing buildings. The location of failure events also signify improved
performance throughout the three experiments, with the later two experiments being able to
Page 4-34
5 5 5
4 Initial Mesh Partial
3 Fracture 3 3
Collapse
2 2 2
1 1 1
0 0 0
-1 -1 -1
-2 -2 -2
-3 -3 -3
-4 -4 -4 Total Collapse
-5 -5 -5
5 5 5
Initial Mesh Low cycle fatigue
4 Inter-storey Drift (%) 4 4 Fracture of main beam bars
Inter-storey Drift (%)
5 5 5
Low cycle fatigue
4 4 4 of main beam bars
3 3 3
2 2 2
1 1 1
0 0 0
-1 -1 -1
-2 -2 -2
-3 -3 -3
-4 -4
-4
-5 -5
-5
beam seated connections for each of the three tests. The seated connection used for the
Matthews (2004) experiment was originally assumed to slide on the concrete seating ledge
relative to the supporting beam. However, the rigidity of the mortar bed seating surface did
not allow the floor unit to slide, and under cyclic loading, the relative rotations between the
supporting beam and floor slab resulted in large cracks forming in webs of the hollow-core
unit. This eventually caused a loss of seating failure of the hollow-core flooring. The floor-
beam connection tested by Lindsay (2004), as a result of the previous poor performance, was
Page 4-35
a pinned type connection, which featured the hollow-core flooring seated on a low friction
bearing strip to allow sliding and a compression backing board. This detail behaved well,
although spalling of the concrete ledge concrete and sliding of the low friction bearing strip
The beam-to-floor connection used in this current experiment rigidly tied the floor into
the beam, possessing a degree of redundancy not offered by the Lindsay (2004) detail.
Improved performance under relative rotation was observed due to the promotion of a critical
section at the beam-slab interface at which the rotation was concentrated. The low friction
bearing strip performed well and allowed the flooring to slide relative to the supporting beam
under the effects of beam elongation, and after the completion of testing, the hollow-core
Compressible material
does not compress
Column
Snapping action
Unit slides
(a) Matthews (2004) seated connection (b) Lindsay (2004) seated connection
behaviour. behaviour.
Rotation centred at Hollow-core units
critical section remain undamaged
Page 4-36
The successive experiments have seen dramatic improvements of the seated connection
performance details. The problems associated with the inadequate behaviour of existing
building details has been remedied in terms of future construction and the seismic adequacy
of the two latter supporting connection details that have been added to the New Zealand
The lateral hollow-core connection throughout the current and previous experiments has been
one of the primary sources of examination and details from the previous tests are shown in
Figure 4-7. A leading source of damage witnessed during the Matthews (2004) test occurred
due to deformation incompatibilities between the perimeter frame and adjoining hollow-core
floor units. The discussion of displacement incompatibility outlines the identified issues and
reasons for the damage. As a solution to these issues, a timber infill solution that allows the
perimeter beams and flooring to displace freely individually was proposed and tested by
Lindsay (2004) and a dramatic improvement was noted. However, the termination of the
starter bar reinforcement prior to the infill-to-floor interface placed high demands on the
ductile mesh reinforcement in the topping slab which fractured at relatively early stages and a
resulting large tear within the topping concrete. To overcome the deficiencies from this
solution, a similar timber infill detail was tested again during the current experiment,
incorporating longer starter bars and conventional reinforcing in place of the ductile mesh
whereby the longer starter bars distributed stresses over a larger area and there was no major
cracking or tearing ensuring that inertia and structural interaction loads from the flooring are
Page 4-37
4.7.3 Overall performance comparisons
Comparisons between the base shear hysteretic behaviour of the three experiments, shown in
Figure 4-18, are able to show the relative overall strengths and stiffnesses of the systems and
energy dissipation characteristics. However, the overall performance of the structure cannot
reinforced concrete frame. Upon examining Figure 4-18, the major improvements made by
the current and Lindsay (2004) experiments are clouded by the similar performances of the
perimeter concrete frames. Despite this, it can be seen that similar performance of the
perimeter concrete frames can be achieved by repairing existing concrete frames (Lindsay,
In addition to the design changes made following the Matthews (2004) test to the end
and lateral hollow-core connection and the corresponding improved performance, there are
also several other aspects of the test structure behaviour to consider. There was little or no
damage to the southern end corners of the hollow-core units signifying the hooked bar,
utilised in the transverse beams to reduce the risk of relocated plastic hinging from occurring,
performed as desired. Other small modifications made from the Lindsay (2004) such as a
second generation bearing strip and seat concrete transverse reinforcing also proved to be
worthwhile. Interestingly, the pattern of damage within the plastic hinges for the current
experiment was different than the Matthews (2004) and Lindsay (2004) experiments.
Significant cracks, one at the beam-column interface and another 100-200mm away from the
column, formed and progressively opened throughout tests. This contrasts with the previous
experiments and expected behaviour of distributed cracking over a conventional plastic hinge
zone length.
Page 4-38
1600 1200
Base Shear
Force (kN)
Base Shear
Force (kN)
1200 900
800 600
400 300
Inter-Storey Inter-Storey
Drift (%) Drift (%)
0 0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -4 -3 -2 -1 0 1 2 3 4
-400 -300
-800 -600
Phase 1
-1200 -900
Phase 3
-1600 -1200
Base Shear
Force (kN)
1200 900
800 600
400 300
Inter-Storey Inter-Storey
Drift (%) Drift (%)
0 0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -4 -3 -2 -1 0 1 2 3 4
-400 -300
-800 -600
Phase 1
-1200 -900
Phase 3
-1600 -1200
Base Shear
Force (kN)
1200 900
800 600
400 300
Inter-Storey Inter-Storey
Drift (%) Drift (%)
0 0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 -4 -3 -2 -1 0 1 2 3 4
-400 -300
-800 -600
Phase 1
-1200 -900
Phase 3
-1600 -1200
Page 4-39
4.8 CONCLUSIONS
The results and discussion provided in this section provide a forensic analysis that enabled an
1. Due to the added torque created by the rigidly reinforced floor-to-beam connection,
torsional hinging was observed to occur at the ends of the transverse beams. The use of
torsional performance of the beam sections, was explained. It was seen that the flexural
plastic hinging of the ends of the transverse beams under Phase 2 transverse loading was
largely the reason for the deterioration of the concrete contribution to the torsional
strength, Tc, and hence led to large torsional twists on the softened cracked beam sections
during Phase 3 loading. It is recommended that future research should further address the
complexities of torsional behaviour when combined with concurrent shear and flexural
actions that are expected to occur under realistic bi-lateral seismic repsonse.
2. The construction details altered and improved from the Lindsay (2004) experiment that
features an articulated topping slab cast over a timber infill demonstrated that it
incompatibility showed good agreement with the experimental results. It was also shown
that extensions of this model could provide a useful tool for analysing and predicting the
3. The strength of the super-assemblage was able to be predicted and good correlation was
seen between the theoretical pushover curve and the experimental base shear force
hysteretic behaviour. The method for accounting for the progressive activation of starter
bars with increasing lateral displacement of the structure matched the actual strength
Page 4-40
increase from the initial onset of yielding and the assumed point of full plastification.
The contribution to the lateral load capacity of the front frame due to the restraint
imparted by the floor slab, the bowstring effect, was quantified and seen to be
significant. The observed crack patterns in the floor slab support the concept of a so-
called bowstring effect. This effect induced a substantial axial thrust force in the
perimeter beams which in turn increases the moment demand on the interior column.
4. A summary of the preceding Matthews (2004) and Lindsay (2004) experiments and
current investigation with respect to the comparative overall structural performance was
made. It was seen that there has been distinct improvements through the successive
experiments and larger displacement demands can now be maintained without significant
in place of mesh, special detailing of the transverse beam, improved bearing strip and
reinforcing of the concrete seat all enhanced the overall performance of the structure.
4.9 REFERENCES
ACI 318-02, Building Code Requirements for Structural Concrete and Commentary,
Bull D.K, 2003, Understanding the Complexities of Designing Diaphragms in Buildings with
Carr A.J, 2003, Ruaumoko 2D User Manual, Computer Program Library, Department of Civil
Dutta A and Mander J.B, 2001, Energy Based Methodology for Ductile Design of Concrete
Page 4-41
Elliot K.S, Davies G, Gorgun H, Adlparvar M.R, 1998, The Stability of Precast Concrete
Fenwick R.C, Davidson B.J, and Lau D, 1999, Strength Enhancement of Beams in Ductile
the New Zealand Structural Engineering Society (SESOC), Vol 12, No. 1, pp 35-40.
Hsu T.T.C, 1993, Unified Theory of Reinforced Concrete, CRC Press, Boca Raton, USA.
Jensen J, 2004, Diaphragm Action Issues in a Jointed Precast Concrete System, Research
Lau D.B.N, Fenwick R.C and Davidson B.J, 2002, Seismic performance of R/C perimeter
frames with slabs containing prestressed units, Proceedings for NZSEE conference,
Lindsay R.A, 2004, Experiments on the seismic performance of hollow-core floor systems in
Matthews J.G, 2004, Hollow-core floor slab performance following a severe earthquake, PhD
New Zealand.
Matthews J.G, Mander J.B, Bull D.K, 2004, Prediction of beam elongation in structural
concrete members using a rainflow method, 2004 NZSEE Conference, Rotorua, NZ,
Zealand.
Page 4-42
Mitchell M.P and Collins D, 1991, Prestressed Concrete Structures, Prentice Hall,
NZS3101, 1995, Concrete Structures Standard, NZS3101, Parts 1 & 2, Standards New
NZS3101, 2003, Draft Amendment No. 3 to 1995 Standard (NZS3101), Standards New
Naito C.J and Cao L, 2004, Precast Diaphragm Panel Joint Connector Performance, 13th
OGrady C.R, 2004, Letters to the Editor, Journal of the New Zealand Structural Engineering
Park R and Paulay T, 1975, Reinforced Concrete Structures, J. Wiley and Sons Inc, New
York.
Paulay T and Preistley M.J.N, 1992, Seismic Design of Reinforced Concrete and Masonry
Priestley M.J, Sritharan S, Conley J.R, Pampanin S, 1999, Preliminary results and
conclusions from the PRESSS five-storey precast concrete test building, PCI Journal,
Taylor L, 2004, Vertical Displacement Incompatibility between Floor Slabs and Seismic
Zealand.
Page 4-43
5. Section 5 Conclusions and Recommendations
5.1 SUMMARY
The research presented in this thesis continued work done by Matthews (2004) and Lindsay
(2004) into examining the seismic performance of precast concrete structures with hollow-
core flooring. The complete failure of the hollow-core flooring observed during the
investigation by Matthews (2004) showed that existing structures using hollow-core flooring
systems pose a serious risk under seismic events. Continuing research by Lindsay (2004)
showed that with simple detailing enhancements, significant improvement in the seismic
performance of hollow-core floor systems can be expected. The present experimental research
were reinforced and rigidly tied the floor into the supporting beam, the first of which has been
proposed in Amendment No. 3 to the New Zealand Concrete Standard (NZS 3101, 2004). A
newly developed test set-up was described which incorporated both the relative rotation
between the supporting beam and hollow-core floor, and tension in the floor representing the
elongation of a parallel moment resisting frame, as the sources of damage. The sub-
assemblage test specimens were tested up to relative rotations of 5.0% and visual and
The main investigation of this thesis involved the testing of a full-scale two bay-by-
one bay concrete frame super-assemblage. The super-assemblage was constructed and several
new detailing improvements recommended from previous research were implemented as well
as the recently proposed, but untested, hollow-core connections with the concrete frame
(NZS 3101, 2004). The full-scale structure was cyclically tested in both the longitudinal and
Page 5-1
transverse directions to inter-storey drifts of 5%. Visual and instrumental observations from
The third major section of this thesis provided a forensic analytical overview of the
experimental observations. The torsion behaviour of the transverse beams was discussed and
analysed using simple calculations. Displacement incompatibility was also described and
results from the experiment were compared with analytical modelling results. The potential
uses of extending the model were also described. A theoretical pushover curve was generated
predicting the strength of the super-assemblage, and the contribution to the strength of the
moment resisting frame of the floor slab, through the bowstring effect, was quantified.
Finally, as a way of completing the current line of experimental research, the present
investigation was compared and contrasted with previous research conducted by Matthews
(2004) and Lindsay (2004). The comparative performances showed that through improved
5.2 CONCLUSIONS
From the research presented in this thesis, the following conclusions can be drawn:
New Zealand Concrete Design Standards (NZS 3101:1995 and amendments) performed
well under simulated seismic loads. The super-assemblage structure was able to sustain
fractured due to low cycle fatigue. The floor system sustained only minor damage, and
Page 5-2
2. The reinforced connection between the hollow-core floor slab and supporting beams
damage suffered by the hollow-core floor units was minimal. Desirable action, where the
relative rotations between the floor and supporting beam were concentrated at the beam-
slab interface was achieved. A second generation low friction bearing strip ensured that
the hollow-core floor slabs slid horizontally relative to the supporting beam and helped
3. The timber infill connection utilised between the lateral load resisting frame and
adjoining hollow-core spanning parallel to the frame performed well during the
experiment. This connection detail allowed the flooring and frame elements to deform
freely. At the same time, the integrity of the connection, in terms of the capability of
transferring inertial forces from the floor to the frame, was maintained. The good
performance of the timber infill detail supports its inclusion into New Zealand Concrete
Design Standards. The displacement incompatibility between the floor and frame was
measured during the experiment and correlations between these results and analytical
4. The lateral load capacity of the super-assemblage was able to be predicted with
satisfactory agreement observed with the experimental results. Although it was seen that
the moment resisting frame dominated the overall super-assemblage strength, the
contribution of the floor slab was noticeable. Compression forces induced in the
longitudinal frame beams due to the bowstring effect were quantified and the progressive
activation of starter bars within the transverse beams was accounted for in calculations of
Page 5-3
5. As a result of the rigidly reinforced connection between the transverse supporting beams
and floor slabs, torsional hinging at the ends of the transverse beams was observed. The
strong floor-to-beam connection caused the floor to remain horizontal and the supporting
beams to stay essentially upright throughout the cyclic lateral loading of the super-
assemblage in the longitudinal direction, and this placed large torsional demands on the
ends of the transverse beams. It was seen that the flexural hinging of the transverse
beams during transverse loading caused deterioration of the ends of the transverse
beams, which accelerated the effects of torsion observed under longitudinal re-loading.
scale super-assemblage test was seen to provide an accurate representation of the three-
dimensional test. A new testing procedure was developed to include the dual sources of
damage of the relative rotation between the supporting beam and slab, and tension forces
in the floor. Although the limitations of the two-dimensional form of test were
acknowledged, the ability to achieve useful and realistic results for less time and
resource constraints make this format of testing a valuable method of investigation floor-
to-beam connections.
7. This overall experimental investigation has shown that a correct level of reinforcement
between the floor slab and supporting beam can lead to good performance. However,
of the sub-assemblage component specimens tested during this current research indicate
therefore imperative that a correct level of reinforcement, such as the reinforcement used
Page 5-4
5.3 RECOMMENDATIONS FOR PROFESSIONAL PRACTICE
One of the main objectives of this research was to validate previously unproven and untested
(NZS 3101, 2004). Recommendations from previous research (Matthews, 2004 and Lindsay,
2004) have also been incorporated to enhance the seismic performance. The following
recommendations have been made for improving professional practice and design standards:
1. Subject to the good performance of the reinforced connection detail used between the
hollow-core floor slabs and supporting beams, it is recommended that this detail is fully
adopted by New Zealand Concrete Design Standards. Figure 5-1 shows a diagram of the
connection detail. Also it can be recommended that the current 1.2% drift design limit
D12 Starter
Bars (Grade 300) 900mm (3*hollowcore depth) D12 Bars @ 300mm each
(300mm spacing) way as topping reinforcing
600mm Lap
2. For cases when an infill slab is used to connect an exterior concrete frame and parallel
spanning precast floor, the detail presented in Figure 5-2 should be used. The starter bars
from the perimeter beams should extend a minimum distance of 600mm into the first
floor unit.
Page 5-5
75mm Concrete Topping
D12 Starters extending 600mm
into the first hollow-core unit Timber in-fill
(lapped with diaphragm reinforcing) 750mm
Hollowcore Units
200mm*25mm Timber boards
Perimeter beam
Figure 5-2 Hollow-core connection detail with a parallel load resisting frame
mesh in diaphragm topping slabs. Grade 300 D10 or D12 reinforcement (the D
decrease the risk of reinforcment fracture and are compatible with conventionally used
4. Low friction bearing strips should be used for the seating of hollow-core floor units.
These allow horizontal sliding of the floor units relative to the supporting beam and help
distribute stresses reducing the chance of concrete spalling from the supporting ledge.
Figure 5-3 shows the layout that the seating of precast units should adhere to.
Floor Unit
Supporting Beam
Low friction
bearing strip
10mm 15mm
50mm
75mm
Figure 5-3 Precast concrete floor seating requirements (close-up view)
Page 5-6
5. Hollow-core units should not be seated on potential plastic hinge zones of supporting
beams. Infill slabs should be used over these regions and additional hooked
reinforcement should be cast in the transverse beams, such as is shown in Figure 5-4, to
force plastic hinging of the supporting beams to occur away from the seated hollow-core
floor units.
Starter bars from the
longitudinal beams Column
(extended into first hollowcore unit)
Hollowcore Units
Timber Infill
Transverse Beam
Hooked Bar
(in addition to the regular reinforcing) Longitudinal Beam
Although this research has addressed several issues that have arisen from previous research
findings, there are still many issues that require research attention. The following lists some
1. Despite the limitations, the effectiveness and ease of conducting proof of concept sub-
testing all floor-to-beam connections that have been used in New Zealand construction,
including other floor types. This research would be able to highlight vulnerabilities
Page 5-7
2. The research conducted in this thesis as well as Lindsay (2004) has successfully
provided validated design solutions for future construction. However the shortcomings
of hollow-core flooring observed during the Matthews (2004) testing indicate that many
existing buildings would be damaged in a seismic event. Future research should look to
find suitable economical and practical retrofit measures for existing structures.
3. As the timber infill detail has been proven as a suitable detail to connect lateral frames
with parallel spanning precast floors, future research should seek to develop a flat
soffit option where the flooring is directly adjacent to the perimeter frame. Research has
discrete floor-to-frame connectors that are able to allow vertical displacements but
dissipate energy and transfer inertial loads is one example of a possible connection
solution. A suitable flat soffit detail would offer designers another option should an
4. As the Matthews (2004) and Lindsay (2004) experiments have shown, hollow-core floor
units are susceptible to damage when seated on plastic hinge zones of their supporting
beams. Future research should seek to develop advanced seating materials that allow
both the horizontal sliding of the floor units relative to the supporting beams and
Page 5-8
In addition to the experimental research recommended, there is also the need for analytical
work in order to better predict and assess the behaviour of hollow-core structures. Ideally, all
experimental research should be coupled with parallel analytical work so that experimental
work can be better explained and understood as well as to offer a useful tool for researchers
and designers. More specifically, areas of analytical research should develop what has been
touched on in this thesis. Modelling of torsion, the contribution of strength from the floor
slabs in terms of the bowstring effect and displacement incompatibility all need to be
extended.
5.5 REFERENCES
Lindsay R.A, 2004, Experiments on the seismic performance of hollow-core floor systems in
Matthews J.G, 2004, Hollow-core floor slab performance following a severe earthquake, PhD
New Zealand.
NZS3101, 1995, Concrete Structures Standard, NZS3101, Parts 1 & 2, Standards New
NZS3101, 2004, Amendment No. 3 to 1995 Standard (NZS3101), Standards New Zealand,
Page 5-9
Appendix A Sub-Assemblage Experiments: Construction &
Results
Full size
test SFD based on
12m span
35kN prototype
SFD based on
6m span
Reaction Frame
Reaction Frame
Hollow-core unit
300 series hollow-core unit
6000mm Horizontal
Support beam and
(beam elongation)
restraints
Actuator
Page A-1
(a) Supporting beam details
Floor Unit
D12 Starter
Bars (Grade 300)
(300mm spacing) Supporting Beam
Low friction
bearing strip
Page A-2
(a) Placement of the low friction bearing strip (b) Cutting of the hollow-core cells.
on the concrete seat.
(c) Supporting beam cast in restraints. (d) Formwork prior to concrete casting.
(e) End view of the loading connections. (f) Top view after concrete casting.
Page A-3
(a) De-bonding of the reinforcing. (b) View of the supporting beam reinforcement
and anchorage of the starter and hairpin bars.
(c) Top view of the reinforcement in supporting (d) Reinforcement details for the connection
beam and topping slab prior to concrete with the loading actuators at the end of the
casting. hollow-core unit.
Page A-4
A.2 MATERIAL TESTS
Tensile tests were conducted to determine the yield and ultimate strengths of the provided
concrete. A summary of the tensile test results are shown in Table A-1, a photograph showing
reinforcement test coupons are shown in Figure A-5 and stress strain relationships from the
Readymix concrete was used for the casting of the sub-assemblage elements. Standard
30MPa structural concrete mixes were specified with a 120mm slump and 19mm aggregate.
Table A-1 also includes the measured compression strengths ascertained by compression
tests.
Page A-5
500 500
450 450
400 400
350 350
Stress (MPa)
Stress (MPa)
300 300
250 250
200 200
150 150
100 100
50 50
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Strain Strain
700 700
600 600
Stress (MPa)
Stress (MPa)
500 500
400 400
300 300
200 200
100 100
0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0.2
Strain Strain
Page A-6
A.3.2 Specimen 1
(a) Continuity crack at the (b) Vertical crack between the (c) Top view of the interface
beam-slab interface at 0.5%. slab and beam at +0.5%. crack at +1%.
(d) Topping cracks at 2%. (e) Widening of the beam-slab interface crack
and exposure of the starter bars.
Page A-7
A.3.3 Specimen 2
(a) Continuity crack at the (b) Vertical crack between the (c) Vertical crack between the
beam-slab interface at 1%. slab and beam at -2%. slab and beam at +3%.
(d) Initial signs of concrete (e) Widening of the beam-slab (e) Large opening of the
seat spalling during the 3% interface crack at +4%. beam-slab interface crack at
cycles. -5% drift.
(f) Progression of spalling observed during the (g) Deterioration of the seat and exposure of
4% cycles. the bearing strip.
Figure A-9 Specimen 2 testing photographs
Page A-8
(a) Vertical dropping of the hollow-core floor (b) Hollow-core unit sliding off and down the
slab during the 5% cycles. damaged seat concrete.
(c) Damage to the topping concrete and extent (d) Failure of the test specimen with a loss of
of vertical dropping during the 5% cycles. seating occurring.
Page A-9
A.4 LOADING PROTOCOL
The following table provides a list of the loading input files for the sub-assemblage testing.
For each increment of vertical drift displacement, the calculated elongation is shown, and the
actual actuator target displacements were listed. Figure A-11 shows the interaction between
Horizontal
Vertical Ram Ram
Half Cycle Drift Amplitude Drift Amplitude Elongation Displacement Displacement
Count % rads mm mm mm
1 0 0 0.00 0.0 0.0
2 0.5 0.005 2.63 -28.5 -1.9
3 0 0 0.00 0.0 0.0
4 -0.5 -0.005 2.63 28.5 -2.8
5 0 0 0.00 0.0 0.0
6 0.5 0.005 2.63 -28.5 -1.9
7 0 0 0.00 0.0 0.0
8 -0.5 -0.005 2.63 28.5 -2.8
9 0 0 0.00 0.0 0.0
10 1 0.01 5.25 -57.0 -3.2
11 0 0 2.63 0.0 -2.6
12 -1 -0.01 7.88 57.0 -7.5
13 0 0 5.25 0.0 -5.2
14 1 0.01 7.88 -57.0 -5.8
15 0 0 5.25 0.0 -5.2
16 -1 -0.01 7.88 57.0 -7.5
17 0 0 5.25 0.0 -5.2
18 2 0.02 13.13 -114.0 -6.7
19 0 0 10.50 0.0 -10.5
20 -2 -0.02 18.38 114.0 -15.3
21 0 0 0.00 0.0 0.0
22 2 0.02 18.38 -114.0 -11.9
23 0 0 15.75 0.0 -15.7
24 -2 -0.02 18.38 114.0 -15.3
25 0 0 15.75 0.0 -15.7
26 3 0.03 23.63 171.0 -10.3
27 0 0 21.00 0.0 -21.0
28 -3 -0.03 28.88 -171.0 -20.6
29 0 0 26.25 0.0 -26.2
30 3 0.03 28.88 171.0 -15.5
31 0 0 26.25 0.0 -26.2
32 -3 -0.03 28.88 -171.0 -20.6
33 0 0 26.25 0.0 -26.2
34 4 0.04 34.07 228.0 -11.5
35 0 0 31.51 0.0 -31.5
36 -4 -0.04 39.32 -228.0 -23.5
37 0 0 36.76 0.0 -36.8
38 4 0.04 36.76 228.0 -36.8
39 0 0 39.37 0.0 -16.7
40 -4 -0.04 36.76 -228.0 -36.7
41 0 0 39.37 0.0 -23.6
Page A-10
300
200
Vertical Ram Displacement Control (mm)
100
0
-40 -35 -30 -25 -20 -15 -10 -5 0
-100
-200
-300
Horizontal Ram Displacement Control (mm)
Figure A-11 Input loading file interaction between the horizontal and vertical actuators.
(Note that negative displacements denotes retraction of the actuator)
A low cycle fatigue theory proposed by Dutta and Mander (2001), enables the plastic rotation
D L p 1 L p
Tp 0.16 1 2 N f
0.5
(A.1)
D' D 2 L
where Tp = plastic drift; L = lever arm of the cantilever column; D = overall member depth;
D = distance between outer layers of steel, 2Nf = the number of reversals to the appearance of
the first fatigue crack with which Nf is the effective number of constant amplitude cycles for a
Miners well-known cycle counting method can be utilised to calculate the equivalent
Page A-11
2
T
Nf i T
(A.2)
ref
Where T = drift for the ith cycle of loading and Tref = reference drift for an equivalent
The following calculations show how the solution was obtained for this case:
T
2
ncycles (T/Tref)
D= 375 mm 0.5 2 0.03125
D' = 280 mm 1 2 0.125
Lp = 150 mm 2 2 0.5
L= 6000 mm 3 2 1.125
Tref = 4 % 4 2 2
Ty = 0.7 % 6 3.7813
This compares well with the experimental observations, where the first fracture occurred at
3.83%.
A.6 REFERENCES
Dutta A and Mander J.B, 2001, Energy Based Methodology for Ductile Design of Concrete
Page A-12
Appendix B Super-Assemblage Construction Photographs and
Design Details
The following pages display the construction drawings of the super-assemblage frame
structure:
B-6 North-West Beam-Column Joint and Beams Precast Unit Plan and Elevation
B-7 North-East Beam-Column Joint and Beams Precast Unit Plan and Elevation
B-8 South-Centre Beam-Column Joint and Beams Precast Unit Plan and Elevation
Page B-1
Page B-2
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand SW Unit Plan Revision: A PU1
Scale: 1:25
Page B-3
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU2
SW Unit Elevations Scale: 1:25
Page B-4
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU3
SE Unit Plan Scale: 1:25
Page B-5
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU4
SE Unit Elevations Scale: 1:25
Page B-6
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU5
NW Unit Plan & Elevation Scale: 1:25
Page B-7
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU6
NE Unit Plan & Elevation Scale: 1:25
Page B-8
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU7
SC Unit Plan & Elevation Scale: 1:25
Page B-9
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU8
Beam Cross Section Details Scale: 1:20
Page B-10
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury
Precast Unit Details Date: 27/11/03
Christchurch, New Zealand SE & SW Column Tops Revision: A PU9
Scale: 1:25
Page B-11
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU10
SE & SW Column Bottoms Scale: 1:25
Page B-12
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury
Precast Unit Details Date: 27/11/03
Christchurch, New Zealand NE & NW Column Tops Revision: A PU11
Scale: 1:25
Page B-13
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU12
NE & NW Column Bottoms Scale: 1:25
Page B-14
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury
Precast Unit Details Date: 27/11/03
Christchurch, New Zealand SC Column Top Revision: A PU13
Scale: 1:25
Page B-15
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Precast Unit Details Date: 27/11/03
Christchurch, New Zealand Revision: A PU14
SC Column Bottom Scale: 1:25
Page B-16
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury
In-Situ Beam Lap Splice Details Date: 27/11/03
Christchurch, New Zealand Longitudional Beams (Beams 1 & 3) Revision: A BS1
Scale: 1:25
Page B-17
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury
In-Situ Beam Lap Splice Details Date: 27/11/03
Christchurch, New Zealand Transverse Beams (Beams 2 & 4) Revision: A BS2
Scale: 1:25
Page B-18
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Beam Lap Splice Date: 27/11/03
Christchurch, New Zealand Revision: A BS3
Cross Section Details Scale: 1:20
Page B-19
Department of Civil Engineering Drawn: CJM Drawing No.
University of Canterbury Back Frame Date: 01/03/04
Christchurch, New Zealand Revision: A BF1
Tie Beam Details Scale: 1:50
B.2 BUILDING CONSTRUCTION PHOTOGRAPHIC LOG
(a) North-west column base (b) Column bases prior to (c) Mid-span splice of east
reinforcement cage casting transverse beam
(c) South east beam column joint (d) Presence of ducting to allow connections of
reinforcement the load frame brackets
(e) Pouring of the column bases (f) Extrusion of the hollow-core units
Page B-20
(a) Lowering of the south-east precast unit (b) Use of a mobile crane for erecting the
precast frame
(c) Moving the north-west precast unit into (d) The frame in place
place
(e) Arrival of the hollow-core floor units (f) Floor units in place
Figure B-2 Erecting of the precast units and placement of the hollow-core floor units.
Page B-21
(a) Grouting of the column reinforcement (b) View of the starter bar reinforcement and
ducting and dry packing the precast interface backing board coverage of the un-filled cores
(c) Placement of the low (d) The supporting beam (e) View of the central column
friction bearing strip on the reinforcement darg bar reinforcement
concrete seat
(e) Topping reinforcement over the timber (f) Diaphragm topping concrete
infill
Figure B-3 Reinforcement detailing of the supporting beams and topping slab.
Page B-22
(a) Concrete pumping (b) Pouring of the concrete topping
(c) Vibration of the poured concrete (d) Manual floating and levelling of the
topping concrete
(e) Kelly float finishing of the topping concrete (f) Curing of the topping concrete
Page B-23
(a) Laboratory before super-assemblage (b) Laboratory after super-assemblage
construction construction
(c) Completed and painted super-assemblage (d) Back reaction frame and tie beam set-up
structure
Page B-24
Appendix C Super-Assemblage Testing and Photographic Log
New Zealand manufactured reinforcement was used for this experimental project. Table C-1
presents the reinforcing properties obtained from the tensile testing carried out using the
Avery testing machine in the Department of Civil Engineering Laboratory. Figure C-1 shows
Column & Beam Transverse Steel YR12 530 714 190 - 0.0005 0.0152
Support Connection R16 336 469 186 4.48 0.0031 0.0200
Starter Bars and Diaphragm D12 307 447 181 3.65 0.0029 0.0218
Drag Bars YD20 580 724 246 3.82 0.0012 0.0156
Page C-1
500 500
450 450
400 400
350 350
Stress (MPa)
Stress (MPa)
300 300
250 250
200 200
150 150
100 100
50 50
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Strain Strain
(a) D12 starter and topping reinforcement (b) R16 hairpin reinforcement
800
800
700
700
600
600
500
Stress (MPa)
500
Stress (MPa)
400 400
300 300
200 200
100 100
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Strain Strain
800
700
600
500
Stress (MPa)
400
300
200
100
0
0 0.05 0.1 0.15 0.2
Strain
Page C-2
C.1.2 Concrete Testing
Standard ready mix concrete was supplied for this project. The mixes were specified with a
30MPa 28-day compressive strength, 19mm maximum aggregate size and 100mm slump. For
the topping slab pour, a pump mix was used with a specified 28-day compression strength of
30MPa and 13mm maximum aggregate size. Compression tests were undertaken at 7-day, 28-
day and the initial day of testing. Table C-2 shows a summary of the compression test data
Page C-3
C.1.3 Hollow-core Properties
Hollow-core concrete units are cast on long casting bed in a single pour and cut to length
following curing and hardening of the concrete. Prestressing strands are tensioned prior to an
extrusion machine travelling along the casting bed extruding the hollow-core units using
zero-slump concrete. Several different types of hollow-core units are available which vary in
depth, with 200, 300 and 400mm depths available, and number of prestressing strands. A
cross section of a typical unit is shown in Figure C-3 and common hollow-core properties are
Page C-4
C.2 INSTRUMENTATION PHOTOGRAPHS
(a) Inclinometers to measure (b) Potentiometers under the (c) Vertical levelling of the
the column drift rotations. timber infill. floor surface.
(f) Sonic transducers measuring the lateral (g) Actuator at base of north west column with
(north-south and east-west) displacements of load-cell and rotary potentiometer.
the columns.
Figure C-4 Instrumentation photographs
Page C-5
(a) Instrumentation of the south-west (b) Instrumentation of the south-centre beam
longitudinal beam and beam column joint. column joint.
(c) View of the computers used for controlling (d) Data logging and controller boxes.
the experiment and data acquisition.
(e) Load frame set-up for longitudinal loading. (f) Load frame set-up for transverse loading.
Figure C-5 Instrumentation and load-frame photographs
Page C-6
C.3 TESTING PHOTOGRAPHIC LOG
(a) Cracking of the infill slab at +0.5% (b) Cracking of the transverse beam (north
east end) at +0.5%
(c) Diagonal cracking within the south centre (d) Continuity crack looking towards the north
beam column joint at 0.5% west column at 0.5%
(e) Cracking over the infill (f) Continuity crack at the (g) Diagonal torsion cracks,
slab at 1% west end at 1% north west corner at 1%
Figure C-6 Phase 1: Damage during the early stages
Page C-7
(a) Cracks extending over the timber infill slab (b) Hollow-core cracking at the south east
(south west bay) at +2% corner at +2%
(c) Web crack forming at south east corner (d) First instance of spalling noticed on east
unit at +2% transverse beam at -2%
(e) Continuity crack along the (f) Crack opening at the south (g) Longitudinal topping
beam-slab interface at the east column at +2% crack between 1st and 2nd
eastern end at +2% hollow-core unit
Figure C-7 Phase 1: Damage during the later stages
Page C-8
C.3.2 Phase 2: Transverse loading
(a) Spalling and exposure of the bearing strip (b) 13mm crack opening at the north east
of the east transverse beam at +1% column at +2%
(c) Large crack at the beam-column interface (d) Top view of the damage at the north east
of the south east corner at +2% corner of the structure at 3%
(e) Damage to the south west (f) Damage close to the (g) South east column
transverse hinge at +3% column face at the north east inclination at 3%
transverse beam end (+3%)
Figure C-8 Phase 2 Transverse loading damage
Page C-9
C.3.3 Phase 3: Longitudinal re-loading
(a) Distribution of topping cracking over the (b) Large crack opening at the south east
south east bay at 2% column at +3%
(c) View of the damage to the south east south (d) Exposure of the reinforcement and vertical
frame hinge at +3% dropping of the west transverse beam (north
end) at +3%
(e) Hollow-core soffit cracking (f) Damage due to torsional (g) Torsional hinging looking
at the east end at +3% hinging at +3% (south east) towards the north east column
Figure C-9 Phase 3: Damage during the early stages
Page C-10
(a) Crushing at the top west (b) Torsional behaviour of the (c) Interface cracking of the
side of the south centre east transverse beam at 3% south east hinge of the south
column at 3% beam at +4%
(d) West transverse beam (e) Damage of the north east (f) Damage of the north west
remaining upright at +4% hinge (east beam) at +4% hinge (west beam) at +4%
(g) Buckling of the bottom reinforcement of the (h) Top view of the damage at the south west
south west beam west hinge at 4% corner of the floor at 4%
Figure C-10 Phase 3: Damage during the middle stages
Page C-11
(a) Damage to the north east (b) Hollow-core soffit (c) Different rotations of the
hinge and evidence of the cracking, 2nd unit, west end at east beam and columns due to
beam dropping at 4% 4% torsion at +5%
(d) The plastic hinges (from west to east and left to right) of the south frame at +5% drift
(e) Evidence of hollow-core pull-off (un- (f) Increasing buckling of the reinforcement of
painted soffit) at +5% the south west beam hinge at 5%
Page C-12
C.3.4 End of Testing
(a) Fracture of the top, outside bar of the south (b) Fracture of the inside and outside top bars
east hinge of the south beams of the south east hinge of the south beams
(c) Cracking over the timber infill of the south east bay at the conclusion of testing
(d) Cracking over the timber infill of the south west bay at the conclusion of testing
Page C-13
C.4 CONCRETE TOPPING DELAMINATION
The occurrence of the concrete topping slab (75mm thick) delaminating from the hollow-core
floor units was recorded throughout testing. A hammer was used to find hollow sounding
areas indicative of delamination. As Figure C-12 shows, delamination was initially noticed at
the 2% peaks during Phase 1, and area of delamination gradually increased throughout the
remainder of testing. Delamination extended along the length of the building at the infill-to-
Unit 1 interface, along the beam-to-floor slab interfaces at the ends, and in the southern
Unit 4
Unit 3
Unit 2
Unit 1
Timber Infill
Page C-14
Appendix D Analytical Results
It is difficult to accurately determine the experimental yield drift, as there is no clear change
in stiffness of the super-assemblage, and it was seen that yielding occurred between around
0.5% drift and 0.7% drift. A yield value of 0.5% drift was used for the formulation of beam
elongation results. The following calculations show the contributions of the beam, column
and beam-column joint towards the calculated yield drift rotation. The calculations follow the
The following relationship from Priestley and Kowalsky (2000) was utilised to calculate the
contribution from the beam. The yield moment was derived from simple mechanics and actual
Hy
I yb 1.70 6.06 u 10 6 (D-1)
hb
where: Iyb = yield curvature of the beam, Hy = yield strain of the beam longitudinal
reinforcement and hb = height of the beam. Knowing the nominal moment capacity of the
beam and the previously calculated yield curvature of the beam, the effective stiffness can be
determined;
M yb 498 u 10 6
EI eff 8.22 u 1013 ( Nmm 2 ) (D-2)
I yb 6.06 u 10 6
where: (,eff = effective flexural stiffness and 0yb = nominal yield moment capacity of the
beam section. Using moment area theorem, the deflection of the beam can be found from the
following relationship:
Page D-1
M yb L2b 498 u 10 6 u 5350 2
Gb 14.45mm (D-3)
12 EI eff 12 u 8.22 u 1013
where: Gb = displacement of the beam at its point of inflection and Lb = distance between the
beam hinges. Finally, using the height of the column (Lc) this displacement can be converted
L2c
G cb Gb 16.58mm (D-4)
0.5 u Lb
To calculate the contribution to the yield drift from the beam column joint, an expression
As f y (d d ' ) L
Jj 38.5 1 0.00136rads (D-5)
BH Ec H Lb
where: Jj = joint distortion, % = width of the column, H = height of the column, (d-d) =
internal lever arm of the beam and L = centreline distance between beam inflection points.
Using this calculated rotation, the contribution of the joint can be found in terms of the
The contribution from the column can be calculated following a similar method used for the
beam. The yield curvature for the column can be determined from the following (Priestly and
Kowalsky, 2000):
Hy
I yc 2.12 7.56 u 10 6 (D-6)
hc
Page D-2
where: Iyc = yield curvature of the column, Hy = yield strain of the column longitudinal
reinforcement and hc = width of the column. The effective stiffness can be found using the
nominal moment capacity of the column, 0yc, and is shown by the following equation:
M yc 892 u 10 6
EI eff 1.18 u 1014 ( Nmm 2 ) (D-7)
I yc 7.56 u 10 6
where: 0c = the column moment at the beam face at yielding of the beam (223kNm) and
Thus, with all three constituent contributing sources calculated the overall yield drift
This calculated value shows good resemblance with the experimental yielding point. It is
likely that the experimental value is higher due to a small amount of cracking of the specimen
prior to testing as well as the cracking that occurred between the precast construction element
interfaces which occurred early in the test. The cracking would influence the effective
stiffness of the section, and the larger overall displacements of the super-assemblage due to
these cracks would result in the yield drift being higher than predicted.
Page D-3
D.2 SUPER-ASSEMBLAGE STIFFNESS
The initial elastic stiffness of the super-assemblage was calculated using the initial results for
each of the three loading phases, and is shown in Figure D-1. It can be seen that the initial
stiffness of structure during Phase 1 loading was 62MN/m (Figure D-1(a)). Figure D-2(b)
shows that the initial stiffness for Phase 2 loading was 34.7MN/m, which is a 44% reduction
of the stiffness exhibited at the start of testing. The cracking of the transverse beams, a large
degree due to torsion reducing the stiffness during Phase 1 loading, can explain this. Cracking
of the floor diaphragm over the timber infill also would reduce the stiffness for the transverse
loading. The stiffness from the early cycles during Phase 3 was 17.5MN/m (Figure D-1(c)).
This is only 28% of the Phase 1 stiffness and is due to the damage accrued from Phase 1 and
Phase 2 loading. The damage of the ends of the transverse beams consequently led to
torsional hinging of these beams, and due to this, the stiffness was significantly reduced. The
widening of the cracks around the north and south corner columns, and continuity crack
between the floor and beams along the east and west bays would reduce the overall super-
600 1500
1000
400 1000
500
200 500
Inter-Storey Inter-Storey Inter-Storey
Drift (mm) Drift (mm) Drift (mm)
0 0 0
-20 -10 0 10 20-30 -20 -10 0 10 20 30-100 -50 0 50 100
-200 -500
-500
Base Shear
Force (kN)
Base Shear
Base Shear
Force (kN)
Force (kN)
-400 -1000
-1000
-600 -1500
Page D-4
D.3 TORSION TEST
A torsion test was undertaken during Phase 2 transverse loading to attempt to determine the
torsional stiffness of the super-assemblage. The test involved displacing the east and west
bays in opposing directions as shown in Figure D-2. For example, positive drifts denote a
positive translation of the east bay (tops of the columns displace north) and negative
translation of the west bay (tops of the columns displace south) and vice versa for the negative
cycle. One reversing cycle to 0.5% was undertaken, and Figure D-3 shows the
corresponding base shear. This was completed after the second cycle to 2%, and a following
third cycle to 2% was undertaken to see if there was any degradation of stiffness due to the
torsion test.
North
Direction Direction
of movement West East of movement
South
S
Force (kN)
N S N
200
100
Page D-5
As Figure D-3 shows, the structure had a higher stiffness in positive drift direction, when the
respective tops of the columns translated northwards, than the negative direction. For
example, for the east bay, the stiffness values obtained from the experimental data for the
positive and negative drift cycles were 15.26MN/m and 10.56MN/m respectively, and similar
behaviour was observed for the west bay. The negative drift cycles correspond to negative
moments of the south east and west hinges, and due to the already highly cracked infill slab
section which under negative moments would be in tension, would require less force to reach
the target displacements. This can explain the lower stiffnesses observed during the negative
drift cycles. During the torsion tests, there was also little noticeable damage to the southern
longitudinal beams and after the resumption of the normal testing regime, there was no
The slipping of the longitudinal top and bottom beam reinforcement passing through the south
central beam column joint was monitored throughout the test by two potentiometers attached
to the reinforcement. Figure D-4 shows the bar slip experimental results for both Phase 1 and
3 longitudinal loading. Figure D-4(a) shows that there was up to 1.9mm of bar slippage
during Phase 1 and during Phase 3 more slipping occurred with a maximum bar slip of
3.82mm being recorded (Figure D-4(b)). The degradation of bond that would allow this bar
slippage would denote a reduction of the stiffness of the super-assemblage during low drift
displacements, which was observed to a small degree. The New Zealand Concrete Design
Standard (NZS 3101:1995) provide design equations to limit the maximum reinforcing size
Page D-6
passing through a beam column joint, which is based on ensuring that sufficient bond is
db f c'
d 3.3D f (D.10)
hc Do f y
where db = the diameter of bar passing through the joint, hc = height of the column section,
Df = factor to allow for one or two way framing, fc = compression strength of the concrete,
Do = overstrength factor for the beam reinforcement, and fy = yield strength of the
reinforcement. Application of this formula gives a maximum bar diameter of 23.5mm, lower
than the actual XD25 reinforcement used. Although this seems to be an oversight of the
design process, it is a smaller bar diameter than commonly used in large concrete frames and
the New Zealand Concrete Code offers another form of the equation:
db D tD p f c'
d 6 D f (D.11)
hc Ds Do f y
where Dt = factor for the amount of fresh concrete cast below the reinforcement, Dp = factor to
allow for the beneficial effects of axial compression forces, and Ds = factor for severe bond
conditions. In this case, these factors are all 1, and the maximum allowable reinforcement
diameter becomes 42mm, and thus the 25mm bars used are suitable, from a design point of
3 3
2 2
1 1
Slip (mm)
Scan N umb er
Slip (mm)
Scan N umb er
0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000 6000
-1 -1
-2 -2
Top
Top
-3 Bottom -3
Bottom
-4 -4
(a) Longitudinal bar slip during Phase 1 (b) Longitudinal bar slip during Phase 3
Figure D-4 Longitudinal beam bar slippage through the central beam column joint
Page D-7
D.4.2 Hollow-core Floor Unit Pull-off
Sprung loaded potentiometers were installed underneath each of the hollow-core units to
measure the horizontal sliding displacements of the floor units relative to the supporting
beam. Figure D-5 shows the relationship between the horizontal pull-off movements and
inter-storey drift. Not surprisingly, and analogous to the beam elongation theoretical
predictions, the amount of pull-off increases with increasing inter-storey drift amplitudes and
is higher for the second cycles. For Phase 1, both the east and west bays showed similar
behaviour, and maximum pull-offs of 12mm and 11.4mm were seen at the second cycle to
12
(mm)
10
2 East
Drift (%) West
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
12
10
6
Displacement
4
(mm)
East
2
West
Drift (%)
0
-4 -3 -2 -1 0 1 2 3 4
(b) Horizontal hollow-core unit pull-off for Phase 3 loading.
Page D-8
16 14
Unit 1East Unit 1West
14 12
Unit 2 East Unit 2 West
Unit 3 East Unit 3 West
12
Displacement (mm)
Displacement (mm)
10 Unit 4 West
10
8
8
\
6
6
4
4
2 2
Scan N umb er Scan N umb er
0 0
0 1000 2000 3000 4000 0 1000 2000 3000 4000
(a) East beam pull-off: Phase 1 (b) West beam pull-off: Phase 1
12 12
10 10
Displacement (mm)
Displacement (mm)
8 8
6 6
4 4
Unit 1 East Unit 1 West
2 Unit 2 East 2 Unit 2 West
Unit 3 East Unit 3 West
Scan N umb er Unit 4 West Scan N umb er
0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
(c) East beam pull-off: Phase 2 (d) West beam pull-off: Phase 2
18 18
16 16
14 14
Displacement (mm)
Displacement (mm)
12 12
10 10
8 8
6 6
Unit 1 West
4 Unit 1 East 4 Unit 2 West
Unit 2 East Unit 3 West
2 Unit 3 East 2 Unit 4 West
Scan N umb er Scan N umb er
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
(d) East beam pull-off: Phase 3 (e) West beam pull-off: Phase 3
Figure D-6 shows the horizontal pull-off displacement measured underneath all of the hollow-
core units, related to time, for each of the three loading phases. The fluctuations show the
effect of the cyclic loading, and for Phase 1, the large increases of the pull-off displacement
Page D-9
signify new inter-storey drifts being experienced. The pull-off was seen to be minimal for
Phase 2, and the slight increase of the pull-off, of less than 1mm, can be explained by
torsional behaviour of the transverse beams. The Phase 3 pull-off relationships indicate a
decreased rate of pull-off, as compared with Phase 1 where the Phase 3 contributed a further
4mm to the already 12mm from Phase 1, and the overall maximum pull-off experienced was
Figure D-7 shows a schematic view of the instrumentation used for the beam plastic hinge
and beam-column joint regions. It was hoped that the set-up used would be able to measure
the relative contributions towards the overall rotation, which would equate to the column
inclination. Potentiometers were arranged in such a fashion that joint shear, hinge shear, hinge
flexure, shearing at the beam-column interface and fixed end rotation could be monitored.
Figure D-8 shows the relative components contributing towards the overall lateral
175mm 700mm
Potentiometers (top and bottom)
Beam-Column Joint Potentiometer measuring the fixed-end rotation
instrumentation - measuring and curvature of the beam plastic
100mm
joint shear deformation 75mm
500mm hinges
100mm
500mm
600mm
600mm
500mm
500mm
Figure D-7 Close-up representation of the beam-column joint and hinge instrumentation
Page D-10
0.025
Rotation Contribution
0.02
Closure Error
(rads)
0.015
0.01
0.005
Inter-storey Drift (rads)
0
-0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025
-0.005
Fixed End Rotation
-0.01 Hinge Flexure
Hinge Shear
-0.015
Joint Shear
Closure Error -0.02
Column Inclination
Total
-0.025
(a) Contributions towards lateral displacement during longitudinal Phase 1 loading
0.025
Rotation Contribution
0.015
0.01
0.005
Inter-storey Drift (rads)
0
-0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025
-0.005
Fixed End Rotation
-0.01 Hinge Flexure
Hinge Shear
-0.015
Joint Shear
-0.02 Column Inclination
Closure Error Total
-0.025
Figure D-8 Components that contribute to the overall lateral displacement of the super-
assemblage
By comparing the total rotations with the column inclination in the relationships shown in
Figure D-8, it can be seen that the instrumentation was able to measure the overall lateral
displacements to a satisfactory level. The closure error indicates the discrepancy, and for the
transverse loading as seen in Figure D-8(b), the instrumentation slightly over-predicts the
amount of lateral displacement of the structure. In terms of the individual components, the
Page D-11
fixed end rotation, measured between the column and beam at the top and bottom appeared to
dominate the rotation contributions. For Phase 2, almost all the lateral displacement occurs
from fixed end rotation with a smaller amount due to plastic hinge flexure. This matches well
with the experimental observations whereby damage to the plastic hinges was predominantly
a large crack at the beam-column interface. As it was expected, there were minimal
contributions from the beam hinge and beam column joint shearing displacements.
At several locations within the super-assemblage topping and floor-to-beam connections, the
reinforcement was strain gauged to be able to monitor the demands imposed on the
reinforcement. One main reason was to monitor the strain development along the length of the
starter bars at the end connections to give an idea as to the possible required length of the
starter bars. This meant that strain gauges were attached at several points up to 2.1m from the
supporting beams. Figure D-9 shows the locations of the strain gauges present within the
structure. Figures D-10 and D-11 provide the strain results for the end connection reinforcing
(starter bars, lapped diaphragm steel and the hairpin bars) for the east and west ends during
Phase 1. Figure D-12 shows the results, for Phase 1 and 2, of the southern beam starter bars
that were cast in the infill topping slab and were embedded half way into the southernmost
hollow-core unit. Two starter bars were strain gauged, one 1m either side of the central
column, approximately where the maximum displacement incompatibility between the floor
and beams was expected. Also, the drag bars that were cast into the south central column
and lapped with reinforcement running across the width of the building, were strain gauged at
various points as shown in Figure D-9. Figure D-13 shows the results obtained from the drag
Page D-12
North
Unit 1 Unit 1
West Infill East Infill
Starter Starter
South
1 3
2 4 5
1 2
300
Hairpin reinforcement
300mm 375mm
375mm
4 3 2 1
(c) Strain gauging of the starter bars passing over the timber infill
Page D-13
12 45
Interface 40 Interface
10 300mm 300mm
900mm 35 900mm
1500mm 1500mm
8 2100mm 30 2100mm
Yield Strain Yield Strain
strain (%)
strain (%)
25
6
20
4
15
2 10
(a) Starter bar strain data for Unit 4 (b) Starter bar strain data for Unit 3
12 18
Interface 300mm 16 Interface
10 900mm 1500mm 300mm
2100mm Yield Strain 14 900mm
1500mm
8 12 2100mm
Yield Strain
strain (%)
6 strain (%)
10
8
4
6
2 4
(c) Starter bar strain data for Unit 2 (d) Starter bar strain data for Unit 1
45 45
40 40
35 35
30 30
Unit 2 Interface
Unit 4 Interface Unit 2 300mm
strain (%)
strain (%)
10 10
5 5
Scan Num ber Scan Num ber
0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
-5 -5
(e) Hairpin bar strain data, Units 3 & 4 (f) Hairpin bar strain data, Units 1 & 2
Figure D-10 Strain data for the west slab-beam connection reinforcement
during Phase 1
Page D-14
45 45
40 Interface 40 Interface
300mm 300mm
35 900mm 35 900mm
1500mm 1500mm
30 2100mm 30 2100mm
Yield Strain Yield Strain
strain (%)
strain (%)
25 25
20 20
15 15
10
10
5 Scan N umb er
5 Scan N umb er
0
0
0 1000 2000 3000 4000 5000
0 1000 2000 3000 4000 5000
-5
-5
(a) Starter bar strain data for Unit 4 (b) Starter bar strain data for Unit 3
2.5 45
40 Interface
300mm
2 35 900mm
Interface
300mm 1500mm
30 2100mm
900mm Yield Strain
1.5 1500mm
strain (%)
strain (%)
2100mm 25
Yield Strain
1 20
15
0.5 10
5 Scan N umb er
0 Scan Num ber
0 1000 2000 3000 4000 5000 0
0 1000 2000 3000 4000 5000
-5
-0.5
(c) Starter bar strain data for Unit 2 (d) Starter bar strain data for Unit 1
45 45
40 40
35 Unit 4 Interface Unit 2 Interface
Unit 4 300mm 35 Unit 2 300mm
Unit 3 Interface Unit 1 Interface
30 Unit 3 300mm 30 Unit 1 300mm
Yield Yield
strain (%)
strain (%)
25 25
20 20
15 15
10 10
5
Scan Num ber 5
Scan Num ber
0
0 1000 2000 3000 4000 5000 0
0 1000 2000 3000 4000 5000
-5
-5
(e) Hairpin bar strain data, Units 3 & 4 (f) Hairpin bar strain data, Units 1 & 2
Figure D-11 Strain data for the east slab-beam connection reinforcement during Phase 1
Page D-15
30 25
Beam/Infill Beam/Infill
25 Mid-Infill Mid-Infill
20
Infill/Unit 1 Infill/Unit 1
20 Unit 1 Unit 1
Yield Strain 15 Yield Strain
strain (%)
strain (%)
15
10
10
5
5
-5 -5
(a) Strain data for the south west starter bar (b) Strain data for the south east starter bar
(1m west of south central column) during Phase (1m east of south central column) during Phase
1 longitudinal loading 1 longitudinal loading
30 15
Beam/Infill Beam/Infill
20 Mid-Infill 10 Mid-Infill
Infill/Unit 1 Infill/Unit 1
Unit 1 Unit 1
10 Yield Strain 5 Yield Strain
strain
strain
0 0
0 1000 2000 3000 4000 5000 6000 7000 8000 0 1000 2000 3000 4000 5000 6000 7000 8000
Scan Num ber Scan Num ber
-10 -5
-20 -10
-30 -15
(c) Strain data for the south west starter bar (d) Strain data for the south east starter bar
(1m west of south central column) during Phase (1m east of south central column) during Phase
2 transverse loading 2 transverse loading
Figure D-12 Strain data for the starter bars passing over the timber infill
As Figures D-10 and D-11 show, significantly higher strains were experienced at the beam-to-
floor slab interface than in the topping slab, where the strains diminished rapidly. Similar
results were observed for each of the floor units and at both the east and west ends. These
results confirm the both the desired and observed results whereby all of the rotational demand
was centred at the support beam-to-floor interface. The lack of strain penetration would
signify low stresses which was matched by a lack of cracking within the topping away from
the supporting beam. The results for the hairpin bars were less reliable due to malfunctioning
Page D-16
strain gauges, although it can be seen that there was higher strains at the beam-floor interface
Figure D-12(a) and (b) shows that the maximum strains experienced by the south
beam starter bars occurred at the interface between the infill slab and hollow-core floor
Unit 1. This can be explained by the vertical deformations caused by the relative displacement
differences between the floor slab and southern beams and the resultant vertical angle at this
interface. Figure D-13 shows there was a small degree of strain penetration along the drag
bars tying the central column into the floor. Higher strain demand was experienced at the
south centre column. Due to the fact that the southern beam starter bars were also anchored
into the flooring, and that no fracture or tear occurred, the demand on the tie bars was
minimal.
4 6
South Centre Column South Centre Column
3.5 Infill/Unit 1 Infill/Unit 1
5
Unit 1/Unit 2 Unit 1/Unit 2
3 Unit 2/Unit 3 Unit 2/Unit 3
Yield Strain 4 Yield Strain
2.5
strain (%)
strain (%)
3
2
1.5 2
1 1
(a) Strain data for the west drag bar during (b) Strain data for the east drag bar during
Phase 1 longitudinal loading Phase 1 longitudinal loading
6 8
South Centre Column South Centre Column
5 Infill/Unit 1 Infill/Unit 1
Unit 1/Unit 2 6 Unit 1/Unit 2
4 Unit 2/Unit 3 Unit 2/Unit 3
Yield Strain 4 Yield Strain
3
2
strain
strain
1 0
0 1000 2000 3000 4000 5000 6000 7000 8000
0 Scan Num ber
0 1000 2000 3000 4000 5000 6000 7000 8000 -2
-1 Scan Num ber
-4
-2
-3 -6
(c) Strain data for the west drag during Phase 2 (d) Strain data for the east drag during
transverse loading Phase 2 transverse loading
Figure D-13 Strain data for the drag bars tying in the south centre column
Page D-17
D.5 WEST TRANSVERSE BEAM TORSION
Five inclinometers measuring rotation in the longitudinal direction were spaced along the
west transverse beam at constant intervals as shown in Figure D-14. Figure D-15 shows the
relative torsion rotations between the column and beam for Phase 3, and Figure D-16 provides
6
5 Phase 1 Phase 3
Angle of Twist (rad/m)
4
3 Loading Loading
2
1
0
-1
-2
-3
-4
-5
-6
-7
-8
Figure D-15 Torsional twist of the south hinge of the west transverse beam
(radians/metre in plastic hinge zone)
Page D-18
2.5
Jo int
2
1
1.5
Torsion Drift (%)
3
1 4
0.5
S c a n N um be r
0
-0.5 0 500 1000 1500 2000 2500 3000 3500 4000 4500
-1
-1.5
-2
-2.5
4
2
5
S c a n N um be r
0
0 1000 2000 3000 4000 5000 6000 7000 8000
-2
-4
-6
For the present experiment, the torsional demand was caused by the moment from the
connection reinforcement between the supporting beam and floor slabs, which occurs under
lateral loading of the structure. The strength of this connection was found during the sub-
assemblage tests undertaken as part of this research, where calculated moment strengths
matched the experimental values well. The positive and negative moment capacity values
were 38kNm and 40kNm respectively for each hollow-core unit. Accounting for the 1.2m
hollow-core unit widths, the torsional moment per metre length of beam is 32kNm/m and
33kNm/m for the positive and negative rotations, which then corresponds to an applied
torsion at the beam ends of 79kNm approximated for both positive and negative rotations.
Page D-19
For the connection detail tested by Lindsay (2004), the presence of the compressible backing
board meant that there was effectively zero moment resistance. Therefore, the torsional
demands arose due to the eccentric loading of the floor. The weight of hollow-core flooring
was taken to be 3.5kPa, the weight of the topping concrete 24kNm-3 and the eccentricity from
the centre of the beam to the centre of the seat was 238mm. This caused a torsional moment
For compatibility torsion, as is the case in this experiment, a reduction of the torsional
moment can occur due to the redistribution of internal forces when cracking first occurs. An
expression from the New Zealand Concrete Code, NZS 3101:1995, was used for calculating
the cracking torsional moment of the beam section, and is the same as the formulation given
by Mitchell and Collins (1991). The cracking torsion, or maximum design torsional moment
f c' Acp2
T *
I (D.12)
3 p cp
in which T* = design torsional moment force at section at the ultimate limit state, or cracking
torsion, I = strength reduction factor (which is omitted in this case for analysis), fc = concrete
compressive strength, Acp = area enclosed by the outside perimeter of the concrete cross-
section, and pcp = outside perimeter of the concrete cross section. Knowing that the overall
concrete cross-section was 400mm wide by 750mm deep, and concrete compressive strengths
obtained from material tests were 35MPa, the torsional cracking moment was 77.2kNm. In
this case, the strength reduction factor was omitted, as this is an analysis of a given section,
not design.
Page D-20
After diagonal cracking has occurred, a truss mechanism of vertical transverse
reinforcing ties and diagonal concrete compression struts forms, and the contribution to the
torsional resistance from the reinforcement can be calculated. Park and Paulay (1975),
provides an expression for the steel contribution to resisting torsion as a function of the
transverse and longitudinal reinforcement, which is shown as Equation D.13. The equivalent
design expression for the nominal torsional strength from NZS 3101:2004 relates to the
amount of transverse reinforcing and stems from information provided by Park and Paulay
(1975).
At f y Al f ly
Ts 2 A0 (D.13)
s P0
in which Ao = area enclosed by the connecting lines between the centres of the longitudinal
reinforcement, fly = yield strength of the longitudinal reinforcing, and Po = perimeter formed
by area Ao.
Here, the interior cross section dimensions of the beam between the centre-lines of the
longitudinal reinforcing are 286mm wide and 582mm deep. The stirrup sets used were XR12
with an leg area of 113mm2 and the longitudinal reinforcement comprised of 3 XD25 bars top
and bottom with a total area of 2946mm2. The stirrup sets were spaced at 150mm and the
yield strengths obtained from material tests for the transverse and longitudinal reinforcing
were 530MPa and 536MPa respectively. This produced a torsional steel contribution, Ts, of
200kNm.
To calculate the torsional twists corresponding to the cracking and ultimate torsion
moments, relationships from Mitchell and Collins (1991) and Park and Paulay (1975) for the
torsional stiffness were used. In this simplified approach, the section is represented by an
Page D-21
equivalent thin walled tube or hollow section with the same overall dimensions. The pre-
cracking stiffness was a function of the section geometry and the post-cracking torsion
stiffness is governed by the deformation of the reinforcing. The pre-cracking stiffness was
calculated using:
4 A02 t
GK G (D.14)
po
Where GK = the pre-cracking torsional stiffness, G = the shear modulus, A0 = area enclosed
by the centreline of the thin walled tube with thickness, t, and po = perimeter of the area. The
3 Acp
t (D.15)
4 p cp
Where Acp = area enclosed by the outside perimeter of the concrete cross section, pcp =
perimeter of the area Acp. Utilisation of Equations D.14 and D.15 produce the following
result:
A ratio of the post-cracking stiffness to pre-cracking stiffness was calculated using methods
included in Park and Paulay (1975). Using the aspect ratio of the section, and a modified
reinforcing ratio expression, the post-cracking stiffness was found to be 4% of the pre-cracked
stiffness (3821kNm2). It is difficult to accurately assess the post cracking torsional stiffness of
a concrete section, and the effect of flexural plastic hinging during Phase 2 loading would
have had a large effect. The deteriorated state of the concrete within the plastic hinge zone
and cyclic loading with which the reinforcing had been subjected to, means that the post-
From the calculated values for the torsional strengths and stiffnesses, the
corresponding twist rotations can be found. The twist rotations were multiplied by the length
Page D-22
of the plastic hinge zone to produce rotations in terms of radians. Considering the torsion
demands and capacity values calculated, Figure D-17 could be generated. The contrast
between the initial stiff pre-cracked response and the post-cracked response can be seen in
Figure D-17. Following Phase 2, the ends of the transverse beams were badly damaged. Thus,
the post-cracked response starts from zero torsional twist in Figure 5-17 and as there is no
concrete contribution due to the state of damage, the ultimate strength is that of the steel only,
as calculated by Equation D.13. The intersection of the torsional demand values gives an idea
as to the torsional twist that would occur, and the trends, more so than the specific values, are
300
250
Torsion Moment (kNm)
200
150
100 Phase I
Phase III
50 Current Experiment Demand
Lindsay (2004) Experiment Demand
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Torsional Twist (rads/m)
The relative displacements between the south longitudinal beams and the southernmost
underneath the timber infill topping slab. In addition to this, levelling was undertaken at the
first cycle peak inter-storey drifts during testing, and this was able to give both the relative
displacements, but also a fixed datum to indicate the vertical displacements with relation to
the ground. The following Figures D-18, D-19 and D-20 show the displaced shapes for the
Page D-23
south beams, the first hollow-core unit (Unit 1) and the relative displacement incompatibility
between the floor and beams. A negative displacement incompatibility equates to the flooring
6
4
2
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
-2
-4
-6
-8
-10
Distance along structure (m m )
Slab -2%
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
-2
-4
-6
Distance along structure (m m )
-2%
4
2
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
-2
-4
-6
-8
-10
-12
Distance along structure (m m )
(c) Vertical displacement incompatibility between the perimeter beams and floor
Figure D-18 Displacement of the south beams and floor slab during Phase 1
Page D-24
4
Beam 1%
3 Beam -1%
Beam 2%
2 Beam -2%
Slab 3%
Slab -3%
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
-5
-10
-15
-20
Distance along structure (m m )
3%
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
-5
-10
-15
Distance along structure (m m )
(c) Vertical displacement incompatibility between the perimeter beams and floor
Figure D-19 Displacement of the south beams and floor slab during Phase 2
Page D-25
20
Beam 2%
Beam -2%
15 Beam 3%
Beam -3%
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
-5
-10
-15
-20
Distance along structure (m m )
-5
Vertical Displacement (mm)
-10
-15
Slab 2%
Slab -2%
-20 Slab 3%
Slab -3%
Slab 4%
Slab -4%
-25
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000
Vertical Displacement (mm)
-5
-10
-15
-20
2%
-25 -2%
3%
-3%
-30 4%
-4%
-35
(c) Vertical displacement incompatibility between the perimeter beams and floor
Figure D-20 Displacement of the south beams and floor slab during Phase 3
Page D-26
The displacement profiles for the beam and floor slab for Phase 1 loading show expected
results. The beams deform in double curvature, and the floor slab in single curvature over the
total span of the building. Figure D-18(c) shows the displacement incompatibility between the
beams and floor, where the maximum relative displacement was 10mm when the flooring was
lower than the frame beams. Reasonably symmetric behaviour was observed between the
positive and negative cycles, and the maximum displacement incompatibility was seen to
occur around 1m either side of the central column (6100mm). Similarly, the results from
Phase 2 (Figure D-19) were expected. The displacement incompatibility relationship shows
that the flooring was a maximum of 5mm higher than the beams during negative drifts when
the transverse beams at the south end were inclining upwards, and during the positive drift
cycles, the flooring was a maximum of 14mm lower than the frame beams.
The results for Phase 3 as seen in Figure D-20 show interesting results. The deformed
shape of the beams at various drift amplitudes shows an expected curvature pattern, but the
displacement of the floors shows a distinct drop of over 16mm. During the experiment, it was
observed that the transverse beams began to drop during Phase 2 loading. Most of the rotation
within the plastic hinges at the end of the transverse beams was centred at the beam-column
interfaces where a significant crack had formed. During the later stages of Phase 2, it was
noticed that the beams were dropping and a vertical dislocation was occurring at these large
end cracks. This is displayed by the vertical displacement results whereupon comparing Phase
1 and 3 (Figure 5-18(b) and 5-20(b)) the amount of the dropping of the floor units is noticed.
Despite this, and as Figure 5-20(c) shows, the same trends of displacement incompatibility
were seen, but the vertical dropping meant that the floor remained predominantly lower than
the southern beams. At the floorings units highest point, the floor was 1mm higher than the
beams, and at its lowest, 33mm below the beams. The maximum incompatibility results were
Page D-27
D.6.2 Analytical Modelling
The analysis involved the construction of an analytical model in the inelastic time history
dimensional model was been set up to model the longitudinal lateral loading of the front
frame, and consists of members representing the frame and floor elements and springs
connecting the floor to the frame. Different spring members were used to model the level of
fixity of the end support conditions, and along the frame between the frame and flooring.
Sensitivity analyses were performed examining the effect of various parameters; different end
support conditions, different lateral connection details, different bay lengths and numbers of
bays and different flooring types. The effects of the different parameters were seen by
examining both the displacement incompatibility patterns as well as the overall system base
shear capacities.
From the sensitivity analyses, it was seen that the end connection detail, through the
stiffness of the end springs, had a large influence on system strength capacity and
degree by the higher ultimate base shear capacity of the current experimental structure than
3500
3000
FIXED k=9.9x108
2500
k=1x108
Base shear (kN)
2000
k=1x107
1500
PINNED k=0,1,1000,100000
1000
500
0
0.0 1.0 2.0 3.0
Inclination (%)
Figure D-21 Modelling the effect of the end connection detail on the base shear capacity
Page D-28
Modifying the side spring characteristics was done by altering the spring stiffnesses or the
yield point representing the point at which cracking or tearing would occur. Adjusting the
stiffness was seen to have a large effect on the displacement compatibility trends, with the
obvious increased deformations for softer springs. The more stiff the springs also
corresponded to an increase in the strength capacity of the system due to the greater ability to
transfer the floor inertia forces to the frame. Figure D-22 shows the effect on the base shear
capacity of the structure for various side connection detail stiffness values.
3500
RIGID k=9.9x108
3000 8
k=1x10
k=1x107
2500
k=0,1,1000 &100000
Base shear (kN)
"SOFT"
2000
1500
1000
500
0
0.0 1.0 2.0 3.0
Inclination (%)
Figure D-22 Modelling the effect of the side connection detail on the base shear capacity
Different bay lengths, using 4m, 6m and 8m showed no major effect on the displacement
trends. Altering the floor system types for 200, 300 and 400 series hollow-core was done by
changing the member stiffness to 0.35EI, 1EI and 2.2EI respectively. This was seen to have a
significant effect with the most severe displacement incompatibility values observed for the
stiff 400 series floor. This can be explained by the fact that the more flexible 200 and 300
series flooring is able to deform with the beams in a double curvature manner (within the
model, but not in reality), whereas the stiff nature of the 400 series means the floor deforms in
Page D-29
Figure D-23 shows the displacement incompatibility results from the experimental tests and
modelling for both the Lindsay (2004) and current experiments. The lateral spring properties
modelling the infill connection were kept the same, but the end spring properties were altered
to model the pinned type seating connection for Lindsay (2004) and the rigid connection in
the present experiment. The results show promising resemblance between the experiment and
analytical model. The overall deformation trends are matched although the model slightly
underestimates the amount of incompatibility that occurs. For the rigid end conditions, the
maximum displacement incompatibility occurred on both sides adjacent to the central column,
as the floor unit deformed in single curvature due to the fixity at the ends. For the pinned end
conditions, which allowed the floor to simply sag, displacement incompatibility peaks were
seen close to the supports as well as either side of the central column, and the displacement
incompatibility values were generally higher than for the fixed end case.
10
8
6
4
Displacement (mm)
0
-2 0 2000 4000 6000 8000 10000 12000
-4
-6 Experiment 1%
-8 Model 1%
Experiment 2%
-10 Model 2%
-12
2
Position along fram e (m m )
0
-2 0 2000 4000 6000 8000 10000 12000
-4
-6 Experiment 1%
-8 Model 1%
Experiment 2%
-10 Model 2%
-12
(b) Observed and modelled displacement incompatibility from the current experiment
Figure D-23 Observed and modelled displacement incompatibility
Page D-30
D.7 BEAM ELONGATION
The phenomenon of beam elongation is explained in Sections 2 and 3 of this thesis. Matthews
occurs. The assumption of the eccentricity values, ecr, is a key part of the theoretical
predictions, and the same values used by Matthews (2004) and later by Lindsay (2004) were
used in these calculations. The following Figures D-24, D-25 and D-26 plot the observed
beam elongation results as well as the predicted response for the three loading phases of the
experiment.
14 14
12 Elongation (mm) Experiment 12
Elongation (mm)
10 Theo ry 10
8 8
6 7 6
Experiment 4 4
Theo ry 2 2
0 0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Int e r- S t o re y D rif t ( %) Inter-Storey Drift (%)
(a) South west plastic hinge elongation (b) South centre west hinge elongation
14 20
18
12
Elongation (mm)
Elongation (mm)
16
10 14
12
8
10
7 6 8
Experiment 6
4
4
Experiment Theo ry
2 2
Theo ry
0
0
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Int e r- S t o re y D rif t ( %) Int e r- S t o re y D rif t ( %)
(c) South centre east hinge elongation (d) South east hinge elongation
45
40
Elongation (mm)
35
30
25
20
15
Experiment 10
Theory 5
0
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Inter-Storey Drift (%)
Elongation (mm)
Elongation (mm) 16 16
14 14
12 12
10 10
8 8
6 6
Experiment 4 Experiment 4
Theo ry 2 Theo ry 2
0 0
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
Inter-Storey Drift (%) Int e r- S t o re y D rif t ( %)
(a) South west plastic hinge elongation (b) North west hinge elongation
20 20
18 18
Elongation (mm)
Elongation (mm)
16 16
14 14
12 12
10 10
8 8
6 6
Experiment 4 Experiment 4
Theo ry 2 Theo ry 2
0 0
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
Int e r- S t o re y D rif t ( %) Inter-Storey Drift (%)
(c) South east hinge elongation (d) North east hinge elongation
Elongation
35
(mm)
30
25
20
15
10
Phase II East Bay
5 Phase II West Bay
Theory
0
-4 -3 -2 -1 0 1 2 3 4
Inter-storey Drift (%)
100
80
60
40
Phase I
20 Phase III
Theory
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
Inter-storey Drift (%)
Page D-32
Theoretical Predictions: Longitudinal Loading
Yield Drift
0.5 %
0.005 rad
ecr: External Hinges
Distance (mm) ecr*Ty
0.425D = 318.75 1.59375
0.475D = 356.25 1.78125
ecr: Internal Hinges
Distance (mm) ecr*Ty
0.225D = 168.75 0.84375
0.275D = 206.25 1.03125
Page D-33
Theoretical Predictions: Transverse Loading
Yield Drift
0.5 %
0.005 rad
ecr: External Hinges
Distance (mm) ecr*Ty
0.425D = 318.75 1.59375
0.475D = 356.25 1.78125
A low cycle fatigue theory proposed by Dutta and Mander (2001) enables the plastic rotation
life capacity of reinforcement to be calculated. The theory has been outlined in Section 2 and
and Section 4 to assess the fracture of longitudinal reinforcement of the main beams during
the super-assemblage experiment. The following calculations outline the derivation of the
Page D-34
The equivalent number of equi-amplitude cycles to the reference drift (5%) experienced
Tref= 5
Nmax= 5.12
q ncycles (T/Tref)2
0.5 3 0.03
1 2 0.08
2 3 0.48
3 2 0.72
4 2 1.28
5 2 1.9
6 4.49
This calculation shows that the reinforcement withstood 4.5 equivalent cycles (NF-Expt) to 5%
during the experiment before fracture occurred. By re-arranging the low cycle fatigue
equation from Dutta and Mander (2001), the theoretical equivalent number of 5% cycles,
NF-Theory, that the reinforcement will be able to sustain prior to fracture can be found from the
following:
2
D L p 1 L p
0.16 1
1 D' D 2 L
N F Theory (D-16)
2 T T y
2
750 511 1 511
0.16 1
N F Theory
1
636 750 2 2675 3.6
2 0.05 0.0064
Therefore, the theoretical number of cycles at an amplitude of 5% required to cause low cycle
Page D-35
D.9 OBSERVED BOWSTRING EFFECT
To account for the presence of the floor slab in calculating the overall strength of the
structure, simple calculations were used to quantify the bowstring effect. Beam elongation is
restrained due to the presence of the floor slab, and this restraint induces large compression
forces within the longitudinal beams. To calculate this compression force, a simplified
approach is used which calculates the eccentricity or depth of the compression arch. From
this, the magnitude of the tension force can be found which is a function of the capacity of the
reinforcement of the hollow-core end connections, the friction of the hollow-core seating and
the amount of diaphragm steel orthogonal to the tension tie. The calculation is as follows:
wl 2 Pl
Te (D.16)
8 4
where e = midspan eccentricity of the tie force with respect to the centroid of the compression
strut, T = combined tie force, l = overall span of arch, P = the tension force provided by the
drag bar reinforcement and w = uniformly distributed load given by the diaphragm reinforcing
orthogonal to the tension tie. As the tie force, T, is a function of the tension reinforcement
2eAs f y
T 2 PWle (D.17)
s
where As = end connection reinforcing area (starter bars and hairpin bars), fy = end connection
for the bearing strip and W = weight of the hollow-core. The uniformly distributed load given
by the diaphragm reinforcing orthogonal to the tension tie, w, can be found by:
Asd f yd
w (D.18)
s sd
where Asd = area of the diaphragm steel, fyd = yield strength of the diaphragm reinforcing and
Page D-36
Using this formulation, and the measured material properties of the reinforcement, an
eccentricity of 2.5m was found. This matches well with the cracking pattern witnessed
throughout testing. The eccentricity value allowed a tension force of 1254kN to be calculated.
By equilibrium, this tension force is resisted by a compression force of the same magnitude,
and this compression force has a significant effect on the capacity of the system and notably
The compression axial force was taken to act at the level of the topping slab and
further simplified to occur at the top extreme fibre. Therefore, the moment capacity of both of
yield strength of tension reinforcement; d = the distance from the extreme compression fibre
to the centroid of the tension reinforcement; d = the distance from the extreme compression
fibre to the centre of the compression block; and N = the axial compression force due to the
bowstring effect. Using this formulation, the moment demand on the interior column, as
found by the sum of the two hinges framing into the south centre column, was found to be
1859kNm. This is equivalent to the nominal moment capacity (1004kNm) when the
Page D-37
D.10 SUPER-ASSEMBLAGE CAPACITY MECHANISM CALCULATIONS
The capacity mechanisms used to calculate the theoretical pushover curves for both the
longitudinal and transverse loading directions, is outlined in this sub-section. The theoretical
curve is defined by two points: the onset of yielding and the point of full plastification.
Yielding was assumed to occur at 0.64% as calculated in Section D.1. Plastification was
assumed to occur at 2% in keeping with the work done by Matthews (2004) and the
observed crack widths during this experiment. The contributions to the overall strength are
from the nominal moment strengths of the beams and for negative moment capacities,
additional strength is derived from activation of the starter bars or end connection
reinforcement and diaphragm reinforcement. The bowstring effect, described in section D.8 is
also recognised.
For the external hinges of the structure, in this case for longitudinal loading, the southeast and
southwest hinges, the negative moment capacity is enhanced by the activation of starter bars
from the transverse beams. These starter bars are progressively activated with increasing
lateral drift. For the yield point, the contributing slab width was taken as the width of the infill
topping slab. At the point of full plastification, the width of activated slab was taken to be
3.05m, or approximately half the span of the transverse beams. At 2% drift the width of the
crack between the beam and slab over the full length of the transverse beam, was over 5mm
signifying that the bars were yielding and contributing to the strength of the system. The
moment capacity values calculated were extrapolated to the column centreline in order to
calculate the overall system base shear, and for this, the plastic hinges were assumed to form
Page D-38
Initial Yield
South central hinges: (accounting for axial compression force due to bowstring effect)
Full Plastification
Also, as the base shear capacity of the entire building is being predicted, the negative moment
contribution from the top bars at the northern ends of the transverse beams need to be taken
into account.
Hinge: NW SW SC SE NE 6M Vb
M
hc
Page D-39
As the experimental results were zeroed at the commencement of testing, the experimental
results do not show the residual or natural moments of the structure. During construction, all
of the beams were propped during the placement of the floors and the casting of the topping
concrete. Upon removal of these supports, the eccentric weight of the floor would have
caused the east and west columns to incline inwards (tops of the columns move towards the
centre of the building, and the bases of the columns move outwards), and thus cause positive
moments within the southeast and southwest hinge zones. This artificially shifted the
hystereses off the true zero moment. By way of a simple elastic analysis using the gross
section properties of the structure, this initial moment was calculated as 120kNm, and the
hysteretic plots for both the southeast and southwest hinges in the longitudinal direction have
Figure D-27 shows the comparisons between the predicted and actual strengths. An
A similar method of analysis was used to calculate the capacity of the super-assemblage in the
transverse direction. In this case, the strengths of the transverse beam hinges and the starter
bars from the south longitudinal beams were the contributing sources of strength. Due to the
presence of the steel tie frame along the northern edge of the structure, no starter bars were
present to enhance the negative moment capacity of the northern hinges, and hence slightly
non-symmetrical behaviour is present. The amount of starter bars activated within the
longitudinal beams was taken to be the same as in the longitudinal loading direction; at initial
yield, 1m was the contributing slab width, and at full plastification, the slab width was half
Page D-40
Initial Yield
Full Plastification
Hinge: NW SW SE NE 6M Vb
M
hc
These strength predictions for Phase 2 along with Phase 1 and 3 longitudinal loading are
shown in the following Figures. Section 4 of this thesis discusses the results and explains the
comparisons between the actual and theoretical strengths, whether the comparisons are good
or otherwise.
Page D-41
1800
1600
Base Shear
Force (kN)
1400
1200
1000
800
600
400
200
Inter-Storey Drift (%)
0
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6
-200
-400
-600
-800
Experiment Phase 3
-1000
Experiment Phase 1
-1200
Nominal Prediction
-1400 25% Overstrength Prediction
-1600 40% Overstrength Prediction
-1800
(a) Phase 1 and 3 Longitudinal loading: Experimental base shear capacity and theoretical
pushover curve.
1200
Base Shear
Force (kN)
900
600
300
-300
Experiment (post-torsion)
-1200
(b) Phase 2 Transverse loading: Experimental base shear capacity and theoretical pushover
curve.
Figure D-27 Base shear capacity: experimental and theoretical relationships.
Page D-42
1200
Moment (kNm)
1000
800
600
400
200
R o t a t io n ( %)
0
-6 -5 -4 -3 -2 -1 -200 0 1 2 3 4 5 6
-400
Phase 1
-600
Phase 3
-800 Nominal Strength
-1000 25% Overstrength
-1200
(a) Moment rotation behaviour for the southwest hinge of the south frame
Moment (kNm)
3000
2500
2000
1500
1000
500
R o t a t io n ( %)
0
-6 -5 -4 -3 -2 -1 -500 0 1 2 3 4 5 6
-1000
Phase 1
-1500
Phase 3
-2000 Nominal Strength
-2500 25% Overstrength
-3000
(b) Moment rotation behaviour for the two centre hinges of the south frame framing into the
south centre column
Moment (kNm)
1200
1000
800
600
400
200
R o t a t io n ( %)
0
-6 -5 -4 -3 -2 -1 -200 0 1 2 3 4 5 6
-400 Phase 1
-600 Phase 3
-800 Nominal Strength
-1000 25% Overstrength
-1200
(c) Moment rotation behaviour for the southeast hinge of the south frame
Figure D-28 Phase 1 and 3 Moment rotation behaviour: experimental and theoretical
relationships.
Page D-43
800
Moment (kNm)
600
400
200
Rotation (%)
0
-4 -3 -2 -1 0 1 2 3 4
-200
Pre-Torsion
-400
Post-Torsion
-600 Nominal Strength
25% Overstrength
-800
(a) Moment rotation behaviour for the north hinge of the west frame
1200
Moment (kNm)
1000
800
600
400
200
Rotation (%)
0
-4 -3 -2 -1 -200 0 1 2 3 4
-400
-600 Pre-Torsion
Post-Torsion
-800
Nominal Strength
-1000 25% Overstrength
-1200
(b) Moment rotation behaviour for the south hinge of the west frame
Figure D-30 Phase 2 Moment rotation behaviour: experimental and theoretical relationships
for the west frame.
Page D-44
800
Moment (kNm)
600
400
200
Rotation (%)
0
-4 -3 -2 -1 0 1 2 3 4
-200
-400 Pre-Torsion
Post-Torsion
-600 Nominal Strength
25% Overstrength
-800
(a) Moment rotation behaviour for the north hinge of the east frame
1000
Moment (kNm)
800
600
400
200
Rotation (%) 0
-4 -3 -2 -1 0 1 2 3 4
-200
-400 Pre-Torsion
Post-Torsion
-600
Nominal Strength
-800 25% Overstrength
-1000
(b) Moment rotation behaviour for the south hinge of the east frame
Page D-45
D.11 REFERENCES
Carr A.J, 2003, Ruaumoko 2D User Manual, Computer Program Library, Department of Civil
Dutta A and Mander J.B, 2001, Energy Based Methodology for Ductile Design of Concrete
Lindsay R.A, 2004, Experiments on the seismic performance of hollow-core floor systems in
Matthews J.G, 2004, Hollow-core floor slab performance following a severe earthquake, PhD
New Zealand.
Matthews J.G, Mander J.B, Bull D.K, 2004, Prediction of beam elongation in structural
concrete members using a rainflow method, 2004 NZSEE Conference, Rotorua, NZ,
Mitchell M.P and Collins D, 1991, Prestressed Concrete Structures, Prentice Hall,
NZS3101, 1995, Concrete Structures Standard, NZS3101, Parts 1 & 2, Standards New
NZS3101, 2003, Draft Amendment No. 3 to 1995 Standard (NZS3101), Standards New
Park R and Paulay T, 1975, Reinforced Concrete Structures, J. Wiley and Sons Inc, New
York.
Priestley M.J.N and Kowalsky M.J, 2000, Direct Displacement Based Design of Concrete
Page D-46