Construction and Building Materials 414 (2024) 134782

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Experimental and statistical investigations of the material properties of FRP

reinforcement in compression
Lukas Bujotzek , Dominik Beck , Eftychia Apostolidi , Danièle Waldmann *
Institute of Solid Structures, Department of Civil and Environmental Engineering, TU Darmstadt, Franziska-Braun-Str. 3, 64287 Darmstadt, Germany

A R T I C L E I N F O A B S T R A C T

Keywords: Internationally leading standards currently do not permit to consider the contribution of reinforcement made of
FRP reinforcement fibre reinforced polymers (FRP) to a concrete member’s compressive load-bearing capacity due to a lack of
Compressive material properties reliable knowledge regarding FRP’s material properties in compression. Thus, there is a very high demand for
Test setup design
carefully specifying an optimised test setup for FRP bar reinforcement that accounts for both theoretical con­
Experimental investigation
Statistical investigations
siderations and established standards from other industry branches. This paper presents an extensive literature
Kolmogorov Smirnov test review on mechanical approaches to describe FRP material properties in compression, existing test setups from
BFRP various industry branches as well as experimental studies that have been conducted in order to research the
CFRP compressive material properties of FRP reinforcement. Based on experimental results from literature a database
GFRP was compiled. Following a reasonable choice of a test setup based on the literature study, an experimental
campaign was setup, in order to investigate the influence of different fibre materials (carbon, glass, basalt),
polymer matrices and bar diameters on the compressive material properties of FRP reinforcement bars. In
addition to the experimental determination of the material properties, the joint evaluation of the total sample
size of n = 95 own experimental test results and the results from the database generated from literature allows for
statistical investigations. Thus, the determination of material scatter and the estimation of the distribution type
by means of a statistical analysis applying the Kolmogorov-Smirnov test provides the basis for further research
towards the reliability of concrete structures reinforced with FRP bars in compression.

1. Introduction properties [1,5,6]. However, several research projects revealed a posi­

tive influence of a consideration of such reinforcement in compression
Reinforcements based on fibre-reinforced polymers (FRP) have been on the load-bearing capacity of a structural member [7–9].
intensively discussed in international research on reinforced concrete FRP consist of fibres from various source materials. These fibres
(RC) structures. Due to their outstanding mechanical properties, espe­ serve to carry the load and are coated by a polymer matrix binding them
cially their high tensile strength and excellent resistance against envi­ together and protecting them against environmental impacts and
ronmental impacts, FRP reinforcement bars have established as a serious stresses perpendicular to the fibre axis [10,11]. Regarding the me­
alternative to conventional reinforcing steel [1–3]. These properties lead chanical properties of FRP, a significant difference between the
to reduced requirements for concrete cover, allowing the design of very compressive and tensile strength of FRP is known from the literature [1,
slender and material-efficient structures. However, the main potential of 10,11]. Tensile failure results from rupture of single fibres. The polymer
FRP reinforcement is the extension of a structure’s service life due to matrix, exhibiting a higher ultimate strain, redistributes stresses to the
improved durability properties. The majority of previous research pro­ adjacent fibres resulting in a successive failure propagation across the
jects focused on the load-bearing behaviour of structural members FRP cross section. Thus, the tensile strength of the FRP is fibre domi­
subjected to flexural loading and the respective application of FRP nated and can be accurately predicted [11]. Regarding failure in
reinforcement in tension [1,4]. FRP reinforcement in compression zones compression, various properties influence the material behaviour lead­
of flexural members as well as in columns and walls is currently not ing to different potential failure modes. The fibre diameter is most
permitted to be taken into account by leading international standards, frequently referred to account for the failure behaviour of different fibre
due to a lack of reliable scientific data on the FRP compressive material materials. Commonly used fibre type such as glass, carbon and basalt

* Corresponding author.
E-mail address: waldmann@massivbau.tu-darmstadt.de (D. Waldmann).

https://doi.org/10.1016/j.conbuildmat.2023.134782
Received 2 August 2023; Received in revised form 2 November 2023; Accepted 25 December 2023
Available online 13 January 2024
0950-0618/© 2024 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

feature approximate maximum fibre diameters of 24, 10 and 30 µm, of glass fibre reinforced polymers (GFRP) [26] as well as carbon fibre
respectively [10–13]. Hahn [14] summarises the different modes of FRP reinforced polymer (CFRP) specimens [25]. Besides the FRP compres­
compressive failure as illustrated in Fig. 1. sive strength, the Elastic Modulus Efc is largely fibre-dominated.
Mode (a) is denoted by longitudinal – splitting and results from a ( )
Efc = Vf ⋅Ef + 1 − Vf ⋅Em (2)
transverse tension failure of the polymer matrix due to the different
Poisson’s ratios ν of the fibres and the polymer matrix [10]. Two possible According to [10–12,27], a differentiation between the compressive
alternatives depending on the stiffness of the polymer matrix exist: (i) in and tensile Modulus of Elasticity of the fibres has not been taken into
stiff matrices, the bond between fibres and polymer matrix fails and (ii) consideration. Therefore, it is assumed that the fibre Elastic Modulus Ef
in soft matrices, the fibres bend until fibre bending failure occurs. Mode in compression is equivalent to that in tension.
(b) called shear crippling (Fig. 1(b)) describes the distortion of the entire According to Carlsson and Pipes [18], the test slenderness ltest / Ø
FRP cross-section due to shear band formation. Shear bands arise from (ltest: free testing length; Ø: specimen diatemeter) influences the
an initial shear angle, attributed to the skewness of the fibres resulting measured compressive strength decisively, as can be seen from the
from the production process. In most cases, the failure plane forms at an compressive strength ffc plotted over ltest / Ø in Fig. 2. The FRP’s distinct
angle of approximately 45◦ to the load direction [14]. Mode (c) is orthotropic material behaviour leads to a disturbance length ld, starting
designated as compressive failure. It describes a progressive failure, from the area of load introduction at both sides of the specimen as a
starting from single fibres failing to stability on a micro level and thus result of imperfectly distributed stresses (e.g. due to clamping the
transferring loads to adjacent fibres, which subsequently fail as well. specimen), as illustrated in Fig. 2 [28]. Within the length ld no uniaxial
Although shear crippling (b) and compressive failure (c) are related [15, stress state can be developed, leading to distorted values for the FRP
16], a distinction between the failure modes can be observed in the compressive strength. Consequently, this effects results in reduced
propagation of the failed pattern along the FRP specimen. Shear crip­ values for ffc in case of a small test slenderness. Secondly, large slen­
pling is more likely to occur for fibre material featuring a low failure derness values lead to second-order effects and eventually result in
strain (e.g. for carbon fibres) and thus exhibit a limited ability to spread buckling failure. Test results from [18] reveal an optimal test slender­
the failure area over large parts of the specimen. As a result, the failure ness of approx. (ltest / Ø)opt ≈ 7 for specimens made of carbon
area is spatially very limited and shear bands at 45◦ represent the fibre-reinforced epoxy resin.
dominating failure mode. On the other hand, fibre materials with a high Several standards for testing the compressive material properties of
ultimate strain (e.g. glass fibres) enable a redistribution of stresses from plastics and fibre-reinforced laminates exist in the polymer and aero­
the initially failing fibre to the adjacent fibres. Thus, a spatially more space technology. The load application via the end faces of the specimen
spread-out failure area occurs leading to mode (c). (Table 1 (a)) is permitted for testing the material properties of unrein­
Analytical models for the determination of the FRP compressive forced polymers, according to EN ISO 604:2002 [29] and ASTM
strength ffc are known from the literature in the field of polymer tech­ D695–15:2015 [30]. On the other hand, standards for testing
nology. The first model based on the shear band theory was provided by fibre-reinforced laminates ISO 14126:1999 [31], D3410:2016 [32] and
Rosen [18] and depends on the shear stiffness Gm, as well as on the DIN EN 2850:2018 [33] prescribe load application via rigidly connected
polymer matrix content (1 − Vf), Eq. (1). shear tabs, (Table 1 (b)), in order to avoid splitting of the specimen in the
Gm load application area. GOST 31938:2012 [34] represents the only
ffc = (1) standard for testing FRP reinforcement bars for concrete applications in
1 − Vf
compression known to the authors. A test slenderness of ltest / Ø = 6.0 is
Rosen’s initial approach has been enhanced by various researchers specified for the experimental determination of compressive strength
over the years regarding the fibre material and the fibre volume content and Modulus of Elasticity. Furthermore, an indirect load application via
[16,19]. Alternative approaches have been developed, e.g. by Lessard grouted anchor sleeves introducing the test force via shear stresses is
and Chang [20], to account for the influence of the fibre diameter as well prescribed (Table 1 (f)).
as by Fleck and Budiansky [21] and Argon [22] to take the plastic shear Several test series by researchers in the field of concrete engineering
behaviour into account. Bazhenov et al. [19] identified a pronounced addressed the question on how to determine the compressive material
correlation between FRP’s compressive and shear strength based on test properties of FRP reinforcement experimentally. The following section
results. Xu and Reifsnider [23] succeeded in developing a novel presents a compilation of the discussed studies in a comprehensive
approach to recalculate compressive strength values from a test series by database (Table 2) for comparative analysis of the different approaches
Lo and Chim [24] and Piggott and Harris [25], consisting of carbon-, and their corresponding outcomes.
glass- and aramid fibre-reinforced samples. A detailed look at the ana­ Wu [35] reported results on GFRP reinforcement bars from three
lytic approach above reveals the need of unique material properties in different manufacturers. The testing force was introduced via the end
order to estimate the FRP compressive strength properly. In engineering
practice, such unique material properties are often unavailable, which
strengthens the requirement for a standardised test method of FRP in
compression.
Several experimental studies from the polymer technology reveal a
linear elastic material behaviour until brittle compressive failure in case

Fig. 2. Influence of ltest / Ø on the compressive strength ffc according to

Fig. 1. Failure modes of FRP in compression following Hahn [14,17]. Carlsson and Pipes [18], [17].

2
 L. Bujotzek et al.
Table 1
Test setups from the literature and existing test standards.
# (a) (b) (c) (d) (e) (f)
Test setup
3

Corresponding EN ISO 604:2002 [29] ISO 14126:1999 [31] none none none GOST 31938:2012 [34]
standard ASTM D695-15:2015 [30] D3410:2016 [32]
DIN EN 2850:2018 [33]

Construction and Building Materials 414 (2024) 134782

Applied Wu [35] none Kobayashi and Fuyisaki [38] Deitz et al. [39] Khorramian Urbanski [48] D’Antino and
by Abed [37] Almerich-Chulia et al. [40] and Sadeghian [46] Pisani [49]
Chaallal and Benmokrane Bruun [41] Fillmore
[36] Tavassoli et al. [42] and Sadeghian [44]
AlAjarmeh et al. [45]
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Table 2
Test results with values from the database.
test series nffc fibre ltest / Ø load application ffc /fft (ffc / fft)m CoVffc nEfc Efc / Eft (Efc / CoVEfc failure
type [-] [-] [-] [-] Eft)m mode**
[-]

Wu (1990): 8 GFRP 1.0 – end faces 0.75 - 0.85 23% 8 - 0.93 6% a

GFRP type I [35] 2.5 0.95
Wu (1990): 12 GFRP 1.0 – end faces 0.26 - 0.53 16% 4 - 0.69 16% a
GFRP type II [35] 2.6 0.57
Wu (1990): 12 GFRP 1.0 – end faces 0.78 - 0.65 33% 2 - 0.75 5% a
GFRP type III [35] 2.7 0.93
Chaallal and Benmokrane 2 GFRP 2.75 end faces 0.73 - 0.76 - 2 1.08 - 1.10 - -
(1993) [36] 0.80 1.12
Deitz et al. (2000) [50] 15 GFRP 4.9 – end faces with steel 0.60 - 0.79 24% - - 1.04 - c / (a)
6.5 cases 1.19
Almerich-Chulia et al. 7 GFRP 11.0 end faces with steel 0.51 - 0.60 9% 7 1.02 - 1.06 9% -
(2012) [40] cases 0.74 1.20
Bruun (2014) [41] 5 GFRP 5.0 end faces with steel 0.56 - 0.60 4% - - 1.00 - c/a
cases 0.62
Tavassoli et al. (2015) [42] 5 GFRP 2.0 – end faces with steel 0.55 - 0.60 17% 5 0.84 - 0.92 13% a
11.0 cases 0.65 0.98
Galen (2017) [47] 24 GFRP 6.0 anchor sleeves 0.41 - 0.56 - - - - - c
0.7
Khorramian and Sadeghian 10 GFRP 2.0 compressive 0.74 - 0.87 6% 10 1.00 - 1.03 1% c / (a)
(2019): coupons 0.99 1.06
GFRP type I [43]
Khorramian and Sadeghian 5 GFRP 2.0 compressive - 0.85 9% 5 - 0.97 2% c / (a)
(2019): coupons
GFRP type II [43]
Khorramian and Sadeghian 20 GFRP 2.0 compressive 0.55 - 0.77 7% 20 1.00 - 1.06 3% c / (a)
(2019): coupons 0.93 1.20
GFRP type III [43]
AlAjarmeh at al. (2019) 54 GFRP 2.0 – 16 compressive 0.33 – 0.76 16% 48 0.95 – 1.04 3% a/b/c
[45] (HM)*** coupons 1.23 1.23
Abed et al. (2020) [37] 12 GFRP 2.0 end faces 0.55 - 0.70 6% - - - - a
0.83
6 BFRP 0.33 - 0.42 5% - - - - a
0.52
Urbanski (2020) [48] 3 B-CFRP* 5.7 – anchor sleeves 0.34 - 0.40 - - 0.99 - 1.06 - c / (a)
13.7 0.46 1.14
D’Antino and Pisani (2023): 20 GFRP 3.9 – anchor sleeves 0.49 – 0.71 9% 24 0.95 – 0.99 2% b / (a)
7.1 0.99 1.16
GFRP thermoset [49]
D’Antino and Pisani (2023): 25 GFRP 3.5 – anchor sleeves 0.39 – 0.56 13% 25 0.86 – 0.97 7% b
GFRP thermoplasts [49] 4.0 0.85 1.12

nffc = 245 nEfc = 160

*Basalt-Carbon Hybrid; **according to Fig. 1: (a) longitudinal splitting; (b) shear crippling; (c) compressive failure; ***HM = high modulus

faces of the specimen at a slenderness of ltest / Ø = 2.5, resulting in a very can be attributed to the small test slenderness leading to a crucial in­
straightforward test setup (Table 1 (a)). Wu described longitudinal crease in disturbing edge effects. Due to the strongly scattering results,
splitting to be the predominant failure mode. Furthermore, a correlation the authors of this paper decided not to integrate the results into the
between the compressive strength and the bar diameter revealing overall compilation in Table 2. Deitz et al. [39] conducted a series of 51
smaller strength values for large bar diameters (Ø = 25.4 mm) was tests on GFRP reinforcement bars in compression at variable test slen­
observed. This observation is due to the higher probability of production derness. The load was introduced directly via the specimen’s end faces
errors and unequal impregnation of the fibres with the polymer matrix, through re-usable, non-grouted steel cases in order to prevent the FRP
when producing large diameters. Chaallal and Benmokrane [36] tested from popping out at high slenderness (Table 1 (d)). The challenging
GFRP bars of different diameters in compression and found a linear process of preparing the area of load introduction perpendicular to the
elastic material behaviour analogous to the behaviour in tension. A bar axis was highlighted by the authors. The observed failure mode of
direct load application via the specimen’s end faces as specified in a the less slender specimens was described as crushing. For the slender test
previous version of ASTM D695–15 [30] from 1991 (ASTM D695–91) specimens, the authors found the test results following a perfect Euler
prescribing a testing slenderness of ltest / Ø = 2.75 (Table 1 (a)) was buckling curve. Similar observations were made by Almerich-Chulia
applied. Furthermore, similar Elastic Moduli were recorded in et al. [40], Bruun [41] and Tavassoli et al. [42], who applied a test
compression and in tension. A similar test setup was applied by Abed setup similar to that used by Deitz [39] (Table 1 (d)) and investigated
et al. [37] who tested GFRP and BFRP specimens and found longitudinal GFRP specimens at different slenderness values in compression. As a
splitting to be the dominating failure mode throughout all test results. A consequence of the load introduction via the specimen’s end faces in
widely cited study by Kobayashi and Fujisaki [38] applied a very elab­ [42], part of the tests showed longitudinal splitting of the whole spec­
orate test setup featuring a CFRP tube embedded into two concrete imen including the area of load introduction. Tests reported by Bruun
blocks at each end of the specimen (Table 1 (c)). Kobayashi and Fujisaki [41] showed longitudinal splitting failure combined by partial crushing
tried to reduce the testing length ltest to a minimum in order to prevent at all levels of slenderness. In addition to that, the authors were able to
the FRP from buckling (ltest / Ø = 0.5). The authors tested FRP with record linear elastic material behaviour until the GFRP specimens failed
carbon, aramid and glass fibres, revealing strongly scattering results that in a brittle manner [41]. Khorramian and Sadeghian [43] and Fillmore

4
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

and Sadeghian [44] carried out tests on GFRP specimens using different industry sectors to select an appropriate test setup. Subsequently, a
types of glass fibres (GFRP type I to III) and applied the load via so-called compilation of test results from the literature serves to extend the large
“coupons”, consisting of a steel plate and a cap at each end of the test number of tests carried out by the authors. The joint evaluation of this
specimen (Table 1 (e)). To avoid the specimen from failing due to load data compilation allows for an assessment of crucial parameters.
introduction, the coupons were grouted with an adhesive epoxy resin. Furthermore, the large amount of test data serves to prepare for reli­
Regarding the free testing length, the authors referred to ASTM ability analyses and the identification of the distribution type of the FRP
D695–15:2015 [30], specifying a test slenderness of ltest / Ø = 2.0. material properties in compression.
Regardless of the GFRP type, an average Modulus of Elasticity ratio of Efc
/ Eft = 1.04 was recorded. Crushing of the specimens within the free 2. Experimental campaign
testing length as well as in the cap region was the dominating failure
mode. Longitudinal splitting failure occurred partially, but was deci­ An investigation of FRP bars made of three different fibre types
sively less pronounced compared to the test results from [39–42]. AlA­ (glass, carbon, basalt) is conducted within the experimental efforts of
jarmeh et al. [45] studied the compressive material behaviour of GFRP this study. To depict the influence of different polymer matrices as well
reinforcement bars applying a test setup similar to that suggested by as different fibre volume contents, FRP reinforcement bars provided by
Khorramian and Sadeghian [46] (Table 1 (e)), but varied the test slen­ different manufacturers are investigated. To evaluate any potential
derness. Specimens that were unaffected by second – order effects impact of the bar diameter, various diameters from the same manufac­
showed a linear elastic material behaviour up to compressive failure. A turer are analysed. The obtained results from the experiments are
test series on GFRP reinforcement bars tested according to GOST interpreted regarding the compressive strength and Modulus of Elas­
31938:2012 [34] (Table 1 (f)) was provided by Galen LLC [47]. The ticity as well as the corresponding failure modes based on the me­
specimens showed failure in a crushing (mode (c)) manner. Urbanski chanical considerations presented in Section 1.
[48] investigated basalt-carbon hybrid bars at varying slenderness. The
load was applied indirectly via grouted anchor sleeves, according to
GOST 31938:2012 [34] (Table 1 (f)). The majority of specimens were 2.1. Tested specimens
found to show crushing failure, with accompanying delamination of
single fibre bundles from the matrix. Linear elastic material behaviour The tests are conducted in the Testing Laboratory of the Institute of
was observed until failure. A recent study by D’Antino and Pisani [49] Solid Structures at TU Darmstadt. A total of n = 95 tests from four
investigated the compressive material behaviour of thermoset and different manufacturers (I, II, III, IV), three fibre types (glass (G), carbon
thermoplast GFRP reinforcement. The authors applied a test setup (C), basalt (B)), two polymer matrix types (vinyl ester (V), epoxy (E))
similar to that suggested by Deitz et al. [39], but grouted the anchor and four nominal diameters (Ø8, Ø10, Ø12, Ø20 mm) are investigated as
sleeves (Table 1 (f)). Linear elastic material behaviour up to crushing summarised in Table 3.
was found for the major part of the specimens that did not undergo The GFRP reinforcement provided by manufacturer I (I-G) show a
buckling. Thus, the authors formulated a constitutive law for the ma­ grooved surface composed of Electric Corrosion Resistant (ECR) glass
terial behaviour of thermoset and thermoplast GFRP which is charac­ fibre reinforced vinyl ester resin featuring a fibre volume content of Vf
terised by a linear elastic behaviour up to failure. = 75%. Three different nominal diameters (Ø8, Ø12, Ø20) are investi­
The test results from the different studies described above are gated for this bar type within this study. The second GFRP reinforcement
compiled into a database for reasons of better comparability and in order type provided by manufacturer III (III-G) made from ECR glass fibre
to prepare for the evaluation in Sections 2.5 and 3.2. Table 2 shows the reinforced epoxy resin shows a helically wrapped surface. The III-G
normalised compressive strength ffc / fft and Modulus of Elasticity in GFRP bars feature a fibre volume content of Vf = 73% and are investi­
compression Efc / Eft, specifying the sample size for each material gated for two nominal diameters (Ø8, Ø12). The CFRP bars provided by
property nffc and nEfc. For the evaluation of the database only test series manufacturer II (II-C) show a grooved surface similar to the I-G GFRP
containing n ≥ 2 test results are considered. Furthermore, the compi­ bars, whereas the CFRP bars provided by manufacturer III (III-C) show a
lation presents the type of fibre material, the range of test slenderness wrapping similar to the III-G GFRP bars. All types of CFRP bars inves­
ltest / Ø and the type of load application according to Table 1. The co­ tigated within this study feature an epoxy polymer matrix. Series II-C
efficients of variation (CoV) provides information on the material scatter and III-C CFRP bars show fibre volume contents of Vf = 68% and 63%
if the sample size allows for it (n ≥ 5). Finally, the predominantly respectively. The investigated nominal bar diameters are Ø10 (II-C) and
observed failure mode (Fig. 1) is noted for each series if available. Tests, Ø8, Ø12 (II-C). Manufacturer IV provides the BFRP bars (IV-B) investi­
which showed failure from stability or second order effects, are removed gated within this study, featuring an epoxy resin as polymer matrix. The
for further evaluation since such failure does not imply to the material BFRP bars exhibit a fibre volume content of Vf = 77%, revealing the
itself. This applies to parts of the studies by Deitz et al. [50], Bruun [41], highest value within the presented test program. The IV-B BFRP bars, of
AlAjarmeh et al. [45] and D’Antino and Pisani [49]. which two different nominal diameters (Ø8, Ø12) are investigated,
Existing studies show strongly varying approaches regarding the feature a sand coated and helically wrapped bar surface. A selection of
experimental determination of the FRP compressive material properties, the experimentally investigated bars is shown in Fig. 3. Note that pre­
making it challenging to interpret the test results independent of the liminary to the tests, the bar diameter Øim was determined by immersion
applied test setup. Furthermore, investigations are limited on the in­ tests according to ASTM D7205/D7205M-06 [5]. The designation of the
fluence of single parameters, such as the chosen FRP type, leading to a individual test series arranges as follows:
reduced comparability regarding decisive variables. Furthermore, the
literature lacks proper evaluation of the generated test results con­
cerning the effects of test slenderness and the observed failure modes
from the polymer literature, resulting in an inadequate understanding of
the behaviour of FRP in compression. The outcome of this study will
provide results from widely arranged experimental investigations
regarding the impact of decisive parameters such as FRP type, fibre
volume content, polymer matrix type and bar diameter on the FRP
material properties in compression. To ensure constant and reasonable Table 3 provides the corresponding tensile strength fft and Elastic
boundary conditions for load application and test slenderness, this study Modulus Eft as well as the associated CoV for each series, to serve as a
evaluates test results from existing studies and standards from other reference for the experimentally derived material properties in

5
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Table 3
Overview of the experimental program.
series n [-] fibre type polymer Vf Øim fft,m Eft,m CoV (fft) CoV (Eft)
matrix type [%] [mm] [N/mm2] [N/mm2]

I-G-V-8 10 glass vinyl ester 75 8.81 1161 49,474 3.1 – 4.6% 1.0 – 3.9%
I-G-V-12 10 75 12.82 1195 52,570
I-G-V-20 10 75 21.11 1195 53,856
10 epoxy epoxy 73 8.78 1074 51,725 n. a. n. a.
III-G-E-12 10 glass 73 12.62 1074 51,725 n. a. n. a.
II-C-E-10 9* carbon 68 9.45 1891 128,608 7.5% 3.8%
III-C-E-8 8* 63 8.76 2467 134,405 n. a. n. a.
III-C-E-12 10 63 12.65 2467 132,499 n. a. n. a.
IV-B-E-8 10 basalt 77 7.41 955 47,611 1.1 – 3.4% 2.3 – 3.5%
IV-B-E-12 8* 77 11.72 950 42,846
total n= 95

* : Specimens were damaged during the production process.

reference values fft and Eft were provided by manufacturer III (III-C) and
adopted from El Ghadioui (2020) [48] (II-C).

2.2. Experimental setup and measurement instrumentation

The testing length was selected to ltest = 6 • Øim corresponding to the

range of test lengths of approximately (5 − 8) ⋅ Ø based on the findings
from the literature [17,18,45,49] as presented above, following GOST
31938:2012 [34].
The chosen test setup is shown schematically and as a picture in
Fig. 4. In order to ensure a secure anchorage of the reinforcing bars
during the test, the ends of each bar are placed into 300 mm long steel
sleeves and grouted with a high-strength mortar as specified in [5]. This
type of end anchoring ensures a concentric positioning of the bars. The
testing force is recorded by an internal load cell with an accuracy of
± 0.05% at a maximum of 650 kN throughout each experiment. Strain
Fig. 3. Selection of the tested samples. transducers (DD1) are mounted on both sides of the bar’s free test
length, with a gauge length of 40 mm. The bar strains are recorded at a
compression. Manufacturer III provides a mean ultimate test force Ftu frequency of 5 Hz.
from tensile tests, enabling the authors to calculate the mean tensile
strength fft,m. Manufacturer I and IV do not provide such values. 2.3. Experimental procedure
Therefore, the tensile strength fft,m was determined experimentally
based on n = 5 test results within preliminary investigations [51] ac­ The tests are performed displacement controlled at a target load
cording to ASTM D7205/D7205M-06 [52]. The manufacturers provide speed of 1.3 ± 0.3 mm/min according to ASTM D695–15:2015 [30] and
values for the tensile Elastic Modulus for all GFRP specimens. The tensile [45,48]. The hydraulic clamping jaws in the transverse direction are

Fig. 4. Schematic test setup (left) and picture of the setup at the laboratory (right).

6
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

positioned in the area of the anchor sleeves to prevent damage to the – 54,671 N/mm2). Besides the fact that the Elastic Moduli in compres­
FRP specimen due to transverse stresses. The first three tests of each sion are consistently at least as high as the corresponding Elastic Moduli
series are carried out without strain transducers to avoid damage to the in tension, the scatter range across the different test series is small. This
equipment from premature failure due to the unknown failure load of observation, in particular, leads to the conclusion that GFRP is well
the specimens in each series. This procedure leads to the determination applicable in compression and underlines the experimental test setup to
of only seven values for the Elastic Modulus per series. For the remaining be appropriate.
specimens per series, the strain transducers are uninstalled at reaching Along with the mean values from the test series, the corresponding
50% of the ultimate load based on the first three tests, as a precautionary scatter is of interest. The CoV(ffc,m) of the compressive strength ranges
measure. Note that after removing the measurement devices, the strains from 3.2% (I-G-V-20) to a maximum of 10% (III-G-E-8). In addition, the
are extrapolated assuming a linear elastic material behaviour based on CoV(Efc) of the Elastic Modulus in compression is determined between
the findings from the literature, as described above [25,26,36,41,45,48, 1.3% (III-G-E-12) and a maximum of 6.3% (I-G-V-20). The test series did
49]. not reveal any notable impact of the bar diameter on material proper­
The deviations between the recorded values of the two displacement ties, despite Wu’s [35] earlier identification of such influence. There­
transducers allow an assessment of the quality of the centric load fore, it is assumed that the bond behaviour between fibres and polymer
application. The illustration in Fig. 5 shows the measurement results of matrix has improved due to an advanced production process compared
the displacement transducers DD1–1 and DD1–2 separately on specimen to when Wu conducted his investigations in 1990. The consequently
I-G-V-8.6. A high quality of concentric load application is observed from more homogeneous distribution of stresses across the FRP cross-section
the minor deviations. The compressive strength ffc is determined as the and the reduction of defects leads to a significantly reduced dependence
ratio of the maximum force Fmax to the cross-sectional area of the rein­ of the FRP material properties on the diameter of the examined spec­
forcement bar from immersion weighting Afi. The compressive Modulus imen. Three different failure modes (Fig. 7) are observed after visual
of Elasticity Efc is determined between the points at 20% and 50% of the inspection as compiled in Table 5. Failure mode a (longitudinal split­
ultimate load according to Eq. (4) as specified in GOST 31938:2012 ting) occurs exclusively for the GFRP bars from manufacturer I and in
[34]. particular for series I-G-V-8 and I-G-V-20, as shown in Fig. 7(a). The
failure pattern is characterised by a vertical crack in the test specimen,
Fmax
ffc = (3) marking a delamination between the fibre and the polymer matrix due
Afi
to transverse stresses. Note that the other specimens from manufacturer
F50% − F20% I assigned to failure modes b and c exhibit partial splitting at the very
Efc = (4) least (Fig. 7(b)).
Afi ⋅(ε50% − ε20% )
The approximately equal Elastic Moduli and compressive strength
values indicate similar fibre material of the specimens from series I-G
2.4. Experimental results and III-G. Therefore, it is assumed that the polymer matrix is the primary
factor causing the differing failure modes. The vinyl ester utilised for the
2.4.1. GFRP specimens I-G samples seems to exhibit a smaller tensile strength compared to the
Fig. 6 illustrated the stress – strain behaviour of the GFRP specimens epoxy resin employed for the III-G samples. Therefore, failure due to
separately for each test series. To provide context for the obtained data, transverse tension (mode a) occurs for the I-G specimens solely. Failure
the authors normalised the mean compressive strength ffc,m and mean of the specimens occurs within the free length for most specimens.
Elastic Modulus in compression Efc,m of GFRP specimens based on However, tests in which failure occurs within the load application area
n = 49 (ffc) and n = 32 (Efc) test results, respectively. The normalisation show comparable strength and stiffness values, indicating the insensi­
was carried out using the corresponding tensile properties fft,m and Eft,m tivity of the glass fibres to transverse stresses. The failure modes of the
resulting in a comprehensive summary presented in Table 4. The ratio III-G specimens are shown in Fig. 7 ((c) and (d)) for direct comparison
between the mean recorded compressive strength ffc,m to the average with the failure pattern observed for the I-G specimens ((a) and (b)). A
tensile strength fft,m is between 0.63 and 0.71 (ffc,m = 720 – 781 N/ more pronounced inclination of the failure joint is identified, leading to
mm2), revealing a minor scatter range. The mean ultimate strain εfc,m for the assumption of failure mode b (shear crippling) in 15 of 20 cases.
the GFRP specimens ranges between 1.29% and 1.43%. The Elastic Note that specimens are partly excluded from further evaluation due to
Moduli ratio Efc / Eft varies between 1.00 and 1.11 (Efc = 51,543 N/mm2 significantly outlying test results and damage incurred during the pro­
duction process. Table 4 provides a record of the final count of valid test
outcomes per series.

2.4.2. CFRP specimens

The stress – strain course of the CFRP specimens is shown in Fig. 8
and Fig. 9. An averaged stress-strain course from all GFRP tests, denoted
as GFRP reference, serves for a better comparability of the following test
series on CFRP and BFRP specimens.
Regarding the stress-strain course of the CFRP specimens, the
significantly higher Modulus of Elasticity Efc compared to the GFRP
reinforcement is noticeable as expected analogously to the corre­
sponding Moduli of Elasticity in tension. Conversely, the strength values
ffc are below the corresponding mean strength of the GFRP reinforce­
ment throughout the whole test series. Normalising the mean CFRP
compressive strength to the corresponding tensile strength reveals ratios
from 0.22 (III-C-E-8) to 0.33 (II-C-E-10). One reason for the compara­
tively small strength ratios ffc,m / fft,m is the very high tensile strength of
CFRP. However, the absolute compressive strength values for the CFRP
specimens range from 548 to 651 N/mm2, which is consistently below
Fig. 5. Deviation between the two measuring points and visualisation of the the corresponding values for GFRP. This observation shows the signifi­
extrapolated measured values. cantly greater sensitivity of the carbon fibres to transverse stresses and a

7
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Fig. 6. Stress-strain behaviour of the I-G GFRP specimens (left) and the III-G GFRP specimens (right).

higher susceptibility to micro buckling due to low individual fibre di­ of the CFRP bars to stresses in the area of load introduction that is more
ameters compared to glass fibres. Moreover, the fibre volume content for pronounced for the specimens featuring a small diameter.
series II-C-E-10 and III-C series is Vf = 68% and 63% respectively, which While the II-C specimens mostly fail within the free testing length,
is lower than that of the GFRP specimens (Vf = 73 − 75%). Conse­ the III-C specimens failed more frequently in the area of load application
quently, CFRP was expected to exhibit smaller compressive strength due to the sensitivity of CFRP in the transverse direction. This different
values, as failure mode (b), observed for the majority of CFRP speci­ failure behaviour is reflected in the corresponding scatter of the
mens, is substantially influenced by the fibre volume content, cf. Eq. (1). compressive strength revealing larger CoVs for the III-C specimens
Nevertheless, it is important to note that the CFRP Elastic Moduli show (CoVffc = 9.6 – 11.6%) compared to series II-C-E-10 (CoVffc = 7.4%).
comparable values to their tensile counterparts. The ratios Efc / Eft range However, the scatter of the Elastic Moduli in compression of CFRP are
from 0.95 (II-C-E-10) to 0.98 (III-C-E-8) and thus indicate the hypothesis consistently at a low level (CoVEfc = 1.9–4.3%).
of equal Elastic Moduli in tension and compression to be accepted for It is particularly interesting to compare the III-C specimens with their
CFRP. Fig. 9 shows series III-C-E-8 and III-C-E-12 in direct comparison GFRP (III-G) counterparts. The influence of the resin as well as the
revealing significantly smaller compressive strength values for the Ø8 manufacturing process can be excluded as unknown parameters since
mm – specimens. This observation is attributed to the high susceptibility the same polymer matrix was used and the quality assurance is assumed

8
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Table 4 similar. The differences can thus be attributed to the fibre type and the
Test results of the series with GFRP specimens. fibre volume content (GFRP: Vf = 73%; CFRP: Vf = 63%). For the III-G
series I-G-V-8 I-G-V-12 I-G-V-20 III-G-E-8 III-G-E-12 specimens (ffc,m = 740 N/mm2), the compressive strengths are 23%
higher than for the III-C specimens (ffc,m = 600 N/mm2), revealing a
Øim [mm] 8.81 12.82 21.11 8.78 12.22
Ønom [mm] 8 12 20 8 12 significantly more favourable behaviour of GFRP in compression
nffc 9 10 10 10 10 compared to CFRP. Furthermore, the GFRP ultimate compressive strain
ffc,m [N/mm2] 750 781 755 720 758 (εfc,m = 1.4%) exceeds the corresponding value for CFRP specimens (εfc,
ffc,m / fft,m 0.65 0.65 0.63 0.67 0.71
m = 0.45%) by 211%. Note that, regarding a concrete member longi­
CoV(ffc,m) 8.7% 4.1% 3.2% 10.0% 7.8%
nEfc 7 7 6 5 7
tudinally reinforced with FRP in concentric compression, this ultimate
εfc,m [%] 1.37 1.34 1.29 1.36 1.43 compressive strain influences the load bearing capacity decisively.
Efc,m [N/mm2] 54,671 57,930 57,381 51,543 52,564 The failure patterns exclusively show specimens that fail along an
Efc,m / Eft,m 1.11 1.10 1.07 1.00 1.02 inclined shear plane (Fig. 10) indicating shear crippling (mode b) as
CoV(Efc,m) 1.5% 3.9% 6.3% 1.8% 1.3%
dominating failure mode, which is consistent with the results from the
literature (Table 7). This failure mode leads to the specimen losing its
entire integrity.
Note that specimens are partially removed from further evaluation
due to strongly outlying test results as well as damage of specimens
during the production process. A documentation of the final number of
valid test results per series is given in Table 6. Due to issues with the
specimens from test series III-C-E-8 as discussed above, only three valid
results can be documented for the Elastic Modulus in compression.

2.4.3. BFRP specimens

Fig. 11 shows the stress – strain course of the BFRP specimens from
series IV-B-E-8 and IV-B-E-12. The average compressive strength of the
IV-B-E-8 specimens is 49% of the corresponding tensile strength fft and
scatters strongly compared to the other test series, featuring a CoVffc of
14.1%. However, series IV-B-E-12 reveals similar strength values
resulting in a normalised compressive strength of ffc,m / fft,m = 50%,
with the CoV being slightly larger than for the GFRP specimens (CoV =
10.3%). Regarding the stress-strain relationship (Fig. 11), lower strength
and stiffness values are observed compared to the GFRP reference.
Fig. 7. Failure modes of III-G GFRP-specimens (left: (a) and (b)) and III-G Nevertheless, the Elastic Moduli in compression of the BFRP specimen
specimens (right: (c) and (d)). show good agreement with the corresponding tensile values revealing a
ratio Efc,m / Eft,m of 0.98 for series IV-B-E-8 (Efc,m = 46,746 N/mm2) and
1.03 for series IV-B-E-12 (Efc,m = 44,321 N/mm2) with a marginal
Table 5
Assignment of failure modes for GFRP specimen.
scatter of CoVEfc = 2.4% and 2.0%, respectively.
Shear failure along the tensioned strand is recognised as the pre­
failure mode I-G-V-8 I-G-V-12 I-G-V-20 III-G-E-8 III-G-E-12
dominant failure mode in the majority of BFRP specimens. In contrast to
mode a 7 1 6 - - the other examined specimens, the BFRP reinforcement bars are heli­
mode b 1 - - 7 8 cally wrapped with a tensioned strand, resulting in preliminary damage
mode c 2 9 4 3 2
to the cross-section. Fig. 12 presents a typical failure pattern indicative
of this trend.
Note that specimens are partially removed from further evaluation
due to damage of specimens during the production process. A docu­
mentation of the final number of valid test results per series is given in
Table 8.

2.5. Integration of the results from the database

In order to verify the findings beyond the efforts of the present study,
a joint evaluation of the results from the database (Table 2) and the
results from this study, regarding the compressive strength ffc, the Elastic
Modulus in compression Efc as well as the corresponding material scatter
is conducted. Beyond the tests on CFRP and BFRP carried out as part of
this study, only Abed et al. [37] and Urbanski [48] examined FRP based
on basalt and basalt-carbon hybrid fibres, respectively. Thus, GFRP is
the FRP type applied for the major part of the investigated studies.
Fig. 13 illustrates the data giving the normalised compressive strength ffc
/ fft, the normalised Elastic Modulus in compression Efc / Eft as well as
the corresponding scatter if available and the sample size allows for
(n ≥ 5).
Fig. 8. Stress-strain behaviour of test series II-C-E-10. The mean value over all strength ffc / fft ratios integrated in the
database based on nffc = 339 test results in total equals 0.68 for GFRP,
which is between the limits (ffc / fft = 0.63 − 0.71) of the values
determined within this study (Table 4). The strength ratios of the BFRP

9
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Fig. 9. Stress-strain behaviour of the test series III-C-E-8 and III-C-E-12.

Table 6
Test results of the series with CFRP specimens.
series II-C-E-10 III-C-E-8 III-C-E-12

Øim [mm] 9.45 8.76 12.65

Ønom [mm] 10 8 12
nffc 9 8 10
fft,m [N/mm2] 1,891 2,467 2,467
ffc,m [N/mm2] 629 548 651
ffc,m / fft,m 0.33 0.22 0.26
CoV(ffc,m) 7.4% 9.6% 11.6%
nEfc 6 3 7
εfc,m [%] 0.49 0.38 0.49
Eft,m [N/mm2] 135,121 137,050 137,050
Efc,m [N/mm2] 128,608 134,405 132,499
Efc,m / Eft,m 0.95 0.98 0.97
CoV(Efc) 4.3% 1.9% 2.2%

results. The mean Elastic Moduli ratio Efc / Eft equals 1.00 regardless of
the fibre type. This observation gives a strong indication to verify the
hypothesis of the FRP Elastic Modulus in compression Efc corresponding
to the Elastic Modulus in tension Eft based on ntot,Efc = 221 test results in
total. Analogously to the observations regarding the compressive
strength, the values for the Modulus of Elasticity from the GFRP tests in
this study (Efc / Eft = 0.97 – 1.16) are in the scatter range of the database
(0.84 – 1.23). None of the studies considered deviates considerably from
the mean value of Efc / Eft = 1.00. Regarding the results for BFRP
specimens obtained in this study, the Moduli ratio ranges from Efc / Eft
= 0.98 to 1.03, revealing a very small range around Efc / Eft = 1.00. Test
results from Urbanski [48] underline this presumption (Efc / Eft = 0.99
–1.14). Since there is no reliable study from literature regarding the
Elastic Moduli in compression of CFRP specimens, the results from this
study stand alone at this point. However, the modulus ratio for CFRP
Fig. 10. Failure modes of the III-C (left) and II-C specimens (right). specimens also corresponds well with the target value of Efc / Eft = 1.00
(Efc / Eft = 0.95 – 0.98).
Regarding the scatter of the compressive strength, test series from
Table 7 Deitz et al. [50], Wu [35] and Tavassoli et al. [42] show the greatest
Number of failure modes CFRP. scattering values, revealing coefficients of variation of CoVffc = 24%;
failure mode II-C-E-10 III-C-E-8 III-C-E-12 23% and 17% respectively. Conversely, the test series by Galen LLC [47]
(CoVffc = 3.6%), Bruun [41] (3.9%), Khorramian and Sadeghian [46]
mode a 1 - -
mode b 8 8 10 (CoVffc = 7%) and the results from this study (Bujotzek et al. (2023))
mode c - - - (CoVffc = 6.7%) feature the smallest CoVs. The mean CoVffc for the
compressive strength based on nffc = 339 in total equals 11.5%. The
overall scatter of the Elastic Moduli in compression of FRP (CoVEfc =
reinforcement obtained within this study ranging from ffc / fft = 0.49 to 3.9%) falls significantly below this value. Results from Tavassoli et al.
0.50 show good agreement with the values from Abed et al. [37] (0.33 − [42] and Almerich-Chulia et al. [40] show strongly scattering values
0.52) and Urbanski [48] (0.34 − 0.46). Unfortunately, no reliable with CoVEfc = 13.4% and 8.7%, respectively. The very small scatter of
studies from the literature are available for comparing the CFRP test

10
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Fig. 11. Stress-strain behaviour of the test series IV-B-E-8 and IV-B-E-12.

Table 8
Test results of the series with BFRP specimens.
series IV-B-E-8 IV-B-E-12

Øim [mm] 7.41 11.72

Ønom [mm] 8 12
nffc 10 8
fft,m [N/mm2] 955 950
ffc,m [N/mm2] 465 472
ffc,m / fft,m 0.49 0.50
CoV(ffc,m) 14.1% 10.3%
nEfc 7 6
εfc,m [%] 0.97 1.02
Eft,m [N/mm2] 47,611 42,846
Efc,m [N/mm2] 46,746 44,321
Efc,m / Eft,m 0.98 1.03
CoV(Efc) 2.4% 2.0%

2.6. Influence of the test slenderness and the setup the FRP compressive
material properties

The test slenderness ltest / Ø as well as the type of test setup according
to Table 1 are the major parameters influencing the compressive ma­
terial properties of FRP apart from the material itself based on the above
considerations. For the following evaluation the GFRP specimens from
[35,37,39–42,45–47,49] are taken into consideration.
As can be observed from the ffc / fft – values plotted over the corre­
sponding test slenderness ltest / Ø in the left part of Fig. 14, the theo­
retically assumed optimum as illustrated in Fig. 2 becomes apparent
based on nffc = 231 test results. The influence of the test slenderness on
the Elastic Modulus as shown in the middle part of Fig. 14 is less pro­
nounced. Nevertheless, the Efc / Eft ratio seems to converge at slender­
ness values ltest / Ø > 4.0. Additionally, high test slenderness values do
not seem to influence the value of the Elastic Modulus in a negative way.
Regarding the CoVs of ffc and Efc plotted over ltest / Ø reveals decisively
less scatter at increasing test slenderness for both, the compressive
strength and the Elastic Modulus. This finding is strongly enforced by
the test series presented in D’Antino and Pisani [49] and AlAjarmeh
et al. [45]. In addition to that, the findings are in line with the theo­
retical assumptions presented above, regarding increased edge effects
for small test slenderness values, leading to a crucial increase in material
Fig. 12. Failure mode of the BFRP along the tensioned strand. scatter.
Fig. 15 illustrates the influence of the chosen test setup according to
the Elastic Moduli in compression determined in this study is particu­ Table 1 on the strength (ffc / fft) as well as the Moduli (Efc / Eft) – ratio.
larly remarkable, ranging between 2.2% (BFRP) and 3.0% (GFRP). The mean ffc / fft – ratio does not seem to be affected decisively by the
Khorramian and Sadeghian [43] (CoVEfc = 2.3%) and AlAjarmeh et al. type of load setup, if the specimens are sufficiently careful prepared. The
[45] (CoVEfc = 3.3%) observed values of a similar small magnitude. test results range from ffc / fft = 0.66 for setup (e) and (f) to 0.69 for

11
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Fig. 13. Results of the normalised compressive strength ffc / fft and compressive Elastic Moduli Efc / Eft from the database.

Fig. 14. Influence of the test slenderness on ffc / fft (left) and Efc / Efc (middle) and the corresponding CoVs (right).

setup (a). Note that the results from the test series on thermoplast GFRP only the test results from Wu [35] were considered. The least scattering
specimens from [49] are not integrated into the compilation since due to results are found for setup (f) (CoVEfc = 2.3%), followed by setup (e)
the comparatively small ffc / fft – ratio of 0.54 this would yield (3.1%) and setup (a) (6.0%). Setup (d) in which the load was introduced
misleading results. The highest scattering values regarding the directly via the specimens end faces by non – grouted anchor sleeves
compressive strength is observed for setup (d) (CoVffc = 16.4%) applying leads to the highest scattering values (CoVEfc = 10.7%).
the load directly to the end faces of the specimen via non-grouted steel
cases, followed by setup (e) (15.0%) and (a) (14.0%). Setup (f) which 3. Statistic evaluation
was applied for the tests conducted within this study as well as in [47]
and [49] reveals the smallest scatter (CoVffc = 6.9%). Regarding the Before the consideration of FRP reinforcement in compression within
influence of the setup type on the Elastic Moduli ratio Efc / Eft shows applicable standards becomes conceivable, an evaluation of the reli­
results corresponding well with the hypothesised value of 1.00 for setup ability requirements of such reinforced concrete members must be car­
(d), (e) and (f). The results corresponding to setup (a) introducing the ried out. In order to prepare for reliability studies of this kind, the
load directly via the end faces of the specimen deviates considerably distribution type of the material properties is checked.
from this value (Efc / Eft = 0.93). Note that for this explicit evaluation

12
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Fig. 15. Influence of the test setup according to Table 1 on ffc / fft and Efc / Eft and the corresponding scatter.

3.1. Kolmogorov-Smirnov test 3.2. Evaluation of the test results

One of the best-known goodness of fit tests is the Kolmogorov- The EDF of the single test results for the compressive strength nor­
Smirnov (KS) test [53]. This test compares the assumed theoretical malised to the mean value of the respective series (ffc,i / ffc,m) and the
Cumulative Distribution Function (CDF) to the Empirical Distribution corresponding CDF as a lognormal distribution are illustrated in Fig. 16.
Function (EDF) from the experimental results. Comparing the maximum The test results obtained within the scope of this investigation (n = 94)
deviation DKS,max between CDF and EDF to a statistic DKS,lim, which can are shown in black, whereas the overall values including the results from
be determined as a function of the sample size n and the chosen confi­ the database (n = 264) are shown in grey for direct reference. Besides
dence level (1 – α) as shown in Eq. (5) leads to acceptance or rejection of the CDF and EDF, the empirical and the theoretical distribution density
the assumed hypothesis. According to Hedderich [54], the following (PDF) is given in the right part of Fig. 16.
equation applies for the selected confidence level (1 – α) = 95%. A good approximation of the EDF by the assumed lognormal distri­
bution (CDF) can be observed from Fig. 16. The highest deviation be­
1.358
DKS, lim = √̅̅̅ for n > 35 (5) tween the two graphs is DKS,max = 0.093 for the test series conducted
n
within this study and DKS,max = 0.082 for the test series including the
For the application of the KS test within the scope of this study, the values from literature. Comparing DKS,max to the critical values for
data is evaluated separately for the compressive strength and the Elastic accepting the null hypothesis for a 95% confidence interval DKS,lim
Modulus in compression. In order to comply with the critical sample size = 0.139 (this study) and 0.0834 (this study with database) leads to the
n > 35, the data of the single test results is normalised to the mean value hypothesis being accepted. Besides the compliance with the limits of the
of the respective series (ffc,i / ffc,m, Efc,i / Efc,m) and jointly evaluated. KS test, a slight positive skew is recognisable for the density functions
Furthermore, it is distinguished between an evaluation of the data from (PDF), which is characteristic of a lognormal distribution. Given the
this study and the data from this study plus the database. With respect to direct correlation between the load share of the FRP reinforcement and
[55,56], there is sufficient reason to believe to appropriately describe its axial stiffness EAf due to the linear elastic material properties, the
the compressive material properties of FRP by means of a log-normal Elastic Modulus in compression is the most crucial material parameter of
distribution. FRP regarding the compressive capacity of FRP reinforced concrete
columns. It is therefore essential to examine the corresponding statistical

Fig. 16. CDF, EDF (left) and PDF (right) for the FRP compressive strength ffc.

13
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Fig. 17. EDF, CDF (left) and PDF (right) for the FRP Elastic Modulus in compression Efc.

characteristics carefully with regard to future reliability analyses. this study serves for a better understanding of the material behaviour
A good agreement of the EDF with the CDF is found ( Fig. 17), of FRP in compression.
regarding the test results of the Elastic Modulus in compression obtained • The presumption of the Elastic Modulus in compression Efc being
within the scope of this study (black), revealing small scatter (CoV = equal to the Elastic Modulus in tension Eft is confirmed based on the
2.9%). According to the result of the KS test (DKS,max = 0.074 < 0.175 = test results and the evaluated database. The experimental values
DKS,lim), the hypothesis of a LN distribution can be accepted at a sig­ from the tests conducted within the scope of this study correspond
nificance level of (1 – α) = 95%. Integration of the values from the well with the results from the literature, although featuring less
database into the statistical evaluation of the Elastic Modulus in scatter. For the first time, the dependence of the compressive mate­
compression leads to a slightly worse agreement between the graphs at a rial properties on the test slenderness, featuring less scatter with
simultaneously increasing scatter (CoV = 4.5%). However, the null hy­ increasing test slenderness is shown based on the test results from the
pothesis of a LN distribution can be accepted at a significance level of (1 literature and tests from this study.
– α) = 95%, (DKS,max = 0.078 < 0.103 = DKS,lim). • The highest compressive strength ratios ffc / fft were obtained for the
GFRP specimens, followed by BFRP and CFRP. The comparison of
4. Conclusions series III-G and III-C reveals a resilient conclusion regarding the
significantly superior material behaviour of GFRP compared to CFRP
Within the current paper, an extensive literature review on previous identifying GFRP as the most appropriate material for a robust design
investigations for the experimental determination of material properties of such reinforced concrete structures in compression.
of FRP in compression as well as on existing test methods was per­ • The test results of the GFRP and BFRP specimens show high values
formed. A widely arranged experimental program addressed the influ­ for the ultimate strain εfc ≥ 1.29% (GFRP); 0.97% (BFRP), signifi­
ence of decisive parameters on the FRP material properties in cantly exceeding the ultimate concrete strain (εcu = 0.35%) accord­
compression based on a reasonably chosen test setup. The obtained re­ ing to EN 1992-1-1 [57]. The ultimate strains for CFRP are also above
sults from test series on GFRP, CFRP and BFRP specimens were pre­ εcu = 0.35% (εfc,CFRP = 0.38 – 0.49%). However, due to the sensitive
sented and put into the context of experimental data obtained from the behaviour of CFRP in compression, the results for CFRP reveal need
literature. Furthermore, the influence of the test slenderness as well as of further investigations.
the test setup on the FRP compressive material properties as well as the • Based on the findings from this study, a log-normal distribution is
corresponding scatter were discussed. Eventually the joint data was identified as suitable to statistically describe the compressive
evaluated regarding the observed distribution type of the material pa­ strength and Modulus of Elasticity of FRP reinforcement independent
rameters. The following conclusions are drawn from the examinations: of the fibre material.

• The experimental results from this study show small scatter and For the design of FRP reinforced concrete members, the contribution
reproducibility of material properties as well as failure modes of FRP to the short-term load-bearing capacity can be generally rec­
regardless of FRP type, bar diameter or fibre volume content. Based ommended. However, further studies should be conducted concerning
on the evaluation of the data from the literature and the data the FRP’s long-term material behaviour in compression as well as cor­
generated within this study the applied test slenderness ltest / d = 6.0 responding reliability studies.
as well as the chosen test setup (f) can be generally recommended for
the experimental determination of FRP compressive material CRediT authorship contribution statement
properties.
• The widely arranged test program enables the authors to compare Beck Dominik: Conceptualization. Apostolidi Eftychia: Writing –
the material behaviour of FRP in compression regardless of the fibre review & editing, Validation. Bujotzek Lukas: Writing – original draft,
type. For the first time, the theoretical findings from the polymer Validation, Methodology, Investigation, Conceptualization. Waldmann
literature, particularly regarding the observed failure modes ac­ Danièle: Writing – review & editing, Validation, Supervision, Re­
cording to Hahn [14], have been analysed in a larger context sources, Methodology, Conceptualization.
including a high number of test results from previous studies. Thus,

14
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

Declaration of Competing Interest [19] S.L. Bazhenov, A.M. Kuperman, E.S. Zelenskii, A.A. Berlin, Compression failure of
unidirectional glass-fibre-reinforced plastics, Compos. Sci. Technol. 45 (3) (1992)
201–208, https://doi.org/10.1016/0266-3538(92)90080-M.
The authors declare that they have no known competing financial [20] L.B. Lessard, F.-K. Chang, Effect of load distribution on the fiber buckling strength
interests or personal relationships that could have appeared to influence of unidirectional composites, J. Compos. Mater. 25 (1) (1991) 65–87, https://doi.
the work reported in this paper. org/10.1177/002199839102500103.
[21] Fleck, N.A. and Budiansky, B. 1991. Compressive Failure of Fibre Composites Due
to Microbuckling. In Inelastic Deformation of Composite Materials, G. J. Dvorak, Ed.
Data availability Springer New York, New York, NY, 235–273. DOI=10.1007/978–1-4613–9109-8_
12.
[22] A.S. ARGON. Treatise on Materials Science & Technology, Elsevier, 1972,
The data that has been used is confidential. pp. 79–114, https://doi.org/10.1016/B978-0-12-341801-2.50007-2.
[23] Y.L. Xu, K.L. Reifsnider, Micromechanical modeling of composite compressive
Acknowledgements strength, J. Compos. Mater. 27 (6) (1993) 572–588, https://doi.org/10.1177/
002199839302700602.
[24] K.H. Lo, E.S.-M. Chim, Compressive strength of unidirectional composites, J. Reinf.
The authors gratefully acknowledge the contribution of Schöck Plast. Compos. 11 (8) (1992) 838–896, https://doi.org/10.1177/
Bauteile GmbH and Deutsche Basaltfaser GmbH by providing FRP 073168449201100801.
[25] M.R. Piggott, B. Harris, Compression strength of carbon, glass and Kevlar-49 fibre
reinforcement bars as well as Galen LLC for providing test results. The reinforced polyester resins, J. Mater. Sci. 15 (10) (1980) 2523–2538, https://doi.
authors also thank the laboratory technicians at TU Darmstadt, J. Wiech, org/10.1007/BF00550757.
E. Rauschenbach and Y. Kibnowski, for providing a significant help in [26] Orth, F. 1992. Statische und dynamische Eigenschaften von
Hochleistungsverbundwerkstoffen. Einfluß der Faser-, Matrix- und
the design of the test setup and the application of the data acquisition
Grenzschichteigenschaften auf die Verbundeigenschaften, Erlangen.
software. Furthermore, special thanks go to Mr. Maximilian Groß, M. Sc. [27] Dr. Elmar Witten. 2013. Handbuch Faserverbundkunststoffe/Composites. Springer
for the good collaboration during the execution of the experimental Fachmedien Wiesbaden, Wiesbaden.
investigations. [28] I. Choi, C.O. Horgan, Saint-Venant’s Principle and End Effects in Anisotropic
Elasticity, J. Appl. Mech. 44 (3) (1977) 424–430, https://doi.org/10.1115/
1.3424095.
References [29] EN ISO. ISO 604:2002. Determination of compressive properties. Beuth Verlag
GmbH, Berlin, 604:2003. https:// aproxy.ulb.tu-darmstadt.de / perinorm/
[1] International Federation for Structural Concrete. 2007. FRP reinforcement in RC document.aspx. Zuletzt geprüft am 7 Oktober 2020.
structures. Bulletin / International Federation for Structural Concrete Technical [30] D20 Committee. D695–15:2015. Test Method for Compressive Properties of Rigid
report 40. fib, Lausanne. Plastics. ASTM International, West Conshohocken, PA.
[2] American Concrete Institute. 2015. ACI 440.1R-15. Guide for the design and [31] ISO. 2000. ISO 14126:1999. Bestimmung der Druckeigenschaften in der Laminatebene.
construction of structural concrete reinforced with fiber-reinforced polymer (FRP) Beuth Verlag GmbH, Berlin, 14126:1999. Zuletzt geprüft am 16 Oktober 2020.
bars. ACI 440.1R-15. American Concrete Institute, Farmington Hills, Mich. 440.1R- [32] D30 Committee. D3410:2016. Test Method for Compressive Properties of Polymer
15, ACI-440.1R-15. Matrix Composite Materials with Unsupported Gage Section by Shear Loading. ASTM
[3] Nanni, A., Luca, A. de, and Jawaheri Zadeh, H. 2014. Reinforced concrete with FRP International, West Conshohocken, PA.
bars. Mechanics and design. CRC Press, Boca Raton. doi: 10.14359/51663912. [33] DIN EN. 2018. DIN EN 2850:2018. Luft- und Raumfahrt - Unidirektionale Laminate
[4] NANNI ANTONIO. Fiber-Reinforced-Plastic (FRP) Reinforcement for Concrete aus Kohlenstoffasern und Reaktionsharz - Druckversuch parallel zur Faserrichtung
Structures, Elsevier, 1993, https://doi.org/10.1016/C2009-0-09136-3. 49.025.40. Beuth Verlag GmbH, Berlin 49.025.40. Zuletzt geprüft am 16 Oktober
[5] Committee. Test Method for Tensile Properties of Fiber Reinforced Polymer Matrix 2020.
Composite Bars . a ASTM International West Conshohocken, PA , 2016 doi: [34] GOST. 2014. GOST 31938:2012. FIBRE-REINFORCED POLYMER BAR FOR
10.1520/D7205_D7205M-06R16. CONCRETE REINFORCEMENT. Interstate Council fpr Standardization, Metrology
[6] Canadian Standard Association. CSA-S806-2. Design and construction of building and Certification (ISC), Moscow, 31938–2012.
components with fibre-reinforced polymers. CSA standard. Canadian Standards Ass, [35] Wu, W.P. 1990. Thermomechanical Properties of Fiber Reinforced Plastic (FRP)
Rexdale Ont., 2007. Bars. Ph.D. Thesis, Morgantown, USA.
[7] N. Elmessalami, A. El Refai, F. Abed, Fiber-reinforced polymers bars for [36] O. Chaallal, B. Benmokrane, Physical and mechanical performance of an
compression reinforcement: A promising alternative to steel bars, Constr. Build. innovative glass-fiber-reinforced plastic rod for concrete and grouted anchorages,
Mater. 209 (2019) 725–737, https://doi.org/10.1016/j.conbuildmat.2019.03.105. Can. J. Civ. Eng. 20 (2) (1993) 254–268, https://doi.org/10.1139/l93-031.
[8] A. Hadhood, H.M. Mohamed, B. Benmokrane, Experimental study of circular high- [37] F. Abed, Z. Mehaini, C. Oucif, A. Abdul–Latif, R. Baleh, Quasi-static and Dynamic
strength concrete columns reinforced with GFRP bars and spirals under concentric Response of GFRP and BFRP Bars under Compression, Composites Part C (2020)
and eccentric loading, J. Compos. Constr. 21 (2) (2017), https://doi.org/10.1061/ 100034, https://doi.org/10.1016/j.jcomc.2020.100034. Open Access.
(ASCE)CC.1943-5614.0000734. [38] Kobayashi, K. & Fujisaki, T. 1995. Non-Metallic (FRP) Reinforcement for Concrete
[9] M. Elchalakani, G. Ma, Tests of glass fibre reinforced polymer rectangular concrete Structures. Proceedings of the Second International RILEM Symposium. Chapman
columns subjected to concentric and eccentric axial loading, Eng. Struct. 151 and Hall/CRC, Boca Raton.
(2017) 93–104, https://doi.org/10.1016/j.engstruct.2017.08.023. [39] D.H. Deitz, I.E. Harik, H. Gesund, Physical properties of glass fiber reinforced
[10] Ehrenstein, G.W. 2006. Faserverbund-Kunststoffe. Werkstoffe - Verarbeitung - polymer rebars in compression, J. Compos. Constr. 7 (4) (2003) 363–366, https://
Eigenschaften. Carl Hanser Verlag GmbH & Co. KG, München. doi.org/10.1061/(ASCE)1090-0268.
[11] Schürmann, H. 2007. Konstruieren mit Faser-Kunststoff-Verbunden. VDI-Buch. [40] A. Almerich-Chulia, P. Martin-Concepcion, J.M. Molines-Cano, J. Rovira-Soler,
Springer-Verlag Berlin Heidelberg, Berlin, Heidelberg. Innovative design of GFRP bars for concrete structures, AMR 457-458 (2012)
[12] Teschner, R. 2013. Glasfasern. Springer Berlin Heidelberg, Berlin, Heidelberg. 553–556.
[13] Hofmann, S. 2021. Rissentwicklung in Betonbauteilen mit [41] E. Bruun, GFRP bars in structural design: determining the compressive strength
Basaltfaserkunststoffbewehrung. Ein Modell zur Berechnung der Rissbreite bei versus unbraced length interaction curve, Can. Young Sci. J. 2014 (1) (2014)
wirklichkeitsnaher Betrachtung des Verbundverhaltens. Dissertation / Institut für 22–29.
Massivbau, Technische Universität Darmstadt Heft 50. Technische Universität [42] A. Tavassoli, J. Liu, S. Sheikh, Glass fiber-reinforced polymer-reinforced circular
Darmstadt, Institut für Massivbau, Darmstadt. columns under simulated seismic loads, Acids Struct. J. 112 (2015) 1.
[14] Hahn HT, W. J. G. 1986 Compression Failure Mechanisms in Unidirectional [43] Khorramian, Koosha and Sadeghian, Pedram. 2018. NEW TESTING METHOD OF
Composites. Composite Materials: Testing and design (seventh conference): a GFRP BARS IN COMPRESSION. Building Tomorrow’s Society, MA7–1 MA7–10.
conference: Philadelphia, PA, 2–4 April 1984. ASTM special technical publication [44] B. Fillmore, P. Sadeghian, Contribution of longitudinal glass fiber-reinforced
893. ASTM, Philadelphia, PA. polymer bars in concrete cylinders under axial compression, Can. J. Civ. Eng. 45
[15] H. HAHN, M. SOHI, Buckling of a fiber bundle embedded in epoxy, Compos. Sci. (6) (2018) 458–468, https://doi.org/10.1139/cjce-2017-0481.
Technol. 27 (1) (1986) 25–41. [45] O. Alajarmeh, A. Manalo, B. Benmokrane, P. Vijay, W. Ferdous, P. Mendis, Novel
[16] N.K. Naik, R.S. Kumar, Compressive strength of unidirectional composites: testing and characterization of GFRP bars in compression, Constr. Build. Mater.
evaluation and comparison of prediction models, Compos. Struct. 46 (3) (1999) 225 (2019) 1112–1126, https://doi.org/10.1016/j.conbuildmat.2019.07.280.
299–308, https://doi.org/10.1016/S0263-8223(99)00098-7. [46] K. Khorramian, P. Sadeghian. Behavior of Slender GFRP Reinforced Concrete
[17] Bujotzek, L., Hiesch, D., El Ghadioui, R., and Proske, T. 2021. Material Properties of Columns. In Structures Congress 2019. Bridges, Nonbuilding and Special
Fibre Reinforced Polymer (FRP) Reinforcement in Compression – A Review. Concrete Structures, and Nonstructural Components, American Society of Civil Engineers,
Structures: New Trends for Eco-Efficiency and Performance. Fédération Reston, Reston, 2019, pp. 88–99, https://doi.org/10.1061/9780784482230.
Internationale du Béton (fib) – International Federation for Structural Concrete, [47] Galen L.L.C. 2017. TEST REPORT 𝒩 ◦К.386–17.B.AKS-6. NATIONAL RESEARCH
Lisbon. MOSCOW STATE UNIVERSITY OF CIVIL ENGINEERING, Moskau.
[18] L.A. Carlsson, R.B. Pipes, Hochleistungsfaserverbundwerkstoffe. Herstellung und [48] M. Urbański, Compressive strength of modified FRP hybrid bars, Materials 13 (8)
experimentelle Charakterisierung, Teubner Studienbücher Mechanik. Vieweg+ (2020), https://doi.org/10.3390/ma13081898. Basel, Switzerland.
Teubner Verlag, Wiesbaden, 1989.

15
L. Bujotzek et al. Construction and Building Materials 414 (2024) 134782

[49] T. D’Antino, M.A. Pisani, Tensile and compressive behavior of thermoset and [54] Hedderich, J. 2016. Angewandte Statistik. Methodensammlung mit R. Springer eBook
thermoplastic GFRP bars, Constr. Build. Mater. 366 (2023), https://doi.org/ Collection. Springer Spektrum, Berlin, Heidelberg.
10.1016/j.conbuildmat.2022.130104. [55] Rempel, S. 2018. Zur Zuverlässigkeit der Bemessung von biegebeanspruchten
[50] Deitz, D., Harik, I.E., Gesund, H. GFRP Reinforced Concrete Bridge Decks 2000 Betonbauteilen mit textiler Bewehrung. IMB, Lehrstuhl und Institut für Massivbau,
doi: 10.13023/KTC.RR.2000.09. RWTH Aachen.
[51] D. Hiesch, T. Proske, C.-A. Graubner, L. Bujotzek, R. El Ghadioui, Theoretical and [56] Joint Committee on Structural Safety. 2001. JCSS. PROBABILISTIC MODEL CODE.
experimental investigation of the time-dependent relaxation rates of GFRP and JCSS.
BFRP reinforcement bars, Struct. Concr. (2023), https://doi.org/10.1002/ [57] DIN EN 1992–1-1:2011–01, Eurocode_2: Bemessung und Konstruktion von
suco.202200212. Stahlbeton- und Spannbetontragwerken_- Teil_1–1: Allgemeine Bemessungsregeln
[52] ACI Committee 318. 2011. Building code requirements for structural concrete (ACI und Regeln für den Hochbau; Deutsche Fassung EN_1992–1-1:2004_+ AC:2010.
318–11M) and commentary. American Concrete Institute, Farmington Hills, MI. Beuth Verlag GmbH, Berlin.
[53] N. Smirnov, Table for Estimating the Goodness of Fit of Empirical Distributions,
Ann. Math. Stat. 19 (2) (1948) 279–281.

16