You are on page 1of 7

Cement & Concrete Composites 26 (2004) 299305

An overview of the fatigue behaviour of plain and bre

reinforced concrete
M.K. Lee, B.I.G. Barr *

Cardi School of Engineering, Division of Civil and Structural Engineering, Cardi University, Queens Building,
P.O. Box 925, Newport Road, Cardi CF24 OYF, Wales, UK
Received 23 July 2002; accepted 26 November 2002

The paper provides a general overview of recent developments in the study of the fatigue behaviour of plain and bre reinforced
concrete (FRC). The fatigue performance of plain concrete and FRC, as reported in the literature, is compared in order to quantify
the inuence of bre inclusion on fatigue behaviour. Despite the conicting information regarding the fatigue behaviour of concrete
reported in the literature, the majority of researchers show that the inclusion of bres can benet the fatigue performance of
2003 Elsevier Ltd. All rights reserved.

Keywords: Fatique behaviour; Concrete; FRC; Structures

1. Introduction internal microcracks, which results in a signicant in-

crease of irrecoverable strain. At the macrolevel, this
The paper provides a general review of recent devel- will manifest itself as changes in the materials me-
opments in the study of the fatigue behaviour of plain chanical properties.
and bre reinforced concrete (FRC). Many structures Fatigue loading is usually divided into two categories
are often subject to repetitive cyclic loads. Examples of [1] i.e. low-cycle and high-cycle loading. Low-cycle
such cyclic loads include machine vibration, sea waves, loading involves the application of a few load cycles at
wind action and automobile trac. The exposure to high stress levels. On the other hand, high cyclic loading
repeated loading results in a steady decrease in the is characterised by a large number of cycles at lower
stiness of the structure, which may eventually lead to stress levels. Hsu presents a wider range of fatigue load
fatigue failure. Although concrete is a widely used spectrum with the inclusion of super-high cycle loading
construction material, the understanding of fatigue [2]. Table 1 summarises the dierent classes of fatigue
failure in cementitious composites is still lacking in loading.
comparison to that of ferrous materials. This incomplete As in the case of static tests, dierent loading ar-
understanding is even more pronounced for composite rangements have been used in fatigue testing, including
materials such as FRC. compression, tension and bending tests. The most
common method of fatigue testing, by far, is via exural
tests. To a lesser extent, compressive fatigue tests have
2. General background also been investigated. In recent years, there has been
more interest in the fatigue characteristics of concrete in
Fatigue may be dened as a process of progressive, tension [35], especially since the introduction of non-
permanent internal structural changes in a material linear fracture mechanics in the analysis of concrete. In
subjected to repeated loading. In concrete, these changes addition, some researchers have studied the eects of
are mainly associated with the progressive growth of combined stresses to the fatigue performance of concrete
[6,7] where it has been found that the fatigue strength of
Corresponding author. Tel./fax: +44-29-2087-4826. concrete in biaxial compression is greater than that
E-mail address: (B.I.G. Barr). under uniaxial compression.

0958-9465/$ - see front matter 2003 Elsevier Ltd. All rights reserved.
300 M.K. Lee, B.I.G. Barr / Cement & Concrete Composites 26 (2004) 299305

Table 1
Classes of fatigue load, after [2]
Low-cycle fatigue High-cycle fatigue Super-high-cycle fatigue
1 2 3 4 5 6
1 10 10 10 10 10 10 107 108 109
Structures subjected to earthquakes Airport pavements and Highway and railway Mass rapid transit Sea structures
bridges bridges, highway structure

Various approaches have been used in the fatigue life there is no known stress level below which the fatigue
assessment of structural elements. A widely accepted life of plain concrete will be innite.
approach for engineering practice is based on empiri- In general, parameters such as loading condi-
cally derived SN diagrams, also known as Wh oler tions, load frequency, boundary conditions, stress level,
curves. In addition, the eects of minimum stress in the number of cycles, matrix composition, stress ratio will
loading cycle may be represented in so-called Goodman inuence the fatigue performance of the concrete spec-
diagrams or Smith diagrams, which are also known as imen. However, the qualitative and quantitative nature
constant life diagrams in the analyses of metals [8]. of these parameters on the fatigue performance of con-
These empirical curves give a graphical representation of crete is yet to be agreed in the literature.
the fatigue performance for certain loading parameters.
Another method is based on fracture mechanics con-
cepts and has been incorporated in a nite element ap- 4. Fatigue of bre reinforced concrete
proach [9,10]. This method is more demanding but
provides an insight into the underlying physical behav- The use of FRC in engineering applications has fur-
iour. thered the need for the study of its behaviour under
fatigue loading. Common applications for FRC include
paving applications such as in airports, highways, bridge
3. Fatigue of plain concrete decks and industrial oors [16], which endure signicant
cyclic loading during their service life. Within these ar-
Concrete is a heterogeneous material which is inher- eas of application, the fatigue characteristics of FRC are
ently full of aws (such as pores, air voids, lenses of important performance and design parameters. How-
bleed water under coarse aggregates and shrinkage ever, there seems to be a gulf in the knowledge of the
cracks). The mechanism of fatigue failure in concrete or fatigue behaviour of FRC in terms of all the inuencing
mortar can be divided into three distinct stages [11]. The variables such as type of loading cycle, strain rates and
rst stage involves the weak regions within the concrete bre parameters.
or mortar and is termed aw initiation. The second stage Generally, it has been observed that the addition of
is characterised by slow and progressive growth of the steel bres can signicantly improve the bending fatigue
inherent aws to a critical size and is generally known as performance of concrete members [1719]. The extent of
microcracking. In the nal stage, when a sucient improvement on the fatigue capacity of FRC can be
number of unstable cracks have formed, a continuous or expected to depend upon the bre volume content, bre
macrocrack will develop, eventually leading to failure. type and geometry. Various combinations of these pa-
Fatigue crack growth can be divided into two distinct rameters will give rise to dierent fatigue characteristics.
stages [12]; the rst stage is a deceleration stage, where However, at the moment, there does not seem to be a
the rate of crack growth decreases as the crack grows comprehensive appreciation of the advantages than can
and the second stage is an acceleration stage, where be attained with bre addition, as there is limited in-
there is a steady increase in the crack growth rate right formation regarding the quantitative inuence and rel-
up to failure [13]. ative importance of bre parameters such as amount,
It has been surmised that dierent loading cycle re- aspect ratio and bre type. In general, the addition
gimes produce dierent failure mechanisms within con- of bres has added a further dimension to the study of
crete. For low-cycle fatigue, the dominant mechanism is fatigue in concrete and has increased the complexity of
the formation of mortar cracks leading to continuous analysis.
cracked networks. On the other hand, high-cycle fatigue As stated previously, the development of fatigue
produces bond cracks in a slow and gradual process [14]. failure in concrete can be divided into three stages. It is
Unlike ferrous metals, concrete does not appear to feasible to retard and inhibit the growth of the aws in
have a fatigue limit. It has been reported that plain the second stage by introducing closely spaced and
concrete subjected to repeated uniaxial tensile stresses randomly dispersed bres as reinforcements. In FRC,
exhibits no fatigue limit under 2  106 cycles [15]. Hence the action of bre bridging and bre pullout dissipates
M.K. Lee, B.I.G. Barr / Cement & Concrete Composites 26 (2004) 299305 301

energy in the wake of the crack tip. This mechanism For uniaxial compression, 30 mm bres increased
plays a dominant role in inhibiting crack growth and fatigue life while 60 mm bres actually reduces it. On the
therefore increases the load carrying capacity of FRC other hand, for exural fatigue tests, it appears that only
specimens. a marginal benet comes from bre addition. To ex-
It is hoped that the addition of bres will endow the plain these observations, it was concluded that the ad-
FRC with a fatigue limit, thus making it a much more ditional aws introduced by bre addition outweighed
attractive material than plain concrete, which appears to the benets for some of the tests carried out.
have no such limit [2]. Li and Matsumoto [20], through The main benet of the addition of bres in the
their model, showed that a fatigue limit exists for FRC. concrete matrix is the increased ability to absorb energy.
Ramakrishnan and Lokvik [21] suggests that FRC Increasing the bre content and aspect ratio increases
reaches an endurance limit at approximately 2  106 the amount of energy spent on crack growth of SFRC
loading cycles. However, it has been proposed that tests under fatigue load [28]. The main bre parameter in-
up to 10  106 cycles need to be carried out to conrm uencing the fatigue performance of FRC seems to be
this conclusion [17]. From the numerous reported nd- the bre content. On the other hand, the aspect ratio
ings, it is obvious that the question of whether FRC has and bre type is secondary in importance [17,29].
a fatigue limit remains unresolved.
The addition of steel bres has been found to sub-
stantially improve load-bearing capacity and resistance
to crack growth [22]. As in the case of static loading, 5. Comparison using results from literature
bres have been found to result in the development of a
large number of small cracks rather than a small number The majority of fatigue life prediction and design of
of large cracks [18]. Furthermore, the inclusion of bres plain and FRC structures have been carried out empir-
produces a more ductile behaviour during fatigue ically. This method involves time consuming testing for
loading [23,24]. a broad range of design cases, which in principle may be
A signicantly higher level of damage in static as well inapplicable to other design cases.
as fatigue testing of SFRC has been found compared to To date, there is no standard procedure for carrying
that observed in plain concrete [19]. This is further out fatigue tests on concrete or FRC. Strictly speaking,
supported by ndings reported in [25] where it is con- fatigue data obtained from a particular test set-up can-
cluded that FRC could undergo larger strains before not be directly compared to data obtained from dierent
failure, compared to plain concrete. loading congurations [13]. However, SN curves are
The addition of bre reinforcement has been found to plotted using strength (or stress) values, which are made
have a dual eect on the cyclic behaviour of concrete. dimensionless by relating them to the static strength.
Fibres are able to bridge microcracks and retard their The dimensionless term, S, in part eliminate inuences
growth, thereby enhancing the composites performance such as specimen shape, watercement ratio, type and
under cyclic loading. On the other hand, the presence of grading of aggregates, concrete strength, curing condi-
bres increases the pore and initial microcrack density, tion, moisture conditions and age at loading etc. Di-
resulting in strength decrease. The overall outcome of mensionless SN curves are thought to represent as close
these two competing eects depends signicantly on the as possible the true behaviour of concrete under fatigue
bre volume [18]. loading [14].
It has been suggested that the presence of bres only There is no agreement whether SN curves may be
help to enhance the composite behaviour in fatigue in used for all types of specimens, loading congurations,
the low cycle region (up to approximately 103 cycles) [7]. testing conditions etc. However, by plotting some of the
Fibres are not seen to provide any improvement for available data in the literature, it is hoped that some
higher number of cycles. This is elucidated by the dif- general trends may be identied. More specically, it
ferentiation between mortar and bond cracking [14]. The would be interesting to see whether the fatigue data
presence of bres is able to increase the fatigue life in the would be able to show the qualitative benets of bre
part of mortar cracking (low cycle region), but is unable addition. Nevertheless, it must be kept in mind that, due
to do so when bond cracking (high cycle region) starts. to dierences in the test programmes, a direct relation-
Consequently, the addition of bres is deemed to be ship cannot be made and this approach only provides a
unable to increase the fatigue limit (if such a thing exist) general comparison.
of concrete. Paskova and Meyer [26] suggest that bres Various researchers have carried out compressive
tend to dissipate more energy at lower stress levels fatigue tests on plain concrete and SFRC [18,26,27,30].
compared to higher stress levels. Similarly, exural fatigue tests on plain concrete and
Most researchers agree that FRC has better fatigue SFRC have been extensively studied [17,27,28,3136].
behaviour compared to plain concrete. However, there For the purpose of the comparison carried out in this
is conicting evidence based on the work of Cachim [27]. review, only work carried out using steel bres with bre
302 M.K. Lee, B.I.G. Barr / Cement & Concrete Composites 26 (2004) 299305

1 1

y = -0.0581x + 0.9979 y = -0.0632x + 0.9987

0.9 2 0.9 2
R = 0.6332
R = 0.6684

0.8 0.8
Stress level, S

Stress level, S
0.7 0.7

0.6 0.6
Paskova and Meyer (1997)
Paskova and Meyer (1997)
0.5 0.5
Grzybowski and Meyer (1993) Grzybowski and Meyer (1993)

Cachim (1999)
0.4 0.4 Cachim (1999)
Do et al (1993)
Run-out specimen Run-out specimen
0.3 0.3
0 2 4 6 8 0 2 4 6 8
(a) Fatigue life, log N (b) Fatigue life, log N

1 1

0.9 0.9

0.8 0.8
Stress level, S

Stress level, S

0.7 y = -0.0497x + 0.9697 0.7

R = 0.2371

0.6 0.6

Plain concrete
0.5 0.5
Paskova and Meyer (1997)
FRC (0.5%)
0.4 0.4
Grzybowski and Meyer (1993) FRC (1%)

0.3 0.3
0 1 2 3 4 0 2 4 6 8
(c) Fatigue life, log N (d) Fatigue life, log N

Fig. 1. (a) SN curve for plain concrete under compression; (b) SN curve for SFRC (0.5% bre content) under compression; (c) SN curve for
SFRC (1.0% bre content) under compression. (N.B. Change of scale for log N ); (d) comparison between SN curves for plain and SFRC (0.5% and
1.0% bre content) under compression.

concentrations (by volume) of 0.5% and 1.0% have been pretation of the trendlines shown in Fig. 1 since the
considered. coecient of determination, R2 , is signicantly less than
Fig. 1(a) shows the SN curve obtained from an unity. For a more meaningful comparison, Fig. 1(d)
analysis by the authors on the test results extracted from shows the linear regression lines for the results shown
the literature [18,26,27,30] for plain concrete in com- separately in the previous three gures. There appears to
pression. On the other hand, Fig. 1(b) and (c) present be a slight degradation in the fatigue life of SFRC rel-
the SN curves for SFRC containing 0.5% and 1.0% of ative to plain concrete under compression loading. A
bres under compression fatigue loading respectively. similar trend has been reported in [27] for only one test
Fatigue test results in the literature shows signicant series. This was attributed to the introduction of addi-
spread in the results and care is required in the inter- tional aws within the concrete matrix by the bres.
M.K. Lee, B.I.G. Barr / Cement & Concrete Composites 26 (2004) 299305 303

1 1
y = -0.0606x + 1.0327
R = 0.7243
0.9 0.9

0.8 0.8
y = -0.0575x + 1.0727
Stress level, S

Stress level, S
R = 0.6062
0.7 0.7
Zhang et al (1996)
Zhang et al (1997)
0.6 0.6
Shi et al (1993)
Bazant and Schell (1993) Singh and Kaushik (2000)
0.5 Oh (1991b) 0.5
Chang and Chai (1995)
Johnston and Zemp (1991)
Johnston and Zemp (1991)
0.4 Cachim (1999) 0.4
Chang and Chai (1995) Cachim (1999)
Run-out specimen Run-out specimen
0.3 0.3
0 2 4 6 8 0 2 4 6 8
(a) Fatigue life, log N (b) Fatigue life, log N

1 1
y = -0.0559x + 1.0854
R = 0.7309
0.9 0.9

0.8 0.8
Stress level, S

Stress level, S

0.7 0.7

0.6 0.6
Plain concrete
Singh and Kaushik (2000)
0.5 0.5
Chang and Chai (1995) FRC (0.5%)

0.4 Johnston and Zemp (1991) 0.4

FRC (1%)
Run-out specimen
0.3 0.3
0 2 4 6 8 0 2 4 6 8
(c) Fatigue life, log N (d) Fatigue life, log N

Fig. 2. (a) SN curve for plain concrete under exural loading; (b) SN curve for SFRC (0.5% bre content) under exural loading; (c) SN curve for
SFRC (1.0% bre content) under exural loading; (d) comparison between SN curves for plain and SFRC (0.5% and 1.0% bre content) under
exural loading.

Fig. 2(a) presents the SN curve for plain concrete loading, there appears to be a signicant benet derived
under exural loading. Similarly, Fig. 2(b) and (c) give from the addition of bres. The improvement is slightly
the SN curves for SFRC containing 0.5% and 1.0% of greater when the bre content is increased from 0% to
bres under exural loading respectively. Similarly the 0.5% compared to the improvement achieved between
R2 values are signicantly less than unity for these test 0.5% and 1.0%. A comparison between the contradic-
results but are slightly better than those observed for the tory trends between SFRC under compressive and
compression test results. Finally, Fig. 2(d) compares the exural fatigue loading suggests that SFRC is more ef-
linear regression lines for all three test results in exure. fective under the latter conditions. This is to be ex-
Contrary to the observations for compressive fatigue pected, since the bres would be able to bridge cracks
304 M.K. Lee, B.I.G. Barr / Cement & Concrete Composites 26 (2004) 299305

and prolong fatigue life. The presence of bres cannot Acknowledgements

display their true eectiveness under compressive load-
ing, as the mode of failure does not induce a signicant The work reported in this paper forms part of the
contribution from the bres. Brite-Euram project Test and Design Methods for
Due to the variety of testing congurations, materials Steel Fibre Reinforced Concrete, contract no. BRPR-
and procedures, the conclusions given above are only CT98-0813. The partners in the project are: N.V. Be-
tentative. However, due to the numerous sources of fa- kaert S.A. (Belgium co-ordinator), Centre Scientique et
tigue data analysed, most systematic variations should Technique de la Construction (Belgium), Katholieke
have been removed in compiling the various test results. Universiteit Leuven (Belgium), Technical University of
Hence, the qualitative trends observed should be gen- Denmark (Denmark), Balfour Beatty Rail Ltd (Great
erally applicable. Britain), University of Wales Cardi (Great Britain),
Fertig-Decken-Union GmbH (Germany), Ruhr-Uni-
versity-Bochum (Germany), Technical University of
Braunschweig (Germany), FCC Construccion S.A.
6. Conclusions (Spain), Universitat Politecnica de Catalunya (Spain).

There is a signicant amount of conicting infor-

mation in the literature regarding the fatigue behaviour References
of cementitious materials. Even for the case of plain
concrete, many contradicting trends have been re- [1] RILEM Committee 36-RDL. Long term random dynamic loading
ported. Therefore, it is hardly surprising that there of concrete structures. Mater Struct 1984;17(97):128.
does not seem to be a general consensus about the [2] Hsu TCC. Fatigue of plain concrete. ACI J 1981;78:292305.
eects of bre inclusion on the fatigue life of concrete. [3] Saito M. Characteristics of microcracking in concrete under static
and repeated tensile loading. Cem Concr Res 1987;17:2118.
The inclusion of bres has added a further dimension [4] Cornelissen HAW. Fatigue failure of concrete in tension.
to the understanding of the fatigue of concrete. Even HERON 1984;29(4):68.
when considering bre parameters alone, there are [5] Zhang J, Stang H, Li VC. Experimental study on crack bridging in
numerous combinations which can inuence the fatigue FRC under uniaxial fatigue tension. J Mater Civil Eng
behaviour of FRC. In addition, there are various 2000;12:6673.
[6] Su ECM, Hsu TCC. Biaxial compression fatigue and discontinu-
combinations of loading frequency, load sequencing, ity of concrete. ACI Mater J 1988;3:78188.
matrix composition, test conguration etc. which can [7] Yin W, Hsu TCC. Fatigue behavior of steel ber reinforced
be expected to change the behaviour of FRC under concrete in uniaxial and biaxial compression. ACI Mater J
cyclical loads. 1995;92(1):7181.
Nevertheless, the majority of researchers have, to a [8] Sendeckyj GP. Constant life diagramsa historical review. Int J
Fatigue 2001;23:34753.
limited extent, found that the inclusion of bres can [9] Reinhardt HW, Cornelissen HAW, Hordijk DA. Tensile tests and
benet the fatigue performance of concrete. However, failure analysis of concrete. J Struct Eng 1986;112(11):246277.
the quantitative nature of this benet is dicult to de- [10] Hordijk DA. Local approach to fatigue of concrete. PhD thesis,
termine. The results analysed in this paper would sug- Delft University of Technology, 1991, 210 pp.
gest that an endurance limit (if such exist) has not been [11] Gao L, Hsu TCC. Fatigue of concrete under uniaxial compression
cyclic loading. ACI Mater J 1998;95(5):57581.
observed before 106 cycles. Further experimental work [12] Horii H, Shin HC, Pallewatta TM. Mechanism of fatigue crack
beyond 106 cycles is required before drawing rm con- growth in concrete. Cem Concr Compos 1992;14:839.
clusions regarding an endurance limit. [13] Kolluru SV, ONeil EF, Popovics JS, Shah SP. Crack propagation
The lack of a well-established test procedure for ex- in exural fatigue of concrete. J Eng Mech 2000;126(9):8918.
ecuting and evaluating fatigue tests makes it dicult to [14] Hsu TCC. Fatigue and microcracking of concrete. Mater Struct
correlate or extend published test results. However, a [15] Saito M, Imai S. Direct tensile fatigue of concrete by the use of
compilation of fatigue data from numerous sources has friction grips. ACI J 1983;80:4318.
been carried out to observe general trends from which [16] Krstulovic-Opara N, Haghayeghi AR, Haidar M, Krauss PD. Use
tentative conclusion have been drawn. The presence of of conventional and high-performance steel-ber reinforced con-
bres does not seem to enhance the fatigue life of con- crete for bridge deck overlays. ACI Mater J 1995;92(6):66977.
[17] Johnston CD, Zemp RW. Flexural fatigue performance of steel
crete under compressive fatigue loading. On the other ber reinforced concreteinuence of ber content, aspect ratio
hand, bre addition benets the fatigue performance and type. ACI Mater J 1991;88(4):37483.
under exural fatigue loading. A possible explanation is [18] Grzybowski M, Meyer C. Damage accumulation in concrete with
that under tensile forces, the bres are able to bridge and without ber reinforcement. ACI Mater J 1993;90(6):594604.
cracks and prolong fatigue life. On the contrary, the [19] Zhang J, Stang H. Fatigue performance in exure of ber
reinforced concrete. ACI Mater J 1998;95(1):5867.
presence of bres cannot display their true eectiveness [20] Li VC, Matsumoto T. Fatigue crack growth analysis of ber
under compressive loading, as the mode of failure is reinforced concrete with eect of interfacial bond degradation.
dierent. Cem Concr Compos 1998;20:33951.
M.K. Lee, B.I.G. Barr / Cement & Concrete Composites 26 (2004) 299305 305

[21] Ramakrishnan V, Lokvik BJ. Flexural fatigue strength of ber [28] Chang D-I, Chai W-K. Flexural fracture and fatigue behavior of
reinforced concretes. In: Reinhardt HW, Naaman AE, editors. steel-ber-reinforced concrete structures. Nucl Eng Des
High Performance Fiber Reinforced Cement Composites: Pro- 1995;156:2017.
ceedings of the International RILEM/ACI Workshop. London: [29] Naaman AE, Hammoud H. Fatigue characteristics of high
E&FN SPON; 1992. p. 27187. performance ber-reinforced concrete. Cem Concr Compos
[22] Morris AD, Garrett GG. A comparative study of the static and 1998;20:35363.
fatigue behaviour of plain and steel bre reinforced mortar in [30] Do M-T, Chaallal O, Atcin P-C. Fatigue behavior of high-
compression and direct tension. Int J Cem Compos Lightweight performance concrete. J Mater Civil Eng 1993;5(1):96111.
Concr 1981;3(2):7391. [31] Zhang B, Phillips DV, Wu K. Eects of loading frequency and
[23] Kwak K-H, Suh J, Hsu TC-T. Shear-fatigue behavior of steel ber stress reversal on fatigue life of plain concrete. Mag Concr Res
reinforced concrete beams. ACI Struct J 1991;88(2):15560. 1996;48(177):36175.
[24] Spadea G, Bencardino F. Behavior of ber-reinforced concrete [32] Zhang B, Phillips DV, Wu K. Further research on fatigue
beams under cyclic loading. J Struct Eng 1997;123(5):6608. properties of plain concrete. Mag Concr Res 1997;49(180):24152.
[25] Rafeeq AS, Gupta A, Krishnamoorthy S. Inuence of steel bers [33] Shi XP, Fwa TF, Tan SA. Flexural fatigue strength of plain
in fatigue resistance of concrete in direct compression. J Mater concrete. ACI Mater J 1993;90(5):43540.
Civil Eng 2000;12(2):1729. [34] Bazant ZP, Schell WF. Fatigue fracture of high-strength concrete
[26] Paskova T, Meyer C. Low-cycle fatigue of plain and ber- and size eect. ACI Mater J 1993;90(5):4728.
reinforced concrete. ACI Mater J 1997;94(4):27385. [35] Oh BH. Fatigue-life distributions of concrete for various stress
[27] Cachim PB. Experimental and numerical analysis of the behaviour levels. ACI Mater J 1991;88(2):1228.
of structural concrete under fatigue loading with applications to [36] Singh SP, Kaushik SK. Flexural fatigue life distributions and
concrete pavements. PhD thesis, Faculty of Engineering of the failure probability of steel brous concrete. ACI Mater J
University of Porto, 1999, 246 pp. 2000;97(6):65867.