ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 105-S67

Evaluation of Moment Distribution in Continuous

Fiber-Reinforced Polymer-Strengthened Concrete Beams
by P. F. Silva and T. J. Ibell

Fiber-reinforced polymer (FRP) strengthening offers many well-

documented benefits for the retrofit of existing reinforced concrete
(RC) structures. The main drawback in using FRP for such
applications, however, is the reduction in ductility capacity after
strengthening. This loss in ductility has led worldwide codes to not
consider redistribution of bending moments in continuous FRP-
strengthened RC beams. This implies that an unstrengthened
continuous RC beam that was previously designed under assumptions
of moment redistribution (MR) and is to be strengthened with FRP
must now be redesigned according to full elastic distribution of
bending moments. This could lead to onerous conditions in such
strengthening applications. This paper sets out a rationale for the
possible appropriate use of redistribution principles for FRP-
strengthened beams. Analytical results show that if a section can
develop a curvature ductility capacity greater than 2.0, it is likely
that redistribution in the order of at least 7.5% can be achieved.
These results are presented and further discussed in this paper.

Keywords: continuous beams; fiber-reinforced polymer; moment redistribution.

INTRODUCTION
Extensive application examples and research exist in the
literature regarding the strengthening of reinforced concrete
(RC) structures using fiber-reinforced polymer (FRP)
composites. It is a well-known fact that strengthening avoids
the need to demolish and replace existing structures, thus Fig. 1—Plastic rotations and moment redistribution in
enabling the design life of current structures to be increased three-span continuous beam.
(Tumialan et al. 2002; Meier et al. 1993; Task Group 9.3
2001; The Concrete Society 2000). The principal advantages
serviceability demands. As such, if it could be shown that
of using FRP composites for such applications have led to
rapid, cost-effective strengthening schemes, so that FRP is sufficient curvature ductility exists in an FRP-strengthened
now a commonly used alternative for the retrofit of concrete region to allow for some level of moment redistribution, it
structures (Teng et al. 2002). One of the main drawbacks in would be advantageous in reducing the required amount of
using FRP is that in many design applications: 1) full FRP-strengthening material.
advantage is not taken of the strength of FRP because Fundamentally there are two redistribution cases that
peeling typically occurs before the ultimate strength of the should be considered and addressed separately. For illustrative
FRP has been achieved; and 2) addition of the brittle FRP purposes, these two cases are defined as Case I and Case II
leads to a reduction in the member curvature ductility in Fig. 1. Case I relates to redistribution of bending moments
capacity (ACI Committee 440 2002; El-Refaie et al. 2003). from an unstrengthened zone, denoted as Zone C in Fig. 1(a),
At first glance, it appears unlikely that moment redistribution and into an FRP-strengthened region, denoted as Zone B in
(MR) should be permitted in FRP-strengthened continuous Fig. 1(a). In this case, it seems logical to consider full MR as
concrete structures. Strengthening of existing continuous in a conventional RC continuous beam, because MR
steel-reinforced concrete structures that were originally depends only on the plastic rotation capacity from which MR
designed according to MR principles, however, could lead to originates. As such, if it can be shown that an unstrengthened
onerous design conditions. If one considers that an existing
zone (Zone C) can rotate sufficiently for the moment to be
RC beam section was designed for a reduced moment due to
redistributed to a strengthened zone (Zone B), the requirements
redistribution, the new strengthened section must now be
designed to resist the originally higher elastic bending for redistribution are met. Furthermore, there is no need to
moment plus the additional elastic moment that is required
for strengthening purposes. Under these circumstances, the ACI Structural Journal, V. 105, No. 6, November-December 2008.
MS No. S-2007-148.R1 received June 6, 2007, and reviewed under Institute
amount of FRP material that may be required to meet the publication policies. Copyright © 2008, American Concrete Institute. All rights reserved,
new load or serviceability demands can be significantly including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the September-
higher than the magnitude increase from the original load or October 2009 ACI Structural Journal if the discussion is received by May 1, 2009.

ACI Structural Journal/November-December 2008 729

 If a structure displays curvature ductility capacity, it is
ACI member P. F. Silva is an Associate Professor of civil and environmental engineering
at George Washington University, Washington, DC. He is a member of ACI Committees certain that it will also display rotation capacity, but the
341, Earthquake-Resistant Concrete Bridges, and 440, Fiber Reinforced Polymer inverse is not necessarily true. Therefore, in this paper, MR
Reinforcement. His research interests include the development of innovative limits for FRP-strengthened concrete structures were
performance-based procedures for the design and retrofit of structures, and use of
fiber-reinforced polymers for the structural rehabilitation of structures. established based on curvature ductility capacity rather than
rotation capacity. In this way, if curvature ductility can be
ACI member T. J. Ibell is a Professor of civil engineering and Chair of the Department demonstrated, it is certain that a safe design based on
of Architecture and Civil Engineering at the University of Bath, Bath, UK. He
received his PhD from the University of Cambridge, Cambridge, UK, and is a redistribution will result.
chartered structural engineer. He is a member of ACI Committee 440, Fiber Reinforced
Polymer Reinforcement. His research interests include the structural assessment of
concrete structures, strengthening of structures, and the use of advanced composites RESEARCH SIGNIFICANCE
for new-build initiatives. Analytical formulations presented in this paper discuss
appropriate redistribution principles for the design of FRP-
strengthened continuous RC beams. Results presented and
discussed in this paper could lead to a rational development
of design guidelines to consider MR in FRP-strengthened
RC beams, very similar to the ACI methodology currently
used for conventional RC beams.

MOMENT REDISTRIBUTION LIMITS IN

CONVENTIONAL RC BEAMS
ACI 318-02 (ACI Committee 318 2002) states that the
level of MR that can be permitted in a conventional steel-
reinforced continuous concrete structure is 1000εt percent up
to a maximum of 20%, where εt is the level of strain in the
extreme tension steel reinforcement, which must be at least
0.0075. Thus, according to ACI 318-02 (ACI Committee
Fig. 2—Conventional RC beam: Strain Levels A and B. 318 2002), the level of MR in percentage is

consider the ductility capacity or the rotation capacity of the 7.5% ≤ MR = 1000εt ≤ 20% (1)
strengthened zone itself (Zone B in Fig. 1(a)).
This principle has been previously demonstrated in an This equation indicates the level of permissible MR that
experimental study by El-Refaie et al. (2003). Their work can be used in the design of conventional RC beams. As
showed that in some strengthened beams, the amount of MR discussed in the next two sections, analytical studies clearly
was equivalent to conventional RC continuous beams show that the MR values based on those specified by Eq. (1)
because redistribution initiated from an unstrengthened are conservative for conventionally RC continuous beams.
region and was redistributed to a strengthened region. Furthermore, it can be shown that for RC beams, the specified
Therefore, it is appropriate that MR into FRP-strengthened steel strain limits are conservative for any values of beam
zones should indeed be permitted, provided that sufficient size, concrete compressive strength, and tensile reinforcement
rotational capacity exists in the structure to make such ratios. Relevant to the study presented in this paper, Eq. (1)
redistribution feasible. does not indicate the level of required curvature ductility that
Case II relates to the redistribution of bending moments a section must possess to develop the amount of plastic rotations
out of an FRP-strengthened region (Zone E). This paper necessary to meet the permissible MR limits. These issues
addresses this issue and suggests levels of redistribution out are discussed in the next three sections and follow some of
of FRP-strengthened regions that could be adopted by the same principles demonstrated by Mattock (1959) for
directly conforming to redistribution principles in existing permissible MR.
design codes around the world (ACI Committee 318 2002;
BS 8110 1997). Research in this area is extremely limited, Required section curvature ductility capacity
although some experimental work has indeed been conducted limits for MR
regarding the MR of continuous RC beams strengthened As stipulated by Eq. (1), the amount of MR that can be
with carbon FRP (CFRP) composites (El-Refaie et al. 2003). allowed in the design of continuous RC beams depends on
In their experimental work, El-Refaie et al. (2003) have also tensile strain limits εt and these can easily be correlated to the
clearly shown that, although lower than in conventional RC curvature ductility capacity of a section. Establishment of
beams, continuous beams strengthened with CFRP sheets this correlation is the main objective of this section.
were able to redistribute moment in the order of 6 to 31%. Parametric studies in support of developing MR limits as
From their experimental work, they have also concluded that a function of curvature ductility were developed in terms of
MR was lower for those sections retrofitted with higher the upper and lower limiting strain levels presented in Eq. (1).
amounts of CFRP reinforcement. This conclusion is also In the rest of this paper and for ease of reference, Strain
consistent with observations obtained from the present Levels A and B (refer to Fig. 2) are defined for a section that
analytical work. Equally important, work has been achieves the tensile strain levels in the extreme reinforcement
conducted into the question of ductility (Burgoyne 1997; of 0.0075 and 0.020, respectively, just as the strain in the
Naaman et al. 2001) and rotation capacity (Lees 1997; concrete in compression reaches the strain εcu value of 0.003.
Casadei et al. 2003) in FRP-strengthened concrete structures, At these strain levels, and assuming a singly-reinforced
which are important considerations for the study of MR in concrete section with an effective depth d, the curvature φu ,
CFRP-strengthened continuous beams. and the neutral axis cu are, respectively

730 ACI Structural Journal/November-December 2008

 ε cu + ε t
φ u = ----------------
- (2)
d

ε cu
c u = ----------------
-d (3)
ε cu + ε t

Under Strain Levels A and B, the area of reinforcement As

necessary to achieve the strain level is given by

f c′ ε cu
- β 1 ----------------
A s = 0.85 ---- - db (4)
f y ε cu + ε t

where fc′ is the cylinder compressive strength, fy is the yield

strength of the reinforcing steel, β1 is the ACI rectangular
stress block depth factor (ACI Committee 318 2002; ACI
Committee 440 2002), εcu is the ultimate strain capacity of
the concrete, d is the effective depth, and b is the breadth of
the section. For completeness, this equation was evaluated as
a function of the concrete compressive strength and results are
presented in Fig. 3(a). This figure clearly shows, as expected,
that the area of steel to impose Strain Levels A and B are
directly dependent on the concrete compressive strength.
Using the area of steel reinforcement As in conjunction
with the first moment area theorem, it is feasible to compute
the depth to the neutral axis cy at first yielding of the reinforcing
steel from

b 2
--- c y = n s A s ( d – c y ) (5)
c

This expression has been evaluated graphically, as shown Fig. 3—Conventional RC beam analysis as function of
in Fig. 3(b). This figure was developed by assuming the yield concrete compressive strength.
strength fy for the reinforcing steel of 400 MPa (60 ksi).
Referring to this figure, it is clear that the ratio cy/d increases
only slightly as the concrete compressive strength fc′
increases, within normal strength values and for the constant
Strain Levels A and B. Defining the onset of section first
yield as being initial yielding of the reinforcing steel (see
Fig. 4), the curvature at first yield φy′ is

εy
φ y′ = ------------
- (6)
d – cy

where cy is computed using Eq. (5). For the bilinear moment

curvature idealization presented in Fig. 4, the curvature
ductility capacity for any imposed tensile strain levels εt is

ε cu + ε t⎞ ⎛ d – c y⎞
Bilinear case: μ φ = ⎛ ----------------
- ------------- (7)
⎝ εy ⎠ ⎝ d ⎠
Fig. 4—Idealized moment versus curvature: conventional
RC beam.
However, if one is to consider MR analysis by ignoring the
post-yield stiffness of RC members, it can be shown that the Fig. 4, the theoretical yield curvature is then defined based
yield curvature must then be modified to reflect the idealized on the following bilinear relationship
elastoplastic response, as schematically depicted in Fig. 4.
This is a reasonable modification because, as further M
discussed in the next two sections, the available curvature φ y = φ y′ --------u (8)
ductility capacity must be reduced for implementation of the M y′
elastoplastic section response in MR analysis. As such,
assuming an idealized elastoplastic response as shown in where the ratio Mu/My′ can be estimated as

ACI Structural Journal/November-December 2008 731

 greater than 3.0, the strain in the tension steel will be at least
0.0075. Similarly, it is clear that for any given normal
concrete compressive strength fc′ , as long as the curvature
ductility capacity at concrete crushing, μφ, is greater than
7.5, the strain in the tension steel is at least 0.020. Further-
more, it can be shown that increasing the quantity of
compressive steel leads to an increase in the curvature
ductility ratio, and the curvature ductility limits previously
mentioned can be applied to any given section regardless of
the level of the compressive steel.
Equation (10) only provides a means for estimating the
curvature ductility capacity of a section as a function of the
steel strain level achieved at concrete crushing; but it does
not yet fully indicate whether the RC section can develop the
permissible MR for continuous beams. As such, another
issue that requires further consideration is the level of plastic
rotation demand in a section required to fully develop the
permissible MR limits. This issue is discussed in the
following by investigating the plastic rotation demand for an
idealized elastoplastic and bilinear moment curvature relation
with post-yield stiffness.

Elastoplastic analysis approach

The plastic analysis of a continuous RC beam is considered
herein to fully quantify the amount of plastic rotation that a
section must possess to meet the requirements for MR. The
elastoplastic approach is mathematically described in terms
of the theoretical yield curvature given by Eq. (8) and the
dashed line presented in Fig. 4. Assuming a three equal-span
continuous beam under a uniformly applied load, wu, (refer
to Fig. 1) the moment at formation of the first hinge may be
designated as the ultimate moment Mu , which is

Fig. 5—Curvature ductility: conventional RC beam. Mu = 0.10(1 – α)wuL2 (11)

where α is the amount of MR allowed and L is the span

β1 cu length. In experimental studies, this expression can easily be
d – ---------- expressed as a function of the elastic Me and ultimate
Mu 2
-------- = -------------------- (9) moment Mu similar to the work by El-Refaie et al. (2003)
M y′ cy
d – ----
3
M
α = 1 – ------u- (12)
Next, dividing Eq. (2) by Eq. (8) and substituting for the Me
appropriate values of cu, cy, and Mu/My′ , the curvature
ductility capacity for any level of tensile strain εt and for an Certainly the plastic rotation demand for a continuous
elastoplastic idealization is beam depends on the number of spans and type of applied
load; however, and only for brevity, a three-span continuous
⎛ d – c----y ⎞ beam was selected for this analytical study, and similar
ε + ε d – c ⎜ 3 ⎟ approaches could easily be developed for a two- or four-span
Elastoplastic case: μ φ = ⎛ ----------------
cu
-t⎞ ⎛ -------------y⎞ --------------------⎟ (10)
⎝ ε y ⎠ ⎝ d ⎠ ⎜⎜ β 1 c u⎟
case. For instance, in a uniformly loaded two- and four-span
continuous beam with constant flexural stiffness along the
⎝ d – ----------
2 ⎠ length of the beam, the first factor in Eq. (11) is 0.125 or
0.1071, respectively. These values lead to smaller bending
This expression only represents the ultimate curvature moments and conversely lower plastic rotation and curvature
ductility capacity of a conventional RC section at a specified ductility demands. Consequently, expressions developed for
steel strain level and concrete crushing. As before with fy at a three-span continuous beam are conservative in relation to
400 MPa (60 ksi), it is feasible to graphically express the the two- and four-span cases.
curvature ductility as a function of the concrete compressive After the first hinge develops, the amount of plastic rotation
strength, as shown in Fig. 5(a), which shows plots for the two demand at the supports is equal to the rotation of a simply
strain levels and curvature ductility capacity computed in supported beam under the applied load αwu, which is
terms of Eq. (7) and (10), bilinear and elastoplastic idealization,
respectively. Referring to Fig. 5(a), it is clear that for any 3
given normal concrete compressive strength fc′ , as long as αw u L
θ P = ---------------
- (13)
the curvature ductility capacity at concrete crushing, μφ , is 24EI

732 ACI Structural Journal/November-December 2008

 This expression assumes that plastic rotations are developed the results for conventional- and/or FRP-strengthened beams
over the plastic hinge region and once again the flexural are significantly different, which would dramatically affect
stiffness is constant between plastic hinges. In addition, Eq. (13) the study of MR in FRP-strengthened RC beams.
was developed based on an elastoplastic approximation. As mentioned previously, assuming a three equal-span
Because the bending stiffness EI is related to the initial stiffness continuous beam under a uniformly applied load, wu, the
up to the yield curvature, φy, and theoretical ultimate first yield moment My′ at formation of the first hinge is also
moment Mu , Eq. (13) may be rewritten as given by Eq. (11). Based on the one-component nonlinear
model formulation (Cheng 2001), it can be shown that the
α 10 amount of plastic rotation demand at the supports is
θ P = ⎛ ------------⎞ ------ φ y L (14)
⎝ 1 – α⎠ 24
3
αw u L ⎞ ⎛
- ------------------------------------⎞
θ P = ⎛ ---------------
3 + r/2
(19)
In addition, the plastic curvature is related to the plastic ⎝ 24EI ⎠ ⎝ ( 8r + 3 ) ( 1 – r )⎠
rotation by
where r is the bilinear factor for the post-yield branch, as
θ defined in Fig. 4. Equation (19) was developed from first
φ P = -----P- (15)
LP principles for a section with a post-yield stiffness. As previ-
ously described, Eq. (19) was developed based on the
This implies that the ultimate curvature φU that a section bilinear approximation defined in Fig. 4. Using the same
must exhibit to develop the moment redistribution α is approach to obtain Eq. (14), Eq. (19) may be written as

α
10 φ y L⎞ θ P = ⎛ ------------⎞ ------------------------------------ ⎛ ------ φ y L⎞
α 3 + r/2 10
φ U = φ y + ⎛ ------------⎞ ⎛ ------ -------- (16) ⎝ 1 – α⎠ ( 8r + 3 ) ( 1 – r ) ⎝ 24 ⎠
(20)
⎝ 1 – α⎠ ⎝ 24 L p ⎠

In this case, the ultimate curvature demand φU is

where Lp is the plastic hinge length. As such, the curvature
ductility demand on an RC section to develop a level of
moment redistribution α is α 10 φ y L⎞
φ U = φ y + ⎛ ------------⎞ ------------------------------------ ⎛ ------ --------
3 + r/2
(21)
⎝ 1 – α⎠ ( 8r + 3 ) ( 1 – r ) ⎝ 24 L p ⎠
α
μ φ = 1 + ⎛ ------------⎞ ---------
10
(17)
⎝ 1 – α⎠ 24λ As mentioned previously, the plastic hinge length was
conservatively estimated as 0.035L. As such, the curvature
ductility demand to develop the required moment redistribution
where λ is defined as the ratio Lp/L. According to the work α for a section with a well defined post yield stiffness r is
by Mattock (1959), it is appropriate to estimate d = L/23 and
Z/d = 5.5, where Z is the distance between points of
α
μ φ = 1 + ⎛ ------------⎞ ------------------------------------ ⎛ ---------⎞
maximum and zero moment. Thus, for the three equal-span 3 + r/2 10
(22)
continuous beam considered herein, λ may be conservatively ⎝ 1 – α⎠ ( 8r + 3 ) ( 1 – r ) ⎝ 24λ⎠
estimated as 0.035L. This is obtained by assuming Lp is
(Mattock 1959)
In this equation, when r is zero, the results revert to the
elastoplastic solution given by Eq. (17). As mentioned
L p = 0.5d + 0.05Z ⇒ λ = 0.035 (18) previously, for an MR of either 7.5 or 20% with a post-yield
stiffness with r of 0.01, the sectional curvature ductility
For an MR of either 7.5 or 20%, Eq. (16) indicates that the demand in terms of Eq. (22) is also nearly 2 or 4, respectively,
curvature ductility demand on a given section is nearly 2 or with results depicted in Fig. 5(b) for different levels of post-
4, respectively. As presented in Fig. 5(a), these numbers are yield stiffness. This figure clearly shows that the curvature
smaller than the available curvature ductility capacity of 3 ductility demand as a function of the permissible MR
and 7.5 required to develop the tensile strain in the steel of decreases as the post yield stiffness increases. In conclusion,
0.0075 and 0.020, respectively, and as such, the permissible it is reasonable to infer that the permissible MR limits given by
MR limits given by Eq. (1) are conservative. This will be Eq. (1) are conservative for any values of r.
further discussed in the next section. Figure 6(a) shows the permissible MR as a function of the
steel strain in terms of Eq. (1), as defined in ACI 318-02
Bilinear analysis approach (ACI Committee 318 2002). Permissible MR limits are
Because MR initiates immediately after first yielding φy′ , plotted with piecewise curves, represented by the solid line.
this section sets out a rationale to compute the amount of In this figure, the dashed plots show the available MR
plastic rotation necessary to develop MR at the onset of the capacity of a section at different levels of steel strain and for
first yield moment, My′ . This approach is of relevance in this concrete compressive strength of 35 MPa (5 ksi). Results
study because, as will be shown in the following, FRP- were also developed for other concrete compressive
strengthened RC beams possess significant post-yield stiffness strengths ranging from 21 to 56 MPa (3 to 8 ksi) and results
compared with conventional RC beams, which should be only vary slightly. These curves were developed by equating
accounted for in MR analysis. It is instructive to compare Eq. (7) to Eq. (22) for the bilinear moment curvature ideali-
these two approaches, namely, the elastoplastic and the zation with r equal to 0.01, and Eq. (10) to Eq. (17) for the
bilinear approach. This comparison will determine whether elastoplastic idealization or r equal to zero. In these plots, the

ACI Structural Journal/November-December 2008 733

 Table 1—Materials properties
Steel Young’s modulus Es 200 GPa (29,000 ksi)
FRP Young’s modulus Ef 210 GPa (30,000 ksi)
Steel yield strength fy 400 MPa (58 ksi)
FRP ultimate strength ffu 3400 MPa (493 ksi)
Concrete compressive strength fc′ 35 MPa (5 ksi)

Table 2—Comparison of results with work

by El-Refaie et al. (2003)
Area FRP, Ratio steel Ratio FRP Redistribution MR MR
ID mm2 (in.2) ρs (ρs′ ), % ρf , % (refer to Fig. 1) exp., % theory, %
H1 — 0.33 (2.1) — C* → B&D 38.40 37.60
H2 25.74 (0.040) 0.33 (2.1) 0.09† E→F 18.72 14.90
H3 77.22 (0.120) 0.33 (2.1) 0.26† E→F 11.19 12.42
H4 128.7 (0.199) 0.33 (2.1) 0.43 † E→F 6.32 8.32
H5 77.22 (0.120) 0.33 (2.1) 0.26 † E→F 18.82 12.42
0.09
H6 25.74 (0.040) 0.33 (2.1) E→F 21.14 14.90
(0.09‡)
*
No strengthening.
†Strengthening top.
‡
Strengthening both sides top and bottom.

Results presented in these sections were also corroborated

with the experimental work conducted by El-Refaie et al.
(2003) and a summary of this comparison is presented in
Table 2. It is important to mention that in the work by El-Refaie
et al. (2003), they used a two-span continuous beam. Results
Fig. 6—Permissible moment redistribution: conventional
presented in Eq. (17) and (22) were substituted by the appro-
RC beam.
priate factors for a two-span beam. These equations, however,
may be updated for a two-span beam by simply substituting
the factor 10 with 8. Results presented in Table 2 for the
conventional beam (H1) shows a good agreement between
the experimental and theoretical results.

MOMENT REDISTRIBUTION LIMITS IN

FRP-STRENGTHENED RC BEAMS
An idealized moment-curvature relationship for an FRP-
strengthened RC beam that has been under a service moment
before strengthening is presented schematically in Fig. 7. At
the time of strengthening, it may be estimated that the strain
in the tension steel bars will be within the values given in
Table 3 and profiles shown in Fig. 8(a), or 33.8 to 20.5% of
the yield strain. These ranges were estimated by assuming
that no live load is acting during strengthening and were
Fig. 7—FRP-strengthened RC beam. based on the following expression

variable α in Eq. (7) and (10) was substituted by 100MR. β 1 c⎞

Required material properties listed in Table 1 have been used ⎛ ⎞ ⎛⎜ d – ------- -
φε 2 ⎟ ε
in establishing these results. ε s = ⎜ --------------------
y
-⎟ ⎜ -----------------⎟ ⇒ 0.205 ≤ ----s ≤ 0.338 (24)
⎜ ⎟ εy
⎝ 1.2 + -------⎠ ⎜⎝ d – ----y ⎟⎠
As expected, it is clear that the permissible MR values 1.6 c
allowable by ACI 318-02 (ACI Committee 318 2002) are κ 3
indeed conservative. These results are consistent with those
reported in the literature and presented in the original work where φ is the flexural strength reduction factor, cu was
by Mattock (1959). In addition, it can be concluded that the computed based on Eq. (3), and cy was computed based on
elastoplastic Eq. (17) and the bilinear approximation Eq. (22) Eq. (5) and Fig. 3. In Eq. (24), the dead load (DL) was
give similar results within normal values of the bilinear factor expressed as a ratio of the live load (LL) (that is, DL/LL = κ)
r. Based on the work presented in the previous two sections, it with the ratio κ varying between 1 and 0.5, as shown in
is possible to rewrite Eq. (1) to express the MR as a function Fig. 8(a). These values specify that κ is 1 when the DL
of the section curvature ductility capacity given by equals the LL and for a κ of 0.5 then the DL is 50% of the
LL. As depicted in this figure, the strain in the steel at the
7.5% ≤ MR = 2.72μφ – 1.04 ≤ 20% (23) time of strengthening increases as the ratio of DL to LL also

734 ACI Structural Journal/November-December 2008

Table 3—Ratio of steel strain at strengthening
versus yield strain (εserv /εy)
Concrete Strain Level A Strain Level B
Ratio of steel compressive εt = 0.0075 εt = 0.02
strain at strength fc′,
condition MPa (ksi) Minimum Maximum Minimum Maximum
28 (4) 0.205 0.321 0.213 0.333
Strengthening
56 (8) 0.212 0.332 0.216 0.338
28 (4) 0.790 0.814 0.792 0.817
Peeling
56 (8) 0.801 0.825 0.802 0.826

increases. This is expected, because as κ increases, so does

the DL acting on the structure and the steel strain at the time
of strengthening.
It is suggested herein that a similar criterion used for
conventional RC beams be imposed on FRP-strengthened
sections for MR out of FRP-strengthened sections. Pending
future research work, it is suggested that MR be limited to
the FRP peeling strain limit, which is taken to be 0.8% (The
Concrete Society 2000), as shown in Fig. 7(b). This figure
presents three strain level cases, where Strain Level C
defines FRP peeling at the onset of concrete crushing, and
the other two cases are self explanatory.
It is suggested herein that as long as the total strain in the
tension steel is at least 0.0075 at peeling of the FRP, then MR
should be permitted out of such sections. At Strain Level C,
the total strain in the tension steel εt is

d–c
ε t = ε s + ⎛ -------------u-⎞ 0.008
Fig. 8—FRP-strengthened RC beam strain levels.
(25)
⎝ h – c u⎠

But, what level of curvature ductility could one hope to

where εs is estimated based on Eq. (24), and cu is the neutral achieve in general from an FRP-strengthened structure and
axis at ultimate. Assuming that h ≅ 1.10d (for most concrete would it exceed the required minimum? In addition, how
structures), and substituting the two limits expressed in would Fig. 6(a) be altered in the case of FRP-strengthened
Eq. (24) for εs into Eq. (26), the strain in the steel bars will RC beams for the tensile stain levels defined according to
reach 0.0075 provided that the depth to the neutral axis in the Eq. (1)? These issues are discussed in the following for
strengthened RC beam, cu, must be at least 0.40d, which is conditions at Strain Level C.
fairly large. This suggests that indeed the value of 0.0075
will be reached in FRP-strengthened beams.
Required section curvature ductility limits for MR
If the depth to the neutral axis is less than this, as is usually
the case, the total strain in the steel bars will definitely exceed Under fully-strengthened conditions, it may be shown that
0.0075 at peeling of the FRP. Analytical results based on the neutral axis depth at first yielding, cy,f , for an FRP-
Eq. (25) are shown in Fig. 8(b) and summarized in Table 3 for strengthened section is
Strain Level C. These results clearly suggest that the steel strain
at FRP peeling will definitely exceed the steel strain of 0.0075 2
necessary for moment redistribution. Thus, it seems likely that – ( η s + η f ) + ( η s + η f ) + 2b ( dη s + hη f )
c y, f = -----------------------------------------------------------------------------------------------------
- (26)
the criterion that the steel strain must be at least 0.0075 appears b
to be met under normal conditions of FRP strengthening.
This is discussed in further detail in the following. where the terms ηs and ηf are used to simplify the aforemen-
This is, of course, for conditions that involve peeling tioned equation, with subscript s referring to steel and f referring
failure of the FRP at near onset of concrete crushing, herein to FRP, and are defined as
defined as Strain Level C. As shown in Fig. 7(b), when
crushing of the concrete occurs well before peeling of the
ηs = ns As; ηf = nf Af (27)
FRP, herein defined as Strain Level D, the strain in the steel
is likely to be less than the minimum recommended strain of
0.0075, and MR is automatically not permitted under where n and A are the relevant modular ratio and area of
conditions set by Eq. (1). Conversely, when peeling of the reinforcement, respectively. Similarly, the curvature at first
FRP occurs before crushing of concrete, herein defined as yield, φ′y,f , for a strengthened section is
Strain Level E, it can be shown the strain in the steel is likely
to be higher than the minimum recommended strain of εy
0.0075, and MR should be considered under conditions set φ y′, f = ----------------
- (28)
by Eq. (1). d – c y, f

ACI Structural Journal/November-December 2008 735

 At Strain Level C, the neutral axis depth at ultimate
conditions, cu,f, for an FRP-strengthened section is

ε cu
c u, f = ------------------
-d (29)
ε cu + ε tf

where the strain in the tension steel εt,f is given by Eq. (25).
The moment at first yield, M′y,f , can be expressed as

c y, f⎞ c y, f⎞
M y′, f = A s f y ⎛ d – -------
- + A f f f, y⎛ h – -------
- (30)
⎝ 3 ⎠ ⎝ 3 ⎠

where cy,f is computed based on Eq. (26), and ff,y represents

the strain in the FRP at first yielding of the tension steel. In
Eq. (30), the FRP stress ff,y at the first yield limit state is

h – c y, f⎞
f f, y = ε y ⎛ ----------------
- – ε bi E f (31)
⎝ d – c y, f ⎠

where cy,f is obtained using Eq. (26) and εbi is

h–c
ε bi = ε s ⎛ -------------y⎞ (32)
⎝ d – c y⎠

In this equation, cy is computed from Eq. (5), the first term

εs represents the tension steel strain at FRP strengthening
and is computed based on Eq. (24), and it is assumed that, at Fig. 9—FRP-strengthened RC beam FRP areas and curvature
strengthening, the concrete has cracked. Assuming that the ductility.
FRP peeling occurs at an FRP strain of 0.8% (The Concrete
Society 2000) and at the onset of concrete crushing, shown
as Strain Level C in Fig. 7(b), the ultimate moment Mu,f is 0.85f c′ β 1 c u, f b – A s f y
estimated based on the following A f = -------------------------------------------------
- (35)
0.008E f

β 1 c u, f⎞ β 1 c u, f⎞
M u, f = A s f y ⎛ d – -------------
- + 0.008 × A f E f ⎛ h – -------------
- (33) where the FRP stress at Strain Level C is 0.008Ef , and once
⎝ 2 ⎠ ⎝ 2 ⎠ again cu,f is given by Eq. (29). Figure 9(a) shows the FRP
reinforcement ratio required to develop the Strain Level C
Before this equation can be used, however, the amount of conditions. This figure shows the level of required FRP area
FRP Af must be limited to meet stringent fire resistance as a function of the concrete compressive strength and for
requirements. Although this issue is not the primary objective different levels of reinforcing steel As. This figure indicates,
discussed in this paper, it is important to understand the as expected, that for RC beams with low levels of steel
effects that fire resistance limits may impose on MR of FRP- reinforcement, Strain Level C will be achieved at higher
strengthened RC sections. Based on the limits set forth by ratios of FRP reinforcement than for heavily reinforced RC
ACI Committee 440 (2002), it is possible to express limits beams. As may be expected, this should affect the curvature
for Eq. (33) with limits set forth by ductility capacity of an FRP reinforced section, which is
discussed in the following. The curvature at the ultimate
condition is given by
⎛ 1.2 + 1.6 ------- ⎞⎟
⎜ κ
M uf = M u ⎜ ------------------------⎟ (34) ε cu
⎜ 1.2 + 0.85 ----------⎟ φ uf = -------
- (36)
⎝ κ ⎠ c u, f

where Mu is the ultimate moment capacity of the where the neutral axis for the FRP-strengthened section cu,f
unstrengthened RC beam. This equation, however, is based is computed in terms of Eq. (29). Note that Eq. (36) applies
on elastic distribution of bending moments, and future only to conditions when peeling failure of the FRP occurs
research is required that can develop other equations based nearly at onset of concrete crushing, herein defined as Strain
on MR principles. Fire resistance limits based on Eq. (34) are Level C, as shown in Fig. 7(b). When crushing of the
discussed in the following sections. concrete occurs well before peeling of the FRP, herein
Following a similar MR procedure outlined for the defined as Strain Level D, or when peeling of the FRP occurs
conventional RC beams, the area of FRP Af necessary to before crushing of the concrete, herein defined as Strain
impose the conditions representing Strain Level C is Level E, Eq. (36) cannot be used; however, as previously

736 ACI Structural Journal/November-December 2008

 Fig. 11—FRP-strengthened beam: cross section.

however, curvature ductility capacity is nearly constant,

which is more representative of the true response of continuous
beams. As such, for the rest of this paper, only the bilinear
conditions will be considered in evaluating the MR of
FRP-strengthened RC beams.
As stated previously, the curvature ductility limits can be
applied to any given section regardless of the quantity of
compressive steel. For MR to be feasible under the permissible
limits set by Eq. (1), the curvature ductility capacity expressed
in Fig. 10(b) must next be related to the curvature ductility
demand requirements stipulated by Eq. (17) and (22).

Moment-curvature relations for

Fig. 10—Curvature ductility: FRP-strengthened RC beam. FRP-strengthened RC beams
The beam section shown in Fig. 11 and the material
properties listed in Table 1 were used to develop example
discussed, the limiting condition for MR of FRP-strengthened moment-curvature plots with and without FRP strengthening.
sections is defined at Strain Level C, and for brevity, only This numerical example serves merely to illustrate the approach
this case is investigated in this paper. As in the conventional and the generality of the approach lies in the analysis. Figure 12
RC beam, at Strain Level C then, the ultimate curvature is important as it presents important parametric information
capacity based on the elastoplastic moment curvature relating to curvature ductility under varying FRP quantities,
idealization (refer to Fig. 7(a)) is aiding the generality of the results.
Figure 12 shows six theoretical moment-curvature plots
φ u1 M′y, f for this cross section under increasing quantities of FRP
μ φ = ------- ------------ (37) strengthening, steel reinforcement ratios, and ratio of DL to
φ y′ M u, f LL. It is clear that as more FRP is added to the section, the
post-yield branch stiffens up considerably, and the failure
For fy of 400 MPa (60 ksi), it is possible to graphically mode changes from FRP delamination to concrete
express the curvature ductility capacity as a function of the compression failure.
FRP reinforcement ratio and concrete compressive strength, In this figure is also shown the limit set by Eq. (34) with
as shown in Fig. 9(b) and 10. Figure 9(b) illustrates the respect to fire-resistance requirements. This limit clearly
variation in curvature ductility capacity as a function of the shows that for sections reinforced with low steel reinforcement
FRP reinforcement ratio and for different ratios of DL to LL, ratios, the amount of FRP that can be added to the original
κ. Figure 9(b) clearly shows that for Strain Level C, the section is very limited. As the section steel reinforcement
ductility capacity of FRP-strengthened RC beams is within ratio increases, however, the allowable fire resistance
3.75 to 4.25, which significantly exceeds the limits for moment increases. These results clearly indicate that limits
redistribution of 7.5%, as previously shown in Fig. 5(b). for fire resistance must be included in MR limits. As this
Unlike Fig. 5(a), where the curvature ductility was nearly paper only describes the approach that designers may use in
constant for different concrete compressive strengths, it is MR limits, fire-resistance limits must be covered as a separate
clear that for FRP-strengthened sections shown in Fig. 10(a), analysis check.
the curvature ductility varies for different concrete compressive Results presented in Fig. 12 were then used to evaluate the
strengths but it is more sensitive for sections reinforced with curvature ductility of the section as a function of the steel
low steel reinforcement ratios and for the elastoplastic strains. As expected, increasing the FRP area decreases the
idealization. For the bilinear idealization depicted in Fig. 10(b), curvature ductility, μφ , and steel strain et , as shown in Fig. 13.

ACI Structural Journal/November-December 2008 737

 Fig. 12—Moment-curvature plots: FRP-strengthened RC beam.

These results were obtained for a concrete compressive

strength of 35 MPa (5 ksi).
Results presented in Fig. 13 are within those values stipulated
in Fig. 10(b). For example, in Fig. 10(b) and for a concrete
compressive strength of 35 MPa (5 ksi), the curvature
ductility capacity for a section with As = 0.004bd is nearly
3.80. In Fig. 13, the available sectional curvature ductility at
Strain Level C (refer to Fig. 7) for As = 0.004bd is nearly 3.7.
This level of comparison can also be inferred to other levels
of As and strain limit levels. This indicates that the mathematical
derivations discussed in the previous section are certainly
within the sectional moment curvature results discussed in
this section. As such, the mathematical expressions developed
in the previous section were used to evaluate the level of
permissible MR for FRP-strengthened RC beams and results
Fig. 13—Curvature ductility capacity. are discussed in the next section.

738 ACI Structural Journal/November-December 2008

 MR out of FRP-strengthened zones presents a more
complex problem than that for conventionally reinforced
concrete sections and detailed analytical expressions were
developed to set limits that can be permitted out of FRP-
strengthened zones. Subsequent conclusions in establishing
rules for MR out of FRP-strengthened zones are as follows:
1. The analytical results generated in this paper are based
on assumptions concerning ductility demands. The results
have been compared with limited experimental data and, as
such, should be considered with caution at this stage until
fundamental experimental research has been conducted in
the area of FRP strengthening of continuous structures;
2. As in conventional RC members, in FRP-strengthened
RC beams, MR develops immediately after yielding of the
longitudinal steel reinforcing is attained; and
3. For FRP-strengthened RC beams, MR should be permis-
sible provided that the section curvature ductility capacity is
above 2.0 and the strains in the tension steel are within the limits
set by Eq. (1). In general, for any given section, if it can be
proven that the beam curvature ductility capacity exceeds 2.0,
then MR in the amount of at least 7.5% can be attainable.

ACKNOWLEDGMENTS
The authors would like to acknowledge the financial support of the
Fulbright Commission for the partial funding of this research and The
University Transportation Center located at the University Missouri Rolla.

REFERENCES
Fig. 14—Permissible moment redistribution. ACI Committee 440, 2002, “Guide for the Design and Construction of
Externally Bonded FRP Systems for Strengthening Concrete Structures (ACI
440.2R-02),” American Concrete Institute, Farmington Hills, MI, 45 pp.
Curvature ductility demands for ACI Committee 318, 2002, “Building Code Requirements for Structural
FRP-strengthened RC beam Concrete (ACI 318-02) and Commentary (318R-02),” American Concrete
As previously outlined for a conventional RC beam, the Institute, Farmington Hills, MI, 443 pp.
levels of MR can be directly expressed in terms of the available BS 8110, 1997, “Structural Use of Concrete: Part 1—Code of Practice
curvature ductility capacity of an FRP-strengthened section for Design and Construction,” British Standards Institution, London, UK,
168 pp.
by equating Eq. (36) to (22) and solving for the moment Burgoyne, C. J., 1997, “Rational Use of Advanced Composites in
redistribution. In Fig. 14(a) and (b), these results are plotted Concrete,” Keynote Lecture, Proceedings of the Third International
for the permissible MR (ACI Committee 318 2002) as a Symposium on Non-Metallic (FRP) Reinforcement for Concrete Structures,
function of the tension steel strain and the available curvature FRPRCS-3, V. 1, Sapporo, Japan, Oct., pp. 75-88.
ductility for an FRP-strengthened RC beam under three Casadei, P.; Denton, S.; Ibell, T.; and Nanni, A., 2003, “Moment
Redistribution in Continuous CFRP-Strengthened Concrete Members,”
different levels of reinforcing steel and initial DL to LL ratio. Proceedings of Composites in Construction 2003, International Conference,
The remaining curves are the same as those depicted in Fig. 6. Rende, Italy, Sept., pp. 307-312.
Referring to Fig. 14(a) and (b), it is clear that for any level Cheng, F., 2001, Matrix Analysis of Structural Dynamics—Applications
of tension, steel present in the original beam and for any and Earthquake Engineering, Marcel Dekker, Inc., New York, NY, 997 pp.
El-Refaie, S. A.; Ashour, A. F.; and Garrity, S. W., 2003, “Sagging and
level of FRP reinforcement, the section exhibits a higher MR Hogging Strengthening of Continuous Reinforced Concrete Beams Using
capacity than the ACI 318 permissible equation for steel CFRP Sheets,” ACI Structural Journal, V. 100, No. 4, July-Aug., pp. 446-453.
strains near εt = 0.0075 and greater. Figure 14(b) also shows Lees, J. M., 1997, “Flexure of Concrete Beams Pre-Tensioned with
the limits for fire threshold. Aramid FRPs,” PhD dissertation, University of Cambridge, Cambridge,
These conditions suggest that as long as an FRP-strengthened UK, Mar.
Mattock, A. H., 1959, “Redistribution of Design Bending Moments in
RC beam possesses a curvature ductility capacity greater Reinforced Concrete Continuous Beams,” Proceedings of the Institution of
than the values depicted in Fig. 14(b), redistribution should Civil Engineers, V. 13, pp. 35-46.
be permissible. Pending future research, analytical results Meier, U.; Deuring, M; Meier, H.; and Schwegler, G., 1993, “Strengthening
discussed in this paper clearly show that if an FRP-strengthened of Structures with Advanced Composites,” Alternative Materials for the
Reinforcement and Prestressing of Concrete, J. L. Clarke, ed., Glasgow,
section can develop a curvature ductility capacity greater Scotland, pp. 153-171.
than 2.0 and the strains in the tension steel are within the Naaman, A.; Park, S.; and Lopez, M., 2001, “Parameters Influencing the
limits set by Eq. (1) redistribution in the order of at least Flexural Response of RC Beams Strengthened using CFRP Sheets,”
7.5% can be achieved provided that limits for fire threshold Proceedings of FRPRCS-5, Cambridge, UK, July, pp. 117-126.
are within those stipulated by Eq. (34). Task Group 9.3, 2001, “Externally Bonded FRP Reinforcement for RC
Structures,” Technical Report, Bulletin 14, Lausanne, Switzerland.
Teng, J. G.; Chen, J. F.; Smith, S. T.; and Lam, L., 2002, FRP-Strengthened
CONCLUSIONS RC Structures, John Wiley and Sons, West Sussex, UK, 266 pp.
This paper has addressed the question of MR in FRP- The Concrete Society, 2000, “Design Guidance for Strengthening
strengthened concrete structures by relating such behavior to Concrete Structures using Fibre Composite Materials,” Technical Report
55, Crowthorne, UK, 72 pp.
the level of ductility at critical sections. In particular, it has Tumialan, G.; Nanni, A.; Ibell, T. J.; and Fukuyama, H., 2002, “FRP
been discussed that MR into FRP-strengthened zones should Composites for Strengthening Infrastructure Around the World,” SAMPE
be permitted under any circumstances. Journal, V. 38, No. 5, Sept.-Oct., pp. 9-15.

ACI Structural Journal/November-December 2008 739