Professional Documents
Culture Documents
State of The Art TRC1
State of The Art TRC1
3.2.2 Scrims................................................................................................................................ 17
4.4.1.5 Shrinkage....................................................................................................................... 35
5.7.1 Spacers............................................................................................................................... 71
5.7.2 Curing................................................................................................................................ 71
6.1.3.5 Comparison of calculations and tests of textile reinforced elements in a pull-out test . 95
7.3.2 Influence of the roughness of the new-old concrete joint ............................................... 174
8.5 Decentralized wastewater treatment plants made of textile reinforced concrete............ 184
8.9 Exterior Insulation and Finish Systems EIFS (C. Aldea) ................................................. 188
8.10.2 Elements of the system: fabric and matrix/joint compound ........................................ 190
ABSTRACT: In this chapter the different materials are characterized ....abstract is missing.
1.1 Definitions
Fabric - A manufactured planar textile structure made of fibres and/or yarns assembled by various
means such as weaving, knitting, tufting, felting, braiding, or bonding of webs to give the structure
sufficient strength and other properties required for its intended use.
Roving
a) In spun yarn production, an intermediate state between sliver and yarn. Roving is a condensed sliver
that has been drafted, twisted, doubled and redoubled. The product of the first roving operation is
sometimes called slubbing.
Textile - originally, a woven fabric; now applied generally to any one of the following:
a) staple fibres and filaments suitable for conversion to or use as yarns, or for the preparation of
woven, knit, or non-woven fabrics;
c) fabrics and other manufactured products made from fibres as defined above and from yarns.
d) garments and other articles fabricated from fibres, yarns, or fabrics, when the products retain the
characteristic flexibility and drape of the original fabrics.
Yarn - A generic term for a continuous strand of textile fibres, filaments of material in a form suitable
for knitting, weaving or otherwise intertwining to a form of textile fabric. Yarn occurs in the following
forms:
8
e) a narrow strip of material, such as paper, plastic film of metal foil, with or without twist, intended
for use in a textile construction.
Note: direct roving and assembled roving are specialised terms, which are typical for glass fibres.
Although I could provide a definition, I did not find any in the dictionary, and therefore I preferred not
to include any on the list.
Tex -
1.1.2 Matrix
Matrix - The matrices used for TRC have to meet special demands regarding production process,
mechanical properties of the composite and durability of the reinforcement material. In most cases
small maximum grain sizes (< 2 mm) are used and hence these matrix systems can be considered as
mortar. On the other hand these matrices offer high performance properties in many respects and are
used for a construction composite material, such that these matrix systems are also called fine grained
concrete (or fine concrete).
The idea of a separate chapter with definitions is an excellent one. But the current list should be
expanded considerably. Some suggestions:
Tex – It is defined in Chapter 3 (Sect. 1.1.3), but should be clearly defined here as well. Also, denier is
often used in the industry and should be defined here.
We should establish a clear terminology regarding what is a fiber, a filament (the same?), a roving,
and a strand (the same?). Then the body of the report should be scanned to make sure, a uniform
terminology is used throughout. Right now, it is far from it.
REFERENCES
9
2 INTRODUCTION (W. BRAMESHUBER)
Working party “Bond of TRC”.
Wolfgang Brameshuber, Institute of Building Materials Research at Aachen University (ibac),
Germany
ABSTRACT: In this chapter the principle bond behavior of a strand .... is discussed ...
10
3 TEXTILES (T. GRIES)
Working party “Textiles”.
Gries; T.; Roye, A. ITA at RWTH Aachen University, Germany
Offermann, P.; Engler, Th. ITB at Dresden University, Germany
Peled, A. Structural Engineering Department at Ben Gurion University, Israel
The use of alkali-resistant man-made glass fibers (AR-glass), carbon and aramide for the design and
fabrication of textile reinforcements essentially meets these requirements (Fig. 3.1). The make-up and
typical properties of these fiber materials are described in the following paragraph. Fig. 3.2 compares
the most important mechanical parameters of these fiber materials. By varying the yarn fineness (see
paragraph 3.1.2, make-up of yarns) the volumes of the examined filament yarns are approximately the
same.
Further, at least partly suitable polymers like polypropylene, polyvinyl alcohol, polyethylene, and
polyacrylnitrile are already known as short fiber reinforcements for concrete. Basalt fibers can also be
11
used for this purpose. Reinforcing fabrics made of metal yarns (steel fiber yarns) are also possible, but
they have a bad cost-performance-ratio. Short steel fibers are nevertheless widely used in steel fiber
concrete.
Fig. 3.2. Mechanical properties of selected high-performance filament yarns (see relevant yarn and
textile fabric properties).
a) b) c)
Fig. 3.3. Yarn construction: a) filament yarn, b) bundled yarn, c) foil fibrillated tape.
Filament yarns are a bundle of elementary fibers, so-called filaments (Fig. 3.3 a). One yarn consists of
several hundreds up to thousands of single filaments. Therefore, the fineness of the yarn, indicated as
tex (gram per 1000 meters), depends on the number of filaments, the average filament diameter and
the fiber density.
12
The stress-strain relations of the filament yarns are significantly influenced by the location of the
single filament in the yarn and by the interactions between the filaments. Ideally all filaments in the
yarn should be parallel and drawn, divergences however may occur depending on the method of
fabrication, quality and intended use. The method of fabrication divides the yarns into assembled and
non-assembled yarns. Assembled yarns are made in a two-step process and consist of several
combined ropes of filaments, whereas non-assembled yarns include only elementary fibers.
The properties of the filament yarns are decisively influenced by the applied size, a device to improve
the processing behavior. Relevant for building purposes is the differentiation between water dispersing
and integrated sizes. The size has a great influence on the quality of the adhesion between the
filaments and it influences the load bearing performance Therefore, intensive research is aiming at the
development of new yarn sizes and coatings. The objective of it is to better exploit the high
mechanical potential of the elementary fibers. Another way to maximize the fiber effectiveness is the
combination of different materials, see section 3.1.6. Fig. 3.3 b shows an example for such a combined
yarn. Further yarn types like polypropylene foil fibrillated tape (Fig. 3.3 c) are used to replace
expensive fiber materials only if needed for manufacturing purposes. For processing on textile
machinery the yarns are stored on bobbins. A typical bobbin for glass filament yarns (roving) is the
roving bobbin, where the yarn is pulled off from inside (Fig. 3.4 a). As a result one twist per round is
applied on the yarn. Cross-wound bobbins with a cylindrical core ensure an almost twist-free
tangential pull-off (Fig. 3.4 b). These bobbins are used for carbon and aramide filament yarns.
a) b)
Fig. 3.4. Bobbin make-up: a) pull-off from inside, b) twist-free tangential pull-off.
The industrial fabrication of AR-glass filament yarns with high zircon shares was first started by
Nippon Electric Glass (NEG) Co., Ltd. in 1975. Since then short fibers (cut filaments) have been used
to reduce crack propagation.
The basic materials silica sand, clay and limestone are melted at temperatures up to 1350 °C and
pulled off the spinning nozzle with a speed between 25 and 150 m/s and diameters ranging from 9 to
27 µm. After spinning a size (0.5 to 1.5 mass-% of the fiber) of organic polymers dispersed in water is
applied on the filaments, 400 to 6600 of whom are combined to yarn without twisting.
The size is important for the following processes and improves the yarn properties as well as the
adhesion and the stress propagation between filaments and matrix material. The basic mechanical
13
properties of AR-glass are - depending on the yarn fineness - tenacity up to 1400 N/mm² linear elastic
elongation up to 2 % with modulus of elasticity between 70 and 80 kN/mm². The density of
2.8 kg/dm³ is relatively high compared to carbon or aramide. Regarding the good adhesion to concrete
AR-glass shows a good cost-performance-ratio at a price of about 24 €/dm³. Table 3.1 compares
commercially available AR-glass filament yarns.
Basic materials are polyacrylnitrile and meso phase pitch. They can be made spinnable by
polymerization and thermal treatment. The anisotropic PAN or pitch fiber from the spinning process
becomes unmeltable in an oxidation procedure at temperatures of 200 to 300 °C. The carbonization
defines the properties of the carbon fibers. Fibers of high tenacity (HT-fibers) are created from PAN at
1500 to 1700 °C and fibers of high modulus of elasticity at 2200 to 3000 °C by graphitization
incandescence. The tenacity of HT-fibers lays between 3000 and 5000 N/mm², the modulus of
elasticity between 200 and 250 kN/mm². HM-fibers range from 2000 to 4500 N/mm² in tensile
strength and 350-450 kN/mm² in modulus. Tests on commercial carbon fiber yarns have shown yarn
tenacities of more than 2000 N/mm² depending on the fineness.
Further positive properties for technical uses are low density (1.8 kg/dm³), very little creeping, good
vibration reduction, low heat expansion and conductivity, good electric conductivity and furthermore
low X-ray absorption as well as high resistance to acid, alkaline and organic solvents. The most
common matrix materials are duroplastic and thermoplastic polymers. Mineral matrix systems like
concrete are special cases for the use of carbon.
The adhesion to concrete is not as good as that of AR-glass. Special sizes have to be implied to
improve the use of filament properties in the yarn, the fabric and the building component. Having
solved this problem carbon will have an increasing importance for reinforcing concrete despite its cost
of more than 40 €/dm³, which is very much depending on the fiber properties. Carbon fibers are much
more difficult to process without deterioration than other fiber materials. Due to the sensitivity to
lateral pressure filament breaks occur more often despite large-scale modifications of the thread-
guides. Since the broken fibers are electrically conductive, all electronic devices and machine parts
have to be shielded. Modern fabrication technology, however, allows the economical production of
carbon fabrics with reproducible and predictable properties.
14
Table 3.2. Selected commercially available carbon fibres
Product type Producer [tex] Diameter [µm] Number of filaments
Tenax HTA 5131 400 tex 400 7 6000
Tenax HTS 5631 800 tex Tenax Fibres 800 7 12000
GmbH
Tenax STS 5631 1600
1600 7 24000
tex
T300J-3000 Toray Industries. 198 7 3000
T300JB-6000 Inc. Japan 396 7 6000
Du Pont began to sell aramide fibers in 1966. It is mostly used for fiber reinforced polymers, electric
insulation, protective clothing as well as for cables, ropes and belts.
Because of their excellent tensile strength and modulus of elasticity aramide fibers also belong to the
high performance fibers. Concerning their properties two types can basically be differentiated: meta-
aramide and para-aramide. Since it provides better mechanical stability, para-aramide is especially
interesting for reinforcements. It exists as normal (N), high tenacity (HT) and high modulus (HM)
version and is easily distinguishable from the white meta-aramide by its golden color. The tenacity is
about 3000 N/mm², the modulus of elasticity between 60 kN/mm² (N-version) and 130 kN/mm² (HM-
version).
Compared to carbon and glass fibers aramide offers a very low density (1.4 kg/dm³), a lower
brittleness and a higher impact resistance. The cost depends on the yarn fineness and the modulus and
is about 42 €/dm³ for the N-version.
Another problem is the negative heat expansion. If the heat expansion of reinforcing material and
concrete are different, tensions arise in the adhesion zone of both components. The use of such
composites is therefore very limited.
A disadvantage is its low resistance against alkaline solutions. But the comparatively new fiber type
Technora® (Table 3.3) offers a much better performance and even shows higher tenacities of about
3500 N/mm².
15
Table 3.3. Selected commercially available aramide fibers.
Product type Producer [tex] Diameter [µm] Number of filaments
Technora T-240 167 tex 167 12 1000
Technora T-241J 167 tex Teijin Twaron 167 12 1000
Twaron T3200 BV 322 12 2000
Twaron 2520 2520 12 k. A.
Other combined yarns can be manufactured by co-extruding. This means that the core yarn is spray-
coated with another polymer. These yarns have a very smooth surface which reduces the adhesion
with concrete. To improve this, additional yarns can be wound around the reinforcing yarn
(Fig. 3.3 b).
3.2.1 Introduction
There are several advantages of reinforcing concrete by means of fibrous materials in the shape of
textile fabrics. Positive from a technical point of view is the suitability of textiles for reinforcement of
even and curved components near to the surface in different load directions. The easy and reproducible
positioning of yarns in the shape of textile fabrics is furthermore an economic profit.
From the variety of textile fabrication processes are only a few left which can be used for concrete
reinforcements. The most important criterion is the possibility to create open structures with high
displacement stability. A good permeability and complete envelope with the concrete is provided by
an open grid structure. To ensure satisfactory handling there must be no displacement of the threads.
Certain manufacturing methods (e.g. warp knitting with insertion of reinforcing threads) are therefore
better suited than others (e.g. braiding).
Fig. 3.5 systematically shows the most common textile fabrics for reinforcing concrete. The largest
amount of fabrics are bi- or multi-axial warp knits since they offer a great flexibility of properties and
are suited for many uses. For a complete overview of textile fabrics and structures see [Ber2000].
16
Fig. 3.5. Overview about the commonly used semi-finished reinforcement products for the
application in concrete.
3.2.2 Scrims
Multi-axial scrims offer a broad variety of properties: drawn threads, any combination of angles
between the layers, any design of layers and a free choice of the weight per unit area. The structure of
a multi-axial scrim consists of several layers of reinforcing thread systems with different orientations
and a knitting thread structure – the warp knit. The knitting thread bonds the thread layers and forms a
warp knitted scrim. By varying the reinforcing thread systems and the mesh structure (fringe, tricot
etc.), the character of the scrim can be adapted to its use.
Fig. 3.6 shows an exemplary lay-up of a multi-axial warp knit. It consists of up to eight layers which
can be adjusted in almost any direction (e.g. 0°, 90°, +45°, -45°). The 0° direction is the machine
direction, the so-called warp direction. The threads laid in other angles are called weft thread systems.
17
3.2.2.2 Circular scrim s
Circular scrims (Fig. 3.7) are tubular textiles. The manufacturing is divided into a laying and a
bonding section. Several bonding principles, like warp knitting, gluing and welding, can be used.
Materials used for ultrasonic welding must contain at least one thermoplastic component. One system
of threads with a thermoplastic element is enough. The other tread system can be made of alternative
polymers. The thermoplastic component is melted by ultrasonic waves. By applying these ultrasonic
waves and pressure the yarn crossings are welded and fixed. The diameter of the textile fabric is
determined by the diameter of the welding unit. Circular scrims are used for reinforcing hollow
sections (tubes, cable pits), cut as planes or formed as tee-sections.
Fig. 3.7. Machine for the manufacturing of circular laminated fabrics by means of ultrasonic
welding at the ITA, textile cross section diameter 120 mm.
These 3-dimensional spacer warp knits are used to reinforce thin concrete elements, whereas both
fabrics are positioned near to the surface.
To produce textile structures with a certain contour, the space between the both fabrics can be
variable. Both fabrics can be manufactured independently. That means that every side can consist of
18
different materials and have different mesh patterns as well as different weft thread densities
(Fig. 3.9).
25 mm 20 mm 25 mm
Fig. 3.9. Spacer fabrics bi-axial weft insertion on both sides and variable fabric distance.
Fig. 3.10 shows a circular warp knit with a bi-axial reinforcement in 0° and 90° direction. By that the
0°-threads are enveloped by a skipped fringe pattern and the crossing points are fixed. With this
method all reinforcing threads are tied up according to the mesh pattern. The diameter of the warp knit
is determined by the size of the needle bar. The diameter in this example is 80 mm.
19
Fig. 3.10. Circular warp knits with bi-axial weft insertion.
There are three basic weaving patterns – plain, twill and atlas weave (Fig. 3.11). Plain weave has the
shortest floating and a very high displacement stability. The floating grows from twill to atlas whereas
the displacement stability decreases which may cause the threads to skip from their position in later
process steps. There is a tendency, however, that with a larger floating the fabrics can be draped more
easily.
For reinforcing concrete elements with woven fabrics technical wovens like flat wovens as well as
multilayer ones can be interesting. Flat wovens of full or half cross weave are mostly used to stop
crack propagation in plaster mortar. Other fields for woven fabrics are the reinforcement of foam,
gypsum and wood elements as well as noise and heat insulating sandwich structures.
Because it is necessary to have an open structure with drawn reinforcing threads the stability of the
cross weaves has to be improved by an anti-displacement finish. Commercially available reinforcing
woven fabrics are made of AR-glass filament yarn. The structure is stabilized by impregnation with a
plastic matrix which also increases the chemical resistance. The fabric can be individually designed
for its use.
Multilayer woven fabrics consist of several layers linked by a binding warp thread. This warp ensures
a defined space between the layers. The pattern of the layers is variable and can be constructed for
every layer separately. Multilayer wovens have thread systems in every dimension.
20
Plain weave Twill weave Satin weave
The textile structures have to be fixed in their ideal position, which is best done during the production
process. Coating, impregnating and the insertion of an additional backing are possibilities of
implanting suitable binding agents. Fig. 3.12 describes the influence of a supplementary fixing on a bi-
axial scrim (weight: 8 g/m², distance between reinforcing threads 7.2 mm). The test was carried out
with the bending test device described in paragraph 3.3 – ??which paragraph is meant??. During
thermo-bonding, the warp knitting threads made of polypropylene are melted under pressure and heat.
Coating and impregnating are based on self-interlacing carboxyl styrene-butadiene copolymers
dispersed in water [Mäd2003].
Fig. 3.12. Width related bending stiffness for different structural strengthening processes Yarn and
textile properties relevant for practical use.
21
filaments of the yarns embedded in the composite material are involved in the load transmission. The
yarn behaviour in the concrete matrix can be estimated if the inner filament friction is activated during
the determination of the yarn strength. The mechanical properties of the textile fibres, yarn and fabrics
are not only used for modelling and developing textile pre forms but also for the further development,
optimisation and quality assurance of raw materials and textile fabrics.
The diameter of a single filament that is needed to determine the mechanical composite properties, e.g.
tensile strength, can be measured by Vibromat ME using the vibration method. The yarn fineness is
determined using a gravimetric method according to DIN EN ISO 2060. The hydrometer is used for
measuring the material density (ambient conditions). The stress-strain-behaviour of the single filament
is measured by the tensile test machine Fafegraph ME made by Textechno Herbert Stein GmbH & Co.
KG. The tensile test is performed according to DIN EN ISO 5079. The free clamping length of the
filaments varies between 60 and 100 mm because of the reduction of the failure near the clamping area
and the reduced influence of the test set-up as a result of the strain measurement.
The determination of the stress-strain-behaviour of alkali-resistant glass filament yarns is based on the
international standard ISO 3341. The modified test set-up for the determination of tensile strength and
breaking strain of glass filament yarns has been developed at the Institute of Textile and Clothing
Technology, cf. Fig. 3.13.
radiussed clamp Test conditions
free gauge length
Fig. 3.13. Test set-up and test conditions for the determination of stress-strain-behaviour of filament
yarns
This experimental set-up allows tensile tests on all high-performance filament yarns. The test is
performed at the Zwick material testing machine, type Z100 with special return clamps and external
strain measuring equipment. The significant advantages of the test set-up and the test realisation
(Fig. 3.13) are:
• high sensibility
22
• direct measurement of the strain
• good reproducibility.
Tensile tests on wetted alkali-resistant glass filament yarns can also be performed under the same test
conditions without the adjustment of the reference atmosphere (relative humidity). In order to wet the
yarns completely the specimens are laid in water for at least one hour. After this, the wetted yarn is
clamped and the wet tensile strength is determined in a standard tensile test. The test set-up is equally
suitable for short-term continuous and alternating loads.
The highest tensile strength of the filament yarn is significantly lower than the total tensile strength of
all single filaments contained (Fig. 3.14). One of the reasons for this is the unequal loading of the
single filaments within the yarn and the existing material defects [Cur1999].
As mentioned earlier, the results of the yarn tests and the tests performed on the composite material
depend on the bond properties between the filaments. The yarn test uses special wire clamps to initiate
loads through friction between yarn and deflection roller or to achieve adhesion between the filaments.
The initiation of loads is realised according to the mechanism of the initiation of tensile loads in
reinforced concrete specimen [Off2001]. This confirms the consistency of the results that are obtained
by both methods, cf. Fig. 3.14.
2000
Stress [ N/mm²]
1500
1000
500
∅
[µm] 13,5 13,5 16 24
0
NEG-ARG NEG-ARG NEG-ARG NEG-ARG
310 tex 620 tex 1100 tex 2500 tex
Fig. 3.14. Dependence of tensile strength on the filament diameter or the yarn fineness in alkali-
resistant glass filament yarns made by Nippon Electric Glass (NEG).
To examine the bending behaviour of the open grid structures a specially developed bending test with
vertical specimen arrangement is used (Fig. 3.15). The specimen is clamped perpendicularly, and the
bending edge is situated in the centre of rotation. The bending results from the rotation of the
specimen against a fixed bar. The pressure load on the bar is measured as a turning moment. Any
influence of gravity is practically avoided by this arrangement. According to Engler et al. 2004
23
[Eng2004] this new method can display the differences between open grid reinforcing textiles better
than the usually applied Cantilever test (DIN 53362). To reduce as much outside influence as possible
an electronically controlled testing device is currently being developed at the ITB.
Other sorts of high performance filament yarn can be tested with this method as well. Main advantages
of this test are high sensitivity, easy sample preparation, no clamp problems, direct elongation
measurement, great number of tests reproducible results with little straggling.
To evaluate the impact of characteristic influences on the tensile strength like length or waviness of
the filaments, the load can be passed into the filament yarns through resin blocks. By this method a
very short and exactly defined free clamping length can be realised (Fig. 3.16).
The weight per unit area is described in DIN 53884. The weight of the warp knitting thread can be
measured with the amount of thread being delivered to the warp knitting machine. To calculate the
area related fiber volume, the glass density of the filament yarn is measured with a hygrometer at
20 °C.
24
bending behavior of open grid structures. The influence of gravity is nearly eliminated with this
sample arrangement.
Another wax to evaluate the handling qualities of textile reinforcing structures is the examination of
the displacement stability (Fig. 3.17). A circular clamp is rotated in the sample which is fixed in its
position by a needle bed and two circular rubber coated clamps. This takes account of the area
character of the textile fabric. The resistance is measured as the moment of torsion at a defined angle.
Recent tests have shown that the displacement test can describe structural differences resulting from
the textile construction and from the bonding properties of the yarns.
a) b)
Fig. 3.17. Experimental determination of the displacement stability, a) non-deformed structure, b)
deformed structure.
REFERENCES
[Ber2000] Bernd, E., Geuer, M., Wulfhorst, B.: Dreidimensionale Textilstrukturen zur Herstellung
von technischen Textilien - Stand 2000. In: Technische Textilien 44 (2001) 4
[Cur1999] Curbach, M.; Jesse, F.: Lecture No. 513: Basic Tensile Tests on Strain Specimens of
Textile-Reinforced Concrete : Techtextil Symposium 1999. - Frankfurt/ Main, 1999,
Lecture No. 513
[Eng2004] Engler, Th.; Franzke, G.; Horlacher, H.-B., Offermann, P. and Schmutterer, Ch., 2004,
“Abrasionsverhalten textilbewehrter Feinbetonschichten,“ Kettenwirk-Praxis, Vol. 38,
No. 2, pp. 17-20
[Gri2004] New developments on manufacturing fibers and textile structures for technical textiles,
Departamento de ingeniería textil y papelera, Universitat politècnica de Catalunya
(Hrsg.): 2004 International Textile Congress: technical textiles; world market and future
prospects, Terrassa, Spanien, 18.-20.10.2004, S. 15-24
[Han2004] Characterization of textile structures for use in concrete – Part 1: strength of 2D structures
and initial stages of simulation of the deformation, Technical Textiles 47, p. E104-E105,
August 2004
25
[Mäd2003] Mäder, E.; Plonka, R.; Schiekel, M.; Hempel, R.: New possibilities: coatings on alkali-
resistant glass fibers for the improvement of concrete : Techtextil-Symposium 2003,
Frankfurt/ Main, 2003, Lecture 4.12
[Off2001] Offermann, P.; Abdkader, A.; Engler, Th.; Schierz, M.: Grundlagen textiler
Bewehrungsstrukturen zur Verstärkung bestehender Bauwerke. - In: SFB 528, Textile
Bewehrungen zur bautechnischen Verstärkung und Instandsetzung (Arbeits- u.
Ergebnisbericht für die Periode II/1999-I/2000) / M. Curbach (spokesman and ed.). -
Dresden: Institut für Tragwerke und Baustoffe der TU, 2001. - p. 25-74
[Off2003] Offermann, P.; Köckritz, U.; Abdkader, A.; Engler, Th.; Waldmann, M.:
Anforderungsgerechte Bewehrungsstrukturen für den Einsatz im Betonbau. In: Curbach,
M. (Edt.): Textile Reinforced Structures. Proceedings of the 2nd Colloqium on Textile
Reinforced Structures (CTRS2), Dresden : Sonderforschungsbereich 528, Technische
Universität Dresden, 2003
[Roy2004a] Roye, A., Gries, Th.: Vom Haustraum zum Traumhaus, Kettenwirk-Praxis 38 (2004),
H. 3, S. 26-27
26
4 CONCRETE/MATRIX (W. BRAMESHUBER)
Working party “Concrete / Matrix”.
Brameshuber, W.; Brockmann, T., Institute of Building Materials Research at Aachen University
(ibac), Germany
Curbach, M., Institut für Tragwerke und Baustoffe, University of Dresden, Germany
Meyer, Ch., Vilkner, G., Department of Civil Engineering & Engineering Mechanics, New York, USA
Mobasher, B., College of Engineering, Arizona State University, USA
Peled, A., Structural Engineering Department, Ben Gurion University, Israel
Reinhardt, H.W.; Krüger, M., Construction Materials Institute, University of Stuttgart, Germany
Wastiels, J., Vrije Universiteit Brussel, Belgium
ABSTRACT: In this chapter different matrices used for TRC elements are characterised. The matrix
design is a complex task as many requirements according to production processes, mechanical
parameters, bonding behaviour as well as durability have to be fulfilled. Hence, within different
research projects and for industry different matrix system have been developed for the use of TRC.
The matrix compositions as well as testing methods and the results for the fresh and hardened
concrete properties are summarised. Also, some extended investigations on the fracture mechanical
behaviour of the newly developed fine grained binder system considered as single component of the
composite TRC are discussed.
4.1 Introduction
The matrices used for TRC have to meet special demands regarding production process,
mechanical properties of the composite and durability of the reinforcement material. In most cases
small maximum grain sizes (< 2 mm) are used and hence these matrix systems can be considered as
mortar. On the other hand these matrices offer high performance properties in many respects and are
used for a construction composite material, such that these matrix systems are also called fine concrete
(or fine grained concrete).
Regarding matrix composition there is some experience for GRC (glass fibre reinforced concrete),
where short fibres are used, but for the use in TRC elements other requirements have to be fulfilled.
One essential aspect is to get full penetration of the technical textiles in order to guarantee for a good
bonding as well as loading behaviour. Hence, the consistency of the matrix has to be adjusted for the
properties of the textile, the geometry of the specimen and the production process. Some technical
textiles may show capillary suction and might show a very thin mesh and therefore highly flowable
consistencies are required. This is also the case for production techniques like injection, while other
production techniques like lamination or pultrusion rather required plastic consistencies. Furthermore,
for industrial series production high initial strength and short demoulding times will be aspired.
Regarding durability chemical stability of the technical textiles within the matrix are essential.
Resistance to frost and other environmental impacts will be demanded for external explication. Also,
mechanical properties like strength and deformation behaviour will depend on the application and are
rather related to the composite. Anyhow, low values of creep and shrinkage will be a further objective
for matrix design. As some of these requirements are conflictive matrix design will always be a best
compromise between all requirement regarding fresh, mechanical and durability aspects as well as
economic aspects for industrial production of TRC elements.
27
Table 4.1. Concrete types – overview.
Matrix systems Cement Type Additives Suitable production
technique /
Application
Mineral based systems
SFB 532 (Brameshuber) OPC Fly ash, All, depending
silica fume, on mixture
superplasticisers composition
SFB 528 (Curbach) OPC Fly ash,
silica fume,
superplasticisers
Glass aggregate concrete OPC Metakaolin Prestressed thin sheets
(Meyer, Ch.)
Friedrich, Th. ? no information Dehydration
Pachow, U. Durapact + e.g. silica fume, Spraying, lamination
varying cement metakaolin, see patent
types specifications
Reinhardt, H.-W. OPC Fly ash, Lamination
silica fume,
superplasticisers
Peled, A. Pultrusion
Alternative binder systems
Polymer modified binder systems OPC Polymers All, depending
on mixture
composition
Calciumaluminate cement (SFB CAC Fly ash, Lamination
532) silica fume,
superplasticisers,
stabilizer
Short-fibres (SFB 532) OPC Fly ash, Lamination
silica fume,
superplasticisers
28
Table 4.2. Matrix compositions of mineral based systems.
Materials Mineral based matrices
others - - - - - - - - -
binder (c+f+s+m) 700 700 1400 2400 700 675 928 979 979
6) 6)
plasticizer 1,5 0,9 2,5 0,6 2 2,5 n.m. n.m. 1,25
polymers - - - - - - -
% of
additives
stabilizer binder -
others -
AR-glass short fibres - - - - - - -
siliceous fines
0-0.125 mm 500 470 118 - 500 927 927 -
aggregates
siliceous sand
kg/m³
0.2-0.6 mm 715 670 168 - 715 550
0-0.6 mm 460 glass
0.6-1.2 mm 920 aggregates
water kg/m³ 280 280 350 228 280 211 309 309 123
water
Typically they show highly flowable consistencies which offer a full penetration of the technical
textiles [Bra2002a]. These special properties are achieved by using a small maximum grain size of
dmax = 0.6 mm, high binder contents, and adding different pozzolanic additives and plasticizers. Also,
they are designed with respect to an increased stability of AR-glass rovings. This increase in stability
is attained by selecting appropriate binder systems in which the level of OPC clinker content is
reduced via substitution with blast furnace slag, natural or synthetic pozzolanic materials, such as fly
ash and silica fume. The alkali ion concentration in the pore solution and the calcium hydroxide
content can be reduced substantially using silica fume. The strong reduction in alkalinity is
accountable to the pozzolanic reaction of the silica fume, in the course of which the Ca(OH)2 which is
present in the pore solution of hardened cement paste reacts with the dissolved SiO2 to form the
strength-enhancing calcium silicate hydrate phases. In order to eliminate the calcium hydroxide
completely, more than 25 % by mass of silica fume in relation to the total binder content (cement +
coal fly ash + silica fume) is required in mixtures containing pure OPC, depending on the water-binder
ratio [Sel1982, Zha1991]. The alkali ion and calcium hydroxide content is reduced in comparison to
pure OPC mixtures adding simultaneously silica fume and fly ash. For reasons of workability, the
amount of silica fume added is limited to moderate quantities, however lower than 10 % by mass of
total binder content. It is not possible to achieve a flowable consistency of these mixtures when larger
quantities of silica fume are added.
In addition to the reference mixture (PZ-0899-01), mixture FA-1200-01 was developed for durability
reasons with a very high levels of fly ash (FA-1200-01). A lowered pH-value and Ca(OH)2 amount
lead to an improved durability performance of AR-glass as described in [Rau2002]. Mixture RP-03-2E
was mainly developed for mechanical reasons in order to enhance the service load of TRC elements
[Bra2003b], like e.g. I-sections or U-profiles, which do not only require high tensile but also high
compressive strengths in the compression zone. Due to a high cement and binder content, which is
double compared to the reference mixture, and a low w/b this mixture offers a dense structure and high
strengths. Mixture MF-02-101 is similar to the reference mixture, but instead of adding silica fume,
the influence of metakaolin on the concrete properties was investigated. Mixture alphadur was
developed by Dyckerhoff, hence the matrix composition cannot be given in detail. Anyhow this binder
system is based on blast furnace slag cement. Other mixtures have been developed and can be found in
[Bra2001a, Bra2004c].
29
Mixture M7, which was developed at Construction Materials Institute, University of Stuttgart,
Germany, for the use in pre-stressed TRC elements, is similar to the mixture PZ-0899-01 (Table 4.2)
in its composition and mainly differs in the max. grain size of dmax = 1.2 mm and a reduced amount of
fines. To gain a high flowability also plasticisers based on polycarboxylate are used. The results of the
fresh and hardened concrete properties are shown in Table 4.8 and Table 4.9.
Also, so-called glass-aggregate concrete has been used mainly for thin sheets prestressed witharamid
fiber mesh [Mey2003, Vil2003a, Vil2003b, Vil2004]. – evtl. section by Ch. Meyer and G. Vilkner
Polymer modifications of the fine grained binder systems within the research project SFB 532 have
been integrated in the investigations as it is well known, that a polymer modification by film forming
additives enhances the flexural strength of the concrete. Because of the small diameter of the dispersed
particles (about 100-200 nm) the polymeric dispersions can penetrate into the spacing between the
filaments before the onset of setting of concrete. Also, the bonding of matrix and reinforcement is
influenced by the polymers (Section xy; bond).
After a screening was carried out, the most effective polymer dispersion was chosen for further
investigation. The characteristics of this polymer are shown in Table 4.3.
30
Table 4.4. Matrix compositions of alternative binder systems developed in industry and research
projects.
Materials Polymer Alternative Short fibres
modified
PZ-PM-20 CAC DP-0101-01 RP-03-3AR
SFB 532 SFB 532 SFB 532 SFB 532
cement c 430 700 600 980
cement type CEM I 52.5 CAC 42.5 CEM III/B 42.5 CEM I 52.5
ready made mixture 300
by Durapact4)
fly ash f kg/m³ 154 210
silica fume s 31 210
metakaolin m - -
binder
others
binder (c+f+s+m) 615 700 1400
1) 1)
plasticizer 1.0 2.0 2.5
5)
polymers 20 - -
% of
additives
2)
stabilizer binder 0.25
others
AR-glass short fibres - 3
siliceous fines
0-0.125 mm 438 520 101
aggregates
siliceous sand
kg/m³
0.2-0.6 mm 626 740 864 145
0-0.6 mm
0.6-1.2 mm
water kg/m³ 245 280 346 350
water
In recent years, phosphate cements have been developed as alternative binders for glass fibre
reinforced composites. These cements are acidic in nature when fresh, and pH-neutral after hardening.
An important drawback is their inherent fast setting properties, which make them unsuitable for other
than repair applications. A specific formulation has however been developed, which is sufficiently
retarded without losing durability: inorganic phosphate cement or IPC [ref1 HPFRCC – reference is
missing]. This cement consist of a powder and a liquid component: the powder is a calcium silicate,
which reacts with the metal phosphates in the liquid component. Its rheology (low viscosity of 4000
mPas) allows the use as matrix material for glass fibre composites. Ordinary E-glass can be used, since
it is not attacked by the matrix [ref 2 M&S – reference is missing].
Within the research project SFB 532 some alternative binder systems based on calcium aluminate
cements have been investigated. A representative matrix composition is shown in Table 4.4. The
investigations on CAC [Bra2001b] have shown the suitability of CAC as binder for TRC in principle,
but the investigations within the SFB 532 have not been carried on to a great extent as the mixtures
were very difficult in practical handling due to a rather complex mixing procedure and high accuracy
requirements for dosage of plasticiser and stabiliser. Even though some of the mixtures based on CAC
31
showed an improved durability performance of AR-glass as verified with SIC-tests (Section 6.3) and
offer high early strengths as wanted for industrial production processes, their rapid setting behaviour
as well as the lack of stable, yet highly flowable consistencies, were of great disadvantage for the
practical use within any production processes. Further research on CAC based binder systems might
lead to optimised matrix compositions with a better practical use for TRC applications.
As already mentioned above short-fibre reinforce matrices are well-known for GRC, where the short
fibres are used as the only reinforcement material. Short-fibres are also used in combination with
technical textiles already for some industrial application, but unfortunately usually matrix composition
and the characteristics of the matrix as single component are not published. Therefore no further
information was available for short-fibre reinforced matrices used in combination with TRC.
The flow of the fresh concrete is determined according to German Standard for mortars [DIN 18555-2
09.82]. The fresh concrete is filled in a slump cone (Fig. 4.1) and the flow is measured on a
horizontally aligned glass plate. Before testing, the glass plate and the inner surface of the cone are
moistened. To measure the flow the cone is centred on the glass plate, filled with the fresh concrete
and then levelled off. Then the cone is lifted slowly and the concrete can flow without any
compaction. The flow then is measured 10 and 30 minutes after mixing. Mean values are taken from
two measurements and are given in mm.
32
Fig. 4.1. Flow, dimensions of cone.
The flow time is measured according to consistency test methods for self compacting concrete. A V-
shaped funnel (Fig. 4.2), standing tight and being fixed in horizontal direction, is filled with the fresh
concrete. The bottom is kept closed until the fine grained concrete is completely filled in and leveled
off. As soon as the outlet is opened the time that the concrete needs to flow through the funnel is
measured i.e. time is stopped when the flow of the concrete is interrupted. The flow time is determined
10 and 30 minutes after mixing and results are given in 0.1 seconds.
33
Four different matrix formulations consisting of three percentages of class F fly ash along with a
control sample were used. The fly ash was used as replacement of the cement as presented in
Table 4.5. In all cases 5 % of silica fume was used and the water/binder ratio by weight was 0.375.
Table 4.5. The mix design (by volume) with the different fly ash contents.
Mix ID Mix design Cement c Fly ash f Silica fume s w/c
% % % -
A Control 95 % 0
B 40 % fly ash 55 % 40 %
5% 0.375
C 60 % fly ash 35 % 60 %
D 80 % fly ash 15 % 80 %
Other investigations regarding rheological properties have been carried out using a rotary viscometer
(Viskomat NT) within the SFB 532, Aachen University. The mixtures PZ-0899-01, HZ-1000-01, FA-
1200-01, FA-1200-11 as shown in Table 4.2 were examined in terms of their rheological
characteristics with a rotary viscometer. The flow curves were measured at different shear rates
(rotational speed) as 25 min with 120 rpm, and afterwards with a rotational speed of 80, 40, and 20
rpm respectively for 1 min each. The viscosity η was derived to evaluate the consistency. Also, a
correlation with flow and flow time was investigates as well as information about the stability of the
mixtures was gained.
34
elastic modulus is determined either on the base of velocity of ultrasound waves (Steinkamp) or
resonant frequency measurements.
4.4.1.4 Carbonation
The concrete prisms (20 x 40 x 160 mm) were stored at a temperature of 20 °C and 65 % R.H.. Slices
of the concrete were split at different testing ages and the internal surface of the slice was sprayed with
a phenolphthalein solution to reveal the non carbonated core. The depth of carbonation, as indicated
by the pink colour, was measured on all sides of the slice and averaged for each specimen.
4.4.1.5 Shrinkage
Shrinkage is measured using prismatic specimens with the dimensions 40 x 40 x 160 mm³ according
to German Standard [DIN 52450 08.85]. These specimens are stored continuously at a temperature of
20 °C and 65 % R.H..
In the research project detailed investigations are carried out to determine the σ-ε-curve under
compression and to investigate possible size effects. For this reason even very small sized specimens
have been investigated, as these were considered to be more representative in respect to the
applications of TRC elements which often show a wall thickness of d = 10 mm only.
The σ-ε-curve and the corresponding compressive strength as well as the Young´s modulus were
determined for cubic specimens of different side lengths d = 10, 20, and 40 mm with a slenderness λ =
1, 2, and 4 each. To meet the accuracy requirements of such small sized specimens precise formworks
are used. The accuracy was verified with a co-ordinate measuring technique using separately produced
small sized specimens. Furthermore a suitable testing method to determine the compressive strength of
such small sized specimens had to be developed and e.g. special gauge blocks were designed to assure
centric loading of the specimens in the middle of the base plate. These developments are described in
detail in [Bra2003a].
The compressive tests presented in this paper are carried out on a universal testing machine
INSTRON 5587 controlled by cross-head displacement at different rates according to the tested
35
specimen dimensions (e.g. 1.4 mm/min for specimen dimensions of d = 40 mm and λ = 2). Hence, the
testing duration is 60 - 90 seconds. The strain is recorded as mean value from three strain gauges fixed
at the specimen sides excluding the casting side.
d = 12.5 15
20
30
100
25 [mm]
40
36
l=240
d=40
a=10
u
l s=200 [mm]
Fig. 4.4. 3-point bending test (notched) for fine grained concrete [Bro2001].
37
The mixture for the pultrusion process should be sufficiently fluid to enable the fabric to transfer
through the cement slurry but dense enough so that it will remain on the fabric when it leaves the
cement bath.
Fig. 4.5 shows a plot of shear strength as a function of the shear rate obtained by this method. Under
the assumptions that the behaviour of the materials can be idealised as a Bingham material, at each
level of fly ash use, the slope of the line represents the viscosity, and the intercept represents the yield
strength. The results in the plot indicate that the viscosity of the control samples increases as the levels
of fly ash increased, indicating that the presence of fly ash causes a higher level of adhesion within the
paste mixture. The yield strength of the sample decreases with the addition of fly ash. This may result
a better penetrate the fabric and develop a better bond strength due to filling of the voids and interfaces
within the fabric, of high fly ash content mixtures. The intercept of the lines in this Fig. 4.5 represent
the shear strength. It is observed that as the fly ash content is increased, the shear strength is
decreased, and the bonding characteristics of the fresh mixture are therefore decreased.
80
Control
40% Fly ash
60% Fly ash
70
Shear Strength, dyne/cm2
60
50
40
3 4 5 6 7
Shear Rate, 1/sec
Fig. 4.5. Shear strength vs. shear rate plots for various levels of fly ash.
It was the objective of the examination with a viscometer to evaluate the fine grained binder systems
in terms of their rheological properties. From the measurements, as described above, a flow curve can
be derived and rheological characteristics like the viscosity and workability of the mixtures have been
evaluated. A correlation between the viscosity and flow and flow time has been derived. In contrast to
the investigations, as described above, not cement pastes but the fine grained binder systems including
aggregates have been investigated. Furthermore, another shear rate was chosen and therefore the
results of these testing series, especially values of shear strength and viscosity, cannot be compared
directly. In Fig. 4.6 the results are shown using different indexes than in Fig. 4.5, but still the slope of
the flow curves gives the viscosity η.
38
shear stress [Nmm]
30
HZ-1000-01
25 η = 0.204
20 PZ-0899-01
η = 0.179
15
FA-1200-01
10 η = 0.135
5 FA-1200-11
η = 0.056
0
0 20 40 60 80 100 120
shear rate [Nmm]
A comparison of the flow curves shows that the mixtures FA-1200-01 and FA-1200-11 with a high
amount of fly ash offer the lowest viscosity and thus the better flowability. Mixture FA-1200-11
shows the lower viscosity despite of its higher silica fume content (20 % by mass of binder content).
Apparently, a high silica fume content in combination with extremely high fly ash does not lead to a
reduced workability, the used mix proportions of 30 % by mass cement, 20 % by mass silica fume,
and 50 % by mass of the total binder content rather provide an advantegeous grading of the fines that
lead to a higher flowability. The mixtures PZ-0899-01 and HZ-1000-01 solely differ in the cement
type. The flow curves in Fig. 4.6 indicate that a higher fineness of grain of the OPC (CEM I 52.5)
leads to a lower viscosity and has a positive influence on the workability of the fine grained concrete.
The results as presented in Section 4.5.1indicate that there is no clear correlation between the acquired
values of flow and flow time. For equal flow spreads, very different flow times were measured.
However, a high flow spread in combination with a short flow indicates a high flowability.
Measurements carried out with the rotary viscometer should show if rheological values like the
viscosity provide further information about the consistency.
Correlation between the flow time t10min and the viscosity η are presented in Fig. 4.7 and indicate a
linear behaviour. The viscosity increases with increasing flow time. However, further examinations
are required to confirm the results, but it can probably assumed that different flowabilities of the fine
grained concrete mixtures can be determined in a way just as well as with the viscosity measurements
with the less complex test method of flow time. Concerning the flow a clear correlation with the
viscosity could not be determined.
39
viscosity η in Nmm*min
0.25
0.20
0.15
0.10 PZ-0899-01
HZ-1000-01
0.05 FA-1200-01
FA-1200-11
0.00
0 1 2 3 4 5 6 7
flow time f10min in sec
A possible sedimentation of the mixtures initially was estimated visually, but could not be quantified.
Hence further tests with the viscosimeter were carried out to measure a possible sedimentation. For
these tests the immersion depth of the mixing paddle was varied. Measurements with a standard
immersion depth (215 mm = height 0) and a reduced immersion depth (175 mm = height 1) were
carried out. Using the standard immersion depth leads to stirring of the whole volume of mixture
during testing, which retards sedimentation. When using the lower immersion depth not the complete
mixture volume might be stirred and sedimentation might not be retarded anymore, which will lead to
different shear stresses. In Fig. 4.8 curves are shown for both immersion depth for two mixtures,
which differ in their stability. The mixture DP-0101-01 was chosen, as it offers a very stable
consistency, while mixture PZ-0899-01oSF. The measured viscosity h is in a similar range for mixture
DP-0101-01 for both immersion depth, which approves of the high stability in consistency. For
mixture PZ-0899-01oSF there is a significant increase in viscosity for the lower immersion depth,
which indicates that there is sedimentation. This testing method such offers the evaluation of demixing
tendencies if the same measurement profile and mixtures comparable in production process are used.
25
10
0 PZ-0899-01 oSF
paddle depth h1
-5
0 20 40 60 80 100 120
shear rate in U/min
40
Others
Other rheological investigation have been carried out in Stuttgart, using a viscosimeter (Visco-Corder,
Fa. Brabender) and are described in [Krü2004, Fre2000]. These investigations also have shown that
the addition of fly ash and silica fume lead to more suitable rheological properties than the use of lime
stone powder. In general, it was found, that the use of fly ash leads to an improved workability and
also offers suitable mechanical properties. As the fly ash properties show a high scatter depending on
the production process and other aspects, initial tests may be carried out to investigate the influence on
the rheological properties.
41
Table 4.9. Basic mechanical properties of alternative binder systems.
Hardened concrete Testing Polymer Alternative binder systems Short fibres
properties age [d] modified
PZ-PM-20 CAC DP-0101-01 IPC RP-03-3AR
SFB 532 SFB 532 SFB 532 SFB 532
Compressive strength 1) water, 20 °C water, 20 °C water, 20 °C sealed
N/mm² 2 15 56 15 39 52
7 22 83 48 52 74
28 35 96 63 68 94
90 42 96 89 80 94
360 49 82 89 n.d. 99
Flexural bending strength 1) water, 20 °C water, 20 °C water, 20 °C sealed
N/mm² 2 4.1 8.6 3.8 6.5 17.2
7 5.6 11.7 6.4 7.5 17.8
28 10.9 11.9 6.9 8.1 19.4
90 12.8 10.7 11.5 10.5 18.1
360 13.9 9.3 12.8 n.d. 18.7
Dynamic modulus of elasticity 1) water, 20 °C water, 20 °C sealed
N/mm² 2 19200 n.d. 16400 n.d. 31500
7 24500 34800 23400 24700
28 28300 34900 28100 26800
90 28500 36100 29600 26600
360 28700 36900 31500 25900
Static Young's modulus - - - - sealed
N/mm² 28 n.d. n.d. n.d. n.d. 21200
Shrinkage 20 °C, 65 R.H20 °C, 65 R.H. 20 °C, 65 R.H. 20 °C, 65 R.H. 20 °C, 65 R.H. 20 °C, 65 R.H.
mm/m 2 n.d. 0.48 0.75 0.02 0.66
7 0.07 0.83 1.75 0.40 1.21
28 0.50 0.95 2.00 1.40 1.40
90 0.82 1.01 2.32 10.50 1.53
360 0.99 1.08 2.45 n.d. 1.63
Carbonation 20 °C, 65 R.H20 °C, 65 R.H. 20 °C, 65 R.H. 20 °C, 65 R.H. 20 °C, 65 R.H. 20 °C, 65 R.H.
mm 2 n.d. n.d. 0.5 This mixture n.d.
7 n.d. 0.8 1.4 does not 0.3
28 n.d. 1.7 3.3 show 1.0
90 n.d. 2.0 4.5 carbonation. 1.4
360 n.d. 3.3 8.6 n.d.
1)
2 d 23 °C/95 R.H., until 7 d water storage (20 °C), then 23 °C/95 R.H.
For selected mixtures the time dependent development of the flexural tensile strength and shrinkage
are shown in Fig. 4.9 and Fig. 4.10 respectively.
42
flexural tensile strength in N/mm² flexural tensile strength in N/mm²
25.0 25.0
PZ-0899-01
FA-1200-01
20.0 RP-03-2E 20.0
alphadur
MF-02-101
M7
15.0 R1* 15.0
R2-17*
10.0 10.0
PZ-PM-20
5.0 5.0 CAC
DP-0101-01
RP-03-3AR
0.0 0.0
2 7 28 90 360 2 7 28 90 360
a) age in days
b) age in days
Fig. 4.9. Flexural tensile strength for a) mineral based, and b) alternative matrices.
1.0 1.0
0.5 0.5
0.0 0.0
2 7 28 90 360 2 7 28 90 360
a) age in days
b) age in days
Fig. 4.10. Shrinkage of tensile strength for a) mineral based, and b) alternative matrices.
The mechanical properties of the mineral based mixtures (Table 4.2) cover a wide range for the
different concrete mixtures. For example, the compressive strengths as listed in Table 4.8 range from
32 to 98 N/mm² at a testing age of 28 days. Similar variations can be observed for the flexural tensile
strength as well as the Young’s modulus. The alternative binder systems (Table 4.4) also show a wide
range of the mechanical properties like e.g. the compressive strength, which ranges between 35 and
96 N/mm² at a concrete age of 28 days. Comparably high flexural strength (about 12 N/mm²) are
obtained for the polymermodified systems as well as for calcium aluminate cements. Higher flexural
strength (about 20 N/mm²) can only be achieved by the addition of short fibres (Table 4.9).
Furthermore, rather high shrinkage strains between 0.30 and 1.47 mm/m are observed for the mineral
based mixtures as well as alternative binder systems (Table 4.8 and Table 4.9). Extremely high
shrinkage strains (> 2.0 mm/m) were obtained for the phosphate cements as well as for mixture DP-
0101-01 (which shows positive durability aspects by a low pH values, but unsuitable consistency as
well as too high shrinkage strains). In principal, large shrinkage strains should be avoided and those
mixtures with high shrinkage strains should still be modified to reduce the shrinkage strains. Still, for
most mixtures the shrinkage strains are higher in comparison to ordinary concrete of similar strengths
and hence, have to be taken into account for chosing the right mixture for the planned application. For
chosing a suitable mixture, all mechanical, fracture mechanical, deformation and durability properties
should be taken into account for the planned application.
43
4.6.1.1 Alkalinity of pore solution
In contrast to steel reinforcement, the corrosion mechanisms of the AR-glass reinforcement are
hampered at a low alkaline pore solution and thus for TRC a low pH value is preferable. Such
durability aspects and corrosion mechanisms of AR-glass and other reinforcement materials are
described in more detail in Section 6.3
Table 4.10 shows the results of chemical analysis of pore solution pressed out from cement pastes of
selected mixtures according to Table 4.2.
44
10 mm 10 mm
In Fig. 4.12 the σ-ε curves are shown exemplary for the specimens with a slenderness λ = 4, as there is
the least influence friction and tranverse strain hindrance, for the mixtures PZ-0899-01, FA-1200-01,
and RP-03-2E. The results of the mechanical parameters derived from these σ-ε curves, i.e.
compressive strength σc, Young´s modulus Ec, and strain at ultimate load ε(σc,max), are listed in
Table 4.11 as mean value of the tested specimens for the mixtures PZ-0899-01, FA-1200-01 and RP-
03-2E for sealed storage at 20 °C and a testing age of 28 days.
120 d = 20 mm
PZ-0899-01
100 d = 40 mm
80
60
40
20 FA-1200-01
0
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00
strain ε in mm/m
Fig. 4.12. σ-ε curves in compression, mixtures PZ-0899-01, FA-1200-01, and RP-03-2E.
45
Table 4.11. Results of compressive tests
Specimen λ2) PZ-0899-01 FA-1200-01 RP-03-2E
dimension
b x d x l 1) σc Ec ε(σc,max) σc Ec ε(σc,max) σc Ec ε(σc,max)
o o o
mm³ - N/mm² N/mm² /oo N/mm² N/mm² /oo N/mm² N/mm² /oo
10 x 10 x 10 1 93.99 - - 42.74 - - 146.91 - -
20 x 20 x 20 94.56 28400 5.736 43.44 15200 5.157 150.27 25700 6.274
40 x 40 x 40 101.18 30400 5.296 44.88 20600 5.168 152.96 28200 6.766
10 x 10 x 20 2 98.90 33600 4.879 47.54 23800 3.820 156.81 28400 7.443
20 x 20 x 40 95.09 30400 5.698 39.75 22600 4.436 153.15 28600 6.816
40 x 40 x 80 89.58 30600 5.485 37.52 22100 3.979 154.24 28900 6.849
10 x 10 x 40 4 77.34 32500 3.870 43.12 21700 4.248 135.24 29200 4.887
20 x 20 x 80 89.13 31100 4.949 38.15 22600 3.375 127.58 28300 5.342
40 x 40 x 160 87.37 31400 4.851 36.51 21900 3.316 143.10 28400 6.276
1)
l – length, b – width, d – thickness of specimen
2)
l – slenderness
Considering the results presented in Table 4.11 no significant size effect for the different specimen
dimension can be observed. Compared to ordinary concrete, where concrete cubes with an increasing
side length (d = 200, 400, and 800 mm) may show a reduction in compressive strength of up to 50 %,
only a slight decrease in compressive strength with increasing specimen size is observed for
specimens with λ = 2, but this is within the scatter of results.
In previous investigations the compressive strength of the fine grained concrete has been determined
using specimen sizes with a thickness d = 40 mm according to standardised testing methods for
mortars [DIN 18555-3 09.82]. However, material properties determined on these specimens might
differ significantly from the stress-strain relations evaluated on much thinner structures (d ≈ 10 mm)
under practical conditions due to size effects. In the classical theories based on plasticity or limit
analysis, the strength of geometrically similar structures is independent of the structure size. But
experimental results show that quasi-brittle materials, like concrete, rocks or ceramics exhibit a
dependence of the nominal stress at ultimate load on the specimen size, the so called size effect.
Different models have been proposed to predict a strength reduction with size increase in concrete
structures, like e.g. pure statistical effects by Weibull´s weakest link theory [Wei1939], the fracture
mechanic size effect law by [Baz1998], and multifractal scaling laws as by [Car1999]. The failure
mode of compressed concrete specimens can be considered as resulting from local tensile
mechanisms, or from a combination of tensile and shear mechanisms, according to the specimen
geometry. Therefore it is the aim of the experiments is to investigate possible size effects on the
compressive strength and the σ-ε-relation of the fine grained concrete. Within the testing programme
size effects due to diffusion based phenomena, like e.g. carbonation, and drying shrinkage, are
negligible as a sealed storage at a constant temperature of 20 °C was chosen. The results presented in
[Bra2003a] show a slight reduction in compressive strength for specimens with a side length d = 40
mm compared to the smaller side lengths, while in the presented investigations no significant size
effect could be observed for the different side lengths of the fine grained concrete specimens. There
are different possible explanations for these results. The so called “wall effect” [Baz1998] describes a
layer of cement paste caused by the concrete formwork and the production process with a reduced
strength at the surface of the specimen. Assuming a similar thickness of this layer for different
specimen sizes a possible strength reduction is more significant for smaller specimen sizes and hence
might have a compensating influence on size effect as described by Weibull [Wei1939]. Another
explanation can be found within the fracture mechanic size effect law of Bazant [Baz2002], which
46
describes the transition from the theory of strength of materials (for small sizes) to the size effect
predicted by linear elastic fracture mechanics (LEFM). For sufficiently small sizes, the nominal
strength is proportional to the material strength, and this represents the failure condition of plastic limit
analysis, characterised by no size effect. The fine grained concrete matrix investigated here with a
very small maximum grain size dmax = 0.6 mm and the chosen small specimen sizes might be within
these limits. Therefore it will be the objective of further investigations to analyse specimen sizes
which approach a region of negligible size effects and to enlarge the amount of experiments for
evaluation of the test results on a statistically secured basis. These results can be used to analyse and
describe the size effect of fine grained concrete with appropriate models.
load in N
600 28 days
PZ-0899-01
sealed storage, 20 °C
500 notch: d/4 = 10 mm
RP-03-2E
400
FA-1200-01
300
200
100
0
0.0 0.1 0.2 0.3 0.4
displacement in mm
47
The corresponding bending tensile strength ftb and the fracture energy Gf are given in Table 4.13. Also,
the characteristic length lch (Table 4.13) has been determined as follows to evaluate the brittleness of
the material. The Young’s modulus and the tensile strength have been used according to Table 4.12.
EG f
l ch = 2
[mm] (1)
ft
Subsequent numerical analysis to derive the softening behaviour as σ-w function often is carried out
using finite element methods (FEM) and usually is based on Hillerborg`s fictitious crack model (FCM)
[Hil1983, Hil1985]. This is used to describe non-linear fracture behaviour and usually requires a load-
displacement curve obtained from a Mode I stable fracture test like e.g. a 3-point bend test on notched
beams as described above. For a given displacement u in the mid-section of the beam known from the
experiment the corresponding displacements can be calculated by means of FEM when the boundary
conditions are met, i.e. no displacement of the rigid supports. The calculated displacement field also
implies the displacements in the mid-section of the beam in x-direction, and hence in the region of the
fracture zone (Fig. 4.14). The corresponding stress distribution is then derived by applying the
material relations of the concrete. Usually a linear σ-ε relation as defined by the tensile strength ft and
the Young´s modulus E is assumed to describe the pre-cracking behaviour, while a σ-w relation is
chosen to describe the softening behaviour of concrete. By assuming a softening curve the resulting
force Psim can be calculated and compared with the force Pexp known from the experiment for each load
step (load-displacement curve from 3-point bending test). In an iterative fitting process the σ-w
relation is determined such that the numerical simulation of the load-displacement curve and the
experiment show a good agreement.
48
P
linear- f(ε)
elastic
ft
FPZ
cohesive stress f(w)
end of
real crack stress transfer
Fig. 4.14. A loaded concrete beam with a crack and a fracture zone.
Apart from the numerical analysis different analytical models, often based on Hillerborg´s FCM, have
been derived in recent years to describe the softening behaviour of concrete. The models proposed by
e.g. [Uch2001, Ulf1990, Ulf1995, Sun1998] all have in common that either by means of FEM or beam
theory the distribution of displacements within the ligament of a concrete beam is calculated. Knowing
these displacements and using the (assumed) σ-w curve allows the derivation of the stress distribution
within the ligament, which finally allows to calculate the force Psim. This output finally is compared
with the experimentally determined force Pexp as already described above for the FE-analysis.
In contrast to these models a new approach is shown in [Bra2004b] where the distribution of
displacements within the ligament of a beam in a 3-point bending test is measured continuously during
testing by a video extensometer and thus already known. Combining the advantages of a step-wise
determination of a multi-linear σ-w relation as proposed by Uchida [Uch2001] and the classical idea
of beam theory, the model allows a straightforward determination of the σ-w-relation. Knowing the
displacements ∆l over the ligament height and applying equilibrium of forces, the determination of the
σ-w relation without the need of using FEM and further optimisation processes is possible.
Fig. 4.15 shows first results of the multi-linear σ-w curves derived with the new model and based on
the results of the 3-point bending test. The derived softening behaviour is compared with σ-w curves
which have been determined by means of FE-analysis. The results of the numerical method are
considered to be reliable as they have been verified for a varying geometry of the concrete specimens
and with numerical simulations of uniaxial tensile tests also [Rau2001]. A comparison of the results of
the softening behaviour approves that the results of the proposed analytical model show a quite
acceptable agreement with known FE methods to derive the tension-softening behaviour of concrete.
To validate these presented results, the proposed model will be enhanced and used for future
investigations. Further mixtures at different testing ages and also possible size effects will be
examined.
49
tensile stress σ [N/mm²]
5.0
4.0
FE-analysis
1.0
Proposed model
0.5
0.0
0.00 0.05 0.10 0.15
crack width w [mm]
Fig. 4.15. σ-w curve derived by FE-analysis and proposed analytical model [Bra2004b].
The need for the determination of the σ-w of fine grained concrete is given as the fine grained
concrete matrices show a different softening behaviour than ordinary concrete, which has been
verified with the numerical analysis. The implementation of mathematical simplifications of σ-w
curves, like e.g. linear, bi-linear, or exponential curves, did not give a good correlation between
experimentally and numerically simulated results. It will be the aim of further investigations to find
more general mathematical solutions of σ–w curves even for the fine grained concrete matrices, which
then can be implemented in further routines to simulate and calculate structural elements of the new
composite material of TRC.
REFERENCES
[Baz1998] Bazant, Z.P. ; Planas, J.: Fracture and size effect in concrete and other quasibrittle
materials. Boca Raton : CRC Press, 1998.
[Baz2002] Bazant, Z.P.: Concrete fracture models: Testing and practice. In: Engineering Fracture
Mechanics 69 (2002), S. 165-205.
[Bra2001b] Brameshuber, W.; Brockmann, T.: Calcium Aluminate Cement as Binder for Textile
Reinforced Concrete. London : IOM Communications Ltd, 2001. - In: Calcium
Aluminate Cements 2001, Proceedings of the International Conference held at Heriot-
Watt University Edinburgh, 16-19 July 2001, (Mangabhai, R.J. ; et al (Ed.)), S. 659-666.
[Bra2002a] Brameshuber, W.; Brockmann, T.; Hegger, J.; Molter, M.: Untersuchungen zum
textilbewehrtem Beton. In: Beton 52 (2002), Nr. 9, S. 424-426,428-429.
50
[Bra2002b] Brameshuber, W.; Brockmann, T.: Betonformulierung – Methodik und Stoffgesetze.
Aachen : Rheinisch-Westfälische Technische Hochschule, 2002. – In: SFB 532:
Textilbewehrter Beton – Grundlagen für die Entwicklung einer neuartigen Technologie,
Arbeits- und Ergebnisbericht 2. Hj. `99-`01, 2. Hj. `02, S. 195-233.
[Bra2003a] Brameshuber, W.; Brockmann, T.: Size Effect on Mechanical Properties of Fine Grained
Concrete Matrices. Dresden : Lehrstuhl für Massivbau, 2003. - In: Textile Reinforced
Structures, Proceedings of the 2nd Colloquium, Dresden, 29.9.2003-1.10.2003, (Curbach,
M. (Ed.)), pp. 161-172.
[Bra2004b] Brameshuber, W.; Brockmann, T.; Banholzer, B.: Analytical Evaluation of the Softening
Behaviour of Fine Grained Concrete. Vail, Colorada : Ia-FraMCos, 2004.- In:
Proceedings of the fifth International Conference on "Fracture Mechanics of Concrete
Structures" (FRAMCOS-5), Vail Colorado, USA 12-16 April, 2004 (Li, V.C.; et al (Ed.)),
Vol. 2, pp. 1145-1153.
[Bra2004c] Brameshuber, W.; Brockmann, T.; Banholzer, B.: Textile Reinforced Ultra High
Performance Concrete. Kassel : Kasseler University Press. - In: Schriftenreihe Baustoffe
und Massivbau (2004), Nr. 3. Ultra High Performance Concrete, Proceedings of the
Intern. Symposium, Kassel, Sept. 13-15, 2004, (Schmidt, M. ; et al (Ed.)), pp. 511-522.
[Car1999] Carpinteri, A. ; Ferro, G. ; Monetto, I.: Scale effects in uniaxially compressed concrete
specimens. In: Magazine of Concrete Research 51 (1999), No. 3, pp. 217-225.
[DIN 1048-5 06.91] Prüfverfahren für Beton, Festbeton, gesondert hergestellte Probekörper, Juni 1991
– German Standard.
[DIN EN 12390-3 04.02] Prüfung von Festbeton. Teil 3: Druckfestigkeit von Probekörpern, April
2002 – German Standard.
[DIN 18555-2 09.82] Prüfung von Mörteln mit mineralischen Bindemitteln. Frischmörtel mit dichten
Zuschlägen, Bestimmung der Konsistenz, der Rohdichte und des Luftgehalts, September
1982 – German Standard.
[DIN 18555-3 09.82] Prüfung von Mörteln mit mineralischen Bindemitteln. Festmörtel, Bestimmung
der Biegezugfestigkeit, Druckfestigkeit und Rohdichte, September 1982 – German
Standard.
51
[DIN 52450 08.85] Prüfung anorganischer nichtmetallischer Baustoffe, Bestimmung des Schwindens
und Quellens an kleinen Prüfkörpern, August 1985 – German Standard.
[Fre2000] Frech, D.: Selbstverdichtender Feinbeton (Mörtel) – Untersuchungen zum Einfluss der
Mischungszusammensetzung auf die Mörteleigenschaften im frischen und erhärteten
Zustand. Universität Stuttgart, Institut für Werkstoffe im Bauwesen, Diplomarbeit, 2000.
[Hil1983] Hillerborg, A.: Analysis of one single crack. Amsterdam : Elsevier Science Publishers,
1983. – Fracture Mechanics of Concrete (Wittmann, F.H. (Ed.)), pp. 43-74
(Developments in Civil Engineering 7).
[Hil1985] Hillerborg, A.: Dimensionless presentation and sensitivity analysis in fracture mechanics.
Amsterdam : Elsevier, 1986. – Fracture Toughness and Fracture Energy of Concrete :
Proceedings of the International Conference on Fracture Mechanics of Concrete, CH-
Lausanne, 1985, (Wittmann, F.H: (Ed.)), pp. 413-421.
[Mey2003] Meyer, C., Vilkner, G.: Glass Concrete Thin Sheets Prestressed with Aramid Fiber
Mesh”, Proc., HPFRCC4, RILEM Publications S.A.R.L., 2003.
[Rau2002] Raupach, M.; Orlowsky, J.; Wolff, L.; Brameshuber, W.; Brockmann, T.:
Untersuchungen zur Dauerhaftigkeit von AR-Glasrovings in Feinbeton. In: Technische
Textilien 45 (2002), No. 2, pp. 93-96.
[Sel1982] Sellevold, E.J.; Bager, D.H.; Klitgaard Jensen, E.; Knudsen, T.: Silica Fume - Cement
Pastes : Hydration and Pore Structure. Trondheim : The Norwegian Institute of
Technology, NTH. - In: Division of Building Materials, Report No. 82.610 (1982),
pp. 19-50.
[Sun1998] Sundara, K.T. ; Iyengar, R. ; Raviraj, S. ; Ravikumar, P.N.: Analytical Study of Fictitious
Crack Propagation in Concrete Beams Using a Bilinear o-w Realation. Freiburg :
AEDIFICATIO, 1998. - In: Proceedings of the Third International Conference on
"Fracture Mechanics of Concrete Structures" (FRAMCOS-3), Gifu, Japan, 12-16 Oct.
1998, (Mihashi, H.; Rokugo, K. (Ed.)), Vol. 1, pp. 315-324.
[Uch2001] Uchida, Y.: Determination of Tension Softening Diagram of Concrete (Draft, 2001),
Japan Concrete Institute, JCI-TC992 Test Method for Fracture Property of Concrete.
[Ulf1990] Ulfkjaer, J.P. ; Brincker, R. ; Krenk, S.: Analytical Model for Complete Moment-
Rotation Curves of Concrete Beams in Bending. Warley West Midland, UK : EMAS,
1990. - In: Proceedings 8th European Conference on Fracture Behaviour and Design of
Materials and Structures, Turin, 1.-5.10.1990, (Firrao, D. (Ed)), Vol. 2, pp. 612-617.
[Ulf1995] Ulfkjaer, J.P. ; Krenk, S. ; Brincker, R.: Analytical Model for Fictitious Crack
Propagation in Concrete Beams. In: Journal of Engineering Mechanics 121 (1995), No. 1,
pp. 7-14.
52
[Vil2003a] Vilkner, G.: “Glass Concrete Thin Sheets Reinforced withPrestressed Aramid Fabrics”,
Ph.D. Dissertation, Columbia University, New York, NY, 2003.
[Vil2003b] Vilkner, G. ; Meyer, C.: Glass Concrete Thin Sheets Prestressed with Continious Aramid
Fibers: Report on Recent Progress at Columbia University. Dresden : Lehrstuhl für
Massivbau, 2003. - In: Textile Reinforced Structures, Proceedings of the 2nd
Colloquium, Dresden, 29.9.2003-1.10.2003, (Curbach, M. (Ed.)), S. 539-550.
[Vil2004] Vilkner, G. ; Meyer, C.: Performance of Unprotected Aramid Fibers in Prestressed Glass
Concrete. Berlin : BAM, 2004. - In: 11th International Congress on Polymers in
Concrete, Berlin, 2nd - 4th June, 2004, (Maultzsch, M. (Ed.)), S. 477-484.
[Wal2003] Walk-Lauffer, B.; Orlowsky, J.; Raupach, M.: Verstärkung des inneren Roving-
Verbundes im textilbewehrten Beton. Weimar: Bauhaus-Universität, 2003. - In: 15.
Internationale Baustofftagung, 24.-27. September 2003, Weimar, Vol. 2 pp. 0281-0290.
[Wei1939] Weibull, W.: A statistical theory for the strength of materials. Swedish Royal Institute for
Engineering Research, Stockholm, 1939.
[Zha1991] Zhang, M.-H.; Gjorv, O.E.: Effect of Silica Fume on Pore Structure and Chloride
Diffusity of Low Porosity Cement Pastes. In: Cement and Concrete Research 21 (1991),
No. 6, pp. 1006-1014.
53
5 PRODUCTION TECHNOLOGIES (W. BRAMESHUBER)
Brameshuber, W.; Brockmann, T., Institute of Building Materials Research at Aachen University
(ibac), Germany
Friedrich, T., Novacret GmbH, Bernkastel-Kues, Germany
Mobasher, B., College of Engineering, Arizona State University, USA
Pachow, U., Durapact GmbH, Haan/Rheinland, Germany
Peled, A., Structural Engineering Department, Ben Gurion University, Israel
Reinhardt, H.W.; Krüger, M., Construction Materials Institute, University of Stuttgart, Germany
Wastiels, J., Vrije Universiteit Brussel, Belgium
ABSTRACT: In this chapter the principle production processes .... are discussed abstract is missing...
5.1 Introduction
In the past production processes have been developed for glass fibre reinforced concrete
(GRC) i.e. a composite consisting of a fine concrete matrix and short AR-glass fibres and for some
applications with additional continuous rovings and in some cases mats made from randomly
distributed short fibres. For textile reinforced concrete (TRC) the reinforcement is additionally
supplemented or primarily provided by technical textiles like e.g. fabrics, warp knits, etc. and also
two-, and three-dimensional products made from different raw materials, see Chapter 3. Depending on
mixture design, form and size of structure and the planned application different production processes
are used. Characteristic for a production process is the way of bringing together matrix and the
reinforcing materials, which may consist of chopped fibres, rovings, technical textiles or combinations
of these. For small series and under laboratory conditions the different production techniques often are
applied as manual technique.
5.3.1 Introduction
The hand lay-up technique is also called contact moulding or hand laminating technique. It is one of
the oldest, simplest and most commonly used manufacturing methods of fibre reinforced composite
materials. Its use with organic resins like polyester or epoxy, and glass fibres, is widespread, and
different products such as boat hulls, wind-turbine blades, aircraft wings, train noses, car and truck
bodies, swimming pools, and even complete minesweepers are manufactured. It is however not
commonly used for cementitious matrix materials. One of the reasons for this is that the reinforcing
fibres must be present in some kind of textile shape, which is not compatible with the traditional
concrete casting technology.
A related manufacturing technique is the so called spray-up technique, where the matrix is projected
on the mould with a spray gun, and the fibres are chopped from a roving by a cutter on the spray gun,
and blown in the matrix jet. The results are quite similar with the hand lay-up technique, but the fibres
are not applied in a textile pre-form.
54
5.3.2 Processing steps
The typical hand lay-up process consists of the following steps:
The surface aspect of the mould is important, since this will determine the aspect of visible finished
side of the product. In order to assist unmoulding of the product after hardening, a release film is
applied to the mould surface prior to laminating, usually by application of a release agent. The precise
application is depending on the type of release agent, and is described in the technical manual of the
manufacturer.
(d) Finishing
After the composite is allowed to harden sufficiently, it is unmoulded, and any machining (openings,
trimming) or assembly can be performed. Eventually post-curing can be executed.
For cement matrix composites, there is one important specific advantage over many other production
techniques: the fibre volume fraction which can be obtained is sufficiently high to obtain a multiple
55
cracking regime after matrix tensile failure, leading to dramatically improved energy absorption,
together with an eventually markedly increased tensile strength. Besides the fact that the production
technique is not belonging to the traditional concrete technology, there are however also two specific
disadvantages: firstly, the rheology of fresh cement – consisting of a dense suspension of particles –
requires an open texture of the reinforcement, in order to let the matrix penetrate it and form the
continuous phase (see 3.2.1 – which section is meant). Although many of the existing glass fibre
reinforcement shapes are not adapted to this need and are too tight, this drawback can be, and is being,
overcome. Secondly, the alkalinity of most cements creates an environment for the glass fibres which
is degrading their reinforcing properties. Since the main aim of the hand lay-up technique is to
improve the tensile strength due to the presence of the fibres, this achievement can be lost with time. It
is still an open debate whether the so-called AR-fibres (see 3.1.3 – which section is meant) are
immune to this degradation.
Placing multidimensional fabrics made of alkali resistant glass (AR-glass) instead of the usual steel
bars in the main load directions of the concrete, might enable the engineer in the future to design and
build light and slender structures with a high load carrying capacity. However, for the production of
three-dimensional thin structured profiles an appropriate production machine was developed within
the research project SFB 532 at the Fraunhofer Institute for Production Technology in Aachen. Within
this quasi-continuous production process an endless U-shaped profile, which incorporates both thin
walls and sharply edged corners, will be produced and subsequently cut down to the desired length.
The dimensions of this profile are given in Fig. 5.1.
10 mm
Textile reinforcement
40 mm
400 mm
Within the production process the highly flowable concrete matrices are pumped and injected in the
closed mould, and then are dehydrated under high pressure in order to gain an appropriate initial
strength for short demoulding periods. After pressing, the profile is shifted and the next part of the
element will be pressed which results in endless members within the quasi-continuous production
process as shown in Fig. 5.2.
56
Quasi-continuous production process U-shaped profile
infeed
infeed
infeed
Fig. 5.2. Quasi-continuous production process for structural TRC U-shaped profile.
For manual production of TRC elements usually fine grained matrices with plastic to flowable
consistencies are used, while for the production process an especially highly flowable consistency was
required for the pumping and injection processes. These highly flowable consistencies which also
offer full penetration of the technical textiles were achieved by using a small maximum grain size
(dmax = 0.6 mm), high binder contents, and adding different pozzolanic additives, and finally
superplasticisers, all of which lead to a more homogeneous and finer structure compared to ordinary
concrete. Table 1 shows the matrix compositions of the reference mixture PZ-0899-01 (Table 4.2)
[Bra2001] and the modified injection mixture for the production process PZ-IN-04. The reference
mixture offers already such a dense packing of the fines that no further compaction with dehydration
was possible. For this reason the modified mixture PZ-IN-04 (Table 4.2) was developed, which has a
higher water content than the reference mixture, and such met all requirements for the planned
production technique, i.e. injection and pumping as well as appropriate dehydration characteristics, yet
stable in consistency.
57
Fig. 5.3. Squeeze pump (left), organic glass mould (middle), detail of injection point (right).
The pumping of the fine grained concrete mixture PZ-IN-04 did not show any difficulties even when a
distributor was used and the pumping was carried out with 4 flexible tubes at once (Fig. 5.3, left).
Also, the flow path of the matrix was investigated using up to four injection points at the same time,
but this procedure lead to air intrusions in the closed mould. Hence, in the production machine valves
for air discharge were integrated, and finally one injection point will be used for the production in the
machine in order to avoid air entraining. This leads to a filling duration of about 1 to 2 minutes for the
U-shaped mould within the production machine.
58
adjustment
guidance
mould
In preliminary tests the influence of the pressure level and duration on the dehydration process was
investigated with the aim of defining the required pressing parameters for the production machine.
Pressed concrete specimens with a thickness of about 20 mm were produced with a pressure of 1.0,
2.0, and 1.5 MPa. The dehydration process was tested by determining the displacement of the press
plunger as this is proportional to the amount of pressed out water. The pressing process was
considered finished if no further displacement of the press plunger was measured. This test set-up was
also used to produce pressed concrete prisms for the compressive and flexural bend tests as described
in the following.
The investigations on the dehydration process with a varying pressure level and duration showed that
for all investigated pressing conditions an appropriate form stability of the pressed concrete specimens
was achieved, see as an example a specimen of the dimension 40 x 40 x 160 mm³ in Fig. 5.5, left.
Fig. 5.5, right, shows the displacement of the press plunger over time for the investigated pressure
grades. It shows that for higher pressure grades the ultimate solidification of the matrix is already
reached after shorter compression times. The amount of water pressed out per second at 2.0 MPa is
almost constant up to a certain point, after which the curve levels off. The fact that only a minimal
amount of water is pressed out after this indicates that the concrete has solidified. Fig. 5.5, right, also
indicates that less water is extracted over time at a lower pressure levels. Hence, these results show
that a high pressure grade leads to stronger solidification and that within the production process at a
pressure level of 2.0 MPa very short pressing periods of only about 30 seconds will be required.
59
Displacement of pressure plunger in mm
2.0
27 sec
20 Pa
1.5 60 sec
15 Pa
10 Pa
1.0 70 sec
0.5
0.0
0 100 200 300 400
Time in seconds
Fig. 5.5. Form stable pressed concrete specimen (left), and influence of pressure grade on the
dehydration process (right).
The results of the kiln drying in Table 5.1 show that at a pressure level of 2.0 MPa an increase in
duration of pressure from 1 to 2 minutes did not lead to any further solidification or pressing out of
water as for both testing conditions a similar water content with a w/b = 0.38 was determined. An
increase in the pressure grade from 2.0 to 4.0 MPa obviously lead to a more significant compaction
with higher water losses, as a lower water content and a reduced w/b = 0.36 were determined.
60
specimens produced under a pressure of 2.0 MPa for 1 and 2 minutes respectively, and a pressure of
4.0 MPa for 1 minute. For the compression tests, specimens of the dimension 160 x 40 x 40 mm³ were
produced, while for the 3-point bend tests specimens with a varying height d = 10, 20, and 40 mm
were produced, i.e. 160 x 40 x d mm³. The compressive and flexural strengths of these specimens were
determined according to [DIN 18555-2 09.82] at a testing age of 28 days (water storage at 20 °C).
The compressive strengths of the pressed concrete specimens were determined at a testing age of
28 days and are given in Table 5.2. These results correspond with the w/b ratios as determined by kiln
drying (Table 5.1). The highest compressive strengths fc = 79.7 N/mm² were determined for the
highest pressure level of 40 MPa corresponding with the lowest w/b = 0.36, while the compressive
strengths for the other pressing levels were similar with values of about 70 to 71 N/mm². These values
were found in the range of the compressive strengths of the reference mixture PZ-0899-01 which has a
similar w/b ratio of w/b = 0.40.
61
Flexural strength ffl of pressed concrete specimens in N/mm²
24
PZ-IN-04 1min 20bar
160 x 40 x d mm³
20 28 days 2min 20bar
water storage, 20 °C
1min 40bar
16
12 unpressed
reference mixture
0
d = 10 mm d = 20 mm d = 40 mm PZ-0899-01
pressure plunger
rubber block
fresh concrete
pressed concrete
Fig. 5.7. Manufacture of a fine grained concrete specimen in stages with a joint.
62
The flexural strengths of the pressed concrete have been investigated as described in the above section,
but furthermore, the influence of the joint which will inevitably be produced within the production
machine on the flexural strength was investigated. In 4-point bend tests of pressed specimens with a
joint flexural strengths ffl = 8.2 N/mm² were determined, i.e. a significant reduction in the flexural
strength was determined in comparison to the specimens without a joint with flexural strengths of
about 12 to 16 N/mm² (Fig. 5.6). However, the flexural strength of the specimen with a joint were in
the same range as the flexural strengths of the reference mixture (unpressed concrete), which generally
is used for the manual production of TRC elements. Hence, within the production process appropriate
mechanical properties will be achieved even if there are joints in the structural elements.
Compression unit
Textile Textile supplying unit Oven
A
concrete A-A
injection conveyor belt +
filter element
element
textile
plunger
A
S formed element
mould seal
63
Rubber stamp Plunger Clamping stamp
Textile
1
High early
Fresh concrete strength
concrete part
2
Furthermore, the injection mixture offered a sufficient form stability immediately after the pressing
procedure. For this mixture according to the results of the tests a pressing duration of about 30 to 60
seconds for a pressure level of 2.0 MPa is expected within the production machine. These properties of
the newly developed fine grained concrete mixture will allow a high production frequency within the
quasi-continuous production process. Due to the step-wise pressing procedure an inevitable joint is
produced, but the results of the flexural strengths of compressed concrete specimens with such a joint
showed sufficiently high flexural strengths, which were even in the range of the reference mixture,
which is used throughout the research project SFB 532 for manual production of TRC elements. Also,
for the compressed mixture PZ-IN-04 appropriate mechanical properties regarding compressive and
flexural strengths were determined for the planned applications like e.g. integrated formwork elements
or similar. The optimisation of machine relevant parameters within the production process will be
subject of further research in the same way as modifications of the machine for a quasi-continuous
production of 3-dimensional hollow TRC structures, e.g. Pi- or box-shaped profiles.
64
laminates. The mixture should be sufficiently fluid to enable the fabric to transfer through the cement
slurry but dense enough so that it will remain on the fabric when it leaves the cement bath. The
pultrusion set up is presented in Fig. 5.11. By using the pultrusion method fabric-cement sheets with
various widths, lengths and thickness of composites can be produced. After forming the sample
through the rollers, additional pressure is then applied on top of the laminates to improve penetration
of the matrix in between the opening of the fabrics and the bundle filaments. The intensity of this
applied pressure is limited and cannot reach high levels, as the matrix is still fresh at this point and
elevated pressures can remove most of the matrix in between the fabric laminates. This results in an
insufficient matrix content to help bind the laminates.
The “FBK-Wellplatte” was the first glass fibre reinforced product to receive in 1987 a technical
approval from “Deutsches Institut für Bautechnik (DiBt)” [DiBt].
65
5.5.3 Module process technology (LPE) (Pachow)
One typical module process technology is the Durapact process technology [Pac1998] which also is a
low pressure extrusion technique. It is designed for the production of flat textile reinforced elements
for rather low capacities and offers a much greater flexibility in product choice and product changes
during production compared to the Wellcrete technology. It is characterised by low investment costs
[Pac1998] and low space requirement as it is based on a modular system. These plants usually consists
of a production belt and a frame supporting a variable station, which is exchanged for a the required
processing step such as positioning of the formwork, matrix (with or without short fibres) extrusion,
compaction, smoothing, or cutting. Stationary plants usually consist of a continuous production belt
fitted with a multi-chamber extruder. The product is built up layer by layer on an extruder throughput
and subsequently deposited on a form as e.g. a flat form sheet.
A roving consists of hundreds or thousands of filaments. In a not impregnated roving these filaments
are loosely packed. If transverse loading is applied the filaments are squeezed out. Further on, the
outer filaments in direct contact to the clamping device do not transfer load to inner filaments due to
their low friction. This means that prestressing with a clamping device is very difficult and almost
impossible.
When the rovings are resin impregnated they can be stressed more easily. In the following the device
is described which has been used for unimpregnated rovings and impregnated ones. Another device is
described which can be useful for impregnated rovings.
66
Side view: Wing bolt M10
Slope for
toothed sheet
Detail A
Top view:
80.00 mm
48.00 mm
50
Bore-hole for
wing bolt M10
18 12
50.00 mm
27.00 mm
6
Borehole 12 mm 12
30
100.00 mm
17.50 mm
Bore-hole for
wing bolt M10
This device is derived from a proposal by Helbing [Hel1976]. The device has been designed such that
a 100 mm wide textile strip can be prestressed. The anchorage length amounts to about 50 mm, the
maximum prestressing force depends on the textile properties and the resin properties. Experience has
been gained that a prestressing force up to 4 kN can be transferred [Rei2002]. The prestressing device
can be used for every type of textile. Fig. 5.13 shows a picture of a clamped textile material.
After the test the screws can be loosened again and the textile can be removed without problem
because the surface was greased and there was no bond between epoxy resin and steel. The clamping
device can immediately be reused again.
A similar device has been used by [Vil2003]. There the textile is anchored in an epoxy block which is
then fixed in a clamping device such as shown in Fig. 5.14.
67
Fig. 5.14. Epoxy end-block holding 15 Aramid rovings (244 tex) [Vil2003].
68
Fig. 5.16. Prestressing frame.
The frame consists of a rectangular rigid steel frame with 40 hydraulic pistons. The pistons generate a
maximum force of 3.5 kN each with an oil pressure of 12 MPa. In the middle of the frame a flat mould
is situated on which the textile reinforced plates can be concreted. Depending on the thickness of the
plate it can be concreted in one layer or two layers. It depends also on the fluidity of the concrete
(mortar). If a very fluid mortar is used one can spread the mortar throughout the mould and finish it by
hand. If the mortar is less fluid one needs a vibrating board in order to compact the concrete. Fig. 5.17
shows a picture where concrete is spread on the textile and vibrated with a vibrating board.
The sides are cast with a magnetic steel strip. This strip enables also the exact positioning of the textile
in the middle of the plate. After concreting the plates are cured 24 hours. Then, the prestressing is
released and the plates can be demoulded.
69
Fig. 5.18. End block and anchorage system [Vil2003].
It was not possible, for example, to anchor 3 layers of aramid fabric with 5 rovings each with a total
target load of 2.2kN (500 lb) without slip between the aramid fibers and the epoxy matrix and pullout
of complete rovings. After an extensive trial-and-error period, it was found necessary to modify the
epoxy with a proprietary filler consisting of very fine particles, which appeared to better penetrate the
interstitial spaces within each roving. With this modified epoxy it was possible to stress 4 layers of 5
rovings to ¾ of the original stress level and the same target prestress force of 2.2kN. This system was
then used successfully for a large number of test specimens.
2. A glass concrete mix design was developed that combines excellent workability with high-
early strength. Using glass concrete as the prestressed matrix will allow to further explore the great
esthetic potential of this novel material in form of prestressed thin sheet applications. Compatibility of
aramid fibers with ordinary concrete matrixes still has to be shown. It is assumed that some of the
mechanisms, which inhibit ASR in glass concrete, also ensure the structural stability of the prestressed
aramid fibers, for their primary degradation mechanism is hydrolysis caused by the same alkaline
environment, which is responsible for the initiation of ASR.
3. Few studies in the field of structural engineering have investigated aramid fibers with regard
to their microstructure. This work gives a detailed overview on what aramid fibers are, how they
deform, and especially what makes them different from solid fibrous materials such as glass or carbon.
Understanding their basic deformation mechanism allows for the utilization of mechanical
preconditioning to greatly reduce tension losses caused by relaxation. This finding will also be of
importance to conventional prestressing applications where aramid fibers are used as FRP.
70
4. Results from bending experiments carried out with prestressed beams with thicknesses of ¼ in
(7 mm) and 1/8 in (3 mm) are very promising. The relatively small rovings develop sufficient bond
with the glass concrete matrix to transfer prestress as large as 2 ksi (14 MPa). The epoxy end-blocks
developed especially for this study could be safely removed after 7 days, prior to the bending
experiments, with no negative effect to the flexural performance. Basic principles of prestressed
concrete theory apply also to prestressed thin sheets. Especially the large deformability of the 1/8 in (3
mm) thin specimens may offer entirely new applications for concrete based materials.
5. Residual strains after bending experiments were found to be very small. Combined with a
permanently compressed matrix very large numbers of loading cycles need to be applied to produce
permanent visible deflections. However, when the specimen is wet during load application,
accumulation of residual strains can be greatly accelerated. The finding is critical for the development
of prestressed thin sheets for outdoor- or wet-room applications.
The usage of aramid fibers for prestressing thin sheets made of glass concrete was at first inspired by
their outstanding success as the basis material for the production of body armor. In this work, their
ability to undergo large deformations by dissipating extensive amounts of energy was successfully
extended to the similarly promising fields of textile-reinforced concrete and Portland cement-based
thin sheet applications [Vil2003].
5.7.1 Spacers
5.7.2 Curing
5.7.3 Others
71
Table 5.3. Production technology for TRC.
Shotcrete short fibres are cut from a a) short fibres (6 - 25 mm) plastic consistency rather thick elements with chimney covers, decorative
a) short fibres and matrix roving and are sprayed with lower strength units
the separately prepared
matrix
3-dimensional forms
b) textile reinforcement +
short fibres and matrix
3-dimensional forms
c) textile reinforcement +
matrix
Injection
1)
a) Injection a) short fibres (6 - 18 mm)
fully-automated b) textiles possible
Injection
Lamination
Dehydration
Pultrusion /
1)
Extrusion a) Extrusion a) short fibres (12 - 50 mm) corrugated sheets
fully-automated b) rovings flat sheets
Wellcrete
Pultrusion /
1)
Extrusion b) Extrusion a) short fibres (12 - 36 mm) profiled panels
fully-automated b) rovings flat panels
Durapact Process c) textiles side forms
technology sandwich panels
Pultrusion /
Extrusion c) Extrusion
input Peled
Lamination + Injection
Prestressing
1)
[Pac2004]
72
REFERENCES
[DIN 18555-2 09.82] Prüfung von Mörteln mit mineralischen Bindemitteln, Teil 2: Frischmörtel mit
dichten Zuschlägen, Bestimmung der Konsistenz, der Rohdichte und des Luftgehalts. –
German Standard.
[DIN 1048-5, 06.81] Prüfverfahren für Beton, gesondert hergestellte Probekörper. – German
Standard.
[Hel1976] Helbing, A.K., Brühwiler, E.: Eine neue Halterung für Zugversuche mit Beton-
Probekörpern. Material und Technik 15 (1976), No. 4, pp 103-107
[Rei1976] Reinhardt, H.W.: On the biaxial testing and strength of coated fabrics. In: Experimental
Mechanics 16 (1976), No. 2, S. 71-74
[Rei2002] Reinhardt, H.W., Krüger, M.: Fine grain concrete panels prestressed with a textile fabric.
In: Proceedings of the JCI Intern. Workshop on Ductile Fiber Reinforced Cementitious
Composites (DFRCC): Application and Evaluation (Takayama, Japan 2002). Tokyo :
Japan Concrete Institute, pp 23-32
[Vil2003] Vilkner, G.: Glass Concrete Thin Sheets reinforced with prestressed Aramid Fabrics.
Dissertation, Columbia University, 2003
73
6 COMPOSITE MATERIALS
ABSTRACT: In this chapter the principle bond behavior of a strand .... is discussed ...abstract is
missing
6.1.1 General
Bond of reinforcing steel in concrete is usually divided into several mechanisms. The weakest
contribution to bond is adhesion which is deteriorated after a first slippage of the steel in concrete.
Then we have friction which is almost constant during the slippage of steel in the concrete. Another
contribution is lack of fit which is mainly due to the irregularities of the cross-section of a strand.
Another one is the Hoyer effect which is generated due to the transverse extension of a wire when the
prestressing force is released. It is only active in a very short part of the anchorage zone of a
prestressing steel in concrete. The most efficient part is the mechanical bond due to the resistance of
the ribs in the concrete.
Textile reinforcement has another bond mechanism. Due to the required reinforcement ratio textiles
are composed of hundreds or thousands of filaments which means that only the outer layer of a roving
can really have an adhesion to the surrounding concrete. This means that only a part of the roving is
anchored in the concrete and the rest can slip easily within the roving at low friction. Fig. 6.1 shows a
schematic of this mechanism.
74
Slave filaments (Aext)
Fig. 6.1. Roving with inner bond < outer bond (e.g.: roving not fully infiltrated with cementitious
matrix).
One can see that the outer filaments are embedded in a hardened cement paste and have a good
anchorage in this paste. The inner filaments are not reached by the hydrated cement paste and thus are
not at all anchored in the concrete. Only the friction between the filaments can generate a certain bond
resistance. To achieve sufficient internal bond a roving could be impregnated by a resin (epoxy resin
or vinyl resin). Then the strand is more or less a rigid composite made up of the filaments and the
resin. Fig. 6.2 shows this situation.
Bond area
Epoxy impregnation
Fig. 6.2. Roving with inner bond > outer bond (e.g. epoxy impregnated roving).
In this case we have an anchorage of the whole filaments in the hydrated cement paste. The filaments
are not allowed to slip anymore within the outer filaments.
This behaviour leads to a certain overall schematic which is depicted in Fig. 6.3. There a load slip
diagram is shown. The lower line shows the case where the inner bond is smaller than the outer bond,
that means the inner filaments can slip with respect to the outer filaments. Then the load-slip curve is
rather flat but at the other hand it has a long tale with a certain pseudo ductility. This effect is due to
the higher slippage of the inner filaments which leads to different stresses in the filaments and thus to
successive failure of the filaments.
75
Internal bond failure (only case b)
Load
External bond failure
Successive filament failure
σ⋅Aa
Case a: Internal bond > External bond
Gf,a
Gf,a = Gf,b
σb,ext⋅Ab,ext
+ σb,int⋅Ab,int
Case b: Internal bond < external bond
Pseudo ductility
Gf,b
Slip
Fig. 6.3. Schematic of pullout slip behaviour of various textiles with infinite embedment length.
Contrary, if the inner bond is larger than the outer bond, i.e. that is true for the impregnated rovings,
we have a larger resistance against slip. The slip-load curve is steeper but falls down immediately after
a certain slip because all filaments fail almost at the same time. This case is almost similar to the case
of smooth bars in concrete. The slip is very short and there is no pseudo ductility anymore. As shown
above bond behaviour of a textile roving and therefore the load bearing behaviour of a reinforced
element is greatly influenced by the inner bond.
The fracture behaviour of unidirectional fibrous composites - in the current case a yarn embedded in a
cementitious matrix - has been tackled with several approaches over the years. One of the simplest
approaches to idealise this system, the so-called one cylinder model (Fig. 6.4), is to assume a
homogeneous and linear elastic yarn loaded under quasi-static conditions and further neglect matrix
deformation [Zas2003]. However, as has been shown in [Ban2004], this simplification of the
composite does not reflect the telescopic and staggered failure of the yarn in any way.
A somewhat more progressive approach had already been introduced some years beforehand by
[Ohn1994], who idealised the yarn / matrix system as a two ring model containing sleeve and core
filaments (Fig. 6.4). However, since the bond stress between the different “rings” is assumed to be
constant and no tensile failure is considered, this model also oversimplifies reality.
Segment models as shown in Fig. 6.4 can likewise be found in a variety of studies, for example
[Sej2002]. They proposed to use periodic unit cells (PUC) which consist of N filaments statistically
equivalent to the original microstructure. To achieve this, fundamental knowledge about the material’s
76
statistics is needed and the original microstructure has to be quantified, e.g. by using a collection of
SEM images. Introducing an n-point probability function Sm defining the chance of finding a
randomly placed filament located in the matrix, a PUC has to be determined with the help of stochastic
optimisation procedures which matches in its Sm function with the original microstructure. Hence the
actual objective is to define the material statistics and properties of the PUC in accordance with the
original composite. However, the bond between the filaments and the surrounding matrix is assumed
as perfect, and the different failure mechanisms occurring during the yarn break down, that is
debonding and tensile failure of the filaments, are not considered in this model. Thus for example the
change in the number of load carrying filaments (load carrying area of the yarn) is neglected and a
staggered breakdown cannot be simulated.
Such a filament breakage and debonding of the filaments from the matrix as well as plasticity of the
matrix are included in the lamina model proposed by [Bey1997], who enhanced an idea introduced by
[Hed1961] as early as in 1961. In this approach the stress distributions are analysed in a sheet of
parallel, high modulus filaments which carry normal loads and are embedded in a low modulus matrix
which carries only shear. Unfortunately, the model implies that the load is introduced only by the
filaments (tensional test) and hence this applied technique cannot easily be transferred to this study.
• the bond versus slip relation τ(s) predominating the bond between AR-glass filament and cement
based matrix (using these τ(s) relations pull-out responses of filaments P(ω) can be simulated for
different embedding lengths Lv, see Fig. 6.5),
• the experimental load versus displacement relationship P(Ω) describing the pull-out response of a
yarn / cement based matrix system (Fig. 6.6), and
• the active filaments versus load relationship NF(Ω) which characterises the failure process of a
yarn during this test (Fig. 6.6).
Pull-out force P in N
3 LV = 1 mm
2 LV = 2 mm
LV = 3 mm
1 Failure range of filaments Ω<ΩNFL LV = 5 mm
LV = 10 mm
Fig. 6.5. Simulated filament pull-out responses with different embedded lengths LV [Ban2004].
77
Pull-out load P in N
Active filaments NF
600 Specimen A: 1500
Experimental P(Ω) NF, m
500 NF, m-1 o Layer m
400 Simulation P(Ω ), m = 5
1000
Simulation P(Ω ), m = 100
300
200 500
100 NF, 1
Layer 1
0 0
0,0 0,5 1,0 1,5 2,0 0 Ω Ωv,f 2.0
m,f
a) Displacement Ω in mm b) Displacement Ω in mm
The principle idea of the YMB - model states that if the pull-out response P(ω) of every single
filament i for 1 ≤ i ≤ NF,m at every single load step is known, the resulting load displacement
relationship and hence the overall response of the yarn P(Ω) can be calculated with a simple
summation procedure similar to [Hed1961]. NF,m corresponds to the number of filaments per yarn, and
Ω to the displacement measured during a yarn pull-out test. Furthermore the model adopts a ring
structure similar to [Ohn1994] which is justified by the telescopic failure visualised in the FILT test
[Ban2004], that is a primary break down of the outer layer of filaments followed by a successive
failure of the adjacent layers until a core of filaments is finally pulled out. Thus the composite is
idealised in the YMB - model as a layered system made up of m layers each with o filaments per layer
v for 1 ≤ v ≤ m (Fig. 6.7) where each layer v responds with an average pull-out response pv(Ω) to the
introduced overall pull-out load P(Ω).
N F, v
Filament 1
Filament ∑ i
i =1 L
Layer m
As the number of filaments in each idealised layer is constant, the active filaments versus
displacement relationship NF(Ω) may be matched by a step function with a constant step height o as
shown in Fig. 6.6. The number of steps used naturally corresponds to the number of layers assumed in
the idealisation; m = NF, m / o. The width of each step refers to the displacement range in which o
filaments of the corresponding layer v fail. The displacements at the end of each step (black squares)
are taken as average failure displacements, that is the layer v is assumed to fail in tension at a pull-out
displacement Ωv, f.
78
Since the filaments in the outer layers reach their tensile strength fv,t and thus also their failure strain
prior to the subsequent layers, it is reasoned that the average extensional stiffness per layer is
gradually decreasing from the outer to the inner layers.
• the matrix which penetrated between the filaments only carries shear,
• the bond between the single filaments and the matrix is basically characterised by the BSR τ(s),
and
the YMB - model further postulates that the main parameter involving this stiffness gradient has to be
the decreasing interlayer contact area (L UC)v towards the core of the yarn. Note, Lv corresponds to the
embedded length and UC,v to the contact perimeter of a layer v. Although this assumption is verified by
experimental findings [Ban2004], it is so far still not possible to determine the exact values of these
geometric micro-structural parameters. In order to still allow an analytical modelling of the pull-out
response of a yarn without adequate knowledge of the contact areas, [Ban2004] proposes the
following proceeding under the assumptions that
• if matrix has penetrated into a layer v at a location i over a length Lv,i all filaments in that layer are
completely embedded in matrix over their circumference π d at that location over that length, and
• all individual embedded lengths Lv,i of a layer may be summed up to a layer dependent embedded
length Lv. Or vice versa, the filaments of a layer v are not embedded over the “free” length Ψv;
compare Fig. 6.8.
Note that only an embedded length of a couple of millimetres is needed to transfer all the forces a
filament can carry in the surrounding matrix (see Fig. 6.5). This kind of idealisation of the
microstructure is also proposed in [Sch2003a] and named “adhesion-bridge model”. Experimental
results show [Ban2004] that the layers and hence the filaments experience a staggered break down, i.e.
the failure locations run from x = L for the sleeve filaments to x << L for filaments near the core. By
assuming that the tensile strength of a filament does not vary over its length, it implies that the load
transfer between filaments and matrix by shear stresses is negligible in the regions Ψv, where the
filaments are not embedded.
Lv = ∑Lv,i
Lv,1 Lv,2 Lv,i i =1 Ψv
Penetrated matrix
Filament
Bulk matrix
79
By additionally implying that the average pull-out force pv of a layer v is governed only in the regions
of embedment Lv and the corresponding displacement Ω of a layer v consists of two components
(II) the elongation ∆ resulting from the deformation of the filaments over the free length Ψv
the YMB - model analytically describes the pull-out behaviour of a yarn in a cement based matrix
under a pull-out load as follows:
m ! m
P (Ω ) = ∑ pv (Ω = ω + ∆ ) ≡ ∑ Pv (ω ) ⋅ o, ∆ =
Pv (ω ) (Ω v, f − ω v , f ) ⋅ EF
(1)
v =1 v =1 EF AF f v ,t
Thereby the average pull-out response pv(Ω) of layer v is equivalent to o times the response Pv(ω) of a
filament in a single filament pull-out test as simulated on the basis of given bond properties for the
corresponding geometric arrangement of the layer v (Fig. 6.5). EF is the Young’s modulus of the
filament and AF the filament’s cross-sectional area. Note that the tensile strength of the filaments may
be layer dependent, that is fv,t. ωv,f is the corresponding pull-out displacement; compare Fig. 6.5.
For a 5 layer idealisation a summation of the individual pv(Ω) responses results in the simulated P(Ω)
relation stated in Fig. 6.6. Additionally the experimentally determined load versus displacement
relationship P(Ω) is presented in the diagram. Increasing now the number of layers to v = 100, the
pull-out response can be simulated in more detail. Note, the higher the number of layers taken, that is
the lower the number of filaments included in a single layer, the more the complex interaction
between the individual filaments is considered due to the fact, that this information is hidden in the
active filaments versus displacement relationship (Fig. 6.6).
Fig. 6.9 shows the situation on principle. Line 1 represents a filament directly in contact with the
matrix. The deformation length L0 is identical with the effective length Lw. Line 2 shows an elastic
bond; as a consequence the deformation length L0 is greater than the effective length Lw. Line 3 shows
adhesive cross-linkages bonding filament to filament.
80
Fig. 6.9. Different deformation lengths L0 due to different positions of cross linkages bonding the
filaments and effective deformation lengths Lw.
Usually the deformation length of a filament depends on the distance to the surface of the roving
wherein it is situated, because the number of adhesive cross-linkages decreases from the outer part to
the inner part of the roving. In the centre of a roving a great number of filaments may remain
unbonded. In Fig. 6.10, left, the effective deformation lengths LW of all filaments of a roving in order
to the length and on the right hand side the stress distributions of all crack bridging filaments of the
roving are shown. With increasing crack width two processes take place simultaneously:
This process continues as long as enough filaments remain in the uncracked state.
Fig. 6.10. Effective elongation length lW of all filaments of a roving (left) and stress distribution
with increasing crack width (right).
The crack bridging behaviour can be predicted as shown in Fig. 6.11 for a particular assumption of the
distribution of adhesive cross linkages in a roving (see enclosed small curve in figure)
81
Fig. 6.11. Example for the prediction of the crack bridging behaviour of textile glass fibre
reinforcement compared to a measured curve.
10 mm
Strand embedded
in epoxy resin block 30 mm
50 mm
Fine-grained concrete
50 mm
The strand is then cast in a fine-grained concrete matrix. The dimensions of the specimens are 50 ⋅ 50 ⋅
L mm3 with L being the embedded length of the strand. L is chosen for the presented tests to be
30 mm (Fig. 6.12).
82
Testing procedure
The pull-out tests are carried out using a universal testing machine at a displacement rate of
0.1 mm/min at 20°C until a maximum displacement of about 1.7 mm is reached. The specimen is
mounted in the machine such that the epoxy resin block is placed through a cut-out of a steel plate and
then clamped in a mechanical grip fixed to the cross head of the machine. The steel plate itself is fixed
by a frame to the bench of the testing machine. All of this creates a set up of the pull-push test, i.e. the
strand is pulled out of the matrix against a restraint. For illustration see Fig. 6.13. The pull-out loads
and the cross head displacements are recorded every 2 N change in force resulting in a load versus
displacement relationship P(Ω).
Load cell
Light source P, Ω
Steel plate
Matrix Epoxy resin
123
Strand P
Digital camera
ω
During the actual pull-out test the specimen is additionally exposed by an artificial light source from
the front via the epoxy resin block as sketched in Fig. 6.13. If a small charge-coupled device camera
(CCD-camera) and a zoom lens is used, the strand can be distinguished on the rear of the specimen
from the surrounding matrix due to its exposure and therefore bright appearance (on the left hand side
of Fig. 6.14 and Fig. 6.15). After a tensile failure a filament is no longer capable of transferring light,
and therefore the bright appearance vanishes for the next load step in the next image.
a) b)
Fig. 6.14. a) Digital image of 5-VET strand, b) Binarised image of the detected filaments using a
numerical image analyzing routine.
If a numerical image analysing routine is applied during this study, the optical image of the CCD-
camera presented in Fig. 6.14 a can be analysed and converted in a binarised image (black and white
pixels) as shown in Fig. 6.14 b. Depending on the chosen resolution and the zoom lens used it is
83
possible to visualise every single filament. For illustration see Fig. 6.15 a and Fig. 6.15 b. However,
care must be taken that the complete strand is recorded during the test. This test method has been
called FILT test (Failure Investigation using Light Transmission properties). Details are described in
[Ban2004].
a) b)
Fig. 6.15. Enlargement of Fig. 6.14 a and 14b: a) Digital image of the strand, b) Binarized image of
the detected filaments using a numerical image analyzing routine.
Fig. 6.16 shows the recorded pull-out load versus displacement diagrams P(Ω) of four specimens. In
general, the principal trend of a pull-out relation can be observed. An almost linearly increasing part
with a subsequent non-linear region until the maximum pull-out load is reached, followed by a
decreasing softening branch.
Pull-out load P in N
600 Specimen A
Specimen B
400 Specimen C
Specimen D
200
0
0.0 0.5 1.0 1.5 2.0
Displacement Ω in mm
As explained in Chapter 6.1.2.1 the images of the FILT test are taken and binarised and recorded
simultaneously with the loads and displacements of the pull-out test. The load versus displacement
diagram P(Ω) recorded for specimen A is presented in the second column of Table 6.1 for different
stages of the pull-out process to clarify this procedure: Ω at maximum load and at 3 subsequent load
steps.
84
Table 6.1. Observed failure process of a 5-VET strand embedded in PZ-0899-01 matrix under a
pull-out load (FILT tests); specimen A.
L-S 1 Load versus displacement diagram P(Ω) FILT test image
1 Pull-out load P in N
600
400
200
0
0.0 0.5 1.0 1.5 2.0
Displacement Ω in mm
2 Pull-out load P in N
600
400
200
0
0.0 0.5 1.0 1.5 2.0
Displacement Ω in mm
3 Pull-out load P in N
600
400
200
0
0.0 0.5 1.0 1.5 2.0
Displacement Ω in mm
4 Pull-out load P in N
600
400
200
0
0.0 0.5 1.0 1.5 2.0
Displacement Ω in mm
1
L-S = load step
For these selected load steps the corresponding binarised FILT images are presented in the third
column of Table 6.1. Remaining intact filaments are visualised by white pixels. The load steps are
labelled in the first column of Table 6.1.
85
a) b)
c) d)
Fig. 6.17. FILT images of specimen a) to d) respectively at the maximum introduced displacement
of about 1.7 mm. Thin white line outlines the perimeter of the strand before the pull-out
test.
It is obviously visible that with an increasing pull-out displacement the amount of intact filaments
decreases. However, this failure process is not consistent over the cross-section of the strand but
different groups of filaments form together, which are finally pulled out. In Fig. 6.17 the FILT images
of all 4 specimens are given at the final strand-end displacement of about 1.7 mm. A comparison of
these FILT images shows the variety of results for these 3 replications, i.e. that different amounts of
filaments at different locations are left intact at the end of the pull-out tests. The results can be
evaluated with respect to the number of broken filaments and the location of the failure.
The results support the idea that the pull-out behaviour is controlled by a strong bonding of the
external filaments in the strand and a slip of the inner filaments, again influenced by the random and
therefore unpredictable penetration of matrix into the core [Maj1974, Bar1987, Cur2003]. In principle
this is not a surprise, although cement grains which measure approximately 10 µm can hardly
penetrate into the spaces (approximately 3 µm wide) between the filaments if the filaments are
assembled in a compact form. However, in many cases the original compact flattened bundle is
loosened during the placing and manufacturing, hence the matrix may penetrate up to a certain degree
into the core of the strand. Nevertheless, the formation of hydration products within the strand is
initially limited.
This uncontrolled penetration leads to a different formation of the inner and outer bond characteristics,
and hence the failure mechanism after exceeding the maximum pull-out load is described as a so-
called “telescopic failure”, i.e. a successive break down layer by layer from the sleeve to the core
filaments. After the tensile failure of the outer filament layers, a core of inner filaments is pulled out of
the strand.
Based on these evaluated results, the complex failure process of a pull-out test on a strand / cement
based matrix system may be sketched as presented in Fig. 6.7.
86
6.1.2.2 Two-sided em bedment test
There is a test which combines the advantage of the pullout test, i.e. to measure clearly the bond-stress
slip relation with the advantage of measuring real material behaviour like it occurs in a crack of a
reinforced element. The test specimen consists of a plate in which a fabric is embedded. The plate is
140 mm long and 70 mm wide and has the thickness of 10 mm. The specimen is cut from the edges in
such a way that only one roving is left intact in the middle of the specimen. The upper side of the
specimen is clamped in a toothed steel plate and the lower part is also clamped in another toothed steel
plate clamping device. The whole specimen is inserted in a tensile testing machine. Fig. 6.18 shows
the testing arrangement.
load cell
clamp
20.0 mm
70.0 mm
140.0 mm
LVDT
Roving
test specimen
Fig. 6.18. Testing arrangement for the two-sided pullout test, left: schematic, right: half-opened.
During the test the crack opening of the specimen is measured by LVDT and can be plotted versus the
tensile load. A typical force displacement diagram is shown by Fig. 6.19.
87
120 Carbon, plain
40
20
0
0 1 2 3 4
Slip ∆s*, mm
Fig. 6.19. Force slip diagram of carbon reinforced specimens with embedment length of 20 mm.
The figure shows a first part which consists of a mostly linear elastic behaviour up to a peak stress and
then a decay to a low value and a friction type remaining behaviour. The first part is due to the tight
embedment of the fabric in the concrete. When the peak stress is reached there is a slip at the shorter
part of the specimen with a final extraction of the roving from the concrete. The figure shows also the
influence of the type of roving and the prestressing of the roving. It can be seen that the specimen
which is epoxy impregnated and sand coated at the surface has the stiffest and greatest bond resistance
in the concrete. All the others are much lower, especially the ones which are not impregnated and not
sand coated.
Similar results have been obtained with AR-glass rovings. Fig. 6.20 shows that the non-impregnated
specimens have a very low maximum pullout force (only 20 N/mm). The impregnated ones have
values up to 60 N/mm and also after the roving starts to slip it has a relatively high residual pullout
force.
65
bond stress per unit length P/(c-∆s*), N/mm
AR-glass
60
no prestressing
55
50 AR-glass, epoxy impregnated
no prestressing
45 prestress 150N/roving
40 prestress 300N/roving
35
30
25
20
15
10
5
0
0 1 2 3 4
slip ∆s*, mm
Fig. 6.20. Bond stress slip diagram of AR-glass reinforced specimens with embedment length of
20 mm.
88
The difference in the two rovings can be seen from Fig. 6.21. The left picture shows that there is a
inner portion of the roving which does not take part in the force transfer between concrete and roving.
The friction between the single filaments is very low and therefore force transmission cannot take
place. On the other hand the external filaments show excellent bond and therefore a small anchorage
length. This means that the descending branch of the not impregnated roving in Fig. 6.20 shows a
combination of friction of internal and fracture of external filaments.
Fig. 6.21. AR-glass roving embedded in concrete, left: non-impregnated roving, right: impregnated
roving.
The right hand picture shows an impregnated roving. It can be seen that there are some voids inside
the roving (green area). On the other hand there is a full bond between concrete and roving and due to
the impregnation with epoxy resin there is also a very good force transmission between the sinlge
filaments.
One parameter that may strongly influence the bond performance is the difference in the coefficient of
thermal expansion from that of steel or concrete. It is also known that transverse pressure improves
bond which is neglected in many bond models. However, this effect seems to be not important for
embedded multi-filament rovings which have not been fully infiltrated with cement due to voids
between the inner filaments. Despite of this the Poisson’s effect becomes significant and influences
the transverse stress field if the roving is impregnated and/or prestressed. Some test results of carbon
and AR glass reinforced and prestressed specimen are illustrated in the following chapter and are also
discussed in [Krü2001a].
89
For numerical studies the bond properties between textiles and concrete, discrete elements were used.
The bond model proposed by Ožbolt et al. [Ozb2002] has therefore been modified for textile
reinforcement and used together with solid finite elements in a 3D FE studies [Krü2002a].
In the numerical studies bond between the textiles and concrete was simulated by discrete bond
element that have recently been implemented into 3D FE code MASA [Ozb2002]. Concrete, which is
discretised by the three dimensional finite elements, is modelled by the microplane model [Ozb2001].
The bond elements connect the concrete finite elements with the reinforcement that is represented by
the truss finite elements (see Fig. 6.22). Only the degrees of freedom in the bar direction are
considered. However, beside the tangential stresses parallel to the bar direction, the radial stresses
perpendicular to the bar direction are generated as well. It is assumed that at a given slip the radial
stress depends on the geometry of the bar and the bar strain as well as on the geometry and the
boundary conditions of the concrete specimen. The interaction between tangential and radial stresses is
accounted for in three different ways: (i) directly, the shear stress depends on the non-local
(representative) radial stress obtained from the concrete elements close to the reinforcing bar, (ii) the
local strain of the bar element and its lateral expansion or extension and (iii) indirectly, in a way that
the larger shear stress (higher bond strength due to larger ribs or roughness of the bar element) cause
higher activation of stresses in the radial direction.
In the present model, splitting of concrete is indirectly accounted for. Namely the interaction between
shear and radial stresses results in corresponding tangential tensile stresses that causes cracking of the
surrounding non-linear concrete elements and, therefore, failure of bond resistance.
Concrete
element
Repeated
Bond element
nodes
(zero width)
Fibre element
Fig. 6.22. Bond elements with zero width [20] – reference is missing.
90
Based on the experimental results [Eli1983, Mal1992] and as well documented by Lowes et al.
[Low2002], the bond slip relationship of steel reinforcement in concrete can be described by the
parameters that are summarized in Table 6.2. The same parameters are used for textile reinforcement,
but in a slightly different manner. The curve of the bond stress versus slip relationship used for the
numerical studies is illustrated in Fig. 6.23.
The parameter τm,0 and τf,0 represent the strength of the mechanical and frictional component (subscript
m and f), respectively, for the case of no confining pressure, no damage and elastically behaving
reinforcing bar element.
Up to the slip s1 at which peak bond strength is reached (see Fig. 6.23), all response curves are defined
by Menegotto-Pinto (MP) equation [Men1973]. The curve defines a curve connecting two line
segments and it reads:
1 R
1
τ ( s ) = τ% ⋅ τ 0 = s% ⋅ b + (1 − b) ⋅ ⋅ τ0 (2)
1 + s% R
where b is the ratio between the target and initial tangents, τ% and s% are normalized stress and
displacement, respectively, and R defines the radius of the curvature. τ0 and s0 are the parameters to
calculate the absolute stress and displacement from the normalized parameters.
k2
b= (3)
k1
s
s% = (5)
s0
91
(k sec ant − k 2 ) (1 − α )
s 0 = s1 ⋅ = s1 ⋅ k sec ant ⋅ (6)
k1 − k 2 k1 − α ⋅ k sec ant
τ 0 = s 0 ⋅ k1 (7)
Bond
stress τ
k2=α·ksecant
τ=τm+τf
k1
τ0
ksecant Monotonic
loading
τm τf=τf,r+τf,v
kunload
τf,v
τf,r
s0 s1 s2 s3 Slip s
Cyclic loading
τf,r
The parameter Ω is calculated as shown in equation (8). Three parameters are considered: ΩS controls
the influence of the yielding of steel reinforcement on the bond response and is set to ΩS = 1 for textile
reinforcement; ΩC accounts for the influence of the lateral stresses between reinforcement and
concrete caused by the stress in concrete and the local strain of the bar element and its lateral
expansion or extension; Ωcyc controls the influence of the loading-unloading-reloading on the bond
response.
Ω = Ω S ⋅ Ω C ⋅ Ω cyc (8)
As shown in equation (9) and Fig. 6.24, the parameter ΩC, which can theoretically vary between 0 and
2, accounts for two different effects. The first is the influence of the lateral strain of the stressed bar
element. The parameter hR is a constant that represents the surface roughness of the reinforcement bar.
Compared to the ribbed reinforcement, hR is close related to the height of the steel lugs. εs is the
reinforcement strain, ds = 2rs the bar diameter and µs is the Poisson’s ratio of the used reinforcement
element. The factor αr controls the influence of the radial concrete stress and for the calculations is set
92
to 1. The parameter αf controls the influence of the roughness of the reinforcement hR on the bond
response. In the present study it was set to 2.
Ωc
2,0
1,0
0,0
Fig. 6.24. Definition of ΩC as a function of lateral stress and strain (left) and representative volume
(right) to calculate average radial stress.
The parameter εp,0 is the strain due to prestressing of reinforcement. Consequently in the case of
prestressing and non external loading the bond is increased only by the radial stress in concrete nearby
the reinforcing bar.
σR 1
Ω c = 1, 0 + tanh α r ⋅ − α f ⋅ µ s ⋅ ( ε s − ε p,0 ) ⋅ (9)
2
0,1 ⋅ f c rs
1 −
( rs + h R )2
The influence of the radial stress in concrete in the vicinity of the reinforcing bar is accounted for by
an average radial stress σr perpendicular to the bar direction. The parameter fc is the uniaxial
compressive strength of concrete. In the finite element analysis the average radial stress associated to
the n-th bar element is calculated as:
1 N N
σr =
VR
∑ σ r ∆Vi
i
with VR = ∑ ∆V i
(10)
i =1 i =1
where ∆ Vi denotes the volume which corresponds to the i-th integration point of the finite element
and σ ir the stress perpendicular to the reinforcement. N is a total number of integration points that fall
into a cylinder of a diameter D (see Fig. 6.24, right sketch). In the presented model D is assumed to be
three times a bar diameter (D ≈ 3 ds). In (10) VR is the representative volume, i.e. the volume of the
concrete cylinder of diameter D that is associated to the truss finite element which represents a
reinforcing bar.
Experiments show that for cycling loading-unloading-reloading the bond strength significantly
decreases with increase of number of loading cycles [Eli1983, Bal1991]. In the present model this
effect is accounted for by the factor Ωcyc that reads:
93
Λ
1.1
where Λ is the accumulated shear energy dissipation and Λ0 is a constant representing the area under
the monotonic bond-slip curve of respective shear component. The above equation has been proposed
by Eligehausen [Eli1983] and it is based on a large number of cyclic test data of steel reinforced
concrete. Similar behavior seems also to be approximately valid for textile reinforced concrete,
however, it is not clarified up to now.
Due to high strength and large deformation capability of the used textiles and small sizes of the tested
specimens, it was considered to use high performance fine grain concrete (fc ≈ 90MPa, Ec ≈ 39GPa).
Criteria for this concrete were high early strength, low shrinkage, good workability (self compacting
and self levelling) and high adhesion and compatibility to the textile fabric. To meet this conditions
the concrete properties were chosen such as shown in Table 6.3. For more experimental detail see
Reinhardt and Krüger [Rei2001].
Textiles
In the tests, three different textiles were used for comparing their bond properties and for seeking a
strong bond effect. The textiles used as reinforcements include carbon, AR glass and aramid textiles
with a warp knitted structure. In this paper we focused our attention on the carbon and AR glass
fabrics.
WARP
In the pull out and bending tests for the textile fabrics, it was found that the strength of textiles
embedded in concrete is much higher than that measured from direct tensile tests for a single roving.
This implies that there should be an embedding effect which could lead to enhance the strength of
textiles. The numerical results that show the strength enhancing effect of the textiles embedded in
concrete will be discussed in another paper.
In order to increase bond strength and improving the bond properties the textiles were impregnated
with an epoxy resin, see Fig. 6.25.
94
Tensile stress, N/mm²
Carbon, 0,04 %/min
4000 Carbon, 10 %/min
Carbon, Epoxy 0,04 %/min
Carbon, Epoxy 10 %/min
Carbon, theor. calc.
Aramid, 1 %/min
3000
Aramid, Epoxy 1 %/min
AR glass, 0,04 %/min
AR glass, 10 %/min
AR glass, Epoxy 0,4 %/min
2000 AR glass, theor. calc.
1000
0
0,0 0,5 1,0 1,5 2,0 2,5 3,0
Elongation, %
Fig. 6.26. Tensile stress-strain relations of textile rovings at different displacement rates.
The main characters of the carbon and AR glass textiles used in the tests are listed in Table 6.3. The
stress versus strain curves of single uncoated and epoxy impregnated carbon and AR glass rovings that
were directly cut out of the textile fabric and measured in direct tension tests are shown in Fig. 6.26. It
can be seen that the measured uniaxial tensile strength is about 1300 MPa for carbon and 2600 MPa
for epoxy impregnated carbon, respectively. This rather low strength is due to a low static fatigue
limit, imperfections in the material, stiffness of the clamping devices and also the damage caused by
the textile production and is not the true strength of the material itself. An equivalent behaviour can
also be observed from tests of the AR glass textiles.
6.1.3.5 Comparison o f calculations and tests of textile reinforced elements in a pull-out test
There are a number of different effects, e.g. specimen shape or production process, that greatly
influences the bond performance of textile reinforcement in concrete. Consequently, a simple pull out
test seems to be inappropriate for analysing the bond behaviour of textile reinforced and also
prestressed concrete. For that reason it was decided to use a double side pull out test with adequate
confinement as can be seen in Fig. 6.27. For experimental details see [Krü2001a, Krü2002b]. Note that
the shown tests do not represent explicit pull out because the perpendicular rovings of the textile are
fixed with epoxy resin to the treated roving. However, this influence becomes mainly effective at a
higher slip value in the post peak area [Krü2001a].
As shown in Table 6.3, the main bond parameters were set to almost constant values for all
calculations. Note that the model is not calibrated at a large data set up to now so the parameters only
qualitatively show the performance of the model. Additionally, it has to be noted that the
reinforcement is assumed to be linear elastic for all the calculations, i.e. ΩS = 1, and no shrinkage is
considered.
95
L1 0.4
C1
G1 0.375
0.35
0.325
Fig. 6.27. Cross section of finite element mesh used for the calculation of pull out tests (with
exemplary volumetric stress).
As shown in the last chapter the bond performance of the model is influenced by ΩC which accounts
for: (i) the influence of radial stress obtained from the surrounding concrete elements and (ii) the
influence of reinforcement strain. To test the effects of these different influences numerical studies
have been carried out with an incremental implementation of these transverse reactions into the bond
model. Due to space limitations only the results obtained by the use of the complete bond model are
shown.
Table 6.3. Input data of material properties used in the numerical studies.
Carbon AR glass
Material properties epoxy impr. epoxy impr. Concrete
CE CE-P AR AR-P
Prestress N/roving 0 625 0 300
Young’s modulus GPa 240 72 39.0
Compressive strength MPa - - 90.0
Tensile strength MPa 2600 1450 4.7
Mechanical bond strength τm MPa 10.1 10.1
Frictional bond strength τf MPa 5.0 5.0
Shear stiffness ksec MPa/mm 159 90
Initial stiffness k1 MPa/mm 500 500
Interlock stiffness k2 MPa/mm 48 27
Radius of the curvature R - 2.5 2.5
Slip S2* mm 0.03 0.03
Slip S3 mm 0.75 2.0
Reinforcement area mm² 0.93 0.95
Perimeter mm 3.418 4.6
Roughness hR mm 0.02 0.02
96
1400
Load, N
1400
Load, N
calculation
calulation
1200 1200 calculation, prestressed
calulation, prestressed
test result
test result
test result, prestressed
1000 test result, prestressed 1000
800 800
600 600
400 400
200 200
0 0
0,0 0,5 1,0 1,5 2,0 2,5 0,0 0,5 1,0 1,5 2,0 2,5
Slip, mm Slip, mm
Fig. 6.28. Calculated pullout load versus slip curves compared to exemplary test results (Carbon
left, AR glass right).
In Fig. 6.28 the results of the pull out stress versus slip are plotted. The input parameters were
calibrated mainly to fit the test results with no prestressing. As can be seen the calculations show good
agreement with the test results. For the analysis of the prestressed reinforced specimen no change in
the input parameters was made, therefore, the results of the numerical simulations show the influence
of ΩC that leads to a higher peak load.
A few numerical studies were carried out in which ΩC was set to constant (i.e. ΩC = 1). These
calculations showed almost the same peak load for prestressed and non prestressed state, however, the
slip was less for the prestressed state.
97
CE (CARBON)
1600
Tensile stress, MPa
AR (AR glass)
1600
Tensile stress, MPa
15 simulated crack
Bond stress, MPa
Fig. 6.29. Calculated tensile stress and bond stress along the embedment depth.
Fig. 6.29 shows the influence of ΩC on the bond stress for unprestressed and prestressed
reinforcement. As can be seen the curves are typical for the bond performance of reinforcement. In the
calculated bond stress slip curves for prestressed reinforcements, shown in Fig. 6.29, the influence of
the reinforcement strain and of the radial stress of the concrete on the bond performance is also
obvious. Due to the prestressing of adhesive type, at initial state the reinforcement lateral strains are
negative (compression). Consequently, the contraction of reinforcement is less than in the case of
unprestressed reinforcement and therefore it does not result in such a large reduction of bond stresses,
or in other words, it leads to a higher peak load.
Another effect can be observed from the numerical simulation shown in Fig. 6.29. This is the effect of
loading-unloading on the bond response. In the proposed bond model, up to the slip of s = s1 bond
performance is assumed to be elastic in case of unloading. If s becomes greater than s1 the descending
branch with the stiffness kunload and resultant residual friction is taken into account for unloading. For
the given input parameters shown in Table 6.3, s1 can be calculated as 0.167 mm for AR glass.
98
Therefore this effect can be clearly seen for the bond stress of AR glass in the post peak area (i.e.
slip = 1.00 mm or slip = 2.14 mm) in Fig. 6.29. Here the bond stress is negative at embedment depth
of about 80 mm, which means that the elastic bond is destroyed.
The benefit of the presented bond model becomes obvious if one considers different types of
reinforcements, e.g. different diameter, surface structures or steel lugs and stress strain properties. It is
a useful tool for study different influences of diverse reinforcements on the bond performance and
consequently on the structural performance of miscellaneous reinforced concrete structures. It is
assumed that the general parameters of the bond stress slip-relation shown in Fig. 6.22 mainly depend
on the concrete parameters and can be set to constant values for a group of reinforcement elements of
the same type. This can be for example a set of steel bars of different diameter or textile reinforcement
type with different Young’s modulus but almost the same surface roughness.
6.1.4.1 Introduction
Fabrics are often viewed as a means for incorporating long and aligned fibers (yarns) in the composite,
and the mechanics of the composite is simply treated in terms of a composite reinforced with
continuous fibers. Thus, the main parameters considered are the bulk properties of the yarn and the
bond of single yarns to the matrix. In a range of studies carried out in recent years it was demonstrated
that the geometry of the fabric can have a significant influence on the nature of bonding, which
sometimes outweighs that of other parameters. These influences are of special nature when the
cementitious matrix is considered. The present document provides a summary of the main influences
characterised in these studies, as well as a compilation of the reports published [Ben1997, Pel1997,
Pel1998a, Pel1998b, Pel1998c, Pel2000a, Pel2003a, Pel2002a, Pel2002b, Pel1994, Pel1995,
Pel2003b, Ben1999, Ben2000].
99
Table 6.4. Bond characteristics of a single straight yarn and yarn in the fabric
Yarn Type Modulus of Number of Bond per unit external bundle Bond of yarn in a fabric
Elasticity filaments in a surface relative to the bond of a
bundle straight yarn**
%
(MPa) (MPa)
Single Straight Fabric
Yarn
7 yarns/cm*
PE (woven) 1760 1 0.17 1.2 700%
5 yarns/cm* 0.73 430%
PP (knit) 6900 100 3.5 2.8 80%
HDPE (knit) 55000 900 11.5 1.8 15%
* Density of yarns perpendicular to reinforcing direction
Note that the knit fabrics are weft insertion type
** calculated per single filament
Weft
Fill Yarns
Varied Density
5, 7, 10 yarns/cm Warp
Warp Yarns
Amplitude
Wave length
Pullout direction
Fig. 6.30. The different fabrics: (a) woven fabric (plain weave), and (b) weft insertion knit fabric.
An additional parameter evaluated in this series of studies was the density of the woven fabrics. The
density of the fabrics was varied by changing the density of the yarns perpendicular to the applied load
(fills): 5, 7 or 10 fills per cm; the warps’ density was kept constant at 22 warps per cm. It should be
noted that the fabric density affects several characteristics: the crimped structure of the warp yarns in
the woven fabric, the number of joints and the penetrability of the cementitious matrix into the fabric.
These characteristics can influence the bond capacity with the cement matrix.
In order to better understand the influences of the yarn itself, individual straight yarns and crimped
yarns untied from the woven fabrics were used to prepare reinforced cement composites.
100
6.1.4.3 Testing
The evaluation of the composite bond characteristics was based on pull-out tests of the original
straight yarn, yarn untied from the fabric and the fabric itself. The pull-out tests were carried out in an
Instron testing machine at a crosshead rate of 15 mm/min. Schematic description of the test set-up is
presented in Fig. 6.31.
Pulled Pulled
15 mm
Yarn Yarn
Embedded
length
Fabric
10 mm
Matrix
Load-slip curves were recorded. In both cases, fabrics and straight yarn, the whole bundle of filaments
was pulled out. The test results are an average of at least 5 specimens. The average shear bond strength
between the yarn and the cement matrix was calculated, assuming the yarn to be a single reinforcing
unit with a diameter equal to that of the whole bundle. The implied assumption here is that there is no
penetration of the cement matrix in between the filaments of the bundle.
The calculated average bond strength values of a single yarn and a yarn in fabric are presented in
Table 6.4. The trends are different in both cases for single yarns and for the fabrics [Pel1998c,
Pel2002a, Pel2003a, Pel2002b, Pel1994, Pel1995, Ben1999, Ben2000, Pel2003b]. For the single
yarns, although the surfaces of the different yarns tended to be similar and smooth, there were marked
differences in the bond values, 11.5 MPa for the HDPE compares with only 3.5 MPa and 0.17 MPa of
the PP and PE, respectively. This behaviour in bond of the single yarns can be attributed to the
influence of the bulk mechanical properties of the fibers, with the yarns with higher modulus
exhibiting higher bond strength (Fig. 6.32).
101
14
10
0
0 20000 40000 60000
Modulus of Elasticity, MPa
Fig. 6.32. Effect of modulus elasticity of the yarn on its bond strength with cement matrix.
This type of behaviour is consistent with mechanism such as clamping and Poisson effects which have
been suggested to account for a variety of bonding effects. On the other hand, in the actual fabric, no
clear correlation between bond and modulus of elasticity could be established (Table 6.4). The bond
strength of the high modulus HDPE fabric is relatively low, only 1.8 MPa compared with the bond
strength of the low modulus PP fabric, 2.8 MPa.
These differences can be attributed to several geometrical characteristics: the crimped shape in the
fabric form which tends to increase bond, and the multi-filament nature in some of the fabrics which
tends to decrease bond due to the limited penetration of the matrix [Pel1998a, Pel1998b, Pel2000a,
Pel2003a]. These characteristics can account for the increase in bond in the woven fabric (430 to
700% in Table 6.4) and for the reduction in the knitted fabric, where the yarns remain straight but in a
multi-filament geometry (80% and 15% in Table 6.4).
For a better understanding of the influence of the crimp geometry of the individual yarn in the woven
fabric on the bond with cement matrix, the pullout behaviour of different crimped yarns untied from
the different woven fabrics was studied [Ben1997, Pel2000a, Pel2003a, Pel2002a, Pel2002b, Pel1994,
Pel1995, Ben1999, Ben2000, Pel2003b]. These crimped yarns had different crimp densities, similar to
the density of the woven fabrics. When considering the effects of the crimp shape of the yarns, two
parameters should be taken into account: the wave amplitude and the wave length (Fig. 6.30 a).
Fig. 6.33 a shows the pullout behaviour of two crimped yarns with different amplitudes, having similar
wave length. The maximum pullout resistance is greater for the crimped yarn with the larger
amplitude, i.e., for the yarn with the 0.13 mm amplitude. Such improvement indicates that the wave
amplitude has a major contribution to the bonding.
A linear correlation could be established when plotting the pullout resistance per one wave versus the
wave amplitude of the different crimped yarns (Fig. 6.33 b) [Ben1997, Pel1998a, Pel2003a,
Pel2003b].
102
(a)
1.5
Maximum Pullout Load per One
1.2
0.9
Wave, N
(b)
0.6
0.3
0.6 N 0.1 N
0
0 0.03 0.06 0.09 0.12 0.15
Wave Amplitude, mm
Fig. 6.33. Effects of wave amplitude on the pullout resistance of untied crimped yarns: (a) load-
displacement response of crimped yarns with different wave amplitude, (b) maximum
pullout load per one wave verses wave amplitude of different crimped yarns.
The different amplitudes were achieved by the different crimps density of the yarns and by applying
pretension on the crimped yarn during the production of the pullout specimens (0.1N and 0.6N). The
linear regression obtained from the data in Fig. 6.33 b is:
Where: P = the pullout resistance per one wave (N), Amp = the crimped yarn amplitude (mm),
0.18 = constant which reflects the contribution of the straight yarn per unit length. The linear
correlation coefficient of this relation is 0.88.
Based on this relation the maximum pullout resistance, Pmax, per unit length can be calculated by
multiple the pullout resistance per one wave in Eq. (12) by the number of waves per unit length:
le le
Pmax = p = (8 * Amp + 0.18) (13)
λ λ
103
When Pmax = the total pullout resistance (N), le = embedded length (mm), λ = wave length (mm).
The maximum pullout resistance, Pmax, was calculated by substituting the geometrical values of the
different crimped yarns into Eq. 13. Comparison of the calculated and the experimental values is
presented in Fig. 6.34 a, showing an excellent agreement for both embedded lengths, 10 mm and 20
mm.
The same empirical approach can be applied to calculate the contribution of the crimped yarns to the
overall bonding of the woven fabric, substituting in the equations the observed geometrical
characteristics of the yarns [Pel1998c, Pel2003a, Pel2003b]. The results of such a calculation are
shown in Fig. 6.34 b. The fact that the calculated values are only slightly lower than the experimental
results, suggests that most of the pullout resistance in the woven fabric can be attributed to the crimped
geometry of its individual yarns. Only a small contribution can be related to the fabric structure, i.e.
perhaps some anchoring offered by the perpendicular yarns [Pel1997, Pel1998c, Pel2003a].
14
12 (a)
Maximum Pullout Load, N
20 mm
10
6 10 mm
2 Experimental Calculated
0
0 5 10 15 20
Crimps Density, 1/cm
15
Experimental Calculated
Maximum Pullout Load, mm
Contribution of
12 fabric structure
(b)
9
Contribution of
Straight yarn crimped geometry
6 of the yarn
Straight
3
0
0 2 4 6 8 10 12
Fills Density, 1/cm
Fig. 6.34. Comparison between the calculated maximum pullout resistance values and the
experimental results of: (a) untied crimped yarns and (b) woven fabrics.
6.1.4.5 Conclusions
Different trends were found when comparing the bond strength values of a single yarn and a yarn in
fabric. A straight forward correlation was found between bond strengths and the modulus elasticity of
the single yarns, with the yarns having higher modulus exhibiting higher bond strength. No clear
correlation between bond and modulus of elasticity could be established with the fabrics.
104
In the case of fabrics other parameters should also be considered such as the geometry of the fabric. It
was shown that the woven fabric structure significantly improves the bonding with cement matrix,
compared to its straight yarn which is not in a fabric form. The crimped geometry of the individual
yarn within the woven fabric was found the have the major contribution to its bonding, due to a strong
anchoring effect.
Pulled out
Fabric Free length
Embedded
Free Fabric Length length
Cement
matrix
Matrix+Fabric
(b)
(a)
No significant influence of the production process on pullout resistance and bond strength was
observed with the glass composites (Fig. 6.36). However, the pullout resistance of the pultruded
systems (impregnated fabrics) was much greater than the pullout resistance of the cast systems (non
impregnated fabrics) for both PVA and PP fabrics. The pultrusion process was found to be beneficial
only when fabrics were used by improving the bond between the fabric and the cement matrix,
whereas not such improvement in bonding was developed when only single bundle was used (not in a
105
fabric form). The results obtained in this work suggest better bond of the PP fabric with the cement
matrix. This improved bonding of the PP fabric system was extremely high when comparing the two
pultruded systems, PP and glass.
This improvement was explained based on the following: when the PP fabric was passed through the
cement chamber during the pultrusion process, the impregnation process helped fill the spaces
between the filaments of the bundled yarns made up the fabrics as well as the loops of the stitches with
the cement matrix, leading to improve mechanical anchoring and bonding. On the other hand, the
bundles in glass fabric were impregnated with epoxy prior to fabric production and the yarns at the
junction points were glued together, these can prevent matrix penetration in between the filaments
even when the pultrusion process was used. Therefore, the pultrusion process did not show a
significant influence on the mechanical performance of the glass composite.
300 300
Pultrusion
Pultrusion Cast (b) PVA
250 250
Cast L=7.6 mm
Pullout Load, N
Pullout Load, N
200 200
150
(a) Glass 150
L=12.7 mm
100 100
50 50
0 0
0 3 6 9 12 0 3 6 9 12
Slip, mm Slip, mm
300
Pultrusion
250
Cast
Pullout Load, N
200
150
100
(c) PP
50
L=12.7 mm
0
0 3 6 9 12
Slip, mm
Fig. 6.36. Pullout load – slip response of the different fabrics: (a) glass, L = 12.7 mm, (b) PVA
L = 7.6, (c) PP L = 12.7 mm.
106
The grooves of the pulled out yarns after the pullout process for cast and pultruded PP systems were
also observed and presented in Fig. 6.38. A relatively smooth surface without substantial damage for
the cement matrix is observed for the cast system (Fig. 6.38 a), while a rough and damaged matrix at
the fabric matrix interface is observed with the pultruded composite (Fig. 6.38 b). This suggests strong
friction forces during the pullout process of the pultrusion system compared to lower friction forces
developed with the cast composites.
These observations suggest good bonding between the pultruded fabric and the cement matrix as also
observed by the pullout tests, for knitted PP fabric.
(a) (b)
(c) (d)
Fig. 6.37. SEM micrographs of reinforcing yarns in PP fabric embedded in cement of the two
processing methods: a) pultrusion and b) casting.
(a) (b)
Fig. 6.38. The groove of the pulled out PP yarn: a) casting, b) pultrusion.
107
6.1.5.4 Pullout mode ling
A model to predict the pullout response of fabric from cement matrix was developed. This model for
prediction of the pullout response was based on a modified shear lag theory approach. The model for
the pullout a fabric from cement matrix is more complicated than that of single straight yarn due to the
restraint offered by the yarns perpendicular to load direction and the nature of the connection of the
yarns in the fabric. Fig. 6.39 presents a schematic model of the anchorage provided by the yarns
perpendicular to load direction. It is expected that the mode of failure in these composites is governed
by a sequential stages of debonding and failure of anchoring points as the cracks propagate from each
junction of the connecting yarns to the next (Fig. 6.39).
Debonded
Matrix fabric
During debonding the following sequence of events are expected to take place as the load is increased:
a) debonding along the length of the loaded yarn until the next junction of the yarns in the fabric is
reached, b) support offered by the yarn perpendicular to the load direction in carrying the load through
bending, and c) failure of the joint of the fabric, followed by extension of the debonding to the next
perpendicular yarn.
The model developed in this study was considered the yarns perpendicular to loading as a beam on
elastic foundation subjected to a concentrated load. The stiffness of the yarns was calculated from
those calculations and used as spring elements carrying load only at the junctions (Fig. 6.39). Various
fabrics were experimentally studied and correlated with this model. Fig. 6.40 presents the comparison
of pullout response of the calculated and experiment data.
108
200 80
(b) PP (c) PE
70 Expt.
Expt.
Math. Yarn
150 Math. Yarn 60 Math. Woven
Math. Woven
50
Load (N)
Load (N)
100 40
30
50 20
10
0 0
0 2 4 6 8 10 12 0 4 8 12 16 20 24
Deformation (mm) Deformation (mm)
REFERENCES
[Bal1991] Balazs, G.L.: Fatigue of bond. ACI Materials Journal 88 (1991), pp. 620-629.
[Bar1982] Bartos, P. 1982. Bond in Glass Reinforced Cements. In: Proceedings of the International
Conference on Bond in Concrete, Paisley College of Technology, Scotland, 60-72.
[Bar1987] Bartos, P.: Brittle-Matrix Composites Reinforced with Bundles of Fibres. London :
Chapman and Hall, 1987. - In: Proceedings of the First International RILEM Congress
from Materials Science to Construction Materials Engineering. Vol. 2: Combining
Materials: Design, Production and Properties, S. 539-554.
[Bas2001] Basche1, H. et al.; Carbon Fibre as an Alternative Reinforcing Element In: Leipzig
Annual Civl Engineering Report 2001, pp.33-38.
[Baz1994] Bazant, Z.P. ; Desmorat, R.: Size Effect in Fiber or Bar Pull-out with Interface Softening
Slip. In: Journal of Engineering Mechanics 120 (1994), Nr. 9, S. 1945-1963.
109
[Ben1988] Bentur, A.: Interfaces in Fibre Reinforced Cements. Pittsburgh : Materials Research
Society, 1988. - In: Bonding in Cementious Composites. Boston, 1987 Materials
Research Society Proceedings Vol. 114 (1988), S. 133-144.
[Ben1997] Bentur, A. Peled, D. Yankelevsky, 1997, “Enhanced Bonding of Low Modulus Polymer
Fibers-Cement Matrix by Means of Crimped Geometry”, Cement Concrete Research
Journal, Vol. 27, No. 7, pp. 1099-1111.
[Ben1999] Bentur and A. Peled, 1999, “Cementitious Composites Reinforced With Textile Fabrics”,
Workshop on High Performance Fiber Reinforced Cement Composites (RILEM)
HPFRCC 3, (Eds. H.W. Reinhardt and A.E. Naaman), Mainz, 16-19 May, pp. 31-40.
[Ben2000] Bentur and A. Peled, 2000, “Optimization of The Structure and Properties of Low
Modulus Polymer Fabrics for Cement Reinforcement” Proceeding of the Sixth
International Symposium on Brittle Matrix Composites (BMC6), Eds. A.M. Brandt, V.C.
Li, and I.H. Marshall, Warsaw, Poland 9-11 October, pp. 430-438.
[Bey1997] Beyerlein, I.J. and Phoenix, S.L. (1997): “Stress profiles and energy release rates around
fiber breaks in a lamina with propagating zone of matrix yielding and debonding” In:
Composites Science and Technology 57, pp. 869-885.
[Bra2000] Brameshuber, W., Banholzer, B., Brümmer, G.: Ansatz für eine vereinfachte Auswertung
von Faser-Ausziehversuchen. Beton- und Stahlbetonbau 95 (2000), Heft 12, pp. 702-706.
[CEB1996] CEB Bulletin 230 (1996): RC elements under cyclic loading. State of the art report, ed.
by T. Telford, Thomas Telford Service Ltd, London, 1996.
[Cur2003] Curbach, M.; Jesse, F.: Festigkeit von Textilbewehrtem Beton mit Bewehrung aus AR-
Glas Filamentgarnen. Dresden : Lehrstuhl für Massivbau, 2003. - In: Textile Reinforced
Structures, Proceedings of the 2nd Colloquium, Dresden, 29.9.2003-1.10.2003, (Curbach,
M. (Ed.)), S. 299-312.
[Eli1983] Eligehausen, R., Popov, E.P., and Bertero, V.V.: Local Bond Stress-Slip Relationships of
Deformed Bars under Generalized Excitations. Report UCB/EERC-83/23. Berkeley:
EERC, University of California, 1983.
[Krü2001a] Krüger, M., Reinhardt, H.-W., Fichtlscherer, M.: Bond behaviour of textile reinforcement
in reinforced and prestressed concrete. In: Otto-Graf-Journal. Vol. 12 2001, pp. 33-50.
[Krü2001b] Reinhardt, H.-W., Krüger, M.: Vorgespannte dünne Platten aus Textilbeton. Proc.
„Textilbeton – 1. Fachkolloquium der Sonderforschungsbereiche 528 und 532“, ed. by J.
Hegger, Aachen, 2001, pp. 165-174.
[Krü2002a] Krüger, M., Ožbolt, J., Reinhardt, H.-W. : A discrete bond model for 3D analysis of
textile reinforced and prestressed concrete elements. In: Otto-Graf-Journal. Vol. 13 2002,
pp. 111-128.
110
[Krü2002b] Krüger, M., Xu, S., Reinhardt, H.-W., Ožbolt, J.: Experimental and numerical studies on
bond properties between high performance fine grain concrete and carbon textile using
pull out tests. In: Beiträge aus der Befestigungstechnik und dem Stahlbetonbau
(Festschrift zum 60. Geburtstag von Prof. Dr.-Ing. R. Eligehausen), Stuttgart, 2002, pp.
151-164.
[Krü2003] Krüger, M., Ozbolt, J., Reinhardt, H.W.: A new 3D discrete bond model to study the
influence of bond on the structural performance of thin reinforced and prestressed
concrete plates. In A.E. Naaman and H.W. Reinhardt (Eds.) „High Performance Fiber
Reinforced Cement Composites (HPFRCC4)“, RILEM 2003, pp 49-63.
[Law1986] Laws, V. et al. 1986. The Glass Fiber/Cement Bond. In: Journal of Materials Science 21,
289-296.
[Leu2000] Leung, C.K.Y.: Interfacial changes of optical fibers in the cementitious environment In:
Journal of Materials Science Volume 35, Issue 24, Dec 2000 Pages: 6197-6208.
[Low2002] Lowes, L.N., Moehle, J.P., Govindjee, S.: A concrete-steel bond model for use in finite
element modelling of reinforced concrete structures. In: Finite Element Analysis of
Reinforced Concrete Structures. ACI Special Publication SP205, 2002, pp. 251-272.
[Maj1974] Majumdar, A.J.: The Role of the Interface in Glass Fibre Reinforced Cement. In: Cement
and Concrete Research 4 (1974), Nr. 2, S. 247-266.
[Mal1992] Malvar, L.J.: Bond reinforcement under controlled confinement. ACI Materials Journal
89 (1992), pp. 711-721.
[Men1973] Menegotto, M., Pinto, P.: Method of analysis of cyclically loaded reinforced concrete
plane frames including changes in geometry and nonelastic behaviour of elements under
combined normal geometry and nonelastic behaviour of elements under combined normal
force and bending. Proceedings of the IABSE Symposium on the resistance and ultimate
deformability of structures acted on by well-defined repeated loads, Lisbon, 1973.
[Nam1989] Nammur, G., Naaman, A.: Bond Stress Model for Fiber Reinforced Concrete Based on
Bond Stress-Slip Relationship. ACI Materials Journal 86 (1989), pp. 45-55.
[Ohn1994] Ohno, S., Hannant, D.J.: Modelling the stress-strain Response of Continuous Fibre
Reinforced Cement Composites. ACI Materials Journal 91 (1994), pp. 306-312.
[Ozb2001] Ožbolt, J., Li Y., Kožar, I.: Microplane model for concrete with relaxed kinematic
constraint. Int. Journal of Solids and Structures 38 (2001), pp. 2683-2711.
[Ozb2002] Ožbolt, J., Lettow, S., Kožar, I.: Discrete bond element for 3D FE analysis of reinforced
concrete structures. In: Beiträge aus der Befestigungstechnik und dem Stahlbetonbau
(Festschrift zum 60. Geburtstag von Prof. Dr.-Ing. R. Eligehausen), Stuttgart, 2002, pp.
239-258.
[Pel1994] Peled, D. Yankelevsky and A. Bentur, 1994, "Bonding and Interfacial Microstucture in
Cementious Material Reinforced by Woven Fabric", Microstruture of Cement-Based
Systems, and Bonding and Interfaces in Cementitious Materials (Eds. S. Diamond et al.),
Material Research Society, Boston.
111
[Pel1995] Peled, D. Yankelevsky and A. Bentur, 1995, "Influences of Yarns Shape in Woven Fabric
on Bonding Performance of Cementitious Composites" Non-Metallic (FRP)
Reinforcement for Concrete Structures, Proceedings of the Second International RILEM
Symposium (FRPRCS-2) (Ed. L. Taerwe), E & FN Spon, Ghent, pp. 192-199.
[Pel1998a] Peled, and A. Bentur, 1998, “Reinforcement of Cememtitious Matrices by Warp Knitted
Fabrics”, Materials and Structures (RILEM) Journal, Vol. 31, October, pp. 543-550.
[Pel1998c] Peled, A. Bentur, D. Yankelevsky, 1998, “Effect of Woven Fabrics Geometry on The
Bonding Performance of Cementitious Composites: Mechanical Performance”, Advanced
Cement Based Materials Journal, Vol. 7, No. 1, pp. 20-27.
[Pel1998d] Peled, A. ; Bentur, A. ; Yankelevsky, D.: Effects of Woven Fabric Geometry on the
Bonding Performance of Cementitious Composites. In: Advn. Cem. Bas. Mat. (1998), Nr.
7, S. 20-27.
[Pel2000a] Peled, A. ; Bentur, A.: Geometrical Characteristics and Efficiency of Textile Fabrics for
Reinforcing Cement Composites. In: Cement and Concrete Research 30 (2000), Nr. 5, S.
781-790.
[Pel2000b] Peled and A. Bentur, 2000, “Geometrical Characteristics and Efficiency of Textile
Fabrics for Reinforcing Composites”, Cement and Concrete Research, Vol. 30, pp. 781-
790.
[Pel2002a] Peled and A. Bentur, “Quantitative Description of The Pull-out Behavior of Crimped
Yarns from Cement Matrix”, Journal of Materials and Civil Engineering, ASCE,
Accepted October 2002.
[Pel2002b] Peled, A. Bentur, 2002, “Textile Fabrics for Cement Composites”, ACI SP-206,
Concrete: Material Science to Application, A Tribute to Surendra P. Shah, Eds, P.
Balaguru, A. Naaman and W. Weiss, pp. 323-340.
[Pel2003a] Peled. A. Bentur, 2003, “Fabric Structure and Its Reinforcing Efficiency in Textile
Reinforced Cement Composites”, Composites, Part A 34, pp. 107-118.
[Pel2003b] Peled and A. Bentur, “Mechanisms of Fabric Reinforcement of Cement Matrices: Effect
of Fabric Geometry and Yarn Properties” 2nd Colloqium on Textile Reinforced
Structures (CTRS2), Dresden Germany, September 29 -October 1, 2003.
[Rei1995] Reinhardt, H. W., Balazs, G. L.: Steel-concrete interfaces: experimental aspects. In:
Selvadurai, A. P. S., Boulon, M. J. (eds.): Mechanics of geometrical interfaces.
Amsterdam: Elsevier, 1995. (Studies in applied mechanics, 42). - ISBN 0-444-81583-X,
S. 255-279.
112
[Rei2001] Reinhardt, H.-W., Krüger, M.: Vorgespannte dünne Platten aus Textilbeton, Proc.
“Textilbeton – 1. Fachkolloquium der Sonderforschungsbereiche 528 und 532”, Ed. By J.
Hegger, Aachen, 2001, pp. 165-174.
[Rei2002] Reinhardt, H.-W., Krüger, M., Grosse, C.U.: Thin plates prestressed with textile
reinforcement. Concrete: Materials Science to Application. A Tribute to Surendra P.
Shah. USA, Farmington Hills, Michigan: ACI SP-206, 2002, pp.355-372.
[RIL1970] Bond test for reinforcing steel, Pull-Out Test. RILEM-Bulletin 1970, Vol. 15, S.175-178
[Sch2004] Schorn, H.: An adhesive cross-linkage model for textile reinforcement in concrete. 2nd Int.
Conference on Structural Engineering, Mechanics and Computation SEMC, Cape Town,
July 2004, pp 1563 – 1567 (CD ROM).
[Som1981] Somayaji, S. ; Shah, S.P.: Bond stress versus slip relationship and cracking response of
tension members. In: ACI Materials Journal 78 (1981), Nr. 1, S. 217-225.
[Zas2003] Zastrau, B. ; Richter, M. ; Lepenies, I.: On the analytical solution of pullout phenomena
in textile reinforced concrete. In: Journal of Engineering Materials and Technology,
Vol. 125, 2003, S. 38-43.
113
6.2 Mechanical Behaviour of Textile Reinforced Concrete
Josef Hegger, Norbert Will, Institute of Structural Concrete, Aachen University, Germany
6.2.1 Introduction
Textile reinforced concrete (TRC) is a new composite material which is reinforced by technical
textiles made out of yarns/rovings. The development of TRC is based on the fundamentals of glass
filament reinforced concrete with short filaments (Fig. 6.41). Similar to ordinary reinforced concrete
structures the filaments are aligned in the direction of the tensile stresses which leads to an increase in
their effectiveness. The load-carrying capacity increases compared to short filament concrete. This
reduces the costs of the still expensive high performance filaments.
The reinforcement of concrete with technical textiles extends its application to completely new fields.
Because of the corrosion resistance of the textile materials, thick concrete covers as known in ordinary
reinforced concrete are no longer needed [Heg2001, Cur2003]. Thus, slender structural members with
a wall thickness of 10 mm are possible. In addition, fine grain concrete matrices guarantee an even and
sharp-edged high quality surface, so that TRC is predestinated for architectural applications. Although
the knowledge about the load bearing behavior of TRC is still limited, applications such as cladding
panels [Heg04 – reference is missing] and integrated framework systems [Bra2003] have already been
implemented.
The used materials and the possible production methods of textile-reinforced concrete are presented in
other contributions of this session. The calculation of elements and structures demands detailed
knowledge of the load-bearing behaviour of the composite material. Indeed this resembles that of steel
reinforced concrete, however, it is influenced more by the bond of the textile reinforcement with the
fine concrete. Therefore, near the material behaviour the cracking behaviour, the load-bearing
capacity, the deformation behaviour and the durability are examined. The calculation models currently
being developed will lead to a better understanding of the failure mechanisms of TRC-structures.
Based on these models, design rules and safety concepts can be set up in the future.
114
6.2.2 Specific features in the load bearing behaviour
Neither the load-bearing performance nor the deformation behaviour has been investigated in detail
yet. Cracks will appear systematically in textile-reinforced concrete because of the low tensile strength
of the concrete. The knowledge of this cracking process is of crucial importance in calculating the
load-bearing capacity, the deformation behaviour and the limiting values in designing the
serviceability. The load deformation behaviour of textile-reinforced concrete under uniaxially loading
(Fig. 6.42) corresponds basically to the reinforced concrete.
textile
single
crack
F F
strain [‰]
The stiffness of the uncracked composite material (state I) corresponds nearly to the E-modules of the
fine concrete. Exceeding the concrete tensile strength the state of first cracking is reached and whole
tension force in the crack is carried by the reinforcement. With an increasing of the tension force
additional cracks occur (cracking formation phase - state IIa). Due to the bond between filaments and
concrete forces are initiated in the concrete, until the tensile strength of the concrete is reached once
more. The cracking distance and the crack width are determined by the reinforcement and their bond
charateristics between reinforcement and concrete. The stress-strain curve shows during the cracking
formation phase a very low increase. In the state of stabilized crack pattern (state IIb) no further cracks
occur. By a load increase the filaments are strained up to the strength of the filaments. The stress-
strain curve proceeds parallel to the naked textile. The difference is called tension stiffening effect.
A ductile deformation area (state III) does not appear in the tensile tests with textile-reinforced
concrete, because the used materials (AR-glass, Carbon) show no ductile capacity. A virtually ductile
capacity which can be observed in tests can be explained by the fact that failure is primarily
determined by a pull-out failure of the textile reinforcement of the concrete and not by the tensile
strength capacity. Thus the postbreak behaviour shows no increase of the load-carrying capacity.
Cracking is determined not only by the stress but also by the bonding action between the textile rein-
forcement and the concrete matrix. Modelling the load-bearing behaviour shall differentiate between
the microscopic bonding behaviour which describes the physical interaction between the raw materials
used (textiles and concrete) discretely and in detail, and the macroscopic bonding behaviour to
describe the influence on the load-bearing performance of whole members.
The present investigations [Heg1998, Heg2001, Cur2003] showed that the tensile strength of the
filaments in the composite material “textile reinforced concrete” can differently than a concrete steel
reinforcement not be fully exploited (Fig. 6.43).
115
Bond Behaviour
Load Bearing Capacity
Slashing Product
Concrete Matrix
Roving / Yarn
Fibre Slope
Composite
Geometry
Filament
Coating
Textile
Fig. 6.43. Comparison of the tensile strength of a fiber bundle (roving) with the tensile load bearing
capacity of a textile-concrete-composite section [Heg01].
Damages of the filament material during the textile manufacturing processes, the bond characteristics
of the filament bundles in the concrete, the filament adjustment and the surface finish of the fibers can
be reasons for this. Thus – deviating from steel reinforced concrete – the bond behaviour of the
filament bundles (rovings) has a determining influence on the load-bearing capacity of the composite
material. Because the rovings form no homogeneous cross section, a direct analogy is not possible to
the bond behaviour of steel in concrete.
6.2.3.1 Testing
ToDo: Aachen University, Dresden University, Prof. Reinhardt
116
Fig. 6.44. Sleeve and core filaments of a roving CemFil 2400.
The textiles consist of yarns/rovings, which are made up in turn of long fibers, called filaments in
following. In principle a distinction between two groups of filaments in a multifilamentyarn/matrix
system has to be made.– sleeve filaments in direct contact with the concrete matrix consist of a better
bond performance than core filaments which are stressed only indirectly by bonding because of the
protection from the concrete by sleeve filaments (Fig. 6.45). So additionally to the bond behaviour of
the reinforcement in the cement stone matrix also the bond between the filaments plays an important
role.
crack
Strain
strain of core
filaments
strain of matrix
After cracking, the tensile stress within the cracked cross-section is carried entirely by the filaments.
The distribution in the strain differs because of the difference in bonding behaviour between sleeve
and core filaments. The sleeve filaments with the higher bond stiffness need a shorter transfer length
in comparison with core filaments in order to reach the same elongation at the crack (Fig. 6.45). It is
the aim of the present research to develop models which are able to describe the different bond
characteristics of an technical textile in a fine-concrete. Besides the experimental investigations by
pull-out, tensile and bending tests numerical models are examined.
117
6.2.3.3 Bond models
Ohno and Hannant [Ohn1994] have developed a mechanical model that takes into account the
displacement in two border layers –the filament-matrix interface and the filament-filament interface as
shown in Fig. 6.45.
Based on this easy model some aspects of the load bearing behaviour of textile-reinforced concrete
can be already explained. Using a few crude simplifications such as a rectangle bond stress–slip
relationship in the contact surface matrix-filament or filament-filament stress-strain curves of tensile
elements can be calculated. For this purpose the proportion of the cross section of the core and sleeve
filaments has to be determined as well as the bond strength in tests. Indeed, only a qualitative
statement is possible to the strains of the core and sleeve filaments.
Present experimental investigations show that the load-bearing behaviour of a multifilament yarn is
determined considerably by his bond behaviour. These multifilament-yarn consist from a huge number
of filaments which show about the length as well as about the cross section varying bond
characteristics to the surrounding matrix as well as to neighbouring filaments (Fig. 6.46).
Through this a high scatter in the results occurs between different tests of the same series which
cannot be eliminated. Brameshuber/Banholzer [Ban2004, Bra2001] comment on these problems. To
characterize the bond behaviour of a multifilament yarn in the pull-out test, the authors divided in
different partial mechanisms, e.g., the bond behaviour between filament and matrix (external bond)
[Ban2004a], the bond behaviour the filaments (internal bond) [Bra2001a] as well as the proportion of
core and sleeve filaments [Bra2001b]. Analytic and numerical models are used here. If all parameters
and factors are known including their statistical distribution, a model with sufficient number of free
parameter can be developed to recalculate a pull-out test..
Based on observations in the scanning electron microscope Schorn [Sch2003] introduces the
“Adhesive cross linkage model” as a bond model (Fig. 6.47).
118
!!! Translation!!!
Rißebene
Lo
Lw schubsteife Haftbrücken
Lo
Lw schubweiche Haftbrücken
Lo
Lw
Haftbrücken zwischen
Filamenten
!!! Translation!!!
Fig. 6.47. Adhesive cross linkage model by Schorn [Sch03].
In this model all filaments of the yarn are not continuously, but by means of adhesive cross linkages of
the cement connected together and with the fine concrete. The adhesive cross linkages exist as a rule
of fine hydration elements of the cement. According to their sort and position the filaments dispose of
differently big deformation lengths between two neighbouring situated adhesive cross linkages. If the
first crack occurs in the fine concrete the filaments with the lowest deformation lengths fail at first,
which reach at first the elongation at rupture and therefore the tensile strength. Their load has to be
carried by the remaining filaments which got still reserves because of bigger deformation lengths.
With increasing crack width further filaments fail successively by tensile failure, if they have reached
their elongation at rupture according to her effective deformation length. The load-deformation
behaviour of the yarns crossing the crack can be explained and predicted by the distribution of all
effective deformation lengths about all filaments in the reinforcement cross section.
119
(!!Translation!!)
Druckkraft
σ Beton
textile
Bewehrung
Feinbeton Zugkraft ε
ρl ε s α
Geometrie Filamentgarn
F F mit Beschichtung F Feinbeton 1
innerer/äußerer Hybridgarne
Verbund Feinbeton 2
ohne
Beschichtung
s s s
(!!Translation!!)
Fig. 6.48. Parameters on the load-bearing behaviour of textile-reinforced concrete.
Extensive experimental and theoretical investigations are currently carried out at the RWTH Aachen
University to determine the load bearing behaviour of textile reinforced concrete. Therefore, the effect
of different fiber materials (e. g. alcali resistant glass (AR Glass), carbon and aramide), different fiber
bundle (roving) and fabric geometries, coatings and concrete properties are examined in tension and
bending tests. Another central aspect is the influence of two-dimensional tension states on the load-
carrying capacity of the textile reinforcement. This occurs in many cases, due to the deviation between
the direction of the reinforcement and the loading which appear, e.g., in the web of profiled beams
(Fig. 6.49).
120
biaxial loading
t
c
F
c
t
c
mati
sche
sloped crack
roving
α
The textile reinforcement fabrics for the test series (Table 6.5) made out of AR-glass, carbon or
aramide fibers were designed and manufactured by the Institute of Textile Technology, RWTH
Aachen University.
They differ in the roving thickness and the mesh size. σmax is determined as the average value of 10
tensile tests on 125 mm long parts of rovings taken from the fabric. The loading has been applied with
a deformation rate of 10 mm/min. The values for σmax are only reference values because the test results
consider on the deformation rate and the length of the specimen. The properties of the fine-grained
concrete used for the specimens are given in Table 4.2 and Table 4.4.
Previous tests [Cur1998, Ohn1994] showed that the tensile strength of the embedded fibers in the
composite material textile reinforced concrete cannot be fully exploited. Reasons for this are damage
of the fibers during the textile manufacturing processes, the bond characteristics of the rovings in the
121
concrete, the fiber adjustment and the surface finish of the fibers. For determining the tensile load
bearing capacity of the composite the test specimen shown in Fig. 6.50 are used.
top view
LVDT
50
F LVDT F
100
50
LVDT
65 135 135 65
200 50 400 50 200
900 [mm]
textile reinforcement
side view
LVDT
10
30
load introduction zone measurement range load introduction zone
Strains were measured directly on the specimens using LVDT‘s. The loading was applied with a
constant deformation rate of 1 mm/min. In order to examine the influence of the fiber orientation, in
one test-series the fabrics were turned around 22.5°, 45.0°, 67.5° and 90.0° with respect to the
direction of the tensile stresses. In addition, four-point-bending tests have been carried out to
determine the load bearing capacity of textile reinforced concrete structures under bending loading.
The influence of the reinforcement quantity on the load bearing capacity and on the effectiveness of
the fibers was examined by varying the reinforcement ratio. The specimens geometry and the test set-
up is shown in Fig. 6.51. The tests have been repeated two times, so that each test-series with its
specific material combination and test set-up consists of three tests.
110
5,6 12
Load cell
Textile
reinforcement 18
120
84
Strain gauge
LVDT
12 5,6
LVDT
6.2.4.3 Results
Tests programme SFB 532 (Aachen University, new data available in December 2004))
122
Flexure Performance ─ Peled/Bentur (data available. results will be worked in 01/2005)
Detailed presentations of the investigations are given in [Heg2001, Heg2002, Cur2002, Cur2003]. The
research results show that the load-bearing capacity is mainly influenced by the interaction between
internal and external bond of the yarns (Fig. 6.45). The direct contact of the filaments with the
concrete matrix delivers on the one hand high bond firmness and, on the other hand, is responsible for
the damaging influence of the alkalinity on glass-filaments. It could be shown that basically the
filaments with direct contact with the matrix determine the bond behaviour. Textile fabrics made out
of yarns with a big diameter showed worse load-bearing behaviour compared to yarns with a relatively
small diameter. Hence, with thinner yarns more filaments had directly contact with the fine concrete
matrix. A mutual impact of the rovings by the multi-layer reinforcement arrangement is not derivable from the
test results.
600
Reinforcement stress [MPa]
MAG-01-03
500
MAG-02-03
400
300
200 MAG-03-03
100
0
0,0 1,0 2,0 3,0 4,0 5,0 6,0
Strain [‰]
Fig. 6.52. Stress - strain curve of textile reinforced concrete in tensile tests.
With a roving embedded in the concrete, the filaments, which are in direct contact with the concrete
matrix, transfer higher bond forces than the filaments which are located inside a roving. Thus, fiber
bundles with a large diameter indicate worse load bearing characteristics compared to rovings with
small reinforcement diameters, which have a more favorable cross-section area to perimeter ratio. It is
obvious that the fabric MAG-01-03 consisting of the finest rovings reaches the highest utilization of
123
the fibers. With increasing roving thickness, the ultimate strength of the textile reinforcement
decreases.
The own tests with textiles 1 and 2, which indicate different refinements of the rovings in 0°-direction
with the same fiber reinforcement degree, confirm the context described before (Fig. 6.53). The
effectiveness of the textile reinforcement is increasing with the increasing ratio of perimeter to area. In
fig. 6 the fiber efficiency is confronted to the quotient from perimeter to cross-section area (Ut =
perimeter of a roving).
0,5 textile 1
textile 2
Fctu / At ⋅ σmax
0,4
0,3
0,2
0,1
0,0
1,0 1,5 2,0 2,5 3,0 3,5 4,0
Ut / At [1/mm]
!!New Summarizing graph !!
Fig. 6.53. Influence of cross section geometry rovings on the fiber effectiveness.
It is recognised that the test samples reinforced with textile 1 allow a substantially higher utilisation of
the fibers.
The four-point-bending tests on samples reinforced with MAG-07-03 showed that there is no effect of
the reinforcement ratio, given in Fig. 6.54 as the percentage of reinforcement cross section area to the
concrete area of the profile on the load bearing capacity and on the effectiveness of the fibers.
124
!!New Summarizing graph !!
0,6
Effectiveness k0 [ - ]
0,4
0,2
0
1
0,96 % 1,44 % 1,93 %
Reinforcement ratio
!!New Summarizing graph !!
Fig. 6.54. Fiber effectiveness k0 from bending tests (AR-glass).
The fiber effectiveness k0, shown as a function of the reinforcement ratio, is calculated as:
σ f ,bt
k0 = (1)
σ max
In this equation σf,bt is the maximum failure stress of the roving in the bending test on the textile
reinforced concrete specimen and σmax is the tensile strength of the roving. For different reinforcement
ratios the fiber effectiveness reaches a constant value of about 40 %. The fiber effectiveness may be
improved by a coating. Coating or laminating the textiles leads to the gluing of the filaments. Thereby,
the bond characteristics of the core filaments between each other are significantly improved and the
effectiveness of the fibers can be more than doubled [Mol2001].
The load bearing capacity of sloped rovings is lower than those of rovings aligned in the load
direction. For fabrics having an equal orthogonal reinforcement the results show symmetry to an angle
of 45°. The rate of the loss of load capacity subjected to the fiber orientation is given by the reducing
factor k0,α as the relation between the load-carrying capacity of the test specimen with skew-angular
reinforcement compared to the value by adjustment of the reinforcement in load direction. The results
of the experiments are shown in Fig. 6.55.
125
!!New Summarizing graph !!
1
F
matrix 0,8
k0,α = Fα / F0 [-]
0,6
delamination
g α
damage rovin 0,4
Carbon
0,2
AR-Glas
0 F
transversal
force 0 15 30 45
Textilneigung α [°]
With increasing fiber slope the effectiveness of the fibers decreases to 61% for a fiber orientation of
45°. Reason for the loss of load bearing capacity is that during the cracking process the sloped fibers
are subjected to additional stresses. The change of the direction of the fibers at the crack edge causes
bending stresses and delaminating of the fibers from the matrix as well as a transverse force pushing
the roving against the crack edge. This leads to fiber failure at the sharp crack edge and can cause a
failure at the matrix edge. Tests by [Mas1990, Bar1982] with sloped rovings showed that the fibers are
not pulled out of the matrix even if the bond length of the fibers is very short. In fact, the member’s
failure is always caused by the fracture of the sloped fibers.
The test results show the strong decrease of the load-carrying capacity of the textile reinforcement
with increasing inclination corner. The values of fabrics made out of carbon were reduced stronger
than those of AR-glass. This can be explained with the higher tensile stress of the roving leading to
higher transverse pressures in the crack.
Influence of a coating
A coating or a laminating of textiles leads to a gluing of the filaments. Thereby the bond
characteristics of the core filaments among each other are significantly improved and the effectiveness
of the fibers increases clearly. To investigate this influence in detail bending tests at profiled beams
were executed. The comparison of the series 2 (textile 2, without coating) and 3 (textile 3, with
coating) as shown in Fig. 6.56 clarifies that the utilisation of the reinforcement cross section is
significantly increasing by a laminating.
126
!!New Summarizing graph !!
300
coated
250
uncoatedt
Mctu [kNcm]
200
150
100
50
0
0 1 2 3 4
At/Ac [%]
!!New Summarizing graph !!
Fig. 6.56. Influence of the coating of textiles on the load bearing capacity of textile reinforced
bending specimens.
The failure moments achieved up to 2.5 times the values of the test series with uncoated textiles. The
reinforcement cross section is homogenised by soaking, so that more filaments take part in the load
support. A coating or laminating of textiles seems to be reasonable, in addition to control directly the
bond characteristics of the sleeve and core filaments. Another attempt to improve the bond qualities is
the development of friction-spun yarns whose fiberglass cores are braided by polypropylen.
Additionally a protection effect of the sensitive glass filaments is realized.
The influence of the concrete matrix on the bond behaviour of the textiles and the break load-carrying
capacity is only low. The deformation behaviour of the uncracked TRC is mainly influenced by the E-
modules of the fine concrete. After cracking no significant influence of the concrete has been
observed.
It is well known that the strength of AR-glass decreases by time. However, numerous investigations in
the fiberglass concrete have shown that the loss of strength is limited and controllable for intervals
usual in the civil engineering. The state of art is presented in [Sch2004]. Another quality not to be
neglected of AR-glass is the dependence of the strength on the duration of the load.
127
Influence of production of composite
6.2.5.1 Introduction
Current investigations show that the failure process of TRC, which is observed at the macro level,
cannot accurately be predicted based on the simple models known from ordinary steel reinforced
concrete. This is primarily due to the inhomogeneous internal structure of the fiber strands (rovings)
consisting of hundreds of separate filaments (Fig. 6.57). Furthermore, non-uniform bond conditions in
the longitudinal direction and within the cross-section of the rovings point out that developing a
material model, capable of accurately describing the load bearing behaviour of TRC is a non-trivial
task.
Therefore, within the SFB 532 the investigations are being conducted at different levels. As shown in
Table 6.6, these are the micro-, meso- and macro-levels.
128
Resolution level Components
Micro-level filament
matrix
bond filament-matrix
bond filament-filament
Meso-level subroving (bond of filaments)
matrix
bond subroving-matrix
Macro-level roving / textile
smeared concrete
smeared bond roving-matrix
At each level, the material components have to be defined and modeled by an appropriate Finite
Element model. Corresponding experiments provide the necessary data for calibrating the numerical
models and obtaining the required mechanical properties of the components. The models at each level
may conceptually either be coupled to an adaptive multi-level computation. Or as an alternative, the
“smeared” material parameters used at one level are predefined at the previous smaller level. In the
following, the main results for each level are described.
6.2.5.2 MICRO-LEV EL
The aim of the micro-level investigations is to completely cover the effects determining the behaviour
of single filaments as well as their mutual interaction and their interaction with the concrete matrix. In
addition to the material properties, such as tensile strength and bond-slip relationship, the random
distribution of several parameters has to be considered. Among these are the following [Chu2004]:
- Geometrical position of filaments within the roving cross-section and the change in the longitudinal
direction (waviness)
- Bonding conditions, i.e., to what degree does matrix contact the filament’s perimeter and how does
this vary in the longitudinal direction
5-10 mm
Concrete elements, EA= ∞
epoxy resin
129
6
bondstress [N/mm²]
5 4,8
4
I
3
II III
1 0,5
0
0 0,1 0,2 0,3 0,4
0,01 0,16 slip [mm]
Fig. 6.59. Bondstress-slip diagram for a single filament embedded in fine concrete.
The interaction between filament and matrix has been studied using pullout tests of single filaments
(Fig. 6.58(a)) and the finite element model shown in Fig. 6.58(b). The resulting bond-slip relation
(Fig. 6.59) is in agreement with the one obtained analytically by Banholzer [Ban2001, Cox1952]
based on the shear lag theory. The bond-slip curve can be divided into three parts. The first part is
determined by elastic adhesional bond. Here, the bond layer is capable of taking up stresses of up to
4.8 N/mm² at a very small slip of 0.01 mm. If the value of the slip exceeds 0.01 mm, the adhesional
bond is lost and only friction bond remains which is about four times smaller than the adhesional
bond. In the second part (slip from 0.01 to 0.16 mm) the friction bond degrades due to “smoothing” of
the interface between filament and matrix. In the third part the friction bond reaches a minimum and
remains constant. This behaviour is very different from steel reinforcement, because the maximum
bond-stress can only be activated at a certain value of slip, i.e., there is no ductility. Therefore, the
maximum force picked up by the filament, cannot be calculated just by the maximum bond stress
times the available anchorage length.
6.2.5.3 MESO-LEVE L
The meso-level investigations aim at describing the behaviour of the rovings within the composite
material. Pullout tests, of which several types exist, are the main tools to study the characteristics of
the bond between roving and matrix (Fig. 6.60).
500
Force F [N]
30 mm
epoxy resin 400
300
200
100
0
0,0 0,5 1,0 1,5 2,0
Roving displacement u [mm]
concrete matrix
Fig. 6.60. Roving pullout test and typical load displacement curve.
In contrast to steel reinforcement, the load-displacement curve does not feature a plastic plateau.
Rupture of filaments or debonding over the entire embedment length lead to a rather quick decrease of
130
the pullout force after the peak-load. In order to simulate pullout tests the so-called bond layer model
shown in Fig. 6.61 has been developed.
decreasing
bond quality
Bond quality
Fig. 6.61. Bond layer model. a) idealized roving x-section, b) bond quality distribution functions.
This model considers a section of the roving idealized as layers and takes into account the
deterioration of the bond with increasing distance between filament and matrix. The bond quality for
each layer is defined by a function of the distance from the matrix. Any changes of these proportions
in the longitudinal direction are neglected.
The parameters to be calibrated for the bond layer model are the bond quality distribution function and
the effective tension strength of the filaments. While the bond quality distribution function determines
the post peak gradient, the maximum pullout force depends mainly on the tensile strength of the
filaments. Preliminary calculations showed that the initial slope of the pullout curve is always
overestimated by the model. Only a decrease in the maximum bond performance can reduce the initial
stiffness, however this leads to a simultaneous decrease of the portion of filament fracture, because
more filaments are pulled out completely. Therefore, the variation range of this parameter is limited,
and the initial stiffness in the experiments cannot be reproduced by solely reducing the bond quality.
As a consequence, this reduction can be explained by the existence of an internal free length between
the macroscopic boundary of the matrix and the first contact of the filaments with the matrix inside the
specimen, i.e., the start of the micro-bonding between filament and matrix, which is illustrated in
Fig. 6.62.
matrix
yarn
Fig. 6.62. Internal free length between matrix boundary and begin of microbonding.
Using a cubic bond quality distribution function, a tensile strength of 1350 MPa and a maximum
internal free length of 6 mm in the center of the roving, the simulation results in the pullout curve
shown in Fig. 6.63.
131
600
400
Experiment
300 Simulation
200
100
0
0 0,2 0,4 0,6 0,8
displacement [mm]
6.2.5.4 MACRO-LEV EL
The tensile tests on composite specimens reinforced with rovings and its simulation play an important
role in the theoretical models. Based on the results of the pullout test described above (meso-level), the
target is to explain the macro-response, i.e., the load displacement curve of TRC under tension load.
Pure rovings are used instead of textiles in order to avoid any influences resulting from the textile
production process, the textile joints, and from the lateral reinforcement. The specimen features a very
satisfactory crack distribution (Fig. 6.64) with an average crack distance of about 14 mm and crack
width of 0.09 mm. This is calculated from the elongation of the measurement length neglecting any
elastic deformation of the non-cracked concrete.
The interpretation of the experimental data raises the following question: Why is there a substantial
increase in the stress while cracking occurs? Normally, one would expect, similar to steel reinforced
concrete, that cracking happens at a stress plateau, which is only determined by the tensile strength of
the matrix.
132
In order to investigate this phenomenon a Finite Element analysis was conducted with the simple
model shown in Fig. 6.65 a).
core elements
Fig. 6.65. FE – model at macro-level. a) FE – model with built-in cracks; b) FE-model with core
added.
The model consists of linear elastic truss elements representing the uncracked concrete and of short
crack elements (l = 1 mm). The number of crack elements is chosen according to the number of cracks
occurring in the experiment (35 cracks).
500
Force F [N]
approximation
400
ft 300
200
100
E
0
2 Gf 0,0 0,5 1,0 1,5 2,0
L ft displacement u [mm]
Fig. 6.66. Material laws of crack elements. a) Concrete tension law b) Approximated pullout curve.
For the crack elements a material law is used which can be split into two parts (Fig. 6.66):
(i) Concrete material law with mesh adjusted softening modulus [Jir2002]. The tension strength varies
randomly (average = 5.0 N/mm², standard deviation = 0.25 N/mm²).
(ii) Material law for the reinforcement, whereby the stiffness and the maximum load (12 x 430 N =
5160 N) are taken from the linearly approximated pullout curve.
133
8
load [kN]
7
3 res_wo_core
res_w_core
2
experiment
0
0 2 4 6
strain [‰]
Because the maximum load that can be picked up by the reinforcement is higher than the crack load,
all 35 cracks open up. Finally, one of the cracks localizes and the reinforcement fails. The strain at
failure of 6 ‰ matches the experiment. However, the model is not capable of predicting the linear load
increase with increasing number of cracks.
In order to model this behaviour, it is necessary to introduce a “core” of filaments, which are activated
gradually with increasing crack number and strain, respectively (cf. [Ohn1994]). This delayed
activation is due to the following:
(i) Filaments in the center of the roving cross-section may have a waviness. Before they pick up load,
they have to be stretched out.
(ii) At the crack edge the core filaments do not have direct contact to the matrix. Therefore, their
anchorage length is much longer than of those (sleeve-) filaments taking up the pullout force.
The bond force of the core is transmitted to the matrix not so much through friction bond between
adjacent filaments as through matrix which has penetrated the roving at several points. These
considerations lead to the modified Finite Element model shown in Fig. 6.65 b). Of course, the
introduction of a core, which is activated through bond elements, actually introduces several new
unknown parameters, such as cross-section, stiffness and tension strength of the core. In addition, the
friction law applied has to be defined. A curve-fitting of the experimental data leads to a portion of
35 % core filaments related to the whole roving cross-section and a tension strength of 600 N/mm².
The increase of bond flow is 65 N/mm per 1 mm of slip. The load strain curve now matches the
experimental curve much better (see res_w_core in Fig. 6.67).
It must be kept in mind that the so far obtained material laws may not be the only solution leading to
these results. Further investigations on tension-tests with different lengths and different degrees of
reinforcement must provide additional data for calibrating the model.
134
6.2.6 Design models / mechanical models
6.2.6.2 Tension
Based on the test results, the tensile load bearing capacity Fctu of the textile reinforcement
cross section embedded in concrete may be calculated as:
n
Fctu = ∑ k1i ⋅ k0i ,α ⋅ Ati ⋅ σ max
i
(1)
i =1
6.2.6.3 Bending
The load-bearing capacity of flexure elements depends basically on the load-bearing capacity of the
textile reinforcement if this determines the failure [Heg1998]. With knowledge of the tensile capacity
of the textile reinforcement the bending capacity of beams can be calculated in analogy to the ordinary
steel reinforced concrete. Indeed, arise by the slope of the rovings in the crack additional losses of load
capacity have to be investigated by experiments. Based on the textile elongation at rupture and the
stress-strain curves of the concrete and the textile reinforcement the moment capacity can be
calculated.
135
6.2.6.4 Shear
The shear capacity of textile-reinforced concrete can be calculated according to ordinary steel
reinforced or prestressed concrete by a concrete capacity Vc and a framework capacity Vf:
V = Vc + Vf (1)
With Vc concrete capacity
Vf framework capacity
Besides, the framework capacity is determined by the textile tensile strength, the crack angle and the
cross section area of the shear reinforcement. Detailed information are given in [Heg2003].
Comparison calculations delivered satisfactory results. With the results of the research projects the
load-bearing capacity of textile-reinforced concrete can be predicted under bending, tension or shear
load.
The mechanical qualities of Textile Reinforced Concrete (TRC) are not known yet enoughly. The load
bearing behaviour of TRC is influenced near material, amount and orientation of the textile
reinforcement also by the fine concrete matrix. In particular the bond behaviour of the filament
bundles (rovings) is important. Known models from the steel reinforced concrete cannot be applied
without additional considerations, because of the inhomogeneous cross sections of the rovings which
show basically other bond behaviour than steel reinforcing bars. However, the attempts have shown
that the tensile strength of the textiles cannot be fully exploited in the composite material TRC
(Fig. 6.43). The purpose of the actual 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 only can be handled if different
resolution levels in the experiments as well as in the simulation are considered. In this research the
behaviour of TRC is investigated at three different levels: micro-, meso- and macro-level. It is shown
how results obtained at one level can be used at the next higher level in order to reduce the number of
unknown parameters. A finite element model is presented, and first results show its capability to
simulate the experimental results. However, systematic experimental research has to provide further
information in order to verify and optimize the models at each level. In addition, simulation techniques
have to be developed at the macro-level taking into account the influences of the textile production
process and the specific geometry of textiles.
Near numerous analogies to the ordinary steel reinforced concrete TRC shows specific differences in
the load-bearing behaviour in some points which cannot be neglected with the practical application.
The material models currently being developed and presented in this paper will lead to a better
understanding of the failure mechanisms of TRC-structures. Based on these models, design rules and
safety concepts can be set up in the future. On the present level of knowledge these points can be
already named, nevertheless, quantitatively they must be determined with experimental investigations.
Such investigations must not hinder the practical application, but also lead to a high optimisation
degree.
136
The authors thank the Deutsche Forschungsgemeinschaft (DFG) in context of the Collaborative
Research Center 532 “Textile Reinforced Concrete”, the Ministry of Labor and Social Affairs in the
German State of North-Rhine/Westphalia (NRW) and the AiF for their financial support.
REFERENCES
[Ban2001] Banholzer, B.; Brameshuber, W.,: Eine Methode zur Beschreibung des Verbundes
zwischen Faser und zementgebundener Matrix. Beton- und Stahlbetonbau, No. 96, pp.
663-669, 2001 (in german).
[Ban2004a] Banholzer, B.; Brameshuber, W.: Tailoring of are glass filament / cement based matrix
bond – Analytical and experimental techniques. In: Proceedings of Sixth RILEM
symposium on filament reinforced concrete (FRC), in 2004, Varenna, brine Como, Italy.
[Bar1982] Bartos, P.: Bond in Glass Reinforced Cements in: Bond in Concrete. Applied Science,
London, 1982, pp. 60- 72
[Bra2001] Brameshuber, W.; Banholzer, B.: Investigations on the High Scatter of Rowing Out test
Results form Multifilament Yarns Embedded in a Cementitious matrix. Lisse: Foreign
Office. Balkema, in 2001. - In: Composites in Constructions, Proceedings of the
international Conference, CCC2001, postage, Portugal, 10-12 October in 2001,
(Figueiras, J.; Juvandes, l.; Faria, R. (Ed.)), pp. 95-100.
[Bra2001b] Brameshuber, W.; Banholzer, B.: Determination of area and Perimeter of Glass
Multifilament Yarns Embedded in a Cementitious matrix. London: Thomas Telford, in
2001. - In: Proceedings of the 5th international Conference on filament-Reinforced
Plastics for Reinforced Concrete Structures, Cambridge, 16-18 July in 2001, (Burgoyne,
C.J. (Ed.)), Vol. 2, p. in 1027-1031.
[Bra2003] Brameshuber, W.; Koster, M.; Hegger, J.; Voss, S.; Gries, T.; Barlé, M.; Reinhardt, H.-
W.; Krüger, M.: Textile Reinforced Concrete (TRC) for Integrated Formworks. 12.
Intern. Techtextil-Symposium, Frankfurt a. M., 7.-10. April 2003.
[Chu2004] Chudoba, R., Vorechovsky, M., Konrad, M., “Stochastic modelling of multifilament
yarns focused on the delayed filament activation and size effect”, in preparation.
[Cox1952] Cox, H.-L., “The elasticity and strength of paper and other fibrous materials”, British
Journal of Applied Science, Vol 3, pp.72-79, 1952.
[Cur1998] Curbach, M.; Hegger, J. et al.: Sachstandbericht zum Einsatz von Textilien im
Massivbau, Deutscher Ausschuss für Stahlbeton (DAfStb), Heft 488, Beuth Verlag,
Berlin 1998
137
[Cur2002] Curbach, M. (Hrsg.): Sonderforschungsbereich SFB 528 – Textile Bewehrungen zur
bautechnischen Verstärkung und Instandsetzung. Arbeits- u. Ergebnisbericht für die
Periode II/1999-I/2002, Institut für Tragwerke und Baustoffe, TU Dresden, 2002.
[Cur2003] Curbach, M. (Hrsg).: Textile Reinforced Structures. Proceedings of the 2nd Colloquium
on textile Reinforced Structures (CTRS2), Dresden, 29.9-01.10.2003.
[Heg1998] Hegger, J., Döinghaus, P.; Will, N.: Textilbeton – Tragverhalten und
Verbundeigenschaften. In: Sachstandbericht zum Einsatz von Textilien im Massivbau”,
Deutscher Ausschuss für Stahlbeton (DAfStb), Heft 488, Beuth Verlag, 1998, pp. 81- 90.
[Heg2001a] Hegger, J.; Schneider, H. N.; Molter, M.; Schätzke, C.: Fassaden aus textilbewehrtem
Beton. Betonwerk- und Fertigteiltechnik, Heft 12, 2001.
[Heg2003] Hegger, J.; Voss, p.: Load-bearing behaviour of textile-reinforced concrete under 2-axial
loading. Textile Reinforced Structures, Proceedings of the 2nd Colloquium on textile
Reinforced Structures (CTRS2), Dresden, 29.9.-1.10.2003, p. 313-324.
[Jir2002] Jirazek, M., Bazant, Z.: Inelastic Analysis of Structures. John Wiley & Sons, 2002.
[Len2002] Lengeler, B. et al.: Parabolic refractive X-ray lenses. In: Journal of synchrotron radiation
9 (2002), 119-124.
[Mas1990] Mashima, M.; Hannant, D.J.; Keer, J.G.: Tensile Properties of Polypropylene Reinforced
Cement with Different Fiber Orientations”. ACI Journal, 1990, Vol. 87, No. 2, pp. 172-
178.
[Ohn1994] Ohno, S.; Hannant, D.J.: Modelling the Stress-Strain Response of Continuous Fiber
Reinforced Cement Composites. ACI Materials Journal, 1994, Vol. 91, No.3, pp. 306-
312.
[Sch2003] Schorn, H.: Ein Verbundmodell für Glasfaserbewehrungen im Beton. Bautechnik 80,
2003, Heft 3, S. 174-80.
138
6.3 Durability (P. Purnell)
Working party “Bond of TRC”.
Dr. P. Purnell, School of Engineering, University of Warwick, UK CV4 7AL
M. Raupach, J. Orlowsky, Institute of Building Materials Research at Aachen University (ibac),
Germany
ABSTRACT: The durability of textile reinforced concrete (TRC) has been the subject of much study
and controversy since the first attempts to reinforce a brittle cementitious matrix with a more ductile
fibre to add residual toughness were made. This chapter will review the relevant literature, starting
with the work on fibre-reinforced concrete (FRC) and particularly glass-fibre reinforced concrete
(GRC) on which extensive literature exists, and then moving on to the special considerations
associated with using textiles as opposed to fibres or tows. It will then report on current research
trends regarding TRC durability and how these concepts will affect the design of future TRC
components with emphasis on the move from decorative and semi-structural into structural, load-
bearing applications. Topics will include: early developments in FRC durability; time dependent
behaviour for various fibre/matrix combinations; accelerated ageing procedures; micro structural
observations of fresh, weathered and aged composites; models of the degradation process and their
impact on accelerated ageing procedures; modifications in fibre/matrix chemistry and composition
designed to improve durability; and issues surrounding hybrid composites and the use of TRC for
strengthening other materials.
6.3.1 Introduction
The durability of fibre–reinforced cements and concretes (FRCs) has been the subject of considerable
academic study and commercial concern for over 30 years. The behaviour of the hydraulic
cementitious matrices themselves is itself still not fully understood; the interaction of these matrices
with a variety of fibre formulations, morphologies and configurations increases this complexity by
another order of magnitude. The interactions are time-dependent - the elapsed time is best measured in
months or even years – and are manifested in the tendency of some FRC components to exhibit
engineering properties that vary significantly during their lifespan. In particular, the tensile/bending
strength, toughness and strain to failure may decline considerably. The physico-chemical mechanisms
responsible occur at the microstructural level and (as is typical for all composites) concern, to a greater
or lesser degree, the fibre-matrix interface. In most FRCs, especially the textile reinforced concretes
that are the focus of this report, interfaces are actually between matrix and bundles, strands or tows of
filaments, into which the matrix does not initially fully penetrate; in fact, this incomplete surrounding
of the filaments is often a design objective. Thus analytical or investigative approaches suited to fibre-
reinforced polymers (FRPs) are not suitable, since in FRP the fibres act as individual filaments
completely encapsulated by polymer matrix. Furthermore, FRP durability studies are generally geared
to their use in high specific value applications e.g. aerospace or military and thus based on a short-
life/high-maintenance paradigm. This is in stark contrast to FRCs, which are used in low specific
value applications such as infrastructure, where of course a long-life/low-maintenance ethos applies.
Thus the study of FRC durability is a law unto itself and frequently different for each fibre-matrix
combination. This is increasingly true as a wider variety of cementitious matrices, chosen for
technical, durability or environmental reasons, are combined with a range of fibres that increases
continuously as new fibres are developed and previously exotic fibres fall in price such that they hove
into the sights of the FRC designer.
This sub-chapter will primarily concern itself with current thinking regarding FRC durability,
concentrating on referencing material published in the last 15 years. For a general overview of the
subject several comprehensive reference sources are available [Maj1991, Ben1990, Maj1992] and
these will be used as primary references for much of the ‘background’ material presented here.
139
6.3.2 Time dependent behaviour
Fig. 6.68 shows the decrease in bending strength (expressed as a fraction of the 28-day strength) for
five types of OPC based matrix FRC, reinforced with either first or second generation alkali-resistant
(AR) glass fibre (see 6.3.4 for a description of the differences), two types of natural fibre, sisal or
coconut (coir), and polypropylene fibre, exposed outside in UK weather. It can be seen that there are
appreciable losses in strength with time, but that the magnitude of the loss is different for each FRC,
even within similar classes (i.e. glass or natural fibres). Thus it is difficult to make generalisations
concerning the ‘durability’ of such systems and each FRC should be evaluated carefully. In general
though, it appears from the available evidence that the durability of common FRC proceeds
polypropylene > 2nd generation AR glass >> 1st generation AR glass >> unprocessed natural fibres.
1 3 5 10 20 years
1.2
Fraction of 28 day 'unaged' strength
0.8
0.6
0.4
AR Glass 2 Sisal
0.2
AR Glass 1 Coir
Polypropylene
0
100 1000 10000
Age / days
Fig. 6.68. Strength vs. time for selected FRC formulations subjected to natural exposure. Key: AR
glass 2, second-generation AR glass/OPC FRC [Maj1991, Col1990]; AR Glass 1, first-
generation AR glass/OPC FRC [Maj1991]; Polypropylene: polypropylene/ OPC/pfa FRC
[Han1998]; Sisal, Coir, OPC FRC [Tol2000]. Curves: best fit of Purnell et al [Pur2001a,
Pur2005] type relationship (S=[1+kt]-0.5). Error bars: ± 1 standard deviation where
available.
Although strength is the most frequently monitored property during ageing, other properties are often
of more interest. Strength in FRC will only degrade to the matrix strength. This is typically around 20-
50% of the original composite strength but in certain composites, where the fibre content is close to
the critical volume fraction and fibres are mainly intended to provide post-peak toughness, the original
composite strength may not be much greater than the matrix strength. In such cases, it is more
appropriate to monitor toughness or strain-to-failure. Since the toughness or strain-to-failure of the
140
matrix is negligible compared to that of the original composite, these properties will effectively
degrade to zero. These properties are, however, rather more difficult to measure than strength and
often display vary large scatter and hence strength remains the most popular metric. For example,
processed natural fibre i.e. cellulose-FRC is normally manufactured such that the fibre volume fraction
is lower than or comparable to the critical fibre volume fraction Vf,crit and it is thus not appropriate to
monitor strength as a degradation metric; indeed, it increases by about half over 5 years outdoor
exposure. However, strain to failure is reduced from about 3% to that of the matrix (i.e. practically
zero) within this period [Ake1989a], indicating that the benefits of fibre reinforcement have been
largely lost after such time even though strength has increased.
Carbon-FRC has up to now been considered an exotic material but in interest in it is increasing as the
cost of carbon fibres falls. Received wisdom was that the carbon fibres were immune to the corrosive
alkaline deterioration mechanisms [Maj1992, Oha1989] and recent investigators agree [Chu2000]
although little if any long-term weathering data exists in the literature. Polymers other than
polypropylene have also been used to reinforce concrete [Zhe1995, Ake1989b, Maj1992]. Most either
do not bond well to cement or are not stiff/strong enough to confer significant benefits and thus no
weathering data exists. Akers & co workers [Ake1989b] studied ageing of PVA-FRC up to 7 years.
No strength loss was reported but it is not clear that the original fibre volume fraction was
significantly above Vf crit and thus this is not a conclusive result. Katz and Bentur [Kat1995, Kat1996]
have noted that the strength and toughness of carbon-FRC reaches an optimum after ~30 days, after
which a slight drop in properties may be observed owing to matrix densification, after which
presumably these properties then stay constant. The magnitude of the effect was dependent on the
matrix formulation i.e. the proportion of silica-fume (see 6.3.4 below) added.
Majumdar [Maj1992] reported that strength was retained in PVA-FRC after 5 years weathering but
again the fibre volume fraction was very low, as evidenced by there being very little difference
between the first-cracking and ultimate strength. Aramid-FRC displayed no loss in properties after 2
years weathering [Wal1978] but more recent commentators express concern over the longer term
compatibility of aramid fibres with cement [Joh2001]. Polyester fibres are not suitable for concrete
reinforcement as they degrade quickly in cement [Zhe1995].
The causes of the time-dependence of properties fall into two broad categories, regardless of the type
of fibre involved. First is fibre corrosion. The nature of most hydraulic cements is alkaline. During
manufacture, the alkali metal compounds present as impurities in the clays or shales used as raw
materials are converted by the high clinkering temperatures into alkali oxides and/or sulphates, which
are incorporated into the structure of the cement clinker. On gauging with water during concrete
mixing, these highly soluble alkalis are released into the mixing water, raising the pH significantly. As
the cement hardens, the mixing water is used up and the alkalis are thus concentrated in the remaining
free water, contained within the capillary porosity of the cement paste, known as the pore solution.
Cement pore solution can reach a pH of up to 13.7, or 700 mmol/l [OH-], the OH- content being
associated mainly with K+ ions but also with Na+ ions; the pore solution is buffered by the Ca(OH)2
(portlandite) produced by the hydrating cement, but contrary to popular belief the Ca2+ content of the
pore solution is generally very low (i.e. <40 mmol/l ) owing to the common ion effect. The integrity of
many fibres is compromised at these very high alkalinities and thus fibre corrosion, leading to a loss of
strength of the reinforcement, is always a composite degradation mechanism that must be considered.
This holds even for fibres which are supposedly immune to alkali attack, since the timescales which
must be considered – i.e. the service lifetimes of typical FRC components, often decades – are much
longer than most exposure experiments. As well as fibre corrosion owing to the alkalinity of the
matrix, in certain applications the ingress of external agents deleterious to certain fibres, such as
chlorides or acidic organics, into the composite must also be considered, although this is unusual.
There is also evidence that some fibres, particularly glass, may be susceptible to physicochemical
attack by direct contact with calcium hydroxide in the matrix to a level over and above that caused by
alkalinity alone [Pro1980]
141
Secondly, a variety of effects caused by the continued hydration of the cementitious matrix also have
the potential to cause composite degradation. Given the continued availability of water, residual
unhydrated cement in the matrix (of which there is always a small amount) will continue to hydrate,
producing more hydration products and increasing the strength of the matrix. The hydrated phase
assemblage itself will also tend to slowly evolve since it is essentially composed of meta-stable
phases. Although this continued hydration and evolution might only involve a very small fraction of
the cement paste, the resultant effects on the fibre-matrix interactions can be profound.
There is a critical volume fraction of fibres Vfc that is required for the composite to display true ductile
behaviour i.e. during a load controlled tensile or bending test, a rising stress-strain curve and multiple
cracking after the matrix first becomes cracked. (NB. Most FRCs will display post-first-crack ductility
during a stroke controlled test even if the volume fraction is less than Vfc, but its nature and magnitude
is critically dependent on the stroke rate of the test). Vfc is dependent on the matrix strength; as the
matrix becomes stronger, a greater proportion of fibres are required to take up the load thrown onto
them as the matrix first cracks and thus becomes unstressed. Thus for composites designed with a fibre
content very close to Vfc, continued strengthening of the matrix may change the composite behaviour
at first crack from ductile to brittle. Careful design of the composite should easily avoid this.
Continued hydration also tends to ‘densify’ the interface between the fibres and matrix. When FRC is
first made, the interface is relatively porous, with little intimate contact between the fibre and matrix.
The interfacial zone is thus relatively weak, which is often beneficial for the composite. At young
ages, cracks travelling through the matrix tend to be diverted through this weaker zone. As the matrix
ages, hydration products (particularly portlandite) fill the spaces in the interfacial zone, reducing the
porosity and increasing the hardness and strength; it becomes energetically and thus statistically more
likely that a crack will travel through the fibre rather than around it. Also, fibres bridging cracks
become bent through tighter radii of curvature as the matrix hardens, since the matrix immediately
adjacent to the point where the fibre emerges from the crack face is less compliant. This induces
greater stress in the fibres and increases the likelihood of them breaking. The bond between the outer
filaments of a unit reinforcing element (strand, tow etc.) will also increase as the interfacial zone
densifies. Many FRCs are designed such that at failure the fibre is intended to pull out of the matrix
rather than break, as this expends more energy and thus imbues the composites with greater toughness.
As the bond strength increases, it becomes more likely that fibre failure, rather than pullout will occur
and the toughness will be reduced (even though the tensile/bending strength may well increase).
A related phenomenon is ‘bundle filling’. In young FRC, there is normally space between the
individual filaments into which the matrix has not penetrated. The reinforcement can thus act in a
rope-like manner, with filaments moving over each other, adding significantly to the toughness of the
composite. As the composite ages, continued hydration and migration of species within the cement
matrix leads to precipitation of hydration products, especially portlandite, between the filaments and
this may eventually fill all the available space. The rope-like behaviour is lost and the toughness of the
composite may be compromised.
The relative importance of each of these degradation mechanisms is dependent on the particular fibre-
matrix combination. Some combinations are particularly prone to specific mechanisms, while in others
the dominant mechanism is still the subject of research and debate. The mechanisms are examined in
more detail in 6.3.5 to 6.3.6 below in the context of the changing microstructure of FRC and
modelling of degradation.
142
durability were advanced, accelerated ageing procedures were used to attempt to validate the
purported improvements and these procedures are still in widespread use today. Two technical types
of test can be identified; hot water ageing and cyclic ageing. The former aims to directly accelerate the
fibre-matrix interactions and thus the degradation by immersing samples of the composite in water at
elevated temperature (typically ≥ 50°C). Hot water is used for two reasons: it provides thermal inertia
and also many FRCs do not appear to suffer degradation owing to ‘dry’ elevated temperature alone;
indoor stored GRC seems to suffer no degradation. The degradation mechanism appears to involve
water in some way, chemical or physical. Other types of test attempt to replicate the cycles of
temperature and/or moisture FRC components are likely to be exposed to during service. These cyclic
tests normally involve alternate and repeated exposure to hot/wet/cold/dry environments and may
involve freeze/thaw cycles or exposure to salts. It should be noted that these tests are not
interchangeable; certain tests are more appropriate for certain FRC formulations e.g. hot water for
GRC and cyclic wet-dry for natural FRC; for ageing of cellulose fibre-reinforced concrete, it is also
considered important that at least one of the steps should promote carbonation of the matrix
[Ben1990].
Tests can also be divided into two other categories: ‘deemed-to-satisfy’ (DS) tests and predictive tests.
DS tests are intended as a quality control measure; if, for example, a coupon of composite is tested
after a fixed ageing period or number of cycles and is found to have properties that exceed some pre-
determined value or percentage of the unaged properties, then it is deemed to have ‘passed’ and the
material fit for purpose. No prediction of the long-term behaviour of the composite in service is
advanced or inferred, only that it is in some way ‘durable’ since it has passed the test. Predictive
testing, on the other hand, attempts to correlate periods or numbers of cycles of accelerated ageing
with longer periods of in-service weathering and thus goes some way towards quantifying the lifetime
that can be expected from the material. Although in theory either cyclic or hot water testing can fall
into either category, in practice hot water ageing tends to be used for predictive testing and cyclic
ageing for DS testing with few exceptions.
Investigators have also frequently subjected fibres (as opposed to composites) to accelerated ageing
regimes and thus inferred their likely behaviour in a cement matrix. These tests generally involve
immersion of samples of fibre in hot alkaline solutions, often of analogous composition to cement pore
solution, for various lengths of time and measurement of resultant reduction in strength. Such tests,
although useful for helping to assess new fibres, should be treated with caution with respect to their
use in predicting durability, since the action of the cement matrix on the fibre is not just controlled by
its alkalinity. Sisal and coconut fibres were shown to completely lose their strength after 300 days in
Ca(OH)2 solution of pH 12 [Tol2000]. Aramid (Twaron) fibres have been subjected to immersion in
concentrated NaOH for 24 hours with no ill effect [Vil2003] but no longer term data is evident. Such
tests have long been used with prospective glass compositions and results are reviewed by Majumdar
& Laws [Maj1991] but new glass or ceramic formulations now tend to be tested in actual FRC
samples [e.g. Che2003, Ma2005] although some fundamental work continues to use just fibres
[Gao2003].
The best known (and most frequently abused) accelerated ageing rationale for FRC is probably that
derived for OPC matrix AR-glass GRC by Proctor and co-workers in the late 1970s [Lit1981,
Pro1982]. A combination of tests on composites aged by hot water immersion and composites exposed
at various locations around the world were combined with data from a specially developed durability
test for fibre strands known as the ‘strand-in-cement’ (SIC) test. The SIC test involves a specimen
comprising a strand of fibres (for GRC, typically about 200 filaments, each of about 14µm diameter)
being partially encased in a small cylinder of cementitious matrix. The specimen is aged by immersion
in hot water for a specified time and the strand subsequently tested in tension with the small cylinder
of matrix still in place. They concluded that a single degradation mechanism with an activation energy
of about 90 kJ/mol controlled strength loss in all these cases and the data was pooled into an
Arrhenius-type relationship correlating the accelerated ageing tests with longer periods of in–service
weathering. This allowed ‘acceleration factors’ to be advanced, e.g. that 1 day of hot water immersion
143
ageing at 50, 60 and 80°C induced the same strength loss as 101, 272 and 506 days of weathering in a
UK climate respectively. This predictive rationale lead to hot water ageing being accepted as the de
facto standard method for validating new GRC formulations with respect to durability, which it
remains [e.g. Orl2003, Hem2003, Bro2002]. However, the method was only derived with OPC
composites in mind, and concern over its application by other investigators to non-standard GRC (and
even FRP) with evidently different ageing characteristics and mechanisms, together with shortcomings
of the model regarding service life prediction and the lack of a micromechanical model lead to the
rationale being updated by Purnell and co-workers [Pur2001a]. The basis and implications of their
model is discussed in more detail in 5.3.6, but it is clear that a unique set of acceleration factors
applies to each matrix formulation used with AR glass [Bed2003, Pur2005]; by implication,
acceleration factors derived for GRC should not automatically be applied to other types of FRC.
Another accepted accelerated ageing test method for GRC is codified in European standard
DD EN 1170-8:1997 Precast concrete products. Test method for glass-fibre reinforced cement. Cyclic
weathering type test. This involves cyclic ageing, where each cycle involves 24 hours immersion in
water at 20°C, 30 minutes of forced drying in air at 70°C and 1 m s-1 airflow, 23 hours in air at 70°C
and 30 minutes of forced cooling in air at 20°C and 1 m s-1 airflow, with samples tested in bending
after 10, 25 and 50 cycles. The test is notoriously difficult to apply, as it requires specialist equipment
and contains no predictive component, only serving as a comparison. As such, it is unpopular with
researchers but has some adherents in industry [e.g. Cia2001].
Some attention has been paid to the accelerated ageing of cellulose and/or natural fibre based-FRC. In
early work, the hot water ageing rationales developed by Proctor & co-workers [Lit1981, Pro1982] for
glass-FRC were used for sisal-FRC [Ber1984] but found to be inappropriate. Bentur and Akers
subsequently established methods for accelerated ageing of cellulose-FRC in a series of papers
[Ben1989a, b; Ake1989a]. Two 24 hour/cycle cyclic regimes were proposed; ambient/elevated
temperature (AE) and CO2 rich/elevated temperature (CE), both followed by three-point bend testing
based on ISO 39611 –1980 E. The AE cycle was 9h in water at 20°C; 3h in air at 20°C; 9h infra-red
radiation 80°C in air; and 3h cooling to 20°C in air. The CE cycle was 8h in water at 20°C,; 1h in oven
at 80°C; 5h in oven at 20°C in an atmosphere saturated with CO2; 9h in oven at 80°C; and 1h cooling
to 20°C. Cycling was continued for 3 months. The CE cycle was found to most closely approximate a
5-year period of natural weathering and was thus considered the most appropriate method. This regime
continues to be used to age cellulose FRC [Kim1999] although other investigators have used slightly
modified carbonation tests in addition to freeze-thaw methods [Mac1999]. All investigators lean
towards the same conclusions: the strength and modulus of the material increases by about 20-50% (as
a result of the carbonation of the matrix increasing matrix strength and possibly bonding) but the strain
to failure is reduced to values similar to that of the unreinforced matrix i.e. <0.1%. The same regime
has been used for PVA-FRC [Ake1989c] but again at very low Vf. Some application-specific tests are
also used. Cellulose-FRC is used for sewer pipes (an application previously filled by asbestos-FRC)
and as such is exposed to external deleterious agents that can cause acidic and/or biological attack on
the fibres and/or matrix. Fisher et al. [2001] exposed cellulose-FRC samples to sewage in aerobic and
anaerobic treatment plants, and also immersed samples in sulphuric acid (pH ~5). It was concluded
that there might be advantages in using cellulose-FRC over steel-FRC in these applications. For FRC
with natural fibres, simple wetting and drying cycles (e.g. 1 day immersion at room temperature
followed by 6 days drying in ambient lab conditions) seems to be considered appropriate, although
since such composites suffer degradation of properties in relatively short time under normal
weathering, it is debatable whether accelerated ageing is necessary [Tol2000, Tol2003].
Accelerated ageing of other FRC types does not appear to have been afforded the same degree of
investigation, with a variety of different ad hoc methods extant in the literature. Polypropylene-FRC
has been assessed using 50 wet/dry cycles followed by impact testing, and 50 freeze/thaw cycles
followed by flexural/compressive testing [Pue2003], which caused some decrease in flexural strength
but left compressive strength unchanged. Recently, Al2O3-based ceramic fibre reinforced concrete has
been immersion aged at 70°C [Ma2005]. Carbon-FRC has been subjected to freeze-thaw ageing based
144
on ASTM C666] for between 30 cycles [Che1996] and 300 cycles [reported in Oha1989] and water
immersion at 75°C for 5 months [reported in Oha1989] with little if any effect on strength. Bentur and
Mindess [Ben1990] discuss how carbon-FRC does not show any long-term response to hot water or
cyclic accelerated ageing and conclude that it is unlikely to suffer any durability problems.
Glass-fibre development is relatively mature. It was clear from the first incarnations of glass-FRC that
E-glass fibre (as used by the FRP industry) would be chemically unstable in the highly alkaline cement
matrix. Alkali-resistant (AR) glass was developed in the 1970s by Pilkingtons, based on silica-soda-
calcia glass with the addition of about 16% zirconia, and marketed as Cem-FIL. Glass-FRC made
from AR fibres was not immune to degradation, however, and further developments were made to the
soluble coating or ‘size’ applied to the fibre (originally applied for manufacturing purposes).
Incorporation of polyhydroxy-phenols in this size modifies the hydration behaviour of the cement
matrix at the interface and significantly improves durability of AR glass-FRC [Maj1991]. AR-glass
fibres with such coatings are known as ‘second-generation’ fibres and are now the industry standard.
From time to time, other glass systems are proposed, e.g. based on strontium [Kar2000] or barium
[Che2003] but these have yet to come to commercial significance. Other fibres tend to be of fixed
composition (e.g. carbon, polypropylene, cellulose) and are thus not amenable to having their alkali
resistance intrinsically increased.
An experimental method of increasing durability is to pre-impregnate the fibre strands with material
intended either to block precipitation of hydration products between the filaments or modify the nature
of hydration such that less deleterious hydration products result. Glass-fibre strands have been
impregnated with microsilica or acrylic polymers [Bar1996]; natural fibres have also been treated with
microsilica [Tol2003]. In both cases, resistance to accelerated ageing and/or weathering was improved
markedly. This approach has yet to be commercially adopted but would appear to be attractive.
By far the most common approach involves modifying the cement matrix, cf. Portland cement, in
order to improve its compatibility with fibres, either by using additives to Portland cement or using
non-Portland cement matrices. A great deal of literature exists on this topic, especially concerning
glass-FRC; pre-1988 literature is summarised in the books by Majumdar and Laws [Maj1991], and
Bentur and Mindess [Ben1990]; the relevant sections are noted, and more recent results will also be
briefly summarised here. Additives are generally intended to reduce the pore solution alkalinity of the
matrix and/or react with the calcium hydroxide produced during hydration; thus pozzolanic materials,
as used to enhance ordinary concrete, are common additives. Condensed silica fume/microsilica (csf),
metakaolin, ground granulated blast furnace slag (ggbs) and pulverised fuel ash (pfa) have all been
investigated for use with most fibre types.
Csf is a by-product of the ferro-silicon industry and consists of extremely fine particles (<1 µm) of
essentially pure silica, which reacts rapidly with Ca(OH)2. It is used as a matrix enhancer for
natural/cellulose-FRC [Tol2003, Mac1999], ceramic-FRC [Ma2005], carbon-FRC [Che1996,
Kat1995, 1996] (where its primary function is to aid fibre dispersion and compaction) and glass-FRC
[Bar1996, Mar1997, Zhu1997]. It is effective in increasing the durability of natural fibre composites
exposed to weathering and cyclic wet/dry ageing, but has limited or variable effect on that of glass-
FRC subjected to hot-water ageing [see also Ben1990, Maj1991]. Addition levels vary between 10%
and 40% replacement of cement, but at higher levels of replacement it has a profoundly negative effect
145
on workability and thus requires higher water/cement (w/c) ratios for manufacture, which is
undesirable.
Metakaolin is a calcined china clay which reacts readily with Ca(OH)2 in the matrix and reduces the
alkalinity of the pore solution [Pur1999]. It has been extensively investigated as an additive for glass-
FRC to improve durability [e.g. Mar1997, Pur1999, Zhu1997, Sou1991, Bal2003, Bed2003].
Normally added at 20-25% cement replacement, almost all investigators agree that it significantly
improves the durability performance of glass-FRC both under accelerated (i.e. hot water) and natural
weathering. The degree of improvement depends on how accelerated ageing results are interpreted
(see 6.3.6 below and [Pur2005]) but it is sufficiently well established for commercial formulations to
be used; the metakaolin is also believed to improve surface finish and resistance to loss of appearance
via surface dusting or efflorescence [e.g. Gil2001]. It does not appear to have been used in conjunction
with other fibre types.
Ggbs has been used with natural-FRC [Tol2003] and glass-FRC [Maj1991] with limited success. It
was reported to have a beneficial effect on glass-FRC at very high cement replacement levels (70%)
but this does not seem to have spurred commercial interest, although some Japanese [Tak1999]
formulations use blastfurnace slag cement in conjunction with other admixtures. The DuraPact matrix
[Pac2001], which purports to have a guaranteed 50-year lifespan, is believed to be based on
blastfurnace slag cement combined with microsilica [Pur1998]. Pfa has been used in polypropylene-
FRC [Han1998] and glass-FRC [Maj1991 pp94-102, Cyr2001]. Polypropylene-FRC with pfa was the
material examined in the long term study by Hannant [Han1998] (see also Figure 1 – Figure is
missing). Such composites appear to retain most of their strength after 18 years of weathering but no
comparison with an unmodified OPC matrix was performed. The addition of pfa to glass-FRC appears
to have some benefit with regard to durability [Maj1991]. The durability improvements reported by
Cyr [Cyr2001] are from a hybrid system including both pfa and silica fume and thus cannot be
attributed to pfa alone. For both pfa and ggbs, it should be noted that slower setting and hardening, and
thus lower matrix strengths at early ages, will results from their addition to the matrix.
Another approach almost unique to glass-FRC is the use of polymer modification of the matrix.
Majumdar and Laws [Maj1991] dedicate a whole chapter to the topic. A wide variety of polymers
have been tried but attention has most recently focused on an acrylic polymer dispersion (‘Forton’).
Originally designed as a curing aid for thin concrete sections to prevent water loss by evaporation
leading to shrinkage cracking, it is added to glass-FRC to improve workability in the fresh state and
improve mouldability. It is also widely believed to improve durability and the ‘5/5’ formula,
comprising OPC-glass-FRC with the addition of 5% Forton polymer solids by volume and 5% AR-
glass fibres by weight is ubiquitous in the glass-FRC literature, so much so that it is often referred to
implicitly as a ‘base’ to which modifiers such as metakaolin are then added [e.g. Gli1993]. The
mechanism by which it enhances durability is different to the pozzolans in that it does not reduce
alkalinity of Ca(OH)2 formation, but is thought to either coat the fibres with an inhibitive layer or
modify hydration at the interface by occupying space that might otherwise fill with Ca(OH)2. There
has been some debate as to whether polymer-modification of the matrix can actually lead to improved
durability as accelerated ageing procedures often yields conflicting results [e.g. Zhu1997, Qia2003].
The most recent research unequivocally demonstrates that hot-water ageing is unsuitable for polymer
modified glass-FRC [Pur2005] and also that the addition of polymer to both plain OPC and modified
matrices confers very significant durability benefits [Bal2003]; over 19 years of natural weathering,
compared to plain OPC matrix glass-FRC, reduction in modulus of rupture (MOR) was negligible and
the degradation of strain-to-failure reduced by 75%. Polymer modification of the matrix is uncommon
with other fibre types but a summary is given by Bentur and Mindess [Ben1990]
Other, non-Portland (nP) common cementitious systems have also been investigated for use with FRC.
Again, this practice is better developed for glass-FRC than for other systems. The use of nP systems
for glass-FRC is summarised by Majumdar and Laws [Maj1991]. High-alumina cement (HAC) and
super-sulphated cement are both less alkaline than OPC and thus candidates for glass-FRC. Worries
146
over the integrity of the HAC matrix at moderately elevated temperatures (the ‘conversion’ reaction),
which led to the banning of HAC for structural use in the UK, have caused glass-FRC producers to be
cautious over its use despite its superior durability characteristics. Super-sulphated cement also
showed improved resistance to ageing but is not a common choice for glass-FRC as it is susceptible to
weakening on carbonation, which is enhanced for the thin sections typical of FRC.
The last ten years has seen an increase in interest in nP matrices based on calcium sulpho-aluminates.
These cements, with 4CaO.3Al2O3.SO4 (ye’elemite or Klien’s compound) as the major active
ingredient, hydrate in the presence of a source of lime to form mainly ettringite, with little or no
calcium hydroxide remaining in the hydrated phase assemblage and a lower pore solution alkalinity
than OPC. Development of such cements has been led by China [Qi2003] and Japan [Tak1999] but
analogues are available in the UK [Gar1991] and the US [e.g. Mol1995] . Little data exists on the
long-term properties of glass-FRC made with such cements but it shows extraordinary resistance to
hot-water accelerated ageing [e.g. Jia2001]; however, caution is required in interpreting such results
(see 6.3.6 below and [Pur2005]). Development of these matrices is continuing and they appear to offer
the best option for completely durable glass-FRC. The use of nP matrices with other fibres has
received little coverage in the literature. Puertas and co-workers [Pue2003] have investigated alkali-
activated blast-furnace slags and pfas (so-called ‘geo-polymers’) reinforced with polypropylene fibres,
subjected to cyclic ageing. Results were unclear and did not show any significant benefit cf. OPC.
The cementitious matrix in FRC is susceptible to reaction with the carbon dioxide in the atmosphere, a
phenomenon known as carbonation. During this reaction, portlandite in the matrix is converted to
Ca(CO)3 and thus the buffering capacity of the pore solution is lost, alkalis are either converted to
carbonates or absorbed in the C-S-H phase (which also carbonates, becoming decalcified) and the
alkalinity of the matrix drops sharply. Its strength also increases. Thus in theory, a carbonated matrix
is more compatible with most fibres. Natural carbonation can take years for FRC components, but
attempts have been made to apply accelerated carbonation to improve their durability. Accelerated
ageing tests for natural fibre or cellulose-FRC include a carbonation component since these materials
tend to carbonate readily when exposed to weathering. There seems to be some confusion in the
literature regarding the difference between carbonation as a treatment or an ageing process but the
recent consensus seems to be that early curing of such composites in a CO2-rich environment improves
their durability [Tol2003, Mac1999]. Purnell and co-workers [Pur2001b, Pur2003a] used super-critical
CO2 to effect carbonation of OPC matrix glass-FRC components within hours. The treatment
significantly increased the resistance of such composites to accelerated ageing, improved mechanical
properties in general and improved their dimensional stability under wetting and drying [Sen2002].
147
findings are summarized in the major reference sources [Maj1991, Ben1990]. This section will focus
on key points relating to mechanisms of ageing.
Fig. 6.69 shows the unaged interfacial microstructure for typical OPC matrix FRC materials. The
interface between the reinforcement unit and matrix is porous and spaces within the unit
(interfilamental spaces and cell lumen for glass and coir-FRC respectively) remain largely free from
hydration product. Fig. 6.70 shows the microstructure of the same FRCs after ageing such that
significant degradation of mechanical properties has taken place. In both materials, much (but not all
in the case of the glass-FRC) of the interfacial and interfilamental/lumen space has been filled with
hydration product, identified as Ca(OH)2 in the glass-FRC and assumed to be the same in the coir-
FRC. This phenomenon is variously known as mineralization, petrification and ‘bundle filling’. It is
generally assumed to be associated with an increase in bond between the fibres and the matrix and, in
cases where the unit reinforcing element is a strand e.g. glass- and carbon-FRC, fibre-fibre bond as
well. This can change the mode at failure from ‘pull-out’ (where, after peak load, fibres are pulled out
of the matrix, expending strain energy and thus increasing pseudo-toughness) to fibre fracture, which
expends far less energy, if the bond passes a critical value; matrix densification effects will also begin
to occur (see 5.3.2 – which section is meant? above). The transition from pullout to fracture has been
observed in glass-FRC [Bar1996], who also observed an intermediate ‘telescopic’ pull-out mode, and
natural/cellulose FRC [Moh2005, Ben1989b]. Enhanced fibre-fibre bond with ageing in glass-FRC
has been directly measured by Zhu and Bartos [Zhu1997] using novel micro-indentation apparatus.
Fig. 6.69. Unaged interfacial microstructure of OPC matrix FRC. Left – glass-FRC, optical thin
section petrography, polars crossed at 75°, R1,2 = porous areas [Pur2000]. Right – coir-
FRC, backscattered electron micrograph [Tol2000].
148
Fig. 6.70. Aged interfacial microstructure of OPC matrix FRC. Left - glass-FRC aged for 56 days at
65°C optical thin section petrography, polars crossed at 77°, CH = calcium hydroxide
crystals [Pur2000]. Right – coir-FRC aged 25 cycles of wetting and drying, backscattered
electron micrograph [Tol2000].
Examination of fibre surfaces exposed in aged or weathered composites rarely shows any significant
fibre corrosion, i.e. loss of section or gross pitting, regardless of the fibre type or the extent to which
properties have been degraded [e.g. Pur2000, Tol2000, Kat1995, Bal2003]. In the short term at least,
in FRC where the fibres are known to be immune to alkali attack i.e. carbon-FRC, loss of strength and
toughness may still be observed with ageing [Kat1995]. Most investigators, therefore, attribute loss of
properties in all FRC to matrix densification, bond enhancement and bundle filling/mineralization
effects. For many FRC composites this is true, but fibre corrosion should not be thus discounted. First,
bundle filling and/or bond increase is not always associated with degradation. In glass-FRC
composites made with sulpho-aluminate based matrices, the interfilamental space is completely filled
with hydration products even in unaged samples with no detriment to strength; in OPC/2nd generation
AR glass-FRC, degradation can occur without complete bundle filling being observed (see below) and
bond can reach a maximum without detriment to mechanical properties [Pur1999]. Secondly, a
transition in failure mode from pull-out to fracture does not necessarily require an increase in bond; a
decrease in fibre strength will also trigger such a transition. In natural fibre composites, pull-out may
be observed as the dominant failure mode even after mechanical properties have been significantly
degraded [Moh2005] and the mode may change from pull-out to fracture without mineralization of the
fibre [Ben1989a]. Thirdly, for glass and carbon, a brief examination of the fracture mechanics of fibres
suggests that their strength is governed by the size and population distribution of surface flaws. The
surface manifestation of critical flaws is likely to be so small (~10 nm) as to be very difficult to detect
on in-situ fibres using the SEM; nucleation and growth of such flaws could thus cause strength loss
without readily detectable fibre surface damage [Pur2005]. Recent work using Weibull analysis and
atomic force microscopy on AR glass fibres subjected to various ageing treatments has established that
the maximum surface defect size (as opposed to manifestation) is ~50 nm and confirmed that flaw size
and population density of surface flaws control the strength of the fibres [Gao2003].
The microstructure of FRC modified for enhanced durability is varied. The size applied to second
generation AR glass fibres reduces the precipitation of portlandite at the fibre-matrix interface and
within the fibre bundle. The ubiquitous monolithic portlandite deposits completely surrounding fibres
in 1st generation OPC matrix glass-FRC after ageing [e.g. Maj1991] are not seen in modern glass-FRC
[e.g. Pur2000]. Reports on FRC modified with csf are variable. Katz & Bentur [Kat1995] did not
observe a significant difference between the interfacial microstructure of OPC and OPC-csf matrix
carbon-FRC using SEM but mercury intrusion porosimetry (MIP) indicated that the csf induced a
significant reduction in porosity after accelerated ageing. Bartos and Zhu [Bar1996] reported that csf
modification (10% cement replacement) to the matrix of glass-FRC did not significantly change the
development of interfacial and interfilamental microstrength with ageing (measured using a micro-
149
indentation technique to ‘push out’ individual filaments). However, matrix modification by metakaolin
(25%) was effective in preventing development of microstrength and this was correlated with reduced
degradation cf. OPC matrix glass-FRC [Zhu1997]. Purnell and co-workers [Pur2000] used thin section
petrography to observe a change in the nature of interfilamental deposits in metakaolin modified glass-
FRC; portlandite was not deposited between the filaments but an amourphous reaction product of
metakaloin was. This was again correlated with reduced degradation during ageing. The
microstructure of glass-FRC made with sulpho-aluminate modified cements is quite different. In
contrast to normal glass-FRC, the matrix completely penetrates the fibre bundles and surrounds all the
filaments even in the unaged condition, without detriment to mechanical properties, and the
microstructure does not change with ageing; fibre pullout is still observed even after ageing (see
Fig. 6.71). However, the matrix contains no portlandite. Thus it would appear that bundle filling is
only detrimental if the precipitated material contains portlandite. SEM examination of the interface in
polymer-modified glass-FRC shows a film of polymer partially covering the fibre surface at young
ages [Ben1990 p262] but this film is not apparent in micrographs of naturally weathered samples
[Bal2003] and the microstrength of the interfacial and interfilamental bond is not affected by polymer
modification of the matrix [Zhu1997]. Given the fragility of the film coating the fibres, it is likely that
polymer modification confers durability by impeding water and thus potential precipitate migration
within the matrix in general with some temporary and/or minor enhancement at the interface. The
microstructure of FRC modified with pfa and ggbs has not received significant recent study and
Majumdar and Laws [Maj1991] noted that under the SEM, the microstructure of glass-FRC so
modified was not markedly different from normal glass-FRC.
Fig. 6.71. Microstructure of sulphoaluminate modified glass-FRC. Left – unaged sample, thin
section petrography polars crossed at 75°. Right – SEM of aged (28 days at 65°C)
fracture surface showing Bartos-style pseudo-ductile ‘telescopic pullout’ [Pur1998].
150
6.3.6 Models of the degradation process
A number of investigators have attempted to model the degradation process in FRC. Models range
from empirical descriptions of the strength loss process under natural and accelerated ageing, to
complex analytical treatments based on consideration of micro-mechanical changes at the fibre-matrix
interface. Some models have attempted to combine the two. The models tend to reflect the debate
surrounding the degradation mechanisms described above in that each of the current models is based
on an assumption concerning the main mechanism in action, i.e. that degradation is caused either by
matrix/interface densification, or fibre strength loss. The focus of this section will be on aspects of
modelling procedures that specifically concern durability; modelling of the general behaviour of FRCs
is covered elsewhere.
The earliest model in which a durability issue can be recognised is the ‘ACK’ model describing the
behaviour of brittle matrix composites [Ave1971]. This theory identifies a critical volume fraction of
fibres Vf, crit required in order that multiple cracking (i.e. pseudoductile) behaviour can be realised. This
quantity is proportional to the matrix strength; thus as the matrix strengthens but the fibre volume
fraction remains constant, the failure mode may change from multiple to single cracking – i.e. brittle
failure – in composites manufactured with a proportion of fibres close to Vf, crit. However, this simple
embrittlement behaviour can be easily avoided by ensuring sufficient fibre content. Assuming that
there is sufficient fibre to avoid single fracture, then after the first matrix crack (i.e. in the multiple
cracking region and beyond) it is a combination of fibre strength and bond strength that control
behaviour; most models can be classified as assuming that one or the other of these parameters
dominates the behaviour.
The model of Katz & Bentur [Kat1996] was developed for carbon-FRC but has general application. It
is based on the principle that “physical densening of the matrix” is the cause of strength and toughness
loss with time. Two concurrent processes were modelled; the increase in fibre-matrix bond with time,
and the reduction in ‘support length’ of an inclined fibre forced to bend while bridging a matrix crack
(which increases the radius of curvature and thus the stress concentration in the fibre, see Fig. 6.72).
The model also takes into account non-uniform fibre lengths caused by breakage during mixing. Bond
and support length were assumed to increase with time at a similar rate to the compressive strength of
the matrix. The bond processes alone were shown to cause a monotonic increase in strength efficiency
of the fibres, while toughness efficiency peaked after ~10 days; both parameters then settled to a
constant value within 40 days. The support reduction process alone caused the ‘active fibre fraction’ to
continuously decrease with time although it is not clear whether this is asymptotic beyond 60 days.
This led to short-term peaks in both strength and toughness efficiencies, which tailed off up to 90
days; again, whether this is asymptotic is not clear. Combining the two effects produced curves that
closely matched the behaviour of real carbon-FRC composites under accelerated ageing for up to 200
days at 60°C (Fig. 6.72). The model was later extended [Kat1997 – Reference is missing] to account
for the effect of variations in fibre modulus and showed that the matrix densification degradation
mechanism increases in severity as fibre modulus or radius increases, and does not apply for ‘ductile’
fibres. This was confirmed by experiments on FRC made with a range of carbon and polymer fibres.
151
Fig. 6.72. Left: Model to describe the bending of an inclined fibre crossing a crack. F = force
required to induce a deflection δ; q = length of support; W = matrix reaction. Right:
Combination of effects of bond and bending (AFF) to strength vs. time behaviour in
carbon-FRC [Kat1996].
The model of Kim et al [Kim1999], developed for cellulose-FRC, also assumes that degradation is
caused by increases in bond with time. Modelling is based around the concept of the fracture
toughness of the material being equal to the sum of the stress intensity factor (sif) caused by loading
and the negative sif caused by closing traction of fibres across a crack. This is an extension of the
concept first presented by Naaman & Shah [Naa1979]. The model requires 14 separate input
parameters including the interfacial chemical debond energy and the fibre snubbing coefficient, some
of which are awkward to measure and is solved numerically. It does predict broad trends in ageing of
cellulose-FRC but rests on a ‘one crack failure’ assumption which is invalid for most FRC with Vf
>Vfc. Also derived from the Naaman and Shah concept is the model advanced by Mobasher and Li
[Mob1995], based on a fracture mechanics/finite element treatment of the change in pull-out
behaviour with interfacial stiffness. The model is designed around steel-FRC. They calculated
increases in interfacial stiffness, and adhesional and frictional bond of roughly 4 and 2 respectively
after 3 days of accelerated ageing.
Purnell and co-workers [Pur2001a, Pur2005] have taken a different approach in a model designed
around glass-FRC but with wider application. They modelled strength loss in fibres (and thus
composites) as being caused by the slow, subcritical growth of surface flaws on the fibre. A stress
corrosion process known as ‘static fatigue’ is assumed to drive this process, which in alkaline
environments can be triggered by very small stresses (e.g. shrinkage or thermal mismatch stresses) and
may be aggravated by the growth of portlandite crystals at the interface. By combining relations
describing the fracture mechanics of fibres, the growth of flaws and ‘rule of mixtures’ theory, they
derived a relatively simple relationship between normalised strength and time controlled by a single
parameter k, which is a function of temperature in an Arrhenius relationship; this allows the hot-water
accelerated ageing process to be rationalised and used to predict strength-time curves for various
glass-FRC formulations. The model was fitted to a very large data set comprised of new data,
literature data and commercial data (Fig. 6.73). This essentially updated the work of Proctor and co-
workers [Pro1982, Lit1981] and showed that the acceleration factors derived by them cannot be
applied to modified matrix glass-FRC as the activation energy of the strength loss process is different
cf. OPC glass-FRC (57-59 kJ/mol cf. 94 kJ/mol); this implies that a different degradation process is at
work, which they conclude is connected to that lack of Ca(OH)2 availability in modified matrices. In
fact, using Proctor and co-workers factors to validate modified matrix glass-FRC can overestimate
durability by almost an order of magnitude at 60°C (Table 6.8).
152
1.2
OPC II 80°C
1
Normalised strength S
OPC II 60°C
0.8
0.6
OPC II
weathering -
0.4 Innisfail
OPC II
0.2
weathering -
Osa ka
0
1 10 100 1000 10000
Time / days
Fig. 6.73. Fit of Purnell model (S=[1+kt]-0.5) to OPC matrix glass-FRC strength vs. time data from
various sources [Purnell & Beddows 1995 – Reference is missing].
Table 6.8. Acceleration factors for glass-FRC with respect to UK weather for various ageing
temperature/matrix combinations, compared with those advanced by Proctor an co-
workers [Pro1982, Lit1981, Pur2005].
Matrix Ageing Temperature
50°C 60°C
OPC 120 340
Sulphoaluminate 20 39
OPC + metakaolin 18 35
Proctor et al 101 272
Other investigators have also used fibre strength loss rather than matrix densification to model
degradation. Orlowsky and Raupach [Orl2003] have examined strength loss in glass fibres and FRC
by assuming that a series of layers are formed by reaction between the glass surface and the cement
matrix, of the order of 20-40 nm thick; this was supported by XPS investigations. They surmise that
strength loss may be due to loss of glass area (rather than flaw growth) [Bro2002] but recent
investigations do not support this since glass-FRC made from filaments of differing diameter lose
strength at the same rate and section loss does not correlate with strength loss [Orl2004]. However, the
effect cannot be discounted entirely and may operate in conjunction with flaw growth.
Which of the models is ‘correct’ is still the subject of much debate, but some general conclusions can
be drawn. Fibre corrosion models are less appropriate if Vf is close to the critical value since they rely
on the post-multiple cracking region being realised. Matrix densification models are less appropriate if
the fibre length is much longer than the critical length since pullout will be minimal at any age
(although fibre bridging effects will still come into play). For ductile fibres, in which neither flaws nor
crack bridging will induce significant stress concentration, consideration of basic ACK theory may be
required. Since it is of course highly likely that matrix densification and fibre corrosion act in tandem
to some degree, it would seem appropriate that a unification of such models be sought.
153
temperature and humidity changes. The shrinkage potential of fibre-reinforced cement thin sheets may
exceed considerably the shrinkage of concrete. Typical shrinkage strains for the latter are usually less
than 0.05%, whereas for the thin sheet composites swelling values in the range from 0.10% to almost
2% have been measured [Bau1994]. It should be noted that this is the situation in thin sheet FRC but
not in fibre reinforced concrete where shrinkage is equal or smaller than that of the unreinforced
concrete. The higher swelling in the thin sheet composites is a result of a variety of influences,
including higher cement content than in a typical concrete mix (i.e. less aggregates) as well as the
presence of fibres that are moisture-sensitive, such as cellulose or natural fibres. In addition, the thin
geometry of many of the thin sheet fibre-reinforced cements leads to higher drying rates.
Volume changes induced during natural exposure by wetting and drying may cause internal damage
due to microcracking. This has been demonstrated for wood particle-reinforced cement [Bec1985] in
which aging in water did not lead to strength loss, whereas drying/wetting cycles resulted in loss in
strength, with the magnitude of loss becoming greater for the more extreme drying cycles.
Degradation was the result of internal damage due to volume changes. Damage of this kind might be
expected to be greater for testing under restrained conditions where internal stresses of greater
magnitude are expected to occur as the result of the restrained volume changes. Thus, reduction in
properties and cracking might be observed even for composites where the volume changes are smaller
than in wood particle-reinforced cement [Bec1985].
Dimensional changes can cause problems not connected with loss of material properties. In some
applications of thin sheet FRC such as cladding, dimensional changes may lead to bowing, micro- and
macro-cracking of panels. The extent of the damage will depend on both the material degradation and
the nature of the joints, fixings and underlying structure; this type of long-term performance problem
can be more critical than the changes in the properties of the material with time [For1983, Hans1990,
Wil1985]. Unfortunately, this problem has not received adequate attention in the open literature,
which usually highlights durability issues related to the properties of the material itself. The required
approach is either to study the restrained shrinkage performance of the material itself [Bec1985], or to
test full-scale panels, including their connections to the structure, as recommended in ISO 8336.
Restrained shrinkage of FRC has been studied more intensively in recent years to examine the extent
to which the presence of a small content of fibre reinforcement can reduce shrinkage and related
cracking in concrete [Ban1993, Sar1993, Kov1993]. These studies, although not directly intended for
the evaluation of long-term performance of thin sheet fibre-reinforced cement, provide a basis for
methodology and scientific background needed to address such issues. Standardization of such tests is
currently underway, using configuration similar to that reported by See et al. [See2003].
There do not appear to be any special considerations regarding the durability of trc compared to FRC.
Researchers who are currently assessing the time-dependant behaviour of trc [e.g. Bro2002, Rau2003]
have not as yet identified any degradation mechanisms in trc different to those discussed above for
FRC. However, the nature of woven textiles compared with isolated strands does lead to some
154
possible durability consequences. The points where the strands cross and are thus in contact might
suffer accelerated degradation owing to abrasion under cyclic load, and may also provide an
preferential nucleation site for Ca(OH)2 crystals. Such effects have yet to be observed and may well be
insignificant.
The use of trc as retro-fitted strengthening to structural elements (in much the same way as FRP is
frequently used [e.g. Concrete Society 2003]) has been investigated and progress was recently
summarised by Wiberg [Wib2003]. No direct assessment of the durability of such systems has been
carried out. The possibility exists that such FRC under load would be susceptible to stress corrosion of
the reinforcing fibres. In glass-FRC, available evidence does not support a macro-stress-corrosion
effect, as samples aged in hot water while loaded at up to 80% of failure load did not degrade
appreciably faster than non-loaded samples. In fact, the bond between the structural element and the
retro-fitted FRC is likely to be the least durable aspect, in parallel with the FRP experience [Concrete
Society 2003].
6.3.9 Conclusions
The durability of trc can be inferred from the substantial literature regarding the durability of FRC, as
no textile-specific durability issues have yet been identified. The durability behaviour of FRC,
particularly glass-FRC, is extremely well characterised compared to other modern composites used in
construction. Durability varies substantially between the different fibre types, with polypropylene-
FRC being the most durable and natural-FRC the least. Glass-FRC made with second-generation
fibres is almost as durable as polypropylene-FRC. Carbon-FRC has excellent durability prospects but
long-term weathering data is not yet available. Substantial improvements in durability can be obtained
by using modified cementitious matrices with many fibre types. The mechanisms of strength loss are
varied but relate either to degradation of the fibres or effects related to the continued densification of
the matrix. The relative effect of each mechanism is dependent on the fibre type and this impacts on
the model that should be used to predict time-dependent behaviour. The choice of any accelerated
ageing regime should be carefully made as acceleration methods or factors relevant for one FRC type
should not be used without question for other types.
The durability of components and structural systems (e.g. full cladding panels) has not received
enough attention, as research has tended to focus on material properties. Research also needs to focus
on unifying the various models of time-dependent FRC behaviour.
Comments by Ch. Meyer / G. Vilkner: - these comments should be regarded in text, mayb own section?
We have performed some tests the results of which may be appropriate for this chapter. We leave it to
the Chapter author to decide where the following section is suited best.
Like all aramid fibers, Twaron fibers have basically good chemical resistance, except in environments
with extreme pH values. Because of this, fibers are often provided with protective coatings such as
PVC, polyester or epoxy. Still, concerns exist that the alkalinity of the pore solution in Portland
cement paste can cause deterioration.
When unstressed fibers were stored in concentrated NaOH solution for 24 hours, no loss of stiffness
nor ultimate strength was found afterwards [Vil2003]. Originally, this outcome was attributed to the
protective coating of the fibers. If the fiber is stretched, however, coating and fiber deformations are
not necessarily compatible. As the fiber stretches, the coating may crack, which introduces areas,
where an alkaline solution can attack the aramid surface. A simple relaxation experiment was carried
out to explore the impact of concentrated NaOH solution on the stress level of an aramid roving. A 6-
hour load-time history is plotted in [Vil2003, Fig 2.16]. Note the high primary relaxation rate, caused
by the fact that the fiber was not preconditioned. After 1.5 hours, a straw, that was aligned around the
fiber and sealed at the bottom, was filled with concentrated NaOH solution. The fiber stress level
155
decreased immediately by a small amount, but the effect decayed in time. Loading the fiber after the
relaxation experiment, however, did not show a strength drop [Vil2003 Fig. 2.17]. Also, the typical
increase of fiber stiffness, due to preconditioning, which occurred during the initial loading to the
relaxation target load, was observed [Vil2003].
Please note that a pdf file of Vilkner’s thesis can be downloaded from www.civil.columbia.edu/meyer
as publication no. 75. We can provide reproducible original figures..
REFERENCES
[Ake1989a] S.A.S. Akers, J.B. Studinka (1989a): Ageing behaviour of cellulose fibre cement
composites in natural weathering and accelerated test, International Journal of Cement
Composites and Lightweight Concrete 11;93-97.
[Ake1989b] S.A.S Akers, J.B. Studinka, P. Meier, M.G. Dobb, D.J. Johnson, J. Hikasa (1989b), Long-
term durability of PVA reinforcing fibres in a cement matrix, International Journal of
Cement Composites and Lightweight Concrete 11;79-91.
[Ake1989c] S.A.S. Akers, J.B. Studinka, P. Meier, M.G. Dobb, D.J. Johnson, J. Hikasa (1989c), Long
term durability of PVA reinforcing fibres in a cement matrix, International Journal of
Cement Composites and Lightweight Concrete 11; 79-91.
[Ave1971] J. Aveston, G.A. Cooper, A. Kelly (1971) Single and multiple fracture, in The Properties
of Fibre Composites: Proceedings of an NPL Conference. UK, IPC Science and
Technology Press, pp15-27.
[Bal2003] H. Ball (2003), Durability of naturally aged gfrc mixes containing forton polymer and
SEM analysis of the fracture interface, in Proceedings of the 13th Congress of the Glass
Fibre Reinforced Concrete Association, October 2003, Barcelona, Spain eds. J.N. Clarke
& R. Ferry. UK, Concrete Society, paper 17, 30pp.
[Ban2003] N. Banthia, N. Nandakumar (2003), Crack growth resistance of hybrid fiber reinforced
cement composites, Cement and Concrete Composites 25;3-9.
[Bar1996] P.J.M. Bartos, W. Zhu (1996), Effect of microsilica and acrylic polymer treatment on the
ageing of grc
[Bau1994] H. Baum and A. Bentur (1994), Fiber Reinforced Cementitious Materials for Lightweight
Construction: Development of Criteria and Evaluation of Their Long Term Performance,
research report. National Building Research Institute, Technion, Israel Institute of
Technology, Haifa, Israel, 1994.
[Bec1985] R. Becker and J. Laks (1985), Cracking resistance of asbestos cement panels subjected to
drying, Durability of Building Materials 3;35-49.
[Bed2003] J. Beddows, P. Purnell (2003) Durability of new matrix glassfibre reinforced concrete, in
Proceedings of the 13th Congress of the International Glassfibre Reinforced Concrete
Association,Barcelona, Spain October 2003. Eds N. Clarke, R. Ferry. UK, Concrete
Society, 2003 paper 16 11pp.
156
[Ben1989a] A. Bentur, S.A.S. Akers (1989a), The microstructure and ageing of cellulose fibre
reinforced autoclaved cement composites, International Journal of Cement Composites
and Lightweight Concrete 11; 111-115.
[Ben1989b] A. Bentur, S.A.S. Akers (1989b), The microstructure and ageing of cellulose fibre
reinforced cement composites cured in a normal environment, International Journal of
Cement Composites and Lightweight Concrete 11; 99-109.
[Ben1990] A. Bentur, S. Mindess (1990): Fibre Reinforced Cementitious Composites; UK, Barking,
Elsevier Science Publishers Ltd., 449pp.
[Ber1984] S.G. Bergström, H-E. Gram (1984), Durability of alkali-sensitive fibres in concrete,
International Journal of Cement Composites and Lightweight Concrete 6; 75-80.
[Che1996] P-W. Chen, D.D.L. Chung (1996), Low-drying-shrinkage concrete containing carbon
fibres, Composites: Part B 27B; 268-274.
[Che2003] J. Cheng, W. Liang, Y. Hu, Q. Chen, G.H. Frischat (2003), Development of a new alkali
resistant coating, Journal of Sol-Gel Science and Technology 27;309-313.
[Chu2000] D.D.L. Chung (2000): Cement reinforced with short carbon fibers: a multifunctional
material, Composites Part B: Engineering 31 ; 511-526.
[Cia2001] D. Cian, B. Della Bella (2001), Structural applications of grc for precast floors, in
Proceedings of the 12th International Glassfibre Reinforced Concrete Congress, May
2001, Dublin, Eire, eds. N. Clarke and R. Ferry. UK, Concrete Society pp41-52.
[Col1990] B.J.B. Cole (1990), Properties of CemFIL 2 composites after 10 years weathering – 6th
series April 1990. CemFIL Technology, Research Support Technology Group, Lathom,
UK.
[Concrete Society 2003] The Concrete Society (2003), Strengthening concrete structures with fibre-
composite materials: acceptance, inspection and monitoring, Concrete Society Technical
Report No. 57. UK, Crowthorne, Concrete Society, 49pp.
[Cyr2001] M.F. Cyr, A. Peled, S.P. Shah (2001), Improving performance of glass-fiber-reinforced
extruded composites, in Proceedings of the 12th International Glassfibre Reinforced
Concrete Congress, May 2001, Dublin, Eire, eds. N. Clarke and R. Ferry. UK, Concrete
Society pp163-172.
[Fis2001] A.K. Fisher, F. Bullen, D. Beal (2001), The durability of cellulose fibre reinforced
concrete pipes in sewage applications, Cement and Concrete Research 31; 543-553.
[For1983] M.W. Fordyce and R.G. Wodehouse (1983), GRC and Building. UK, Seven Oaks,
Butterworths, 1983. pp?
[Gao2003] S.L. Gao, E. Mäder, A. Abdaker, P. Offerman (2003), Environmental resistance and
mechanical performance of alkali-resistant fibers with surface sizings, Journal of Non-
Crystalline Solids 325;230-241.
157
[Gar1991] G.C. Gartshore, E. Kempster, A.G. Tallentire (1991), A new high durability cement for
GRC products, in Proceedings of the 8th Biennial Congress of the Glass-fibre Reinforced
Cement Association, Maastrict, Netherlands 1991 pp3-12. UK, GRCA.
[Gil2001] G.T. Gilbert, P.J. Ridd (2001), Continuing premix spray developments in the USA, in
Proceedings of the 12th International Glassfibre Reinforced Concrete Congress, May
2001, Dublin, Eire, eds. N. Clarke and R. Ferry. UK, Concrete Society pp189-196
[Han1991] D.J. Hannant (1991): Durability of cement sheets reinforced with fibrillated
polypropylene networks: 10 year results, Magazine of Concrete Research 43;257-264.
[Han1998] D.J. Hannant (1998): Durability of polypropylene cement composites: 18 years of data,
Cement and ConcreteResearch 28; 1809-1817.
[Hans1990] N.W.Hansen, J.J. Roller, J.I. Daniels, and T.L. Weinman (1990), Manufacture and
installation of gfrc facades, in Thin Section Fiber Reinforced Concrete and Ferrocement,
ACI SP-124 ed. J.I. Daniels and S.P. Shah. American Concrete Institute, Detroit, MI,
1990. pp. 182-213
[Joh2001] C.D. Johnston (2001), Fiber-reinforced Cements and Concretes (Advances in Concrete
Technology Volume 3). Netherlands, Overseas Publishers Association/Gordon and Breach
Science Publishers, 364pp.
[Kat1995] A. Katz, A. Bentur (1995), Effect of matrix composition on the aging of CFRC, Cement
and Concrete Composites 17;87-97.
[Kat1996] A. Katz, A. Bentur (1996), Mechanisms and processes leading to changes in time in the
properties of CFRC, Advanced Cement Based Materials 3;1-13.
[Kim1999] P.J. Kim, H.C. Wu, Z. Lin, V.C. Li, B. deLhoneuk, S.A.S Akers (1999),
Micromechanics-based study of cellulose cement in flexure, Cement and Concrete
Research 29; 201-208.
[Kov1993] K. Kovler, J. Sikuler, A. Bentur (1993), Restrained shrinkage tests of fibre reinforced
concrete ring specimens: effect of core thermal expansion, Materials and Structures
26;231-237.
158
[Lit1981] K.L. Litherland, D.R. Oakley, B.A. Proctor (1981): The use of accelerated ageing
procedures to predict the long-term strength of grc composites, Cement and Concrete
Research 11;455-466.
[Ma2005] Y. Ma, B. Zhu, M. Tan (2005), Properties of ceramic fiber reinforced cement composites,
Cement and Concrete Research 35;296-300.
[Mac1999] R. MacVicar, L.M. Matuana, J.J. Balatinecz (1999), Aging mechanisms in cellulose fiber
reinforced cement composites, Cement and Concrete Composites 21; 189-196.
[Maj1991] A.J. Majumdar and V. Laws (1991), Glass Fibre Reinforced Cement. BSP Professional
Books, Oxford, 197pp.
[Maj1992] A.J. Majumdar (1992) Fibre reinforced cement – thin sheets, in Proceedings of the 9th
International Congress on the Chemistry of Cement, New Delhi, India, ed. National
Council for Cement & Building Materials. New Delhi, NCB, 1993 pp737-774.
[Mar1997] S. Marikunte, C. Aldea, S.P. Shah (1997) Durability of glass fiber reinforced composites:
effect of silica fume and metakaolin, Advanced Cement Based Materials 5:100-108.
[Mob1995] B. Mobasher, C.Y. Li (1995) Modeling of stiffness degradation of the interfacial zone
during debonding, Composites Engineering 5;1349-1365.
[Moh2005] B.J. Mohr, H. Nanko, K.E. Kurtis (2005), Durability of kraft pulp fiber-cement
composites to wet/dry cycling, Cement & Concrete Composites 27; 435-448.
[Mol1995] H. Molloy, T. Harmon, J. Jones, H. Sone (1995), Thin concrete panels produced with AR
glass chopped strand and scrim, in Proceedings of the Glassfibre Reinforced Cement
Association 10th Biennial Congres, October 1995, Strasbourg, France. UK, GRCA, paper
1/7 ppI-X.
[Naa1979] A.E. Naaman, S.P. Shah (1979), Fracture and multiple cracking of cementitious
composites, in Proceedings of the Twelfth ASTM National Symposium on Fracture
Mechanics, ed. S.W. Frieman. USA, Pa., ASTM Special Technical Publication 678
pp183-201.
[Orl2004] J. Orlowsky, P. Purnell (2004), private communications October 2003 - May 2004.
[Pac2001] U. Pachow (2001), New manufacturing technologies for grc, in Proceedings of the 12th
International Glassfibre Reinforced Concrete Congress, May 2001, Dublin, Eire, eds. N.
Clarke and R. Ferry. UK, Concrete Society pp197-202.
[Pro1980] B.A. Proctor, B. Yale (1980), Glass fibres for cement reinforcement, Philosophical
Transactions of the Royal Society London A294; 427-436.
[Pro1982] B.A. Proctor, D.R. Oakley, K.L. Litherland (1982): Developments in the assessment and
performance of grc over 10 years, Composites 13;173.
159
[Pue2003] F. Puertas, T. Amat, A. Fernández-Jimenez, T. Vásquez (2003): Mechanical and durable
behaviour of alkaline cement mortars reinforced with polypropylene fibres, Cement and
Concrete Research 33;2031-2036.
[Pur2003] J. Beddows, P. Purnell, (2003), Durability of new matrix glassfibre reinforced concrete,
in Proceedings of the 13th Congress of the Glass Fibre Reinforced Concrete Association,
October 2003, Barcelona, Spain eds. J.N. Clarke & R. Ferry. UK, Concrete Society paper
16, 11pp.
[Pur2005] P. Purnell, J. Beddows (2005), Durability and simulated ageing of new matrix glass fibre
reinforced concrete, Cement and Concrete Composites in press accepted March 2005.
[Pur1998] P. Purnell (1998), The Durability of Glass Fibre Reinforced Cement made with New
Cementitious Matrices, PhD Thesis, Aston University, UK, 1998.
[Pur1999] P. Purnell, N.R. Short, C.L. Page, A.J. Majumdar, P.L. Walton (1999) Accelerated ageing
characteristics of glass-fibre reinforced cement made with new cementitious matrices,
Composites: Part A 30;1073-1080.
[Pur2000] P. Purnell, N.R. Short, C.L. Page, A.J. Majumdar (2000) Microstructural observations in
new matrix glass fibre reinforced cement, Cement and Concrete Research 30;1747-1753.
[Pur2001a] P. Purnell, N.R. Short, C.L. Page (2001a): A static fatigue model for the durability of
glass-fibre reinforced cement, Journal of Materials Science 36 5385-5390.
[Pur2001b] P. Purnell, N.R. Short, C.L. Page (2001b) Supercritical carbonation of glass-fibre
reinforced cement Part I: Mechanical testing and chemical analysis, Composites A
32;1777-1787.
[Pur2003a] P. Purnell, A.M.G. Seneviratne, N.R. Short, C.L. Page (2003b), Supercritical carbonation
of glass-fibre reinforced cement Part II: Microstructural observations, Composites A
34;1105-1112.
[Qi2003] C. Qi, B. Tianyou (2003), A review of the development of the grc industry in China, in
Proceedings of the 13th Congress of the Glass Fibre Reinforced Concrete Association,
October 2003, Barcelona, Spain eds. J.N. Clarke & R. Ferry. UK, Concrete Society paper
37, 11pp.
[Qia2003] X. Qian, B. Shen, B. Mu, Z. Li (2003) Enhancement of aging resistance of glass fiber
reinforced cement, Materials and Structures 36;323-329.
[Sar1993] M. Sarigapbuti, S.P. Shah, K.D. Vinson (1993), Shrinkage cracking and durability
characteristics of cellulose fiber reinforced concrete, ACI Materials Journal 90;309-18.
160
[See2003] T.H. See, E.K. Attiogbe, M.A. Miltenberger (2003) Shrinkage cracking of concrete using
ring specimens, ACI Materials Journal 100;239-245.
[Sen2002] A.M.G. Seneviratne, N.R. Short, P. Purnell, C.L. Page (2002), Dimensional stability of
super-critically carbonated glass-fibre reinforced concrete, Cement & Concrete Research
32;1639-1644.
[Sou1991] P. Soukatchoff, P.J. Ridd (1991), High durability glass fibre reinforced modified
cementitious matrix, in Proceedings of the 8th Biennial Congress of the Glass-fibre
Reinforced Cement Association, Maastrict, Netherlands 1991 pp37-44. UK, GRCA.
[Tak1999] Y. Takeuchi, M. Hayashi, T. Tamaki, H. Tanaka (1999), Glass fibre reinforced concrete
using super low contractile admixture, in Proceedings of the 2nd Asia-Pacific Special
Conference on Fiber-Reinforced Concrete, Singapore, August 1999 eds. T.S. Lok, K.
Tseng.
[Tol2000] R.D. Tolêdo Filho, K. Scrivener, G.L. England, K. Ghavami (2000): Durability of alkali-
sensitive sisal and coconut fibres in cement mortar composites; Cement and Concrete
Composites 22; 127-143.
[Tol2003] R.D. Tolêdo Filho, K. Ghavami, G.L. England, K. Scrivener (2003), Development of
vegetable fibre-mortar composites of improved durability, Cement and Concrete
Composites 25; 185-196.
[Vil2003] G. Vilkner (2003), Glass Concrete Thin Sheets Reinforced with Prestressed Aramid
Fabrics, Ph.D. Dissertation, Columbia University, New York, NY, 2003.
[Wal1978] P.L Walton, A.J. Majumdar (1978) Properties of cement composites reinforced with
Kevlar fibres, Journal of Materials Science 13;1075-1083.
[Wib2003] A. Wiberg (2003), Strengthening of Concrete Beams using Cementitious Carbon Fibre
Composites, Doctoral Thesis, Royal Institute of Technology, Stockholm, Sweden.
(TRITA-BKN Bulletin 72, 2003).
[Wil1985] G.R. Williamson (1985), Evaluation of glass fiber reinforced concrete panels for use in
military construction, in Durability of Glass Fibre Reinforced Concrete, proceedings of
international symposium, ed. S. Diamond. Prestressed Concrete Institute, Chicago, 1985.
pp. 54-63
[Xu1998] G. Xu, S. Magnani, D.J. Hannant (1998): Durability of hybrid polypropylene-glass fibre
cement corrugated sheets, Cement and Concrete Composites 20;79-84.
[Zhu1997] W. Zhu, P.J.M. Bartos (1997), Assessment of interfacial microstructure and bond
properties in aged grc using a novel micro-indentation method, Cement and Concrete
Research 27;1701-1711.
161
6.4 Fire Resistance (M. Krüger)
Remark: This is a first draft of the chapter “Fire resistance” for the state of the art report on textile
reinforced concrete. Due to only a few research results only a short overview could be given. The
authors have only knowledge of tests conducted at the University of Stuttgart. If anyone knows certain
investigations concerning fire resistance of textile reinforced concrete please let us know.
M. Krüger, H.W. Reinhardt, Institute of Building Materials Research at Stuttgart University (IWB),
Germany
6.4.1 Introduction
Textile reinforced concrete becomes comparative if there is the demand for lightweight or exposed
concrete elements of high quality. The needed concrete cover of textile reinforced elements is very
small so thin-walled elements could be produced. However, if thin-walled textile reinforced concrete
will be used e.g. for facade elements or other structural parts, fire resistance and fire performance
becomes of importance. In particular the temperature and the rise of temperature in such elements are
high in case of fire that needs to be investigated. On the one hand the fire performance of the textile
materials itself is of interest. On the other hand the fire performance of the fine grained concrete as
well as of the textile reinforced composite has to be examined. Thus this chapter will give a short
overview about some aspects regarding fire resistance and fire behaviour of textile reinforced concrete
elements and the fibres used for textile reinforced elements.
The knowledge of the fire resistance of textile reinforced elements is still quite small because only a
few test results exist up to now. Most fire tests given in the literature were conducted at FRP-elements
or FRP-wrapped concrete elements [Bis2004]. Some fire tests conducted at concrete elements only
reinforced with textiles were enforced at the University of Stuttgart for instance [DBV212, DBV229],
[Iak2003, Rei2000]. The research was focussed on the fire behaviour without loading and the
remaining load bearing capacity after enforcing fire tests.
There were diverse rules and standards like the German DIN 4102 in which requirements for the fire
resistance of structural elements were regularised. Some of these regulations identify a remaining load
bearing capacity after a fire test, the limitation of deformations and the load bearing capacity during a
fire test, or are aiming at thermal conductivity to anticipate fire spread. Especially for the load bearing
capacity of textile reinforced elements during a fire test only a few test results exist. Therefore some
test results of instationary fire tests are presented in this chapter.
162
temperatures greater than 400°C. However, the tensile behaviour of glass fibres at high temperatures is
comparable to the temperature resistance of steel reinforcement. The temperature resistance of most
used types of aramid fibres (see Aromat. Polyamid in Fig. 6.74) is less than that of glass fibres and a
decomposition temperature is usually given to 400°C. Though there were some special Aramid fibres
types developed to obtain better temperature resistance most Aramid fibres will fail quickly in case of
combustion.
The best performance at high temperatures is achieved by carbon fibres. Different types of carbon
fibres, e.g. high modulus or high strength fibres, are commercially available. These different types
were carbonized and then graphitized at special temperatures of about 2000°C up to 3000°C that
results also in different temperature resistance. However, usual carbon fibres show only little decrease
in tensile strength at temperatures of up to 900°C.
The loss of strength, however, has different reasons. All kind of glass fibres melt at high temperatures
whereas carbon, aramid and other polymeric fibres decompose and/or carbonize at high temperatures.
The deterioration of the microstructure of conventional concrete due to micro cracking increases
rapidly at temperatures of about 450°C, whereas micro cracking already starts at about 150°C
[Kor1999]. However, fine grained concrete used for textile reinforced concrete elements has very low
pore volume, high binder content and small aggregate size. Hence micro cracking is expected to be
163
minor. However, the structure of hydrated cement changes with increasing temperature, because
physical and chemical linked water will be yielded. This process still starts at about 100°C and the free
water in the concrete will transform into vapour. If the permeability of the concrete does not allow the
water exhausting, vapour pressure will occur. It is known that concrete spalling occurs when vapour
pressure becomes higher than the tensile strength of the concrete [Mey1972].
Table 6.9. Test specimen and test parameter for fire tests.
Load bearing Constant load Specimen
Name Textile (2 layers) capacity from during fire age at fire
bending test test test
7-03 MAG-7-03 (AR Glass) 10 kN 3.2 kN 110 d
5-03-1 MAG-5-03-1 (Carbon) 12 kN 4.2 kN 112 d
5-03-2 MAG-5-03-2 (Carbon) 12 kN 4.2 kN 110 d
6-03-B01 MAG-6-03-B01 (Carbon, SBR coated) 15 kN 5.0 kN 106 d
Mainly the tensioned zone of the test specimens were exposed to fire according to the ISO
temperature-time curve (ETK) given by DIN 4102. A maximum burning length of 90 minutes was
considered for the tests. During the fire tests the test specimens were loaded with a constant load (see
Fig. 6.76). The load was calculated to approximately one third of the load bearing capacity that was
determined at bending tests enforced separately at equivalent specimens. During the fire test the
temperature at the surface of the web as well as the deflection in the third points were measured with
LVDTs.
164
Textile
Concrete
12
T3
5,6
120
84
T2
18
T1
5,6
12
110
[mm]
Fig. 6.75. Cross section of the test setup with the test specimen partly exposed to fire.
30 30 30
[c m ]
10 0
C o n s ta n t L oa din g
D e f o rm a t io n m e a s u re m e n t
C o m b u s t io n C h a m b e r
Temperaturemeasurement
C a. 5 6
165
determined. Fig. 6.77 shows the temperatures measured during the fire test and in Fig. 6.78 the change
of deflection of the AR glass reinforced specimen is plotted against the burning length. It can be seen
that the specimen failed after 44 minutes. During the test the deflection increased due to the
temperature induced lengthening of the tensioned area of the element and the different temperature
distribution (see Fig. 6.77).
1000
temperature, °C
900 ETK
800 T1
T2
700 T3
600
500
400
300
200
100
0
0 10 20 30 40 50 60 70 80 90
burning length, min.
Fig. 6.77. Temperature plotted against the burning length (7-03, AR glass).
10
deflection, mm
0
0 10 20 30 40 50 60 70 80 90
burning length, min.
Fig. 6.78. Change of deflection plotted against the burning length (7-03, AR glass).
Fig. 6.79 shows the AR glass reinforced test specimen after the fire test. The specimen failed abruptly
and a large crack occurred. However, a lot of small cracks were visible at the surface area that was
exposed to fire. At the fractured surface it could be observed that the AR glass textile has melted in the
lower part. In the upper part of the specimen it was almost not affected. The softening temperature of
AR glass is about 800°C to 850°C [Cem-FIL]. However, if the temperature resistance of glass fibres
presented in Fig. 6.74 is considered, a large decrease of strength becomes apparent at temperatures of
about 400°C. It can be seen from Fig. 6.77 that the temperature in the lower web is greater than 400°C
already after 25 minutes causing significant strength reduction. As a conclusion the fire behaviour of
the AR glass fibres is crucial for the specimen failure.
166
Fig. 6.79. Specimen after fire test (7-03, AR glass).
At total of three fire tests with carbon reinforced elements were conducted. The results of these tests
are shown in Fig. 6.80 for the plain carbon fabrics and in Fig. 6.81 for the SBR-coated fabric
compared to test specimen 5-03-1. The test specimens 5-03-1 and 5-03-2 were reinforced with the
same carbon textile. Nevertheless the fire behaviour of the two specimens is different, which will be
described in the following.
Test specimen 5-03-1 has shown an increase in deflection in the first 25 minutes and then a decrease in
deflection until the 60th minute. After that a large crack occurred and the deflection increased rapidly
until failure at the 75th minute. The axial coefficient of thermal expansion of carbon at room
temperature is given to -0.4⋅10-6K-1. That means that the carbon fibre gets shorter if it is heated.
However, the coefficient of thermal expansion of high strength fine grained concrete is about 14 to
20⋅10-6K-1 [Bor2002, Kor1999]. This has lead to different thermal expansion of the carbon fibres and
the concrete during the fire test. At the beginning of the fire test the concrete elongation has
overbalanced the fibre strain. After 20 minutes this situation has changed and the fibre strain has
overbalanced the concrete strain that has lead to a decrease in deflection. Due to the different
coefficient of thermal extension compressive stresses in concrete and tension in the carbon textile
occurred. At a certain stress level, which was reached after about 60 minutes, the strength of the
carbon fibre was exceeded resulting in successive failure of the filaments and then to the collapse of
the whole element.
10
deflection, mm
2
failur eof test specimen after
1 75 minutes (5-03-1)
spalling of the concrete cover (5-03-2)
0
0 10 20 30 40 50 60 70 80 90
burning length, min.
Fig. 6.80. Change of deflection plotted against the burning length (5-03-1 compared to 5-03-2).
The fire behaviour of test specimen 5-03-2 was different; hence the same type of textile and concrete
was used. During the fire test a successive spalling of the concrete cover was recognized beginning at
the 12th minute. The deflection increased during the whole burning length. The test was stopped after
60 minutes, because most of the concrete of the lower flange spalled off (see. Fig. 6.81). Although the
167
fibres were exposed directly to fire after 12 minutes the strength of the carbon fibres was high enough
to carry the load until the 60th minute. The deflection measurement was continued until the 75th minute
showing that the specimen still was able to carry the load.
Fig. 6.81. Bottom view of specimen after fire test (5-03-2, Carbon).
Fig. 6.82 shows the test results of the specimen reinforced with the SBR coated carbon fabric (6-03
B01) in comparison to the test specimen 5-03-1. The deflection increased rapidly at the beginning of
the test supposed to be caused by the weak bond between the SBR coated carbon and the concrete.
After that the deflection increased more and more until test stop after 54 minutes. The element has
shown no concrete spalling. But it was recognised that the bond between the textile and the concrete
failed resulting in a fibre pullout at a length of approximately 30 cm. The bond failure has lead to the
large increase of deflection after 54 minutes. The failure of the bond is due to the melting and
decomposition of the SBR coating. SBR is a thermoplastic that softens at about 90°C and decomposes
at about 250°C. The measured temperature curves at the bar of specimen 6-03 B01 were similar to the
measured curves at specimen 7-03 in Fig. 6.77 thus the graph is not shown here. However, if the
temperatures of Fig. 6.77 were considered, it is obvious that the melting point of the SBR was reached
after about 5 minutes and the decomposing point at about 15 minutes.
10
deflection, mm
4 Failure due to
debonding
3
0
0 10 20 30 40 50 60 70 80 90
burning length, min.
Fig. 6.82. Change of deflection plotted against the burning length (5-03-1 compared to 6-03-B01,
Carbon).
168
elements have a thin-walled design. Therefore heat conduction becomes very important to anticipate
fire spread as well as that the heat of the element quickly becomes very high. However, heat
conduction of textile reinforced elements is low compared to structural elements e.g. made of steel, so
a proper design of the elements will be up to the standard. With respect to the structural design ainly
three aspects have to be accounted for. First of all one has to choose a sufficient textile. AR glass
textiles show similar fire behaviour compared to steel, but due to the very high temperature of the thin-
walled elements in case of fire tensile strength of AR glass will decrease rapidly. This is also true for
textiles made out of aramid that also have a low decomposition temperature of about 300°C to 400°C.
Carbon fibres show comparatively very good fire behaviour and in particular a high strength at high
temperatures. However, the coefficient of thermal is negative that could lead to thermal constrained
stresses. Therefore different failure modes could occur which is fibre failure due to exceeding the
ultimate tensile strain as well as concrete crushing due to compressive stresses or spalling of the
concrete cover due to transverse tensional stresses.
The use of coatings like SBR can influence the fire behaviour of textile reinforced elements, because
their fire resistance usually is less than of the concrete or the textile material. In the shown test
thermoplastics like SBR were used as a coating that leads to rapid bond failure if heated. Although
resins like epoxy show higher temperature resistance, the decomposition temperature is low compared
to the temperature that occurs in a thin-walled element in case of fire. However, there are some special
resins commercially available, which are optimized for high temperatures of up to 400°C or even
more. Such resins might provide better fire behaviour, but further tests are needed.
Concrete spalling seems to be a general problem mainly caused by vapor pressure that can occur in a
case of fire. Concrete spalling depends on the pore volume, the matrix composition and other matrix
characteristics. Even so, this problem also has to be further investigated for thin-walled elements.
6.4.5 Conclusions
The tests have shown that thin-walled textile reinforced elements show a quick temperature increase if
exposed to fire. Therefore the fire resistance of a textile reinforced element is primarily affected by the
fire behaviour of the textile material. Tests have shown a fire resistance period of AR glass reinforced
elements of up to 45 minutes. It is supposed that the fire resistance period of AR glass reinforced
elements could not be greatly enlarged if low concrete cover is considered, because the temperature of
the textile increases rapidly. The same is true if aramid textiles were used, because the fire resistance
of Aramid is also poor. Textile reinforcement made out of carbon could withstand very high
temperatures, so such fibres achieve the potential for excellent fire resistance if used as reinforcement.
In one test a fire resistance period of 75 minutes was achieved, which is quite a long period for a thin-
walled structural element. The collapse of the test specimen was caused by carbon fibre failure due the
different coefficient of thermal expansion compared to concrete resulting in high thermal fibre stresses
in case of fire. It is suggested that an optimization of the reinforcement ratio could further improve the
fire resistance.
Resins or coatings could improve the tensile strength of the textile as well as the bond between textile
and concrete. But it has to be kept in mind that the decomposition temperature of most resins is even
low and therefore bond failure will occur at comparatively low temperatures leading to element
failure. However, if the resins will be further optimized, it is assumed that the fire behaviour also
could become better compared to concrete elements reinforced with plain fabrics.
169
REFERENCES
[Bis2004] Bisby, L.A., Kodur, V.K.R., Green, M.F.: Numerical parametric studies on the fire
endurance of fibre-reinforced-polymer-confined concrete columns. Canadian Journal of
Civil Engineering, V 31, No. 6, Dec. 2004, pp. 1090-1100, http://irc.nrc-
cnrc.gc.ca/ircpubs.
[Bor2002] Bornemann, R., Schmidt, M., Vellmer, C.: Brandverhalten ultra-hochfester Betone In:
Beton 50 (2002), H. 9, S. 418 – 422.
[Cem-FIL] http://www.cem-filreinforcements.com/pdf/tech%20data.pdf
[Kle2004] Kleineberg, M., Herbeck, L., Brosinger, A.: CFRP APU INTAKE DUCT for
MEGALINER. SAMPE Europe Conference & Exhibition, 2004.
[Kor1999] Kordina, K., Meyer-Ottens, C.: Beton Brandschutz Handbuch (2. Auflage). Verlag
Bau+Technik, 1999.
[Mey1972] Meyer-Ottens, C.: Zur Frage der Abplatzungen an Betonbauteilen aus Normalbeton bei
Brandbeanspruchung. Heft 23 des iBMB der TU Braunschweig, 1972.
[Rei2000] Reinhardt, H.W.: Integral Formwork Panels Made of GFRC. In: High-Performance Fiber-
Reinforced Concrete Thin Sheet Products. ACI SP-190, 2000, S. 77-85.
170
7 TRC FOR REHABILITATION (M. CURBACH)
Working party “Bond old-new”.
Manfred Curbach, Institute of Concrete Structures at Technische Universität Dresden, Germany
ABSTRACT: In this chapter the bond between old concrete and new layers made of textile reinforced
concrete is discussed. – abstract is missing
7.1 Introduction
If old concrete members are strengthened with a new layer, the bond between the old concrete
and new applicated strengthening layer is of special interest. The transfer of the bond forces from one
layer to the other through the bonding joint has to be guaranteed to ensure a full composite action of
the strengthened RC-member.
A loss of composite action leads to minor than the maximum reachable ultimate loads at full
composite action. Therefore it is necessary to know the weak points of the bond to avoid an anchoring
failure. Two Examples of such weak points are a subsequently applied flexural strengthening layer of
slabs, which are installed in the structure, which ends in front of the supports (Fig. 7.1, top) and a
shear strengthening layer of T-beams, which does not reach the compression zone (Fig. 7.1, right).
Fig. 7.1. Weak points of the bond anchoring at differently strengthened RC-members.
The questions regarding force transfer mechanisms between the textile-reinforced strengthening layer
and the old concrete can be investigated using a separate bond test. The bond tests give information
about the relationship between shear loading and deformation, as well as the necessary bond length
and the transferable ultimate bond forces. Details of such a test set-up are described in [Ort2003].
There is no comparable research into the bond between old concrete and a textile reinforced fine
grained concrete layer found in other literature beside the research at the University in Dresden.
171
7.2.1 Bond failure modes in general
A bond failure of flexural strengthened RC-members can generally start at different points. Depending
on the position, where the failure starts, the following failure modes can be distinguished (Fig. 7.2).
The bond of shear strengthened RC-members may fail at the same failure modes as mentioned above.
Only the failure mode I mostly starts at the upper end of the strengthened beam web (Fig. 7.1, right)
172
Fig. 7.3. Theoretically possible layers of bond failure between old and new layer.
If a correct sandblasting of the old concrete’s surface is guaranteed, only the failure layers a) and c)
are possible. In this case, the failure layer inside the new-old concrete joint is avoided by the strong
interlocking between the old concrete and the strengthening. More detailed information concerning the
failure layers are given in [Ort2004].
A fourth bond failure mode can occur at the anchoring range of a textile reinforced concrete
strengthening to the old concrete: a pullout of fibres from the finegrade concrete layer (Fig. 7.4). This
mode of failure occurs primarily if fabrics made of very thick multifilament yarns are used. The bond
of these thick yarns and the surrounding matrix can only be activated across a relatively large fibre
length. But this mode of failure does not directly concern a bond failure between the old and new
layer, rather a failure of the inner bond of the fibre bundle inside the finegrade concrete matrix.
Fig. 7.4. Inner bonding failure inside the new concrete layer.
As a conclusion, the bond between textile reinforced concrete strengthening layers and the old
concrete may fail by failure modes that are not known from traditional adhesive bonded strengthening
material, such as steel plates or FRP-strips. All these possible failure modes must be kept in mind
when developing a design model to draw the outline of the failure.
173
7.3 Load carrying capacity of the bond
The load carrying capacity of the bond is dependent on the load carrying capacity of the individual
layers, that can cause a failure. The influence of these tree possible layers for the bond failure between
the old concrete and the TRC-strengthening layer will be shown next.
Fig. 7.5. Influence of the matrix area on the load carrying capacity taken from adhesive tensile
tests.
174
7.3.3 Influence of the properties of the old concrete
The properties of the old concrete cannot be influenced by the TRC-strengthening, because they
depend on the behaviour of the existing concrete member, that has to be strengthened. During his
research on the bond of steel plates, Holzenkämpfer found a dependency of the deformability of the
old concrete on the maximum aggregate diameter [Hol1994]. A larger maximum aggregate diameter
of the old concrete member leads to a bigger interlocking effect between the aggregates though a
developing bond crack. Therefore, a higher bond stress can be transferred to such an concrete member.
Because of the above mentioned facts, the well known assumption of constant strain across the
thickness of the strengthening layer and slip occurring within the bonding joint, as with adhesive
bonded steel plates or CFRP strips, cannot be confirmed. The cracking has an enormous influence on
the deformation or slip in the bond length region. The interrelations of the cracking and deformation of
the bond are described in [Ort2003, Ort2004] in more depth.
175
7.5 Bond models
Currently, the developed adhesive bonding laws for additional strengthening with steel plates or CFRP
strips hold only for the following three important assumptions:
1. In the basic model for bonding laws, an adhesive bond is assumed. The bond model is
developed from the selected bond law by solution of the differential equation of bond slip.
2. The strain across the thickness of the steel plate or CFRP-strip, respectively, is assumed to be
constant. According to this, the slip only appears in the adhesive joint which functions as the
interface between the old concrete and the adhesive bonded strengthening layer (Fig. 7.6, top).
This assumption is based on the fact that the Young’s modulus of the steel plates with
ES = 210,000 N/mm² or CFRP-strips with ECFRP ≈ 75,000 N/mm² [Biz1999] is much higher
than that of the adhesive layer with EA ≈ 6,000–11,000 N/mm² [Han1996].
3. The shear stresses for the reinforcing material are calculated by using the difference between
the measured strains and their corresponding stress-strain relationships. Therefore it assumes
that the strain across the plate length continually increases.
In conclusion, the existing bond models for additional strengthening are basically adhesive bond laws
with the assumption of slip e. g. in the adhesive joint. Because the special material properties of the
TRC-strengthening (see section 7.4) this models cannot be applied to the bond of this material.
7.5.2 Strut-and-tie-model
The forces inside the bond region can be described with a simplified strut-and-tie-model (Fig. 7.7).
This model allows to establish the interrelations between the tension force in the TRC-strengthening
layer FL and the resulting peel-off force in perpendicular direction Ti, which can easily be analyzed by
adhesive tensile tests.
176
7.6 Conclusions
The bond between TRC-strengthening layers and the old concrete may fail by failure modes
that are not known from traditional adhesive bonded steel plates or FRP-strips. All these possible
failure modes must be kept in mind when developing a design model to draw the outline of the failure.
The existing bond models for additional strengthening are basically adhesive bond laws with the
assumption of slip in the adhesive joint, but because the special material properties of the TRC-
strengthening this models cannot be applied to the bond of TRC-strengthening layers. A design model
for the bond of TRC-strengthening layers is currently not yet available.
REFERENCES
[Biz1999] Bizindavyi, L; Neale, K. W.: Transfer lengths and bond strengths for composites bonded
to concrete. In: Journal of composites for construction 3 (1999), v. 4, p. 153–160
[Bla1998] Blaschko, M,, Niedermeier, R. and Zilch, K., ‘Bond failure modes of flexural members
strengthened with FRP’, in Proceedings of the Second International Conference on
Composites in Infrastructures, Tucson, 1998 (Saadatmanesh, H. and Ehsani, M. R., 1998)
315–327.
[Hol1994] Holzenkämpfer, P.: Ingenieurmodelle des Verbunds geklebter Bewehrung für Beton-
bauteile. Braunschweig, Technische Universität, Fakultät Bauingenieurwesen, iBMB,
Diss., 1994
[Ort2003] Ortlepp, R. and Curbach, M., ‘Bonding Behaviour of Textile Reinforced Concrete
Strengthening’, in ‘High Performance Fiber Reinforced Cement Composites –
HPFRCC 4’, Proceedings of the Fourth International RILEM Workshop, Ann Arbor,
June 2003 (Naaman, A. E. and Reinhardt, H.-W., RILEM Proceedings PRO 30, Bagneux,
2003) 517–527.
[Ort2004] Ortlepp, R., Ortlepp, S. and Curbach, M., ‘Stress Transfer in the Bond Joint of
Subsequently Applied Textile Reinforced Concrete Strengthening’, in ‘Sixth RILEM-
Symposium on Fibre Reinforced Concrete (FRC) – BEFIB 2004’, Proceedings of the
Sixth RILEM-Symposium on Fibre Reinforced Concrete (FRC), Varenna-Lecco,
September 2004 (Editor, N1. and Editor, N2., RILEM Proceedings PRO xx, Verlag,
2005) xxx–xxx.
[Ran1997] Randl, N.: Untersuchungen zur Kraftübertragung zwischen Alt- und Neubeton bei
unterschiedlichen Fugenrauhigkeiten. Innsbruck, Universität, Fakultät für Bauingenieur-
wesen und Architektur, Diss., 1997
[Ula2003] Ulaga, T.: Betonbauteile mit Stab- und Lamellenbewehrung: Verbund- und
Zuggliedmodellierung. Zürich, ETH, Institut für Baustatik und Konstruktion, Diss., 2003
177
8 APPLICATIONS OF TEXTILE REINFORCED CONCRETE
Josef Hegger, Norbert Will, Institute of Structural Concrete, Aachen University, Germany
8.1 Introduction
In the architecture of the last two decades can be observed among other things, that materials
and her natural surface conditions are in the field of vision of the architects. Under these aspects the
development of the textile-reinforced concrete shows a significant technological and creative
contribution in the civil engineering. This new composite material allows by the production of filigree
concrete elements a completely new architectural appearance of the building material concrete. High-
quality exposed concrete surfaces with sharp edged appearance can be realized with the new material.
The reinforcement of concrete with technical textiles extends its application to completely new fields.
Because of the corrosion resistance of the textile materials, thick concrete covers as known in ordinary
reinforced concrete are no longer needed. Thus, slender structural members with a wall thickness as
low as ten millimetres are possible. In addition, fine grain concrete matrices guarantee an even and
sharp-edged high quality surface, so that TRC is predestinated for architectural applications. Although
the knowledge about the load bearing behaviour of TRC is still limited, applications such as cladding
panels [Heg2001b] and an integrated framework system [Bra2003] have already been implemented.
Textile-reinforced concrete offers already today a wide spectrum of applications, either as an
alternative to customary building materials or on account of his favourable characteristics for entirely
new ranges of application. In the following some of these projects are introduced.
Cladding panels made of TRC are used for the first time for the extension building of the Institute of
Structural Concrete, Aachen University. The existing hall with a span of 12.0 m has been extended by
four axes of 5.4 m each, resulting in an additional usable floor space of 21.6 m x 12.0 m. On the
longitudinal side of the hall, curtain wall panels are used instead of hitherto natural stone, which would
have been the typical choice. The upper part of the facade is cladded with panels. The design of the
curtain wall panels is shown in Fig. 8.3. At this time the sun-protection in Fig. 8.3 b is only added by a
computer. It will consist of textile reinforced concrete lamellae and will be attached during the next
time.
Textile reinforced cladding panels have been used for the extension building shown in Fig. 8.1 of the
Institute of Structural Concrete, RWTH Aachen University. The existing single-aisle hall with a span
of 12.0 m has been extended by four axes of 5.4 m spacing each. Curtain wall elements were used
178
except for the lower part (socket) of the building, where sandwich plates of 35 mm thick facing shells
were placed (Fig. 8.2).
Fig. 8.1. New extension of the testing hall of the Structural Concrete Institute.
sandwich elements
with facing shell of
textile concrete
facing shell of textile-
structural reinforced- reinforced concrete
concrete shell
loadbearing
structure
thermal insulation air gap
thermal insulation
Fig. 8.2. Sandwich elements and cladding panels out of textile reinforced concrete.
Innovative, textile reinforced concrete components have been developed for this purpose. On the
longitudinal side of the hall, 2685 x 325 x 25 mm curtain wall panels as shown in Fig. 8.3 have been
applied instead of hitherto natural stone, which would have been the typical choice. The high cost of
the natural stone and its manufacture restricts its use to high quality administrative buildings. Textile
reinforced cladding panels are notably less expensive and are therefore a cost efficient alternative for
residential and commercial structures.
179
a) View b) Detail c) Section
Fig. 8.3. Curtain wall construction of the Structural Concrete Institute.
Using fine-grained concrete with a maximum particle size of approximately 1.0 to 2.0 mm, smooth
surface finishes as well as sharp-edged profiles and joints could be achieved. The results are concrete
surfaces with a completely new look.
The design and calculation of the required textile reinforcement is based on the information from the
knowledge of the Collaborative Research Center 532 in Aachen [Heg2001]. The dimensions and
reinforcement is shown in Fig. 8.4.
50 168,5 50
7 18,5 7
y
32,5
x
2,5
268,5 [cm]
a) Dimensions and support conditions of panels (in cm).
50 168,5 50
7 18,5 7
32,5
268,5
20 20 [cm]
laminated alkali resistant
vertical extra layer glass fiber fabrics vertical extra layer
MAG 04-01
0°-direction horizontal
90°-direction vertical
180
b) Arrangement of textile reinforcement.
Fig. 8.4. Dimensions and reinforcement of textile reinforced concrete panels used.
The dimensions of panels are 268.5 x 32.5 x 2.5 cm³. The reinforcement layer in longitudinal direction
ranges about 4 mm from the surface of the panel. This leaves a concrete cover of at least 3 mm. In the
horizontal direction the panels are clamped at four points. In the vertical direction they are supported
by two thrust bearings, the other bearings only admit horizontal forces. Thus, no stresses due to
temperature changes are generated by the support conditions. The facade is subjected to wind load.
The design of the panels was made under the condition of no cracking under service loads.
Considering the results of the described experimental investigations a coated alkali resistant glass fiber
fabric (2200 tex and 8.33 mm mesh size in longitudinal direction, and 320 tex and 8.4 mesh size in the
transverse direction) has been chosen for the reinforcement of the panels. The coating doubles the
effectiveness k0 of the uncoated fabric, thus, k0 increased to 80 %. In addititon, using a coated fabric
leads to a better manageability because of the improved stability of the fabric structure. The fabric
having a maximum tensile strength σmax of 627 MPa and including a material safety factor γt = 1.5 for
the textiles, the required area is 65.8 mm2/m. The used mesh has a cross section area of At = 97.4
mm²/m in the longitudinal direction (direction of the tensile stresses). The fabrics are arranged in two
layers close to the surface, thus, providing an upper and lower reinforcement layer. In the area of the
bearings an additional layer has been provided in the vertical direction as shown in Fig. 8.4 b. The
load bearing capacity has been checked in four-point-bending tests as shown in Fig. 8.5.
1350
140 280 500 280 140
LVDT
25
LVDT
The result of the tests was a safety factor against collapse under service loads higher than 4.
Furthermore, the large deflection values measured in the tests indicate the ductility of the panels.
For the production of the panels a self-compacting fine-grained concrete having an optimum
consistency capable of fully soaking the textiles was used. The panels were produced lying
horizontally as shown in Fig. 8.6.
181
a) b)
Fig. 8.6. Production of the textile reinforced concrete panels. a) Reinforcement inside formwork,
b) concreting of elements
First, a 4 mm thick concrete layer is poured in the formwork, followed by the placing of the first layer
of textile reinforcement. Then a 17 mm thick layer of concrete is poured followed by the placing of the
second reinforcement layer, and finally the remaining 4 mm concrete layer is poured. The difficulties
in the exact positioning of the textile reinforcement layer are one disadvantage of this very simple
production process.
An agraffe fixing device shown in Fig. 8.7 is used for the fixing of the curtain wall panels.
Vertical Substructure
(plugged at steel- Agraffe
reinforced wall)
Horizontal Profiles
Agraffe
Dowel in
cone-shaped
borehole
Panel
The vertical aluminum substructure (skeleton) of the device is plugged into the steel-reinforced wall.
The agraffes are fixed to the textile-reinforced panels using special dowels. These are positioned in the
panel inside cone-shaped boreholes. Pull out and shearing tests have been carried out in order to check
the load bearing capacity of the dowels. The results showed that the dowels can resist more than seven
times the load they are actually subjected to in practice even if they are positioned in cracked concrete.
182
8.3 Sandwich elements
The concreting of the sandwich elements is performed in analogy to the production of ordinary
reinforced concrete elements. The textile reinforcement (in this case a contoured profiled spacer
fabric, consisting of two cover layers and yarns in between as shown in Fig. 8.8 a) is placed in the
formwork, which is then filled with highly liquid fine concrete. Before that, the anchors connecting the
facing shell elements with the reinforced concrete bearing elements are put in place. After the
hardening of the facing shells, the reinforced concrete bearing element is poured. In Fig. 8.8 b) a
completed sandwich element is shown.
a) b)
Fig. 8.8. Sandwich elements made out of textile reinforced concrete. a) profiled spacer fabric, b)
completed sandwich element
183
Fig. 8.9. Prototype of the parapet sheet.
The ordinary steel reinforcement was substituted in the 20 mm panel by an biaxial textile
reinforcement made from AR-Glass. The edge stiffening is reinforced by fiber bars. The dimensions
and the dead weight of the element were reduced to a minimum. Besides the lower transport and
mounting weights advantages exist also for the anchorage constructions.
Due to their high load-bearing capacity and lower-cost fabrication, tanks made of reinforced concrete
have the biggest market share. Steel-reinforcement is normally arranged to provide its load bearing
capacity and tightness. Depending on the type of the plant (aerobic or anaerobic) and the composition
of the wastewater the reinforcement or the concrete may be damaged by corrosion. The risk of a
corrosive attack by biogeneous sulfuric acid in anaerobic parts of a wastewater treatment plant is
usually higher than in aerobic parts, but even there it may occur, e.g. due to insufficient service.
Decentralized wastewater treatment plants made of textile reinforced concrete offer many advantages
such as a lower wall thickness and a higher durability against corrosion attacks. Because no
experiences were present for the construction and the application of decentralized wastewater
treatment plants made of textile reinforced concrete, suitable fine concrtes and textiles were developed
at the RWTH Aachen University within the scope of an AiF research project as well as calculation and
production procedure for this application [Heg2003, Heg2004b]. The practical applicability of the
research results was proved by the realization of a prototype shown in Fig. 8.10 in cooperation with
the industrial partner Mall GmbH, Donaueschingen, Germany, a manufacturer of prefabricated
decentralized wastewater treatment plants within the scope of the project.
184
a) Geometry b) Prototype
Fig. 8.10. Prototype of a decentralized wastewater treatment plants made of textile reinforced
concrete.
The geometry of the planned tank is defined according to custom and usage requirements and to the
demands of the available German Standards. The wall of the basin has a thickness of 4 cm, which is
less than half of the thickness of a steel-reinforced tank. The textile reinforcement will consist of two
layers which are arranged one at each side of the wall. The chosen biaxial carbon fiber fabric has a
weight of 160 g/m² in each direction. Fig. 8.11 shows the prepared reinforcement with arranged
concrete spacers.The concrete cover is about 5 mm.
concrete spacer
carbon layer
spacer fabric
The 1650 tex-rovings have a spacing of 10 mm. The carbon fiber fabrics will be glued on a spacer
fabric that ensures the distance between the inner and the outer layer of the reinforcement. The
concrete cover will be achieved by gluing concrete pieces of about 5 mm thickness on the
reinforcement. The concrete pieces ensure the distance between reinforcement and formwork during
the pouring of the concrete. Today there are no textile reinforcements available which include
185
applicated spacers. Therefore, preparing the reinforcement has to be done manually and is time
consuming.
In the final state an additionally steel reinforcement carries the tensile forces. In addition, the
necessary concrete cover can be reduced by the application of the fine concrete matrix, so that a better
exploitation of the cross section becomes possible. The industrially precast elements can be moved
due to their low weight by hand on the building site. The durability of the textile fiberglass
reinforcement is not so important, because the textile reinforcement has to take over a load bearing
function according only during the erection state. Near the optimisation of the formwork cross section
and the first tips for the rational production of the elements are the cooperation of formwork and
concrete as well as the fire behaviour the main investigation targets.
186
Cross section of compact prototypes with Orthogonal textile reinforcement with worked in
reinforcing bar elements in the ribs and in the reinforcing bar elements
edge stiffening frame
The ordinary steel reinforcement was substituted in the 20 mm panel by an biaxial textile
reinforcement made from AR-Glass. The ribs of a total height of 70-mm were reinforced by fiber bars.
187
The structure is mainly stressed by pressure. Only with eccentric load cases, e.g., wind effect on the
long sides bending moment occur. The whole construction is stabilized by the diagonal position of the
curves without additional elements in longitudinal direction. Furthermore the complex geometry of the
structure can be composed with very easy elements shown in Fig. 8.15.
Fig. 8.15. Jointing of diamond-shaped elements and spatial appearance from the inside.
The diamond-shaped element, with dimensions of 1000 x 600 mms, a height of 160 mms and a
thickness of 25 mm has a weight of approx. 23 kgs. The approximated curve form is generated by
coupling the elements in each joint with an offset of 5º.
The example shows that already today textile reinforced concrete can be used for structures. Easy
connecting technologies, flexible textiles and the first static calculation models form the bases for the
development of structures which point up the constructive and creative characteristics of the new
material like slimness of the elements, sharp edged appearance and excellent concrete surfaces.
EIFS are used as exterior retrofits for existing buildings to improve the insulation value. There is also
a large emphasis on the use of EIFS for new designs and construction. EIFS has proven cost effective
for a wide range of buildings including office structures, retail establishments, industrial plants,
schools and residences.
Buildings manufactured using an EIFS are 30% more energy efficient than traditional methods where
the insulation is on the inside of the building. The more even surface temperature distribution inside
the building increases the comfort of the occupants.
188
• Low cost
• Wide range of finish colors and textures offered for retrofit as well as new construction
• Many installation options from field applied to prefabricated panels
• Major code compliance: ICBO; BOCA; SBCCI
The EIFS is a non-load bearing exterior wall system consisting of layers of the following components:
an insulation board which is adhesively or mechanically attached to a substrate such as plywood; a
polymer modified cement base coat reinforced with a fiberglass mesh; and a finish coat. The insulation
board is generally composed of polystyrene or polyisocyanurate foam. The cement base coat and
finish coat serve as water barriers and also as a matrix to transfer loads to the fiberglass mesh. The
fiberglass mesh is needed to prevent cracking of the base coat and finish coat due to thermal loads on
the panel as well as to protect the panel from impact damage. In some instances, two layers of mesh
are used for high impact resistance, or meshes of higher areal weight. Reinforcing meshes are
available in the following typical configurations:
189
17) Radiant Heat
The installation methods are unique to each systems supplier and they are not universal. Little
information is on file here in that regard. In general, the cement base coat is applied to the foam board
to a closely controlled depth; the fabric is then embedded into the wet base coat such that it is
encapsulated entirely. The decorative finish coat is applied after the base coat has cured.
Joint finishing products are an integral part of gypsum board systems. The main function of the joint
tape and compound system is to render the surface of the wall smooth, uniform and allow for an
aesthetic finish.
There are two basic types of joint compounds on the market that are used with reinforcement tapes:
drying type compounds and setting type compounds. The main ingredient of joint compound is
calcium carbonate mixed with other ingredients such as clay, talc, mica, thickeners, resin/latex, perlite,
preservatives and water to produce creamy, easily spreadable paste. Calcium sulfate is used only in
setting type compounds. Setting type compounds give a harder, stronger binding material compared to
drying type compounds.
190
8.10.3 Typical tests
• Physical testing of joint compound, paper joint tape, glass-mesh tape and an assembly of joint
compound and joint tape: ASTM C474.
• Joint treatment materials: ASTM C475. Note: The joint treatment materials described in these
standards are for use with gypsum board installed in accordance with specification ASTM C840.
• tests for glass mesh tapes: ASTM C474;
o tensile strength
o width
o thickness
o skewness
o method for evaluating tensile properties of gypsum panel joints
• An adhesive test for evaluating ability of self-adhesive tape to stick to gypsum board panels is
in review with ASTM.
• specifications for glass mesh tape; ASTM C475
o Tensile strength – shall be not less than 30 lbf/in (5.25N/mm) in the cross direction
o Width – shall be not less than 1 7/8 in. (48mm) nor more than 2 ¼ in. (60mm)
o Thickness – shall be not more than 0.012 in. (0.30mm).
o Skewness – shall be not more than 8.75% when tested in accordance with method
D3882
8.10.4 Installation
A self adhesive tape is directly applicated to the gypsum board surface (A compound is spread over
top of the taped joint with sufficient pressure to allow the compound to penetrate through the mesh
and into the joint space.
After the first coat of compound is dried, one or more coats may be applied and sanded after drying.
The number of coats of compound applied and amount of sanding done will depend on the desired
level of aesthetic finish to be achieved.
8.11 …..
Search for realized / planned applications still in progress
Infomation available: Durapct, Novacret, Dr. Peled, Prof. Bartos, C. Meyer, Prof. Reinhardt,
Fachvereinigung Faserbeton…..
191
8.12 Summary and view
Todo: Aachen University
REFERENCES
[Bra2003] Brameshuber, W.; Koster, M.; Hegger, J.; Voss, S.; Gries, T.; Barlé, M.; Reinhardt, H.-
W.; Krüger, M.: Textile Reinforced Concrete (TRC) for Integrated Formworks. 12.
Intern. Techtextil-Symposium, Frankfurt a. M., 7.-10. April 2003.
[Cur1998] Curbach, M.; Hegger, J. et al.: Sachstandbericht zum Einsatz von Textilien im
Massivbau, Deutscher Ausschuss für Stahlbeton (DAfStb), Heft 488, Beuth Verlag,
Berlin 1998
[Cur2003] Curbach, M. (Hrsg).: Textile Reinforced Structures. Proceedings of the 2nd Colloquium
on textile Reinforced Structures (CTRS2), Dresden, 29.9-01.10.2003.
[Cur2003b] Curbach, M.; Ortlepp, S.; Brückner, A.; Kratz, M.; Offermann, P.; Engler, T.: Entwiclung
einer großformatigen, dünnwandigen, textilbewehrten Fassadenplatte. Beton- und
Stahlbetonbau 98, Heft 6, 2003, S. 345-350.
[Heg1998] Hegger, J., Döinghaus, P.; Will, N.: Textilbeton – Tragverhalten und
Verbundeigenschaften. In: Sachstandbericht zum Einsatz von Textilien im Massivbau”,
Deutscher Ausschuss für Stahlbeton (DAfStb), Heft 488, Beuth Verlag, 1998, pp. 81- 90.
[Heg2001b] Hegger, J.; Schneider, H. N.; Molter, M.; Schätzke, C.: Fassaden aus textilbewehrtem
Beton. Betonwerk- und Fertigteiltechnik, Heft 12, 2001.
[Heg2003] Hegger, J. et al., Decentraliced Wastewater Treatment Plants made of Textile Reinforced
Concrete, Tagungsband zum Tech-Textil Symposium, Frankfurt, 2003, CD-ROM, Nr.
4.28
[Heg2003a] Hegger, J.; Voss, p.: Load-bearing behaviour of textile-reinforced concrete under 2-axial
loading. Textile Reinforced Structures, Proceedings of the 2nd Colloquium on textile
Reinforced Structures (CTRS2), Dresden, 29.9.-1.10.2003, p. 313-324.
[Heg2004] Hegger, J., Schneider, H. et al., “Exterior Cladding Panels as an Application of Textile
Reinforced Concrete”, accepted for ACI-Journal 2004.
192
[Heg2004b] Hegger, J., Brameshuber, W.; et. al.: Kleinkläranlagen aus Textilbewehrtem Beton.
Betonwerk- und Fertigteiltechnik, Heft 1, 2004.
[Mok1968] Mokk, L.: Montagebau in Stahlbeton. Budapest, Berlin: Ungarischer Verlag für Technik
1968, S. 107-109, 115.
[Ner1957] Nervi, P.L.; Rogers, E.N..; Joedicke, J.: Pier-Luigi Nervi – Bauten und Projekte.
Stuttgart: Hatje Verlag, 1957, S. 37-43.
193