Materials and Structures (2006) 39:765–776

DOI 10.1617/s11527-005-9039-y

ORIGINAL ARTICLE

Load–bearing behaviour and simulation of textile

reinforced concrete
J. Hegger · N. Will · O. Bruckermann · S. Voss

Received: 14 February 2005 / Accepted: 29 July 2005 / Published online: 19 July 2006

C RILEM 2006

Abstract The new composite material Textile 1 Introduction

Reinforced Concrete (TRC) is a promising devel-
opment which may open up entirely new fields for Textile reinforced concrete (TRC) is a composite mate-
the application of the construction material concrete. rial which combines the advantages of fibre reinforced
The possible more filigree structures with high concrete and ordinary steel reinforced concrete (Fig. 1).
quality surfaces make TRC an attractive choice for Due to the corrosion resistance of fibre materials
the architect and give the engineer more freedom (e.g. alkali-resistant glass (ar-glass), carbon, aramid),
in design. However, the use of TRC requires design the concrete cover is no longer needed as a chemical
rules which are currently being developed at RWTH protection [1, 2]. The thickness of structural members
Aachen University, Germany. In this article, recent depends primarily on the necessary value to ensure a
experimental results as well as modeling techniques proper anchorage of the reinforcement and to avoid
are described. splitting failure. Modern producing processes provide
textiles which can be specially tailored for even com-
Résumé plex geometries of structural members [3]. With these
Le nouveau matériel composite “béton armé en tex- techniques the rovings can be aligned in the direction
tile” (TRC = Textile Reinforced Concrete) constitue un of the expected tensile stresses, leading to an increase
développement prometteur qui ouvre de nouvelles per- in their effectiveness and load-carrying capacity com-
spectives dans l’utilisation du béton comme matériau pared to fibre reinforced concrete. This reduces the
de construction. Le TRC permet de construire des struc- costs of the still expensive high performance fibres.
tures filigranes de haute qualité, une alternative atti- Despite these advantages, the application of TRC
rante pour l’architecte et une plus grande liberté dans failed to extensively appear up to now. The main rea-
la conception pour l’ingénieur. Cependant, l’usage du son for this is the lack of design rules for TRC. In
TRC exige la connaissance de ses règles de conception, recent years, especially within the scope of two collab-
qui actuellement sont sujet de recherche à l’université orative research centers in Germany, great efforts were
d’Aix-la-Chapelle, Allemagne. Cet article présente des made to develop models describing the load bearing
récents résultats expérimentaux, ainsi que des tech- behaviour of TRC and to provide the required design
niques de modélisation du matériel. rules. Thereby, experimental research on TRC mem-
bers is being conducted in order to gain information on
the load bearing behaviour and to develop design meth-
J. Hegger · N. Will · O. Bruckermann · S. Voss
Institute of Structural Concrete, RWTH Aachen University, ods in a straightforward manner (see section 2). Struc-
Germany tural members already put into practice are usually
766 Materials and Structures (2006) 39:765–776

Fig. 1 Reinforcing systems

of concrete.

based on such test series targeted to the particular ap- two-dimensional ar-glass fibre fabric with the name
plication [4, 5, 6, 7]. MAG-07-03. The fabric consists of 2400 tex rovings
At the same time, development of consistent numer- which have been produced by Saint-Gobain Vetrotex.
ical and analytical models is a main theme of research. This roving consists of about 1300 fibres (filaments)
These models require the consideration of the mi- which have a diameter of approximately 29 μm. The
crostructure of the fibre strands (rovings), see section 3. mesh size of the fabric is about 8 mm × 8 mm (Table
1). Its binding leads to different geometrical and me-
chanical properties of the rovings in 0◦ -direction (warp)
2 Experimental investigations on TRC and in 90◦ -direction (weft) (Table 2). The rovings in
members 0◦ -direction have a higher tensile strength because the
compression of the filaments due to the binding in-
Knowledge about the load bearing behaviour under uni- creases the friction between them and because of the
axial tension is an important input for the design engi- lower damage during the manufacturing process com-
neer. pared to the rovings in 90◦ -direction. The test method
The investigations described are limited to tensile for the determination of the strength of the rovings is
tests on specimens reinforced with an ar-glass fabric. described in [9].
In doing so, the effects of different reinforcement ra- A fine grained concrete with a maximum grain size
tios, different roving geometries caused by different of 0.6 mm was used for the investigations. Super plas-
bindings and different impregnations were examined. ticizer and fly ash were added to achieve a very good
Another central aspect is the influence of local transver- flowing capability of the concrete in order to ensure a
sal stresses on the load-carrying capacity of the textile
reinforcement. Such stresses occur, when the direction Table 1 Structure of the ar-glass fabric
of the rovings deviates from the direction of the crack. Mesh
This is the case for instance in the shear zone of profiled size Cross-section
beams. Name Roving [mm] area [mm2 /m]
The material properties, the test set-up and impor-
tant results of tensile tests are described below. Further MAG-07-03 0◦ 2400 tex 8.3 112
tests have been carried out in order to determine the 90◦ 2400 tex 8.4 112
load-bearing behaviour of textile reinforced concrete
structures under bending, shearing and torsion loading,
e.g. see [8]. Table 2 Mechanical properties of the ar-glass fabric

Tensile Young’s
2.1 Materials Name strength ft [MPa] modulus [MPa]

MAG-07-03 0◦ 974 60500

The specimen for the tensile tests which are de-
90◦ 533 62100
scribed in this paper have been reinforced with a
Materials and Structures (2006) 39:765–776 767

Table 3 Mechanical properties of fine grained concrete (PZ-

0899-01)
Young’s
Compression Tensile Modulus
strength [MPa] strength [MPa] [MPa]

78.3 4.4 34000

Fig. 2 Geometry and test set-up of the tensile tests.

Fig. 3 Photogrammetry measurement system.
r Fibre orientation (0◦ to 45◦ in steps of 5◦ and 90◦ );
proper penetration of the small gaps of the fabrics. The r Lamination of the fabric with epoxy resin and acry-
mechanical properties of the concrete are given in Table
late respectively.
3. Further information and other fine grained concrete
mixes suitable for TRC can be found in [10]. Each particular test with its specific material com-
bination and test set-up was conducted three times.
2.2 Test set-up and series of experiments
2.3 Results
The geometry and test set-up of the tensile tests is
shown in Fig. 2. The specimens have been produced in
Figure 4 shows the load-strain curve for a two-layered
an upright formwork, i. e. the concrete has been filled
reinforcement with the ar-glass textile MAG-07-03.
in from the top side. Using this production process the
The diagram contains both the continuously recorded
curvature of the specimen due to non-uniform shrink-
data of the LVDTs and the strains calculated from the
age of both sides of the specimen can be prevented.
In addition to the conventional LVDT measurement,
a photogrammetry measurement system was used in
some of the tests (Fig. 3). Thereby, the actual displace-
ment field was recorded frequently during the test, from
which the crack pattern within the measurement range
could be gained afterwards. The system consists of
three digital cameras which map the deformations in
adjustable intervals of 5 to 30 s. The precision of the
measurement data amounts to about 3 to 5 μm [11].
The load was applied with a constant deformation
rate of 1 mm/min. The following parameters were var-
ied in the test program:
Fig. 4 Load-strain curves and number of cracks in tensile test
r Reinforcement ratio (1.1% to 3.3%); (2 layers MAG 07-03).
768 Materials and Structures (2006) 39:765–776

Fig. 5 Crack width in tensile test (2 layers of MAG-07-03).

Fig. 6 Strains of inner and outer filaments.

displacement field provided by the photogrammetry

measurement system at the indicated load steps. In ad-
dition, the corresponding number of cracks is given.
The curve features the typical linear-elastic behaviour
as long as there is no cracking of the matrix. After
passing the bend-over point, multiple cracking occurs
and the load simultaneously increases. There is no dis-
tinct post cracking-region; new cracks form up to a load
of about 7 kN. The maximum crack width amounts to
about 0.12 mm, a short time before the failure (Fig. 5).
Fig. 7 Load-strain curves of specimen with 2 layers of MAG-
A typical phenomenon is the increase of the crack 07-03.
widths at the sides compared to the middle of the mea-
surement range. Furthermore, the failure occurs in this
region. This is a result of the different bond perfor- In contrast to the tensile tests on rovings (Table 2) in the
mance of the filaments depending on their position in tests of the composite, the rovings of the 90◦ - direction
the cross section of the roving. The outer filaments achieved about 20% higher strength than the ones in
(sleeve filaments) have direct contact to the matrix and 0◦ -direction. Obviously, the more flat cross section and
therefore feature better bond performance compared the loose binding of the 90◦ -direction rovings result
to the inner filaments (core filaments), which have no in better bond performances of the inner filaments and
direct contact to the matrix because of the low pene- to a more homogenous activation of the rovings total
tration depth. As a result the stress distribution over cross-section. The result is a higher effectiveness of the
the total cross section of the rovings is inhomoge- reinforcement. Hence, although the tensile tests of the
neous, whereby the outer filaments get higher strains rovings are an important method to control the material
and therefore higher stresses in the cracks than the in- quality, the strength measured in these tests cannot be
ner filaments (Fig. 6). Further explanation will be given transferred to the load-bearing capacity of the compos-
in section 3. ite without consideration of further parameters which
influence the bond performance, e.g. the binding of the
roving.
2.3.1 Influence of the binding

The effect of the different cross section geometries of 2.3.2 Influence of the reinforcement ratio
the rovings in 0◦ - and 90◦ -direction of the fabric has
been investigated in tests on specimens reinforced with The tensile tests have been carried out on specimens
two layers of MAG-07-03. The load-strain curves of that were reinforced with one, two and three layers of
two representative tests of each series are shown in Fig. the ar-glass fabric respectively. The reinforcement with
7. Obviously, the different geometry of the rovings due one layer of the fabric (At /Ac = 1,1%) was just able
to the binding leads to different load-bearing behaviour. to bear the first crack load of the specimen.
Materials and Structures (2006) 39:765–776 769

Fig. 9 Crack with inclined roving.

Fig. 8 Tensile strength of textile in specimens with different
reinforcement ratio.
radius of curvature a bigger filament diameter re-
sults in higher flexural stresses.
The results of the tests show that the strength of the (b) The inner filaments get lower strains because they
reinforcement decreases with increasing reinforcement do not align in the load direction as do the outer
ratio (Fig. 8). Obviously, the reinforcement layers in- filaments. This effect leads to a further increase
fluence each other and cause reduction in bond perfor- of the difference between the stresses of the outer
mance. and inner filaments, respectively. As a result, the
The discrepance between the ultimate strength of the effectiveness decreases.
roving taken from the fabric ft and the strength in the (c) The transversal stresses lead to a better bond per-
component σ max can be expressed by the effectiveness formance particularly of the inner filaments.
factor k1 which is defined as the ratio between both
values: It becomes evident that the situation resulting from
an inclined reinforcement direction is very complex.
σmax Up to now, the influence of each effect cannot be deter-
k1 = (1) mined separately. It is only possible to give the global
ft
reduction factor k0,α for the ultimate strength taking
into account these effects. The factor k0,α is defined
The effectiveness of the reinforcement arranged in as the ratio between the load–carrying capacity of the
one or two layers was about 0.49, and it decreased down reinforcement with and without inclination:
to 0.43 for the three-layer reinforcement.
σmax,α
k0,α = (2)
σmax
2.3.3 Influence of the fibre orientation

Two 20 mm long notches in the middle of the mea- The results of the experiments with the ar-glass tex-
surement zone reduced the width of the cross-section tile MAG-07-03 are given in Fig. 10.
of the specimens to 60 mm. The photogrammetry mea- With increasing fibre slope, the effectiveness of the
surement system was used to detect the crack opening fibres decreases to about 50% for an angle of 45◦ . It has
and the parallel crack shift. The maximum crack width to be considered that the influence of the fibre orienta-
in the tests was about 0.3 mm which is only about 15% tion may be affected by the properties of the rovings,
of the roving diameter. This has to be considered for so that the described results cannot be generally trans-
the representation of the geometrical properties in the ferred to other ar-glass fibre fabrics.
crack area (Fig. 9) and for the illustration of the main
effects, which are: 2.3.4 Influence of impregnation

(a) The transversal action and the bending result in In order to achieve a better activation of the inner fil-
damage of the filaments and a decreasing load- aments the ar-glass fabric has been impregnated with
bearing capacity. Thereby, the diameter of the fila- epoxy resin and with acrylate respectively. As it has
ment has a significant influence because at the same already been known from prior investigations [12] the
770 Materials and Structures (2006) 39:765–776

with
Fctu tensile load-bearing capacity
k1 factor for effectiveness of the textile
k0,α factor for the orientation of the rovings
At cross section area of the rovings
ft ultimate strength of the reinforcement
In case of biaxial tension loading the reduction of
the load-bearing capacity of the reinforcement has to
be considered [13]. Further experiments with different
loadings like bending and shearing have been carried
out in order to develop design methods for TRC struc-
tures [8].
Fig. 10 Reduction factors k0,α .

3 Modeling textile reinforced concrete

In general, a textile reinforcement features the follow-

ing three main differences compared to the load bearing
behaviour of an ordinary steel reinforcement in uniaxial
tension:

(a) The common textile materials, e.g. ar-glass or car-

bon, exhibit perfect brittle failure. There is no plas-
Fig. 11 Load-strain curves of tensile tests with coated textile
tic deformation capacity.
reinforcement.
(b) Since there is no profiling of the reinforcement
which anchors it mechanically to the concrete ma-
impregnation with epoxy resin leads to a significant in- trix, the bond solely relies on adhesion and friction.
crease in the ultimate strength of the fabric (Fig. 11). (c) If not glued together by an additional lamination
Furthermore, the epoxy resin impregnated reinforce- process (see for example Fig. 11), the hundreds of
ment revealed good bond properties which were de- filaments in the roving do not pick up the same
tectable from the multiple cracking and the low crack stresses when loading is applied to the composite
distances. The failure of the specimen occurred due to material; this is due to their different bond condi-
the splitting of the concrete in the load introduction tions.
zone along the reinforcement. Hence, the concrete was
not able to bear the bond stresses. In recent years, issues (b) and (c) have been cen-
The impregnation with an acrylate dispersion tral subjects of research. Because of the diversity of
yielded an increasing load-bearing capacity, too. But fibre materials, their coating (size) and their process-
the impregnation reduced the outer bond performances ing into textiles, it is not possible to give universally
which resulted in big crack spacings and big crack applicable parameters for TRC. It is rather necessary
widths. The failure was caused by the breaking of the to investigate each combination of reinforcement type
reinforcement. and concrete matrix in order to calibrate the parame-
ters of a chosen model. Therefore, the main purpose of
modeling TRC, at least until the current state of the art,
2.4 Design method is not the prediction, but the explanation of its load-
bearing behaviour and to provide data which cannot
Based on the test results, the uniaxial tensile load– be experimentally measured. In the scope of the joint
bearing capacity Fctu of the textile reinforcements cross research project “Collaborative Research Centre 532:
section embedded in concrete may be calculated as: Textile Reinforced Concrete - development of a new
technology” a consistent numerical modeling strategy
Fctu = k1 · k0,a · At · f t (3) has been developed. This strategy covers the micro-,
Materials and Structures (2006) 39:765–776 771

Fig. 13 Cross sections of embedded ar-glass rovings left: Plain

roving from bobbin right: Roving with textile binding.
Fig. 12 Bond-stress vs. slip of single filament [14].

3.2 Bond of a roving

meso- and macro levels. On the micro-level, the be-
haviour of a single filament embedded in the concrete The present investigations [1, 2, 17] showed that the
matrix is analysed. Based on these results, meso level tensile strength of the filaments in the composite ma-
models (Fig. 14) are used to simulate pull-out tests terial “Textile Reinforced Concrete” cannot, unlike a
of rovings and to draw conclusions about their crack- steel reinforcement, be fully exploited. The main rea-
bridging capability. In order to simulate TRC on larger son is that the filaments within the roving are, first,
scales, one needs a simplified model which takes into not continuously embedded and, second, not in contact
account the special characteristics of TRC in a smeared with the matrix along their whole perimeter as can be
manner. In the following, the three levels are briefly de- seen for instance in the micrographs Fig. 13. That is,
scribed whereby the transition from the meso level to only very few filaments can actually transfer the bond
the macro level is discussed in more detail. stresses determined in the single-filament pull-out tests,
or in other words, have a bond quality of 100%. In gen-
eral, their bond-conditions get worse from the outside
3.1 Bond of a single, perfectly embedded filament to the core of the roving. The distribution of the bond
quality across the cross section corresponds to the abil-
The adhesional and frictional bond of textile fibres ity of the matrix to penetrate the roving, which depends
was investigated by BRAMESHUBER et al. [14]. They on the following:
pulled single filaments out of a concrete matrix and
calculated slip vs. bond-stress relations for different r the rheologic characteristics of the matrix,
types of filaments using an analytic procedure mainly r the diameter of the filaments and
based on COXs shear-lag theory [15]. Fig 12 shows r the type of textile binding.
the result for a filament taken from a Vetrotex R
roving
(VET-RO-ARG-2400-01-03). There is adhesion up to A smaller diameter of the filaments leads to smaller
a relative displacement between filament and matrix gussets between them and therefore the penetration of
of 0.0004 mm. This type of filament with its special matrix is limited. E.g. carbon rovings with a four times
coating features only a relatively small stress-drop af- smaller diameter than ar-glass exhibit a much worse
ter loss of adhesion, i.e. there is a smooth transition to matrix penetration. The textile binding has a strong
the frictional bond range. The transferrable bond-stress influence as well. Plain rovings or those with a flat
of about 3.3 N/mm2 remains nearly constant with in- and loose binding feature “matrix-bridges” and have a
creasing slip between filament and matrix. relatively even distribution of the bond quality across
For the modeling on the meso- and macro-levels, the cross-section as it can be seen in Fig. 13 (left).
the measured bond-stress relation was idealized as A tight binding leads to a denser arrangement of the
elastic/ideal-plastic and parameterized by the critical filaments so that most filaments are insulated from the
slip s crit and the bond quality, which is a scaling fac- matrix, Fig. 13 (right).
tor reducing the transferrable bond-stress, if a filament The change of these conditions in the longitudinal
is not perfectly embedded [16]. direction of the rovings has not been examined yet by
772 Materials and Structures (2006) 39:765–776

Each of the three presented models qualitatively

provides the same stress distribution across the cross-
section of the roving. In all cases there is a successive
failure of the filaments starting from the outside and
propagating to the inside of the roving. In the models
Fig. 14 (c–d) some of the inner filaments may not break,
but be completely pulled out of the matrix, depending
on the embedment length.

3.3 Macro-level simulation

From a macroscopic point of view, only the load at

which the failure of the reinforcement starts is of major
interest, but not so much the exact shape of the descend-
Fig. 14 Bond models for rovings on meso-level.
ing branch. Therefore, the simulation of the load bear-
ing behaviour of TRC in uniaxial tension on the macro-
level does not require to model the successive failure of
any imaging technique. A qualitative sketch of a section the individual filaments in detail as described in the pre-
along a roving is given in Fig. 14 (a). vious section. In order to obtain realistic stress-strain
Figures 14(b–d) shows three models accounting for relations, crack distances and crack widths it is much
these characteristics so far proposed by different re- more important to describe the total load transfer be-
searchers. All of them consider the different free defor- tween matrix and reinforcement until the first filaments
mation lengths of the individual filaments to be the pre- break. For this purpose it is sufficient to consider only
dominating influence on the crack bridging behaviour two layers/groups of filaments (subrovings), which is
of a roving in a tensile test. In the Adhesive Cross Link- the minimum to describe the different bond conditions
age Model [18], Fig. 14 (b), the strain and stress dis- of the filaments (Fig. 15). This approach is similar to
tributions of the filaments are calculated by the crack the models proposed in [21, 22].
opening and the free deformation lengths only. As the Both the sleeve and the core subroving are contin-
crack width reaches a certain value, the outermost fil- uously linked to the matrix by zero-thickness bond-
aments with the lowest deformation lengths fail first, elements. The transferrable bond-forces per unit length
followed by a successive failure of the filaments with (bond-flow) are calculated as
longer deformation lengths. BANHOLZER [19] used a
very similar model in order to analytically describe A Rov
the load-displacement curve of roving pull-out tests. In T s (s) = τ f il (s) · U f il · · (1 − η) · q s (sleeve) (4)
A f il
contrast to the Adhesive Cross Linkage Model he does
not consider the linkages as fixed, but allows the fila- A Rov
T c (s) = τ f il (s) · U f il · · η · q c (core) (5)
ments to delaminate from the matrix. Thereby, it is as- A f il
sumed that all filaments are embedded around their en-
tire perimeter, i.e. the bond quality is 100%, Fig. 14(c).
The most advanced model is the Bond Layer Model,
Fig. 14 (d), developed by KONRAD et al. [20], which is
actually a further improvement of the two other models.
This uses a distribution for the bond-quality in the zone
with possible debonding (the bond-quality decreases
from the outside to the inside of the roving). In ad-
dition, the delayed activation, taking into account the
waviness of the filaments within the free length, has
been introduced. Thus, before the filaments pick up
load they have to overcome a certain amount of strain. Fig. 15 Two Subroving Model.
Materials and Structures (2006) 39:765–776 773

with a continuous one. The behaviour of the Two Sub-

roving Model is therefore dominated by bond-stress slip
relations in contrast to the meso-level models which
are dominated by the free deformation lengths of the
filaments. A macro-level simulation with two subrov-
ings, but based only on free lengths according to Fig.
14 (b) is not possible, because such a model would
not give any information about the stress transfer from
the roving to the matrix, which is necessary to sim-
ulate multiple cracking and to obtain realistic crack
distances. The mixed free-length/continuous-bond ap-
proaches (Fig. 14(c–d)) are not suitable as well, since
Fig. 16 Material-laws for the subrovings. the positions of cracks developing during a tensile test
are not known a priori. At each crack one would have to
whereby: introduce the free lengths by an expensive remeshing
T = bond-flow procedure. Compared to the analytical model of OHNO
τ (s) = bond-stress (see Fig. 12) and HANNANT [21] there are several differences and
U f il = perimeter of one filament advantages, respectively:
A f nil = cross section of one filament
A Rov = cross section of one roving r The scatter of the tensile strength of the matrix can
η = fraction of the core-subroving be taken
q = bond-qualities of the core and the sleeve, into account.
respectively. r Arbitrary bond/slip-relations can be applied.
In the model, the continuous, “real” stress distri- r The core-subroving is connected directly to the ma-
bution is replaced by two stress blocks representing trix, according to the meso-level models.
the same total force. Thereby, the two subrovings are r The delayed activation can be considered.
assumed to feature a linear-elastic behaviour. The ini- r The cracks are not a priori supposed to be axes where
tial waviness of the filaments is taken into account by core and sleeve share the same displacement.
a strain ε 0 (delayed activation) which has to be over-
comed before stressing the core-subroving. The sleeve- The latter is for example not the case in the ar-
subroving is stressed directly, because lateral adhesion eas between measurement and load introduction zones.
and contact pressure respectively keep them in their There, a transition from the regular tensile behaviour to
initial position (Fig. 16). pull-out behaviour takes place (Fig. 6). As it has already
Since the stress of the sleeve-subroving is smaller been described in [23], this leads to an increase of the
than the stress of the outmost filaments, it cannot be stresses of the sleeve-subroving whereas the stresses of
used to determine the point at which the succesive fail- the core-subroving decrease.
ure starts. For this reason, an additional single-filament
with perfect bond (100% bond-quality) was introduced
in the model. This filament has no influence on the sim- 3.3.2 Homogenization of meso-level parameters
ulated load-strain curves, because it has only a very
small cross-section. It is rather used to find the point of In order to verify that the macro-level model is in prin-
failure, which is assumed to happen when the outmost ciple capable to provide the same relation between the
filaments break. crack opening and the crack bridging force as the Bond
Layer Model, a series of numerical simulations has
3.3.1 Discussion of the two-subroving-model been conducted. First, pullout tests with an embed-
ded length of 30 mm have been calculated with the
Besides the consideration of only two subrovings, the Bond Layer Model applying various parameters for the
main simplification of the model is the replacement of maximum delayed activation and the maximum free
the discontinuous connection to the matrix (Fig. 14 (a)) length while the distribution of the bond quality was
774 Materials and Structures (2006) 39:765–776

Fig. 17 Pull-out test simulations.

Fig. 19 Critical slip of the core in the Two Subroving Model vs.
the maximum delayed activation and maximum free length in
the Bond Layer Model.

Knowing the relations between the models it is pos-

sible to predict the influence of a change of the condi-
tions at the meso level on the load bearing behaviour
of TRC on the macro-level, i.e. in tensile tests. Hence,
the models provide important information about how
to modify the textiles, for instance by chemical mod-
ifications of the filament-surface or by changes of the
textile binding in order to get a different penetration
behaviour of the matrix.
The Two Subroving Model can of course also di-
rectly be calibrated by the results of the tensile-tests.
Fig. 18 Bond quality of the core in the Two Subroving Model
vs. the maximum delayed activation and maximum free length
This is still necessary, because the Bond Layer Model
in the Bond Layer Model. has not been calibrated yet to all types of rovings and
especially not for rovings after the processing of the
fabric, which show a different penetration of the ma-
kept constant. The simulations were controlled by the trix (Fig. 13). As a proof of the capability of the Two
pull-out displacement u (Fig. 17(a)). Subroving Model to simulate the behaviour of TRC-
Second, an automated calibration routine was used specimens in uniaxial tension (Fig. 2), the results of
to determine the bond law to be applied for the sleeve the FE-model and the experimental data are compared
fibre as well as the bond law and delayed activation for in Fig. 20 and Fig. 21. The specimen were reinforced
the core fibre in the Two Subroving Model (Fig. 17(b))
which lead to the same results. Thereby the displace-
ment was increased up to 0.05 mm, which is about half
the failure crack-width observed in tensile tests. The
fraction of the core-subroving was chosen as 75% of
the cross-sectional area of the roving.
As a result of the analysis, the dependencies of the
parameters of both models are resolved. Fig. 18 is an
example showing the bond quality while Fig. 19 il-
lustrates the critical slip of the core fibre in the Two
Subroving Model depending on the maximum delayed
activation and maximum free length in the Bond Layer Fig. 20 Measured and simulated load-strain curves (2 layers of
Model. MAG-07-03).
Materials and Structures (2006) 39:765–776 775

investigations is the characterization of these factors to

increase the load-bearing capacity.
The explanation of the load–bearing behaviour of
TRC is a task that can only be handled if different res-
olution levels in the experiments as well as in the sim-
ulation are considered. In this research the behaviour
of TRC is investigated at three different levels: micro-,
meso- and macro-level. It was shown how results ob-
tained at one level can be used at the next higher level.
Fig. 21 Measured and simulated crack-development (2 layers This way of proceeding can reduce several uncertain-
of MAG-07-03). ties of the straightforward calibration of parameters on
each level separately. The Two Subroving Model was
presented, and initial analysis proves its capability to
simulate the experimental results. However, systematic
experimental research has to provide further informa-
tion in order to verify and optimise the models at each
level.
In addition, simulation techniques have to be devel-
oped taking into account the influences of impregna-
tions and of the fibre orientation. Both issues have so
far been analysed experimentally yet.
Fig. 22 Maximum-stress of the single-filament with 100% Near numerous analogies to the ordinary steel rein-
bond-quality. forced concrete, TRC shows specific differences in the
load-bearing behaviour in some points which cannot be
with two layers of MAG-07-03. In addition, the cal- neglected within the practical application. The material
culated maximum stress of the single-filament with a models currently being developed and presented in this
bond-quality of 100% is given in Fig. 22. Right before paper will lead to a better understanding of the failure
the whole tensile specimen fails, the single filament mechanisms of TRC-structures. Based on these mod-
reaches its ultimate strength of about 1400 N/mm2 . The els, design rules and safety concepts can be set up in the
calculation was conducted with a fraction of the core- future. At the present level of knowledge these points
subroving of 90%, a bond quality of the core of 5.5% can be already named, nevertheless, quantitatively
and a bond quality of the sleeve of 15%. they must be determined with experimental investiga-
tions. Such investigations must not hinder the practi-
cal application, but also lead to a high optimisation
4 Summary degree.

The mechanical qualities of Textile Reinforced Acknowledgements The authors thank the Deutsche
Forschungsgemeinschaft (DFG) in context of the Collaborative
Concrete (TRC) are not sufficiently known yet. The Research Center 532 “Textile Reinforced Concrete” for their
load-bearing behaviour of TRC is influenced by ma- financial support.
terial, amount and orientation of the textile reinforce-
ment and the fine concrete matrix. In particular the
bond behaviour of the textiles is important. Models References
known from the steel reinforced concrete cannot be
applied without additional considerations, because of 1. Hegger J (2001) Textile reinforced concrete. Proceedings
the inhomogeneous cross sections of the rovings which of First Colloquium of Collaborative Research Centres 528
and 532, (RWTH Aachen University) “only available in
show basically other bond behaviour than reinforcing
German”.
bars. However, the attempts have shown that the ten- 2. Curbach M (2003) Textile reinforced structures. Proceed-
sile strength of the textiles cannot be fully exploited in ings of the 2nd Colloquium on textile Reinforced Structures
the composite material TRC. The purpose of the actual (CTRS2), (Dresden, 2003) “only available in German”.
776 Materials and Structures (2006) 39:765–776

3. Roye A, Gries T, Peled A (2004) Spacer fabrics for thin 13. Hegger J, Voss S (2004) Textile reinforced concrete under
walled concrete elements. Proceedings of the Sixth Rilem biaxial loading, proceedings of 6th RILEM symposium on fi-
Symposium on Fibre Reinforced Concrete (FRC), Varenna, bre reinforced concrete (FRC), BEFIB 2004, Varenna (Italy),
Italy, 20–22 September, 1505–1514. 20–22 September. 1463–1472.
4. Brameshuber W, Koster M, Hegger J, Voss S, Gries T, Barlé 14. Brameshuber W, Banholzer B (2003) Bond characteristics of
M, Reinhardt H-W, Krüger M (2003) Textile reinforced con- filaments embedded in fine grained concrete. in Textile Re-
crete (TRC) for integrated formworks. Proceedings of the 12. inforced Structures, Proceedings of the 2nd Colloquium on
Intern. Techtextil-Symposium, Frankfurt a. M., 7–10. April Textile Reinforced Structures (CTRS2), Dresden, Germany,
2003. pp. 63–76.
5. Hegger J et al. (2003) Decentralized wastewater treatment 15. Cox H-L (1952) The elasticity and strength of paper and other
plants made of textile reinforced concrete. Proceedings of the fibrous materials. British Journal of Applied Science 3:72–
12. International TechTextil-Symposium, Frankfurt, 2003, 79.
CD-ROM, Nr. 4.28. 16. Hegger J (2002) Collaborative Research Center 532–Textile
6. Hegger J, Schneider H et al. (2004) Exterior cladding pan- Reinforced Concrete–development of a new technology. Re-
els as an application of textile reinforced concrete, in ‘Thin port of the periode II/1999-I/2002’ (Institute of Structural
Reinforced Cement Based Products and Construction Sys- Concrete, RWTH Aachen University) “only available in Ger-
tems’, SP-224, (American Concrete Institute, Farmington man”.
Hills, Michigan, USA) 55–70. 17. Hegger J, Döinghaus P, Will N (1998) Textile reinforced
7. Hegger J, Brameshuber W, et al. (2004) Small sewage treat- concrete—load bearing behaviour and bond properties. in
ment plants made of textile reinforced concrete. Concrete Progress report on the application of textiles in structural con-
Plant + Precast Technology 70(1): 52–60. crete (German Committee for Reinforced Concrete (DAfStb)
8. Hegger J, Voss S (2003) Load-bearing behaviour of textile- (488)) 81–90 “only available in German”.
reinforced concrete under 2-axial loading. in Textile rein- 18. Schorn H (2003) A bond model for glassfiber-reinforcements
forced structures, Proceedings of the 2nd Colloquium on in concrete Bautechnik 80(3):174–180 “only available in
Textile Reinforced Structures (CTRS2), Dresden, pp. 313– German”.
324. 19. Banholzer B (2004) Bond behaviour of a multi-filament
9. Hanisch V, Roye A, Gries Th (2004) Characterization of yarn embedded in a cementitious matrix. Phd-thesis, RWTH
textile structures for use in concrete - Part 1: Strength of 2D Aachen University, 2004.
structures and initial stages of simulation and deformation. 20. Konrad M, Chudoba R (2004) The influence of disorder in
Technical Textiles 47: 104–E105. multifilament yarns on the bond performance in textile rein-
10. Brockmann T (2001) Requirements and properties of cement forced concrete. Acta Polytechnica 44(5–6):186–193.
based fine grained concretes. in Textile Reinforced Concrete. 21. Ohno S, Hannant DJ (1994) Modelling the stress-strain re-
Proceedings of First Colloquium of Collaborative Research sponse of continuous fibre reinforced cement composites.
Centres 528 and 532, RWTH Aachen University, 2001, 83– ACI Materials Journal 91(3): 306–312.
98, ISBN 3-980 7302-1-2. “only available in German”. 22. Bartos P (1987) Brittle-matrix composites reinforced with
11. Benning W, Lange J, Schwermann R, Effkemann C, Görtz bundles of fibres. Proceedings of the First International
S (2004). Monitoring crack origin and evolution at con- RILEM Congress from Materials Science to Construction
crete elements using photogrammetry, in ‘ISPRS Proceed- Materials Engineering. (2): Combining Materials: Design,
ings XXXV Part B’, Proceedings of XXth congress of IS- Production and Properties, 539–554.
PRS (International society for Photogrammetry and Remote 23. Hegger J, Bruckermann O, Chudoba R (2004) A smeared
Sensing), Istanbul (M. Orhan Altan, 2004) 678–683. bond-slip relation for multi-filament yarns embedded in fine
12. Krüger M (2004) Prestressed textile reinforced concrete, concrete. Proceedings of the Sixth International RILEM
Ph.D.-Thesis, Institute of Construction Materials, University Symposium on Fibre-Reinforced Concretes (FRC) BEFIB
of Stuttgart, 2004 “only available in German”. 2004, Varenna, Italy, pp. 1453–1462.