Sie sind auf Seite 1von 14

Materials and Structures (2012) 45:433–446

DOI 10.1617/s11527-011-9775-0

ORIGINAL ARTICLE

Influence of experimental setups on the apparent uniaxial


tensile load-bearing capacity of Textile Reinforced Concrete
specimens
J. Hartig • F. Jesse • K. Schicktanz •

U. Häußler-Combe

Received: 21 December 2010 / Accepted: 29 August 2011 / Published online: 11 September 2011
Ó RILEM 2011

Abstract An important parameter for dimensioning Keywords Textile Reinforced Concrete 


of Textile Reinforced Concrete (TRC) structures and Tensile behaviour  Testing  Load-bearing capacity 
structural elements is the tensile load-bearing capacity Simulation
of the composite. Respective values are usually
derived from uniaxial tensile tests with overcritically
reinforced TRC specimens. In this paper, influences
from specimen geometry, e.g. plane and waisted 1 Introduction
specimens, and load application design, e.g. stiff glued
steel plates or soft clamping constructions are inves- For the design and dimensioning of structures or
tigated. Therefore, experimental results regarding the structural elements, values of the mechanical proper-
load-bearing capacity of the composite are statistically ties of the applied materials have to be provided to
evaluated. The experimental observations are sup- design engineers. These values are usually determined
ported by results of numerical simulations with a one- by means of appropriate laboratory tests. The tests
dimensional model based on the Finite Element should be designed such that they can be accurately
Method. These simulations provide stress distributions reproduced with low effort and preferably low
in concrete and reinforcing fibres as well as the tensile expense. New materials or material combinations
load-bearing capacity. Based on these results existing often have mechanical and structural properties devi-
test setups for the derivation of the load-bearing ating from existing materials, which often necessitates
capacity of the composite for dimensioning are modification of existing or development of new test
assessed. As a result, plane plate specimens with a setups.
load application by means of friction is recommended The objective of this paper is to assess the most
for experimental determination. suitable methodology for determining the load-bear-
ing capacity of the novel cement-based composite
Textile Reinforced Concrete (TRC) out of a number of
existing experimental setups. These investigations
support and provide background data to the recom-
J. Hartig (&)  F. Jesse  K. Schicktanz  mendation of ‘‘RILEM Technical Committee
U. Häußler-Combe 232-TDT’’ [24] for performing uniaxial tensile tests
Institute of Concrete Structures, Faculty of Civil on TRC.
Engineering, Technische Universität Dresden,
01062 Dresden, Germany TRC is a composite consisting of a fine-grained
e-mail: jens.hartig@tu-dresden.de concrete matrix and continuous reinforcement fibres,
434 Materials and Structures (2012) 45:433–446

which are applied as fibre bundles, see Fig. 1. The multiple


bundles are called roving or multi-filament yarn and cracking post-cracking state
consist of several hundred or even thousand elemen-
tary fibres called filaments, which possess diameters of uncracked
state
a few microns. The designation ‘‘textile’’ in TRC Vf >> 1.3 V crit

mean stress
means that the reinforcement is applied as 2D or 3D
structure of suitable reinforcement material produced
by means of textile technology. The choice of the
material and the properties of the textile depend on the
Vf 1.3 V crit
expected loading. For structural applications such as Vf < V crit
bridges [9] and beams [22] or as strengthening
overlays for existing RC structures [2, 5, 6, 28] fibres
of alkali-resistant (AR) glass and carbon are preferred, mean strain
because they provide high tensile strength and a
Fig. 2 Schematic mean tensile stress–strain behaviour of TRC
sufficient Young’s modulus in case of AR-glass and a
superior Young’s modulus in case of carbon.
If the concrete cracks under tensile loading due to elastic behaviour until brittle failure. Bundles of brittle
an exceeding of the concrete tensile strength, the fibres fibres exhibit different and more complex failure
need the ability to resist the loading; otherwise the mechanisms, e.g. due to varying strength along each
composite fails. The amount of fibres that is required fibre. If one filament reaches its tensile strength at a
to resist the cracking load is called critical fibre weak position, it fails. The released load of the
volume or critical fibre reinforcement ratio Vcrit. If the filament has to be redistributed to other filaments. This
applied fibre volume Vf is higher than the critical fibre behaviour can be modelled with so-called fibre bundle
volume, TRC shows multiple cracking under tensile models (FBM) based on the initial work of Daniels
loading and strain-hardening behaviour after stabilised [10]. Although FBM’s neglect the interaction of the
cracking. The described load-bearing behaviour is filaments with matrix, the effect that a small number of
shown schematically in Fig. 2 where also an estima- failed filaments might lead to the collapse of the entire
tion for the ultimate loads depending on Vf according yarn can be transferred to the composite behaviour [7,
to Jesse [18] is given. 8, 26].
The mechanical properties of the reinforcement Failure prefers to start at the filaments with highest
fibres have a significant influence on the failure utilisation. In TRC, reinforcement yarns are usually
behaviour of the composite. Single fibres of both not entirely penetrated with concrete matrix due to
AR-glass and carbon show an almost perfect linear filtering effects caused by small distances between the
filaments. Thus, essentially two bond zones exist
inside a yarn. In an outer ring zone, also referred to as
fill-in zone or sleeve zone, matrix intrudes facilitating
relatively strong adhesive bond between matrix and
filaments. However, in the yarn core primarily weak
frictional load transfer at the contact areas of the
filaments is possible. The inner bond can be increased
applying additional coating or impregnation penetrat-
ing the core of the yarns.
In cracked TRC where the reinforcement yarns
bridge the concrete cracks, the highest loaded fila-
ments are usually situated in the fill-in zone at the
multi-filament
cracks because of the strong adhesive bond with the
yarn
filament concrete. The differences in the bond quality lead also
matrix
to differences in the stress transfer length necessary for
Fig. 1 Multi-filament yarns embedded in matrix complete activation of the filaments in the fill-in and
Materials and Structures (2012) 45:433–446 435

core zones of the yarns. As the stress-transfer length is

10
limited by crack spacing also differences between glued steel sheets
single and stabilised cracking states occur. Investiga-
tions by Hegger [17] and Bruckermann [4] showed

100
that local stress elevation in single-type cracks lead to
a reduction of the load-bearing capacity of the 100 300 mm 100
500 mm
composite compared to the case where a regular
multiple crack pattern develops. (a)
Another failure mechanism can occur associated

8
with carbon reinforcement, which has a very high
rubber interlayers
Young’s modulus. In case of wavy textiles, this can steel sheets
lead to considerable transverse stresses in the concrete

100
exceeding the concrete tensile strength, which finally
results in splitting failure of the composite [16, 21]. In 100 300 mm 100
the following, sufficiently straight yarns are assumed 500 mm
such that splitting failure due to yarn waviness can be (b)
neglected.
Besides the bond conditions and the crack devel-

7
opment local concentrations of fibre stress might be glued steel sheets

also induced by the test setup, e.g., the way the load is

105
applied to the specimen during testing or the shape of

60
the specimen. This is investigated in the present 6 5 75 220 mm 75 6 5
500 mm
contribution. The importance of this issue arises from
(c)
the fact that for dimensioning of structural elements,
the load-bearing capacity of the composite is a crucial

6
property, which ideally should only depend on the
material but not on the test setup used for determina- 60

105
tion. Various strategies exist to reduce these stress
concentrations at the clamping, which will be briefly
summarised in the next section. Besides the reduction 125 250 mm 125
of artificial stress concentrations, it is also desirable to 500 mm
have a test setup, which is easy to use and where the (d)
specimen production as well as the preparation can be
performed with minimum effort. This might facilitate 30
10

the testing of a larger number of specimens necessary steel sheets

for statistical significance of the results.


100

200 50 400 mm 50 200


900 mm
2 Existing experimental setups (e)
Fig. 3 Selection of available setups for the determination of the
Various test setups for the experimental determination uniaxial tensile behaviour of TRC. a Setup by Dugas et al. [11],
of the load-bearing properties of TRC have been b setup by Jesse [18], c setup by Brockmann and Raupach [3],
developed by different researchers [3, 11, 18, 22, 23], d setup by Orlowsky and Raupach [23], e setup by Molter [22]
see Fig. 3. These methods vary, e.g., in concrete
technology for specimen production, dimensions and
shape of the specimen as well as design of the load the influence of the specimen shape and the load
application by the testing machine. All these factors application design.
influence the measured tensile load-bearing capacity The most basic specimens are the plate specimens
of the composite. The present investigations focus on in Fig. 3a, b, which cause also the lowest effort in
436 Materials and Structures (2012) 45:433–446

specimen production. More effort has to be spent for is realised by gluing steel plates with epoxy resin
the dumbbell specimens as shown in Fig. 3c, d, which to the specimen (Fig. 3a, c) or by placing a steel
have offsets in the reinforcement plane. The most sheet inside the specimen (Fig. 3e). The main
complicated specimen with regard to production is the transfer mechanism between specimen and clamp-
specimen in Fig. 3e where offsets out of the rein- ing is adhesive tension and shear. No slip occurs
forcement plane are applied. These offsets or wais- between the steel sheets and the specimen.
tings are usually intended to reduce stress – Type B—Soft clamping: Load transfer between
concentrations at the transitions between the sup- specimen and clamping is based on Coulomb
ported and the free testing lengths of the specimen. A friction (Fig. 3b). Concrete cracks are facilitated
drawback of waisted specimens is that often concrete in the supported parts of the specimen. Slip occurs
cracks occur in the waisted zones when concrete between specimen and clamping if static friction is
cracking has already finished in the zone of constant exceeded.
thickness. Thus, an influence of the specimen geom-
The setups shown in Fig. 3b, d with soft clamping and
etry arises in the experimentally determined tensile
frictional load transfer allowing for gradual load
response of the material, which is not desired.
application and significant deformations of the spec-
Also for the load application to the specimen
imen in the load transfer zones are recommended by
different techniques were developed. A common
‘‘RILEM Technical Committee 232-TDT’’ for uniax-
technique is to attach steel sheets by means of
ial tensile tests on TRC.
adhesives to the specimen as, e.g., performed in the
setups shown in Fig. 3a, c. This leads to consider-
ably reduced deformations in the supported parts of
3 Experimental investigations
the specimen compared to the free testing length and
is associated with a suppression of concrete cracking.
3.1 Concept
A different technique is used in the setup shown in
Fig.3e where the load is applied by means of a bolt.
Experimental investigations were carried out to quan-
Therefore, perforated steel sheets are embedded in
tify differences in the apparent load-bearing capacity
the specimen to distribute the concentrated load. The
of the composite determined with a stiff type A load
steel sheets also prevent concrete cracking in these
application and a soft type B clamping corresponding
parts of the specimen. A setup, which allows
to Fig. 3a, b, respectively. Moreover, two specimen
cracking within the supported part of the specimen,
series with different types of textile reinforcement
is given with the setup shown in Fig. 3b where the
were tested: one series in which fabrics were uncoated
load is applied by means of a clamping construction.
and a second series where a polymer-coated fabric was
Rubber interlayers are placed between specimen and
applied. For a statistical analysis of the results, a
steel plates to prevent local stress concentrations
sufficient number of specimens is needed. Each series
caused by the uneven concrete surface. Thus, the
comprised of a total of 44 specimens. In the test setup
load is primarily applied by friction. Another
representing stiff type A load application, steel plates
clamping technique is used in the setup shown in
were glued to the clamping area. The increased
Fig. 3d where steel flanges fitting to the waisted
stiffness in the parts of the specimens with the steel
shape of the specimen are used for load application.
sheets prevented reaching the tensile strength of the
This technique supposedly leads to multi-axial stress
concrete matrix and, therefore, the development of
states in the specimen, at least in the vicinity of load
cracks in the clamping area. Half of the specimens
application, which might be undesirable. Summaris-
from each series was tested using this configuration.
ing, two major types of load application can be
The other specimens were attached to the testing
classified:
machine by means of clamping at both ends such that
– Type A—Rigid load application: Interface as well forces are transferred to the specimens by means of
as load application is stiff compared to TRC under Coulomb friction corresponding to type B load
tension. Hence, concrete cracking is prevented application. This facilitated cracking of the concrete
within the supported range. In most cases, this type in the clamping zone.
Materials and Structures (2012) 45:433–446 437

3.2 Materials Table 1 Properties of textile reinforcement


Parameter Value
A special fine-grained concrete with a maximum
aggregate size of 1 mm and CEM III cement as binder Warp thread, 0°-direction
was used for the cementitious matrix. For more Material Alkali-resistant glass
information regarding the composition of the concrete, Yarn count 1200 tex (nominal), 1274 tex (actual)
the mixing sequence and hardened concrete properties Spacing 7.2 mm
(see [18]). Weft thread, 90°-direction
The reinforcement textiles, see Fig. 4, were pro- Material Alkali-resistant glass
duced of commercial AR-glass filament yarns by Yarn count 640 tex (nominal), 665 tex (actual)
Saint-Gobain Vetrotex. The geometrical specifica- Spacing 10.2 mm
tions of the textile are given in Table 1. The test series Textile
1 was reinforced with this structure without additional Weight per unit area 242 g/m2 (253 g/m2 with coating)
coating, while additional coating was applied to the Type of binding Double tricot
textiles of series 2 specimens. An aqueous suspension Needle thread Poly-propylene textured (15 dtex)
containing styrol-butadien-co-polymers was used as Stitch length 2.0 mm
coating [12], which was thermally cured using infrared
radiation after application in textile processing [19].
In all specimens, the warp threads were aligned in approximately 1.9% in loading direction. After the
loading direction. concrete matrix had set, the plates were covered with
wet cloths and foil and allowed to harden for 3 days. The
3.3 Specimen production and test setup plates were then removed from their moulds and cured
in a water bath at room temperature until an age of
A hand lamination process was utilised to produce the 7 days. Afterwards, the plates were cured/stored in a
specimens. Thin concrete layers and textile reinforce- climate chamber at 20°C with 65% RH until testing at an
ment were applied in an alternating manner on a age of 28 days.
stainless steel formwork, which had a size of 1.20 m 9 Before testing, the plates were cut into specimens of
0.70 m. The formwork had adjustable guides for dimensions of 400 mm 9 30 mm using a water-
varying height levels to ensure accurate thickness of cooled diamond saw. It was ensured that reinforce-
each concrete layer. Two reinforcement layers were ment aligned in the loading (longitudinal) direction
applied resulting with a plate thickness of approxi- was not truncated. Specimens for types A and B were
mately 7 mm in a reinforcement ratio Vf of then systematically tiled (like black and white squares
on a chessboard) into two groups of 22 specimens each
to minimise systematic error. The specimens intended
for type B setup were immediately ready for testing
after being sawed. The load application areas of the
type A specimens were modified by gluing steel plates
with dimensions of 120 mm 9 30 mm 9 1 mm on
both sides of the specimens. A commercially available
silica-filled epoxy resin adhesive was used for gluing.
The tests were performed force-controlled in a
standard servo hydraulic testing machine with a
warp loading capacity of 25 kN. For both, the type A and
direction B setups, the specimens were clamped between
weft scored/roughened steel grips on a length of 120 mm
direction from their respective ends. A 0.5 mm rubber layer was
inserted between specimen and steel grips to prevent
local stress concentrations and to compensate the
Fig. 4 Applied textile reinforcement roughness of the specimen surfaces. Two M16 screws
438 Materials and Structures (2012) 45:433–446

on each clamp, with a specified torque of 5 Nm per The tensile load-bearing capacity can be described
screw, ensured uniform contact pressure between steel by stress values rur or ruc, which might be defined as
grips and specimens. The connection between clamps Fu Fu
and testing machine was designed as a hinge to avoid rur ¼ and ruc ¼ ; ð1Þ
Ar Ac
bending and transverse loading of the specimens. The
following data was measured during testing: tensile respectively. While rur is the ultimate force Fu related
load via load cell; cylinder displacement via linear to the cross-sectional area of the reinforcement Ar in
variable displacement transducer (LVDT); specimen loading direction, ruc is related to the cross-sectional
deformation using a 100 mm clip-on extensometer. area of the concrete Ac, respectively. As these stress
values correspond to the ultimate state of the com-
3.4 Results and discussion posite they might by also referred to as composite
strength. For the evaluation of the utilisation level of
3.4.1 Material and failure behaviour the reinforcement, rur might be more useful. For
dimensioning of TRC structural elements considering
Typical results of the experiments described in the
previous section are mean stress-mean strain relations,
see Fig. 5. The mean stress is the measured force (load
20
cell) related to the cross-sectional area of the concrete type A
and the mean strain is the measured deformation (clip-
mean stress [N/mm2]
on extensometer) related to the measurement length. 15
type B

These relations show three distinctive states: the


uncracked state with high stiffness, the state of
multiple concrete cracking with low stiffness and the 10

post-cracking state with an intermediate stiffness, see


also Fig. 2. This so-called strain-hardening behaviour 5
is described in more detail in Jesse [18]. type A
During testing, cracks developed in the specimens, type B
as expected, in the clamping zones in type B load 0
0 0.5 1.0 1.5 2.0
application. With type A load application, no cracks mean strain [%]
were visible in the clamping zones. In all specimens, (a)
numerous cracks with crack spacing of approximately
15–20 mm developed in the free length of the 20
type A
specimens. Furthermore, failure of the specimen most
often occurred right at the transition from the
mean stress [N/mm2]

type B
15
supported to the unsupported part of the specimen
when type A load application was used. In contrast,
type B specimens tended to fail less often at the 10
transitions to the supported parts and more often in the
central region of the free specimen length.
Failure of the composite occurs as a result of the 5
type A
tensile failure of the reinforcement fibres. It is worth to type B
note that in the case of uncoated reinforcement, the 0
core fibres where often entirely pulled-out of the 0 0.5 1.0 1.5 2.0

concrete in the supported range of the specimen mean strain [%]


indicating quite poor inner bond. In contrast, the (b)
length of fibres pulled out in case of coated textiles
Fig. 5 Experimental mean stress–means strain relations and
were considerably shorter than the available anchor-
photographs of typical examples of failed specimens. a Series 1
age length indicating an effective improvement of with uncoated AR-glass fibres. b Series 2 with polymer-coated
inner bond. AR-glass fibres
Materials and Structures (2012) 45:433–446 439

only the integral load-bearing behaviour of the com- corresponding to the samples provided by the exper-
posite, ruc is supposedly more suitable. The relation iments. Due to linearised representation of the Weibull
between both values is given with ruc = (Ar/Ac)rur distribution this can be performed by a linear regres-
where the conversion factor (Ar/Ac) = Vf is the sion (LR). Alternatively, a maximum likelihood (ML)
reinforcement ratio. estimation can be performed. The values provided by
means of both methods are given in Fig. 6 as well as
3.4.2 Statistical evaluation the respective linear functions. Significant differences
between both estimates appear in all cases for m while
A comprehensive assessment of the effects of the type the values of r0 differ not significantly. Presumably,
of load application and the influence of additional the values based on ML estimation have higher
coating on the reinforcement is facilitated by means of significance as this method takes the underlying
a statistical evaluation of the experimental results. For distribution function into account. Still, it has to be
the failure analysis of tension members often an considered that the Weibull distribution, though
analysis applying the Weibull distribution [27] is useful, is only an assumption. Theoretical investiga-
performed. The Weibull distribution is based on the tions [e.g. 25] indicate that strength distributions for
assumption that failure of a specimen occurs when its composites usually consist of a combination of several
weakest link fails. The cumulative distribution func- distribution functions. For the following evaluations,
tion (CDF) of the two-parametric Weibull distribution the ML values are, however, used.
is given with
  m 
r σur; σ [N/mm²]
FðrÞ ¼ PðS  rÞ ¼ 1  exp 
r0 ð2Þ 2
300 400 500 600
F F type A type B
0  r  1: 0.99
0.95
1 400 mm 30 mm
0.90
ln ln (1/(1-F))

where m is the so-called Weibull modulus and r0 is a 0


LR: m = 11.2; 0.50
characteristic strength corresponding to a failure
-1 σ0 = 455.7 N/mm²

F
probability of 63.2% (corresponding to the case ML: m = 12.1;
r = r0). The CDF specifies the probability P that a -2 σ0 = 454.7 N/mm² LR: m = 11.9; 0.10
σ0 = 518.8 N/mm²
random variable S representing all possible strength -3 ML: m = 13.2;
0.05
values takes a value less than r. Equation 2 can be σ0 = 518.0 N/mm²
-4
linearised by means of some logarithm operations 5.6 5.7 5.8 5.9 6.0 6.1 6.2 6.3 6.4 6.5

yielding ln σur; ln σ
  (a)
1
ln ln ¼ m lnðrÞ  m lnðr0 Þ: ð3Þ σur; σ [N/mm²]
1  FðrÞ
500 600 700 800 900
2
Thus, m can be also interpreted as the slope while r0 F F type A type B 0.99
1 0.95
corresponds to the root of the Weibull distribution 400 mm 30 mm
0.90
ln ln (1/(1-F))

function in its linearised form. Diagrams correspond- 0


0.50
ing to Eq. 3 are given with Fig. 6a, b. In these
-1
F

diagrams, the samples rur,i with 1 B i B n correspond- LR: m = 15.9;


σ0 = 712.6 N/mm²
ing to the n experimentally determined ultimate forces -2
ML: m = 18.2; LR: m = 15.3; 0.10
are plotted. Therefore, the rur,i values are sorted in σ0 = 711.2 N/mm² σ0 = 760.3 N/mm²
-3 0.05
ML: m = 17.3;
ascending order. The position of the sample on the -4 σ0 = 759.3 N/mm²
ordinate is determined as 6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9

  ! ln σur ; ln σ
1 1 (b)
ln ln ¼ ln ln i : ð4Þ
1  Fðrur;i Þ 1  nþ1
Fig. 6 Weibull-plot for failure stresses rur (failure position:
open circle near the clamping, open square at the clamping).
It is now possible to estimate parameters m and r0 of a Series 1 with uncoated AR-glass fibres. b Series 2 with
the assumed continuous Weibull distribution function polymer-coated AR-glass fibres
440 Materials and Structures (2012) 45:433–446

Figure 6a shows Weibull-plots of the rur values 4 Numerical investigations


of series 1 specimens comprised of uncoated AR-
glass fabric and Fig. 6b of series 2 specimens with 4.1 Model
polymer-coated AR-glass fabric. In both specimen
series, the characteristic strength values r0, which For theoretical analysis of the influence of different
are a measure for the expected value, were signif- load application types and specimen shapes on the
icantly higher for type B than for type A load load-bearing capacity of the composite, an essentially
application. Moreover, the differences in r0 are one-dimensional finite element model is used, which is
smaller for the coated (subscript c) reinforcement described in detail in Hartig et al. [14]. The model
(r0,A,c/r0,B,c = 0.937) compared to the uncoated consists of bar element chains representing the axial
(subscript u) reinforcement (r0,A,u/r0,B,u = 0.878). stiffness of the constituents of the test setup, i.e.
Furthermore, r0 values are larger for the coated concrete, AR-glass and steel, and bond elements
reinforcement compared to the uncoated reinforce- representing the load transfer in between, see Fig. 7.
ment (r0,A,u/r0,A,c = 0.639; r0,B,u/r0,B,c = 0.682). The bond elements act like shear springs. Linear
The shape parameters m, which are a measure for elastic material behaviour with limited tensile strength
the variance, differ only insignificantly with regard is assumed for reasons of simplicity for all materials.
to the loading types A and B. Nevertheless, between Consideration of concrete tension softening has espe-
coated and uncoated reinforcement considerable cially in the ultimate state of the composite negligible
differences appear. The variance is smaller for influence on the results and remains unconsidered. The
coated reinforcement corresponding to a larger applied material properties are summarised in Table 2.
value of m compared to the uncoated reinforcement, In addition to Hartig et al. [14], the model is
compare Fig. 6. enhanced concerning the representation of the load
The elevation of expected strength and the reduc- application to the specimen, which was performed in
tion of variance due to coating of the reinforcement Hartig et al. [14] simply by prescribed displacements
can be explained by improved inner bond. This leads at the end nodes of the concrete bar element chain. For
to a homogenised stress distribution in the yarn, a more realistic representation of the load transfer by
which, in turn, results in higher applicable stresses means of glued steel sheets (type A) or friction-based
(larger r0) with relatively low scatter (larger m). Vice clamping (type B), additional bar elements are used
versa, the larger stress concentrations in the sleeve representing steel sheets, see Fig. 7. At corresponding
filaments lead in case of uncoated yarns associated element nodes, bond elements link these bar elements
with poor inner bond to an earlier failure of these to the concrete elements. In a first approach, linear
fibres. Nevertheless, this results due to the lower load elastic bond laws hcs are applied to these bond
level also in a higher probability that load can be elements. The different stiffness of the interfaces in
redistributed to less utilised filaments without failure type A and B load application is taken into account
of the entire reinforcement. with different bond moduli G. While for the case of
Figure 6 contains also information about the rubber interlayers in the clamping constructions (type
failure position of each specimen. The majority of B) G = 100 MN/m3 is chosen, for the case of glued
specimens fail at the transitions between clamping steel sheets (type A) G = 1 9 1010 MN/m3 is used.
and free length. Only for specimens with coated The reinforcement is modelled with a ring model
reinforcement and type B clamping a significant consisting of five layers. The outermost layer, referred
number failed distant to the load application in the to as sleeve layer in the following, represents the
free length of the specimens, see Fig. 6b. However, filaments in the fill-in zone, which is assumed to
a direct correlation of the failure position right at contain 20% of the total cross-sectional area of
the transitions between clampings and free length of the reinforcement. The four inner layers represent
the specimen to low strength can not be drawn. the filaments in the yarn core. The determination of the
Furthermore, outliers could neither be explained by geometrical properties of the layers is described in
the location of the failure section nor the resulting Hartig et al. [14]. Between concrete and sleeve layer
Weibull plots. the bond law hcr is applied representing initially
Materials and Structures (2012) 45:433–446 441

Yarn representation Discretization


Transverse Longitudinal direction:
direction: Δu
plain specimen waisted specimen
steel plate
hcs
matrix
hcr
sleeve layer
(20% Ar) hrr
core layers hrr
(80% Ar) ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
hrr

reinforcement element matrix element node


bond element steel element
Constitutive relations
σ-ε relations for steel, concrete and Bond law hcr : Bond law hrr :
reinforcement; τ−s relation hcs:
10 10
stress σ; bond stress τ

bond stress τ [N/mm²]


bond stress τ [N/mm²]

ft
8 bond 8
degradation
6 6
G
E
4 friction 4 unloading friction
1
1
good inner bond
2 2
unloading
poor inner bond
εt 0 0.2 0.4 0.6 0.8 1.0 1.2 0 0.2 0.4 0.6 0.8 1.0 1.2
strain ε; slip s slip s [10-5 m] slip s [10-5 m]

Fig. 7 Mechanical model and constitutive relations

adhesive bond between the concrete and the filaments the filaments of 10 lm results in the linear-elastic
merging to friction due to bond degradation. At the bond stress–slip relation to a slope of G = 3 9
inner interfaces the bond law hrr is used based on the 108 MN/m3. The parameters of the other bond
assumption of primarily frictional load transfer laws were estimated based on investigations in Hartig
between the filaments, see Fig. 7. et al. [14].
For the inner bond, modelled with bond law For the parametric studies, a geometrical configu-
hrr, three cases are considered. The first case repre- ration corresponding to Fig. 3b was chosen. The
sents poor inner bond, e.g. no matrix penetration and reduction of specimen size as in the experiments
no impregnation, with a low frictional bond stress, the presented in Sect. 3 is not performed. The material
second case represents good inner bond, e.g. some parameters of concrete and reinforcement as well as
matrix or coating penetration, with increased frictional the reinforcement ratio of approximately 1.9% as
bond stress and the third case represents perfect inner summarised in Table 2 correspond to the experiments
bond assuming perfect adhesion between the fila- described in Sect. 3. For the bar elements, a length of
ments, e.g. due to polymeric coating or impregnation, 0.2 mm was chosen to represent multiple cracking of
with a linear-elastic bond law similar to hcs, see Fig. 7. the concrete and the load transfer between the
For the latter case, the bond stiffness was estimated constituents of the setup appropriately. Besides plane
based on investigations on a nano-clay polymeric specimens also waisted specimens are taken into
coating by Gao et al. [13], which revealed a Young’s account. If considered, waisting was applied on
modulus of about 6000 N/mm2. Assuming a Poisson’s lengths of 50 mm before the clampings with ele-
ratio equal to zero yields a shear modulus of 3000 N/ ment-wise linearly increasing element thickness from
mm2. Assuming further an effective distance between 8 to 24 mm, see Fig. 7.
442 Materials and Structures (2012) 45:433–446

Table 2 Material and geometrical properties of bar elements Fig. 8 Stress distribution for different types of load applica- c
tion, specimen shape and inner bond laws (left: poor inner bond,
Steel Concrete AR glass centre: good inner bond, right: perfect inner bond). Plane
specimens with a glued steel plates and b soft interlayer.
Cross-sectional 1000 800 (free 15.34 c Waisted specimens with glued steel plates. d Comparative
area A (mm2) length) simulations of single cracks
Young’s modulus Et 210,000 28,500a 74,450b
(N/mm2) respectively. The stress amplitudes are the largest in
Tensile strength ft 500 (not 5a 1231b the outermost reinforcement layer directly connected
(N/mm2) reached)
to the concrete and decrease towards the innermost
a
According to Jesse [18] core layer. Nevertheless, it is observable in Fig. 8 that
b
According to Abdkader [1] differences in the stress amplitudes decrease with
improving inner bond. For the cases referred to as
perfect inner bond (right column in Fig. 8), stress
differences between the reinforcement layers vanish.
The boundary conditions are given with prescribed Figure 8a shows stress distributions for type A load
displacements at the nodes in the centre of the two bar application and a plane specimen with different
element chains representing the steel plates. While the qualities of bond in the reinforcement core. In case
displacement of one of these nodes is prescribed equal of poor inner bond (Fig. 8a, left), the overall maxi-
to zero, the displacement is incrementally increased at mum value of reinforcement stress develops in the
the other node, see Fig. 7. The nonlinearities intro- sleeve layer in the cracked cross section at the
duced by the material laws necessitate the application transition between the supported and the free range
of an incremental-iterative solution procedure; see at x = 0.1 m. It results from the much slower activa-
[14] for details. tion of the core layers compared to the sleeve layer
associated with large stress transfer lengths in the
4.2 Results and discussion reinforcement core. Thus, the core layers participate
less on crack bridging compared to the regular stress
The model is able to provide the stress distributions in distribution in the centre of the free specimen length
concrete and reinforcement, which can not be deter- leading to an increased loading of the sleeve layer.
mined in the experiments. Thus, the results of the Towards the centre of the specimen the local stress
simulation might shed some light on the mechanisms maxima of the reinforcement decrease and are almost
of the failure behaviour of the composite. In Fig. 8, constant starting from the third crack distant to the
stress distributions in the concrete and the reinforce- clamping zone. In case of good inner bond (Fig. 8a,
ment on a length of 200 mm consisting of 100 mm for centre), this stress elevation at the transition to the
each a clamped end and a part of the free length of the supported range is considerably reduced due to
specimen are shown. The stress distributions corre- improved compatibility of deformation between the
spond to load levels right before ultimate failure of the yarn layers associated with shorter stress transfer
composite reaching the tensile strength in one of the lengths in the core layers and vanishes for the case of
reinforcement elements. Figure 8 contains different perfect inner bond (Fig. 8a, right).
cases of specimen shape, load application type and For a plane specimen with soft type B clamping, see
inner bond quality of the reinforcement. Fig. 8b, it turns out that the stress elevation in the
Concrete crack positions are indicated in Fig. 8 in sleeve layer at the clamping is slightly reduced for the
the illustrations of the specimens and can be also case of poor inner bond compared to type A clamping.
identified in the stress distributions by concrete This results from an additional crack of the concrete in
stresses equal to zero in most cases. As the reinforce- the supported range of the specimen, which leads to
ment bridges the cracks, the stresses in the reinforce- larger exploitation of available stress transfer length in
ment layers have, in general, maximum values at the the sleeve layer (from crack position to end of the
cracks. Due to stress transfer by means of bond, the specimen). In turn, this means that stress elevation in
stresses in the concrete increase in the uncracked parts the sleeve layer will always occur when the available
starting at the crack and decrease in the reinforcement, anchorage length is shorter than the stress transfer
Materials and Structures (2012) 45:433–446 443

15 1500
stress in concrete [N/mm²]

stress in reinforcement
12.5 negligible no stress 1250
stress stress elevation

layers [N/mm²]
elevation elevation
10 1000

7.5 750

5 500

2.5 250

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
x[m] x[m] x[m]
(a)
15 1500
stress in concrete [N/mm²]

stress in reinforcement
12.5 negligible no stress 1250
stress stress elevation

layers [N/mm²]
elevation elevation
10 1000

7.5 750

5 500

2.5 250

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
x[m] x[m] x[m]
(b)
15 1500
stress in concrete [N/mm²]

stress in reinforcement
12.5 stress no stress 1250
elevation elevation

layers [N/mm²]
stress
10 elevation 1000

7.5 750

5 500

2.5 250

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
x[m] x[m] x[m]
(c)
15 1500
stress in concrete [N/mm²]

stress in reinforcement

12.5 1250
layers [N/mm²]

10 1000

7.5 750

5 500

2.5 250

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
x[m] x[m] x[m]
(d)
444 Materials and Structures (2012) 45:433–446

length necessary for complete activation of the sleeve through of the core layers in the first crack. Full
fibres. The additional cracks in the supported parts of activation of the core layers is only reached after a
the specimen lead to a better exploitation of available certain number of further cracks. For the cases of
anchorage length and can result in an almost complete improved inner bond, the core layers are already at the
disappearance of stress elevation as shown in Hartig first crack almost completely activated.
et al. [15]. Correspondingly, the stress elevation in the Finally, the load bearing-capacity is quantified by
sleeve layer diminish for the case of good inner bond means of respective stress values. For the determina-
(Fig. 8b, centre) and vanish again for the case of tion of these values, the failure mechanism is simpli-
perfect inner bond (Fig. 8b, right). fied and it is assumed that failure occurs when the first
In Fig. 8c, the results for waisted specimens with element in the sleeve layer reaches its predefined
type A load application are shown. In case of poor tensile strength. The respective stress values are
inner bond, the stress elevation in the sleeve layer at determined corresponding to Eq. 1 as the integral
the crack close to load application has a similar extent force over a cross section divided by either the cross-
as in the previous cases. However, the position of the sectional area of the concrete resulting in ruc or the
respective crack is situated at the transition of the cross-sectional area of the reinforcement resulting in
waisted part of the specimen to the part with constant rur. These values are compiled in Table 3. As
thickness. In the waisted parts of the specimen reference values, the ultimate stresses are determined
concrete cracks do not develop because the concrete based on exceeding of the tensile strength of the
tensile strength is not reached in the respective reinforcement elements on a range of 0.2 m in the
elements before the reinforcement fails. Nevertheless, centre of the model where stress concentrations due to
also if cracks in the waisted part of the specimen would the load application are assumed to be diminished.
develop, stress elevations in the sleeve layer at the Thus, the earlier failure of the reinforcement due to
respective cracks would occur as corresponding boundary effects is neglected in the reference ultimate
simulations revealed [15]. The reason is again the stresses.
large difference in stress transfer lengths between Table 3 reveals that in general the load-bearing
sleeve layer and core layers. For the cases with capacity of the composite increases with the improve-
improved inner bond, a crack in the waisted specimen ment of inner bond. This coincides with investigations
part developed. For the case of good and perfect inner presented in Hartig et al. [14] and Konrad and
bond, the stress elevation in the sleeve layer in Chudoba [20]. The case of a single crack (e.g.
particular cracks diminishes. double-sided yarn pull-out in undercritically rein-
The stress distributions in the reinforcement in the forced specimens) is probably the lower limit case
vicinity of the first cracks adjacent to or in the load regarding the influence of stress distribution in the
application regions of the specimens show signifi- reinforcement on the load-bearing capacity of the
cantly different courses than the cracks in the centre of composite. The reference case of a perfect strain
the free length. While the reinforcement stress distri- specimen is the upper limit for the composite strength
bution in the regular multiple crack pattern in x- for each conducted bond quality.
direction is symmetric to the cracks, it is not for the The general tendencies regarding the influence of
cracks adjacent or in the load application region. It is test setup on the apparent load-bearing capacity
rather observable that towards the load application composite observed in the experimental study pre-
region the stress distribution is similar to the case of a sented in Sect. 3 coincide with the simulations. Thus,
single crack. Corresponding results of simulations the model seems to be valid and can serve for
where only one concrete element is allowed to crack explanation of the effects observed in experiments.
and the load is applied to end nodes of the concrete are Nevertheless, differences between experiment and
shown in Fig. 8d. Especially for the case of poor inner simulation exist—not only because of different geo-
bond, a large difference in the reinforcement stresses metrical properties. The simulated results overesti-
between the sleeve layer and the core layers is mate experimental results. This is caused supposedly
observable, which results from different load transfer by an overestimation of the actual strength of the
length in the sleeve layer and the core layers. This reinforcement yarns after textile processing and spec-
leads in the case of multiple cracking to a certain pull- imen production, which lead to additional flaws in the
Materials and Structures (2012) 45:433–446 445

Table 3 Load-bearing capacity of the composite due to different load application types and specimen shapes expressed as ruc and
rur values (N/mm2)
Bond quality Reference Single crack Type A/plane Type B/plane Type A/waisted

Poor inner bond 20.7; 1078 (100%) 15.1; 788 (72.9%) 16.0; 835 (77.3%) 16.5; 859 (79.7%) 15.7; 819 (75.8%)
Good inner bond 22.3; 1163 (100%) 22.0; 1147 (98.6%) 22.3; 1163 (100%) 22.3; 1163 (100%) 21.9; 1144 (98.2%)
Perfect inner bond 23.5; 1226 (100%) 23.4; 1222 (99.6%) 23.5; 1226 (100%) 23.5; 1226 (100%) 23.5; 1226 (100%)

filaments reducing strength. Moreover, also the reinforcement fibres adhesively connected with the
assumed bond properties are only estimates, which concrete. This effect is only pronounced in case of
were not specially calibrated. Statistical effects poor inner bond in the reinforcement, which results in
remained also unconsidered in the simulations, but large differences in the stress transfer length between
will influence the simulated results. sleeve and core fibres. The soft type B clamping
facilitates concrete cracking in the load application
zones leading to a better exploitation of available
5 Summary and conclusions anchorage length for the reinforcement and a reduc-
tion of stress elevation in the sleeve fibres. The
In the present contribution, experimental and theoret- simulated and experimental results show that the
ical investigations regarding the influence of the influence of the load application on the apparent load-
experimental setup on the apparent tensile load- bearing capacity of the composite decreases with
bearing capacity of TRC as determined in experiments increasing inner bond. Since the application of textile
are presented. The knowledge of the load-bearing reinforcement without secondary coating for control-
capacity is important for a reliable design of structures ling and improving inner bond, structural stability and
and structural elements. handling on site is not recommended for practical
The experimental results revealed a strong depen- applications, the problems that arise with poor inner
dency of the load application type to the specimen on bond at the clampings should not be overrated.
the measured ultimate load and failure position of the Besides the advantage of lower sensitivity to stress
specimen. Tests with rigid load application (type A), concentrations, test setups of type B have some
e.g. by means of stiff glued steel sheets, showed lower advantages that make them favourable for testing.
ultimate loads and frequent failure of the specimens The specimens are straight and of constant thickness.
adjacent to the transition between the supported and Thus, the specimens can be easily produced and also
the free parts of the specimens. In contrast, soft load cut from larger plates, which reduces the effort for
application by means of friction (type B) led to higher production. This enables these specimens also for
ultimate loads and less frequent failure of the speci- quality testing on construction site or from serial
men at the transition between the supported and the production in precast plants as they can be removed
free parts of the specimen. The experimental results with low effort from existing structural elements, at
confirmed also the expectation that the improvement least as long as they have sufficiently large areas with
of inner bond in the reinforcement increases the load- planar geometry. Another aspect is the preparation and
bearing capacity of the reinforcement. These charac- fixation of the specimen in the testing machine, which
teristics can be explained by means of the stress can be executed with type B clamping with small
distribution in the specimen, which are, in general, not effort as gluing, curing and removing of epoxy resin
accessible with experimental methods. from the steel plates after testing is not necessary.
Therefore, theoretical investigations using a numer-
ical model were carried out. The simulations showed Acknowledgments The authors gratefully acknowledge the
how rigid (type A) and soft (type B) load application financial support of this research from Deutsche Forschungs-
gemeinschaft DFG (German Research Foundation) within the
influence the stress distribution in the composite and
Sonderforschungsbereich 528 (Collaborative Research Center)
the apparent load-bearing capacity. Especially rigid ‘‘Textile Reinforcement for Structural Strengthening and
type A clamping can lead to stress elevations in the Retrofitting’’ at Technische Universität Dresden, Germany.
446 Materials and Structures (2012) 45:433–446

References 15. Hartig J, Jesse F, Häußler-Combe U (2010) Evaluation of


experimental setups for determining the tensile strength of
1. Abdkader A (2004) Charakterisierung und Modellierung Textile Reinforced Concrete. In: Brameshuber W (ed) 2nd
der Eigenschaften von AR-Glasfilamentgarnen für die ICTRC—Textile Reinforced Concrete: proceedings of the
Betonbewehrung. Dissertation, TU Dresden international RILEM conference on material science
2. Bösche A, Jesse F, Ortlepp R, Weiland S, Curbach M (2008) (MatSci), vol 1. RILEM Publications S.A.R.L., Bagneux,
Textile Reinforced Concrete for flexural strengthening of pp 117–127
RC-structures—Part 1: Structural behavior and design 16. Hausding J, Lorenz E, Ortlepp R, Lundahl A, Cherif C
model. In: Aldea CM (ed) ACI SP-251 Design & applica- (2011) Application of stitch-bonded multi-plies made by
tions of textile-reinforced concrete. ACI, Farmington Hills, using the extended warp knitting process: reinforcements
pp 19–40 with symmetrical layer arrangement for concrete. J Text
3. Brockmann T, Raupach M (2002) Durability investigations Inst. doi:10.1080/00405000.2010.515729
on Textile Reinforced Concrete. In: Proceedings of the 9th 17. Hegger J, Will N, Bruckermann O, Voss S (2006) Load-
international conference on durability of building materials bearing behaviour and simulation of Textile Reinforced
and components (CSIRO2002), Brisbane, Australia, paper Concrete. Mater Struct 39:765–776. doi:10.1617/s11527-
no. 111 005-9039-y
4. Bruckermann O (2007) Zur Modellierung des Zugtragver- 18. Jesse F (2004) Load bearing behaviour of filament yarns in a
haltens von textilbewehrtem Beton. Dissertation, RWTH cementitious matrix. Dissertation, TU Dresden (in German)
Aachen 19. Köckritz U (2007) In-situ Polymerbeschichtung zur Struk-
5. Brückner A, Ortlepp R, Curbach M (2006) Textile Rein- turstabilisierung offener nähgewirkter Gelege. Dissertation,
forced Concrete for strengthening in bending and shear. TU Dresden
Mater Struct 39(8):741–748. doi:10.1617/s11527-005- 20. Konrad M, Chudoba R (2009) Tensile behavior of cemen-
9027-2 titious composite reinforced with epoxy impregnated mul-
6. Brückner A, Ortlepp R, Curbach M (2008) Anchoring of tifilament yarns. Int J Multiscale Comput Eng
shear strengthening for T-beams made of Textile Rein- 7(2):115–133. doi:10.1615/IntJMultCompEng.v7.i2.40
forced Concrete (TRC). Mater Struct 41(2):407–418. doi: 21. Lorenz E, Ortlepp R, Hausding J, Cherif C (2011) Effi-
10.1617/s11527-007-9254-9 zienzsteigerung von Textilbeton durch Einsatz textiler Be-
7. Chudoba R, Vořechovský M, Konrad M (2006) Stochastic wehrungen nach dem erweiterten Nähwirkverfahren. Beton
modeling of multi-filament yarns. I. Random properties Stahlbetonbau 106(1):21–30. doi:10.1002/best.201000072
within the cross-section and size effect. Int J Solids Struct 22. Molter M (2005) Zum Tragverhalten von textilbewehrtem
43(3–4):413–434. doi:10.1016/j.ijsolstr.2005.06.063 Beton. Dissertation, RWTH Aachen
8. Curbach M, Schicktanz K, Jesse F, Hartig J (2006) Sto- 23. Orlowsky J, Raupach M (2006) Modelling the loss in
chastische Eigenschaften der Zugfestigkeit freier und strength of AR-glass fibres in textile-reinforced concrete.
zementös eingebetteter Filamentbündel aus AR-Glas. In: Mater Struct 39:635–643
Ruge P, Graf W (eds) 10. Dresdner Baustatik-Seminar, 24. Brameshuber W et al (2010) Uniaxial tensile test—test
Neue Bauweisen—Trends in Statik und Dynamik. TU method to determine the load bearing behavior of tensile
Dresden, Dresden, pp 91–100 specimens made of textile reinforced concrete (Proposal for
9. Curbach M, Weiland S, Michler H (2008) Textile Rein- a recommendation by RILEM TC 232-TDT to be published
forced Concrete—the bridge between the textile industry in Materials and Structures)
and the construction sector. Text J 125(4):58–69 25. Vořechovský M (2010) Incorporation of statistical length
10. Daniels HE (1945) The statistical theory of the strength of scale into Weibull strength theory for composites. Compos
bundles of threads. I. Proc R Soc Lond A183:405–435 Struct 92(9):2027–2034. doi:10.1016/j.compstruct.2009.
11. Dugas M, Weise S, Curbach M, Hempel R, Offermann P, 11.025
Franzke G (1998) Force-deformation behaviour of tensile- 26. Vořechovský M, Chudoba R (2006) Stochastic modeling of
loaded specimens made of Textile Reinforced Concrete. In: multi-filament yarns. II. Random properties over the length
Techtextil symposium 1998, Lyon, pp 143–152 and size effect. Int J Solids Struct 43(3–4):435–458. doi:
12. Gao SL, Mäder E, Plonka R (2004) Coatings for glass fibers 10.1016/j.ijsolstr.2005.06.062
in a cementitious matrix. Acta Mater 52(16):4745–4755. 27. Weibull W (1951) A statistical distribution function of wide
doi:10.1016/j.actamat.2004.06.028 applicability. J Appl Mech 18:293–297
13. Gao SL, Mäder E, Plonka R (2007) Nanostructured coatings 28. Weiland S, Ortlepp R, Hauptenbuchner B, Curbach M
of glass fibers: improvement of alkali resistance and (2008) Textile Reinforced Concrete for flexural strength-
mechanical properties. Acta Mater 55(3):1043–1052. doi: ening of RC-structures—Part 2: Application on a concrete
10.1016/j.actamat.2006.09.020 shell. In: Aldea CM (eds) ACI SP-251 Design & applica-
14. Hartig J, Häußler-Combe U, Schicktanz K (2008) Influence tions of textile-reinforced concrete. ACI, Farmington Hills,
of bond properties on the tensile behaviour of Textile pp 41–58
Reinforced Concrete. Cem Concr Compos 30(10):898–906.
doi:10.1016/j.cemconcomp.2008.08.004

Das könnte Ihnen auch gefallen