You are on page 1of 14

1286

Evaluation of ductility and allowable moment


redistribution in reinforced concrete structures
Adnan Shakir and David M. Rogowsky

Abstract: Designers can use moment redistribution to reduce the design bending moment envelope. Code provisions
for moment redistribution are not entirely rational. They neglect the effects of important parameters on permissible mo-
ment redistribution and can be very conservative. To establish a realistic limit on permissible moment redistribution,
one needs a rational model for predicting the plastic rotation capacity of critical sections (plastic hinges). This paper
presents a model for computing the plastic rotation capacity, q p , and permissible moment redistribution, b , in rein-
forced concrete beams. Important parameters, affecting q p and b , are identified and incorporated in the model. The
model is validated against experimental results and shows good agreement. A comparison of the moment redistribution
limits is made between the model and CSA A23.3-94. Although the code provides a reasonable estimate of b for unfa-
vourable combinations of parameters, the code can be very conservative when conditions are favourable for moment re-
distribution. Deeper beams with closely spaced stirrups allow significantly more moment redistribution than that
predicted by the code.

Key words: moment redistribution, ductility, plastic rotation capacity, bond–slip, shear cracking, reinforced concrete
beams, c/d, ultimate concrete strain.

Résumé : Les concepteurs peuvent utiliser la redistribution de moments pour réduire l’enveloppe du moment de flé-
chissement de conception. Les dispositions du code pour la redistribution de moments ne sont pas entièrement ration-
nelles. Elles négligent les effets d’importants paramètres sur la redistribution de moments permise et peuvent être très
conservatrices. Pour établir une limite réaliste de la redistribution de moments permise, un modèle rationnel pour la
prédiction de la capacité de rotation plastique de sections critiques (pivots plastiques) est requis. Cet article présente un
modèle pour le calcul de la capacité de rotation plastique q p et de la redistribution de moments permise b pour des
poutres en béton armé. Les paramètres importants, affectant q p et b , sont identifiés et incorporés dans le modèle. Le
modèle est validé contre des résultats expérimentaux et montre une bonne concordance. Une comparaison des limites
de redistribution de moments est accomplie entre le modèle et la norme CSA A23.3-94. Bien que le code fournit une
estimation raisonnable de b pour des combinaisons non favorables de paramètres, le code peut être très conservateur
lorsque les conditions sont favorables à la redistribution de moments. Les poutres épaisses avec des étriers étroitement
espacés permettent significativement plus de redistribution de moments que ce qui est prédit par le code.

Mots clés : redistribution de moments, ductilité, capacité de rotation plastique, lien-glissement, craquement dû au cisail-
lement, poutres en béton armé, c/d, déformation ultime du béton.

[Traduit par la Rédaction] Shakir and Rogowsky 1299

Introduction duce the design moment envelope and the amount of


Because of the ease of computer-based analysis, structures required flexural reinforcement.
are now generally analysed and designed for more load Structural design codes (CSA A23.3-94; ACI 318-95; BS
cases than in the past. This leads to an expansion of design 8110-85; CEB Model code 1990; JSCE 1986; DIN 1045-78;
bending moment envelope. Larger moments produce higher DS411-1986) recognize the nonlinear behaviour of rein-
construction cost. Moment redistribution can be used to re- forced concrete structures and allow limited redistribution of
elastic moments. Different formulae for the redistribution of
elastic moment are proposed by various codes of practice.
Received September 22, 1999. These code provisions consider permissible moment redistri-
Revised manuscript accepted June 2, 2000. bution as a section property and relate it to the relative depth
A. Shakir and D.M. Rogowsky.1 Department of Civil and
of the compression zone at failure (c/d) or the reinforcement
Environmental Engineering, University of Alberta, Edmonton, index (w = Asfy/bd fc¢ ) only. Although c/d or w can fairly well
AB T6G 2G7, Canada. represent the combined effects of stress–strain characteris-
tics of the materials, the geometry of the cross section, and
Written discussion of this article is welcomed and will be the amount of tensile and compressive reinforcement, it is an
received by the Editor until April 30, 2001. oversimplification to relate moment redistribution to c/d or w
1
Author to whom all correspondence should be addressed only. Note that for a given cross section, c/d and w are re-
(e-mail: dmrogowsky@civil.ualberta.ca). lated to each other. The most important factors influencing

Can. J. Civ. Eng. 27: 1286–1299 (2000) © 2000 NRC Canada


Shakir and Rogowsky 1287

permissible moment redistribution are the plastic rotation ca- of permissible moment redistribution. The Japanese and the
pacity and plastic rotation demand, which in turn depend German codes are the most conservative. The most liberal is
upon many other factors. Thus, to establish a realistic limit the Danish code (not shown in Fig. 1), which allows a maxi-
on permissible moment redistribution, a realistic estimate of mum moment redistribution of 66%, with the only restric-
plastic rotation capacity is required. tion that the reinforcement index is less than the balanced
Plastic rotation capacity, qp, of structural concrete mem- value.
bers has been studied by Chan (1955), Cohn (1964), Mat- The large differences between the compared code require-
tock (1964, 1983), Corley (1966), Baker and Amarakone ments and the deficiency of code provisions in incorporating
(1964), Naaman et al. (1986), Riva and Cohn (1994), and the main parameters show the necessity for further examina-
others. The first expressions for qp in the literature were of tion of the code provisions for moment redistribution. The
the type following section examines the Canadian code (CSA A23.3-
94) provision for moment redistribution in detail.
[1] qp = (fu - fy)Lp
where fy and fu are the curvatures at the critical section at Moment redistribution in CSA A23.3-94
yield and ultimate, respectively, and Lp is the effective hinge
length. In these expressions the plastic curvature is assumed CSA A23.3-94 requires that the permissible moment re-
to be constant over the effective hinge length. In reality, the distribution at a section must not exceed the lesser of 30–
plastic curvature varies along the plastic hinge length, de- 50(c/d) or 20%. This can be restated as
pending upon the member stiffness, loading distribution,
c 1
bond–slip relationship, and other factors. It has been estab- [2] (max) = (30 - b)
lished through experimental and theoretical studies (Dilger d 50
1966; Bachmann 1970; Langer 1987) that the presence of in-
where b is the change in moment expressed as a percentage
clined shear cracks causes a shift in the tensile force, thereby
of the elastic moment. To ensure ductility, the c/d ratio must
increasing the plastic hinge length and the plastic rotation
not exceed the value given by eq. [2] for a given amount of
capacity.
moment redistribution, b. Note that this need only apply to
Although numerous attempts have been made to deter-
locations where moment redistribution reduces the moments
mine an expression for qp, it appears that none of the pro-
below the elastic bending moment.
posed formulations for qp can be considered completely
To assess the code limit of c/d, an alternate expression for
satisfactory. A summary of the expressions for available
c/d can be derived in terms of required plastic rotation, using
plastic rotation is given in Table 1. Most of them are based
the plane-section compatibility analysis and eq. [1], and is
on tests of simply supported beams with point loads, in
given as
which only a limited number of parameters governing the
plastic rotation capacity of a member are considered. In par-
c éq f /E ù
ticular, the formulations neglect the effects of loading distri- [3] (max) = e cu ê preqd + y s ú
d ­ ê K1 1 - (cy / d) úû
bution (uniformly distributed vs. concentrated) and structural ë
configuration (continuous vs. simple spans). Riva and Cohn
(1994) developed a nonlinear analysis model, which consid- where e cu is the ultimate concrete strain at extreme compres-
ers the influence of these parameters. Their model, however, sion fiber, fy is the yield strength of steel, Es is the modulus
neglects the favourable effect of shear cracking on plastic ro- of elasticity of steel, cy is the depth of neutral axis at first
tation capacity. yield, qpreqd is the required plastic rotation at a hinge, and K1
The objectives of this paper are to recognize the differ- is a constant that defines the plastic hinge length Lp (Lp =
ences among the various design codes on the limits of per- K1d). For the purpose of the present analysis, a value of 1.0
missible moment redistribution, to demonstrate that the code is assumed for K1 to compute plastic hinge length. Figure 2
(CSA A23.3-94) equation for moment redistribution is not shows the graphical solution of eq. [3].
entirely rational, and to develop an analytical model for the To compare the allowable c/d limits from eq. [2] and
assessment of plastic rotation capacity and permissible mo- eq. [3], moment redistribution is performed on the continu-
ment redistribution. The influence of major parameters on ous beam shown in Fig. 3. The beam is analysed for six load
rotation capacity, especially shear cracking and concrete– cases to obtain the maximum moments at critical sections
steel bond–slip, is included in the model. with patterned live loads. The factored elastic moments and
the service elastic moments are given in Table 2.
Moment redistribution in design codes Moment redistribution was done for each load case in
turn. Controlling elastic moments were reduced by up to
Design codes (CSA A23.3-94; ACI 318-95; BS 8110-85; 20% with non-controlling moments adjusted to satisfy static
CEB Model code 1990; JSCE 1986; DIN 1045-78; DS411- equilibrium. In some cases (e.g., end spans) both positive
1986) have proposed different formulae for the redistribution and negative moments control and statics dictated less than
of elastic moments. Figure 1 gives a graphical representation 20% moment redistribution. Table 2 shows the reduced de-
of these formulae. All these formulae consider b, the permis- sign moments and values of b at critical sections. Reinforce-
sible percentage of moment redistribution, to be a function ment was provided for these reduced moments at critical
of c/d or w only. The comparison shows that significant dif- sections. Moment redistribution reduced the total flexural re-
ferences exist, among various design codes, on the amount inforcement required by 14% in this example.

© 2000 NRC Canada


1288 Can. J. Civ. Eng. Vol. 27, 2000

Table 1. Summary of available expressions for plastic rotations.


Model Expression for plastic rotation capacity
æL ö
Riva and Cohn (1994) q p = ç p ÷ fp z
è zø
For 1.0 £ fp /f y < 7.0,
-(0.9 - 0.8 g) (-D / 640 w 2 )
Lp æ B ö æ fp ö æ bö
= çA - ÷ ç ÷ ç ÷ f (g)
è 800wø çf ÷ è bw ø
z è yø
For fp /f y > 7.0,
Lp E æM öæ bö H

= (1.25w) F + G çç u - 1÷÷ ç ÷
z fp / f y è M y ø è bw ø
The constants A, B, C, D, E, F, G, and H and the function f(g)
depend upon bending moment distribution
1.05 - (w - w¢ ) L p
Naaman et al. (1986) qp = (upper bound)
850(w - w¢ ) - 35 d / 2
1.05 - 1.65 (w - w¢ ) L p
qp = (lower bound)
1300(w - w¢ ) - 40 d / 2
where Lp = d/2

0.004
FIP (1984) qp =
c/ d
æ Lö æ wb ö
q p = 0.0086 ç1 + 0.1 ÷ çç ÷
d ø è w + w p - w¢ ÷ø
Mattock (1983)
è

æ zö
Corley (1966) q p = ( fu - f y ) ç 0.5d + 0.2 d ÷
è dø

q tu æ z ö é æ w - w¢ ö d ù
Mattock (1964) = 1 + ç1.14 - 1÷ ê1 - ç ÷ ú
qu è d ø êë è w b ø 16.2 úû
z
Baker and Amarakone (1964) q p = 0.8 ( e cu - e cy ) k1 k3 ( k1 k3 = 0.5)
d

The required plastic rotations were obtained by perform- sections 3, 4, and 5, the required plastic rotations and c/d
ing a hinge-moment analysis. In hinge-moment analysis the limits from basic theory are different. In other words, eq. [3]
structure is modified and analysed for each load case by in- suggests that there is no unique relationship between b and
troducing hinges and the corresponding reduced design mo- c/d as implied by the code. Since the required plastic rota-
ments at the appropriate critical sections as shown in Fig. 4. tion is obtained from the hinge-moment analysis, it takes
Figure 4a shows the beam with the load case that causes into account the effects of loading and structure configura-
maximum moment at the centre support (node 5). Figure 4b tion in addition to the section properties and material proper-
shows the elastic bending moment diagram and the reduced ties. The code equation does not take into account the effect
design moments at all critical sections. A hinge will form at of all these factors. Although eq. [3] is fundamentally more
section 5 because at this section the factored elastic moment correct than the code equation for moment redistribution, it
is greater than the moment resistance. Figure 4c shows the does not account for the presence of shear cracking, tension
modified structure with a hinge at section 5. The reduced de- stiffening, and other factors of practical significance.
sign moment is applied to the ends of the members framing The conservative nature of the code and its inability to ac-
into the plastic hinge. The difference in the resulting mem- count for different parameters affecting plastic rotation ca-
ber end rotations is the required plastic hinge rotation. The pacity and plastic rotation demand points to the need for a
procedure is repeated for each load case. The required plas- comprehensive analytical model for computing permissible
tic rotations are given in Table 2. limits of qp and b.
The allowable c/d ratios, for given amounts of moment re-
distribution from Table 2, at the critical sections are given in Identification of important parameters
Table 3. The code c/d limit is very conservative for most of
the sections. It is also important to note that even though the To develop an analytical model for computing qp and b, it
required moment redistribution percentage is the same at is essential to recognize the important parameters affecting

© 2000 NRC Canada


Shakir and Rogowsky 1289

Fig. 1. Comparison of code provisions for moment redistribution. Fig. 3. Example beam: (a) beam geometry and location of criti-
cal sections and (b) beam cross section, material properties, and
loading.

Table 2. Design solution.

Section 2 3 4 5
Melastic (kN·m) 379 –493 264 (–36) –400
Mservice (kN·m) 271 –357 186 (–14) –286
Mreduced (kN·m) 353 –394 211 (–88) –320
Fig. 2. Allowable c/d ratio vs. plastic rotation. b (%) 6.8 20 20 20
q preqd (rad) 0.0029 0.00357 0.00355 0.00278

(b) geometric parameters: shape of the section, reinforce-


ment index (w), stirrup percentage;
(c) structural parameters: structural layout, members’ slen-
derness ratios (L/d), support conditions;
(d) loading parameters: load intensity and distribution;
(e) type of plastic hinge: flexural crack hinge, shear crack
hinge. A flexural crack hinge is one that contains flexural
cracks only; a shear crack hinge is one that contains in-
clined shear cracks in addition to flexural cracks.
For the purpose of the present study, the following param-
eters are considered of primary significance: compressive
strength of concrete, tensile strength of steel, the bond–slip
relationship, reinforcement index, member slenderness, trans-
verse reinforcement, and shear cracking in the vicinity of
plastic hinge.

Constitutive relationships
The constitutive relationships used in the development of
the model are illustrated in Fig. 5. For concrete, a modified
nonlinear behaviour of reinforced concrete structures. These Park et al. model (1982), shown in Fig. 5a, is used to incor-
parameters are the following: porate the beneficial effects of confinement, if any. The prin-
(a) material parameters: stress–strain laws for concrete and cipal modification is that the ultimate concrete strain is
steel, concrete–steel bond–slip law, material strengths; limited to 0.005. For steel, a bilinear stress–strain relation-

© 2000 NRC Canada


1290 Can. J. Civ. Eng. Vol. 27, 2000

Fig. 4. Hinge-moment analysis: (a) loading, (b) comparison of Fig. 5. Constitutive relationships: (a) concrete stress–strain law,
elastic moments and reduced design moments, and (c) modified (b) steel stress–strain law, and (c) moment–rotation relationship
beam with modified loading. for hinge.

Table 3. Comparison of the code c/d limit and the theoretical c/d
limit from eq. [3].
Section 2 3 4 5
c/d (limit from A23.3-94) 0.464 0.2 0.2 0.2
c/d (theory) 0.548 0.5 0.505 0.558

ship is adopted to include the effect of strain hardening, as


shown in Fig. 5b. The concrete and steel constitutive rela-
tionships lead to the moment–rotation curve shown in
Fig. 5c.
Based on the test results (Mattock 1964; Park et al. 1982),
the extreme concrete fibre strain limit is taken as 0.005
rather than 0.0035 as used in A23.3-94. In Mattock’s tests
the experimental concrete strains at ultimate were found to reinforcement, if present, increase the maximum concrete
be in the vicinity of 0.008 with few as high as 0.02. In the strain at ultimate moment resistance. On the basis of the ex-
tests of Park et al., conducted on a series of axially loaded perimental evidence, the assumption of an ultimate concrete
columns, the maximum concrete strains recorded were be- strain of 0.005 would be conservative for typical rectangular
tween 0.016 and 0.026 while the load–deformation curve beams. However, it must be noted that for some specific
was still rising. Higher values of experimental peak strains cases, such as box or tee beams and beams with constant
would have been obtained had the tests continued further. moment region, the limit of 0.005 might be unconservative.
The concrete strain at first visible crushing was at least For the latter cases, the designer should use an appropriate
0.005. In flexural members the strain gradient and the shear value for the ultimate concrete strain.

© 2000 NRC Canada


Shakir and Rogowsky 1291

Fig. 6. Cracking in the vicinity of plastic hinge. Fig. 7. Bond model (Sigrist and Marti 1994).

orientation of cracks will be opposite to that in the negative


moment region.
It was Dilger (1966) who first recognized the influence of
shear cracking on the tension steel force and proposed an ex-
pression that allows one to compute the shift in tensile force
due to shear cracking. Based on the same idea, the tensile
Analytical model force distribution can be determined by considering various
free body diagrams from Fig. 6. The variation of tension
The elongation of tension steel is the primary cause of force, Tx, within the fan region is given as
plastic rotation. One, therefore, needs a model that accu-
M u Vu x 2
rately predicts the elongation of tension steel. The proposed [4] Tx = - 2
model is based on the bond stress model of Sigrist and Marti dv 2dv
(1994) to account for tension stiffening. The beam is as-
sumed to be composed of a number of cracked elements. Outside the fan region, the variation of Tx is given as
The concrete deformations between the cracks are assumed M x Vx
to be small and are neglected. [5] Tx = +
dv 2
The model is entirely general. For the purpose of this pa-
per, it is applied to a two-span continuous beam subjected to where Mx and Vx are the moment and shear at the section
uniformly distributed load. Similar models can be developed being considered. In the derivation of the above expressions,
for other loading types and continuity conditions. it is assumed that the cracks are inclined at an angle of 45°,
The tension zone of the beam consists of a number of there is no contribution of concrete in resisting shear, and
cracked tension elements. These elements consist of the steel the support reaction is concentrated at a point. Thus it is as-
and the concrete surrounding it as shown in Fig. 8a. The sumed that the entire shear is resisted by transverse steel and
elongation of the steel depends upon the distribution of ten- the width of support is neglected. These assumptions under-
sion steel stress within the cracked elements. The stress dis- estimate the plastic hinge length and the tension steel force
tribution between cracks, in turn, depends upon the and would provide a conservative estimate of plastic rotation
magnitude of tensile forces at the cracks, stress–strain law capacity.
adopted for the steel and the concrete, and the bond–slip For beams with flexural cracks only, the distribution of
law. tension force is given as
Mx
[6] Ft =
Computation of tension steel force dv
Figure 6 illustrates the cracking pattern in the vicinity of a
plastic hinge. At the interior support of a beam, the diagonal Stress and strain distribution within cracked elements
cracks, instead of being parallel, tend to radiate from the The bond stress model of Sigrist and Marti (1994) is used
compression zone at the reaction point, forming a fan- to determine the stress and strain distribution within the
shaped region. The fan region is assumed to extend a dis- cracked elements. The model uses a stepped rigid perfectly
tance dv from the face of the support, where dv is the dis- plastic bond stress – slip relationship, as shown in Fig. 7.
tance between the centroid of the tension steel and the When the steel stress is less than the yield stress, the bond
centroid of the compression stress block. Outside the fan re- stress is given as
gion the cracks are parallel to each other and inclined at an
assumed angle of 45°. In the positive moment region, the [7] t b1 = 0.6 fc2/3

© 2000 NRC Canada


1292 Can. J. Civ. Eng. Vol. 27, 2000

Fig. 8. Application of bond stress model for determining steel are the distances in which the steel stress drops from the
strain distribution within cracked element: (a) cracked element, yield value to the minimum value. These values have been
(b) bond shear, (c) steel stress, and (d) steel strain. derived for the four cases and are given in the Appendix.
(Notations used are the same as those given in Fig. 8.)
The procedure for computing the plastic rotation capacity
consists of two parts. In the first part the expression for qp is
developed by considering the elongation of tension steel. In
the second part the expression for qp is developed by consid-
ering the stiffness, geometry, and loading on the structure.
The correct solution requires iteration in which the values of
loading and the steel strain at failure are adjusted to give the
same value of qp from the two approaches.

Part A — Computation of ␪ p from elongation of


tension steel
The following steps are involved:
(1) Divide the tension zone of the beam, in the vicinity of
plastic hinge, into a number of cracked elements.
Usually it is sufficient to consider a length d to each
side of the plastic hinge.
(2) Calculate the yield moment, My, corresponding to first
yielding of steel at the interior support.
(3) Calculate the corresponding load, qy,
8M y
[9] qy =
L2
(4) Assume a value of ultimate load, qu, greater than qy.
(5) Assume a value of steel strain above the support, corre-
sponding to ultimate load, such that the following rela-
tionship is satisfied:
æd ö
[10] e sm = e cu ç - 1÷
èc ø
where e sm is the mean steel strain in the first cracked
element, e cu is the ultimate strain in concrete, and c is
the depth of the neutral axis.
The procedure is to assume a value of e s at the crack at
the support and compute c and Mu at the support.
When the steel stress is greater than the yield stress, the Knowing qu and Mu the distribution of moment and
bond stress is given as shear force is determined by statics. Equation [4] or
eq. [6] gives the tensile force at the ends of the first
[8] t b2 = 0.5t b1 = 0.3 fc2/3 cracked element. The bond stress model gives the steel
stress (s sx) and strain (e sx) distribution within the first
where fc is the effective compressive strength of concrete. cracked element from which the mean steel strain (e sm )
For purposes of this study, fc is taken as fc¢ , the specified can be computed as
compressive strength of concrete. The rate of change of the
steel stress along the bar is 4t b/db, where db is the diameter æ Sm ö
of the reinforcing bar and t b = t b1 for the pre-yield region [11] e sm = ç ò e sx dx ÷ sm
ç ÷ ­
and t b2 for the post-yield region. For a tension element è 0 ø
loaded with Ft at both ends, typical distributions of t b, s s,
and e s are given in Fig. 8. If the end forces are not equal, the Finally, compare the values of mean strain e sm from
stress and strain distributions become nonsymmetric but can eqs. [10] and [11]. If the two agree, the assumed value
be readily obtained. There are four cases that need to be of e s was correct; if not, a new value of e s is assumed
considered depending upon the magnitude of the steel stress and the procedure repeated until the two values agree.
at crack locations. To determine the steel stress and strain (6) With qu from step 4 and Mu from step 5, determine the
distribution within a cracked element, the distances xL2, xR2, moments, shears, and tension steel forces along the
xL1, and xR1 and the magnitude of minimum steel stress member.
within the cracked element must be known. xL2 and xR2 are (7) Obtain the steel stress and strain distribution within
the distances in which the steel stress at the left and right each cracked element. Only those elements that contain
cracks respectively drops to the yield value, and xL1 and xR1 plastic deformations need to be considered. The plastic

© 2000 NRC Canada


Shakir and Rogowsky 1293

hinge length is limited to the region in which s s at Fig. 9. Determination of q p from structure geometry and load-
cracks is greater than or equal to fy and can be obtained ing: (a) two-span beam with uniformly distributed load,
from the stress distribution along the length of the (b) statical system after hinge formation, (c) distribution of elas-
beam. The curvature (fx) distribution can be obtained tic and inelastic bending moments, and (d) deformations at yield
by dividing the steel strains by the distance (d – c). and ultimate.
e sx
[12] fx =
d-c
(8) Integrate the curvatures to obtain the ultimate rotation
of the hinge:
Lp
1
[13] qu =
d-c ò e sx dx
0

(9) Use steps 6 through 8 to calculate the yield rotation (qy)


for e sy, qy, and My. All cracked elements within the
hinge length need to be considered. Since e sy, qy, and
My are known, the calculation of qy does not require a
trial-and-error procedure.
(10) Calculate the plastic rotation of the hinge:
[14] qp = q u - q y

Part B — Computation of ␪ p from structure geometry


and loading
Figure 9a shows a two-span beam, subjected to a uni-
formly distributed load. Assuming uncracked sections and a
constant flexural stiffness, the elastic bending moment at the
interior support is
q u L2
[15] Me =
8
where qu is the ultimate load on the beam. When Me is
greater than Mu, the nominal moment of resistance of the
section, moment redistribution occurs. The ratio b, called the
redistribution factor, is defined as
Me - Mu
[16] b=
Me

The yielding of reinforcement, at a moment of My, leads to


the formation of a plastic hinge at the interior support. The
corresponding load qy is given in eq. [9].
With hinge formation, the statical system changes to that 12 EI é Mu - My ù
[19] k= ê ú
shown in Fig. 9b. The plastic hinge is modelled as a rota-
êë (q u - q y)L - 8 (M u - M y) úû
2
L
tional spring with a rigid, linearly hardening plastic M–qp re-
lationship. The plastic hinge rotation, qp, can be derived as
With qy, My, qu, and Mu known from the first part of the
(q u - q y)L3 analysis, compute the value of k using eq. [19]. Calculate qp
[17] qp = again using either eq. [17] or eq. [18].
12 EI + 8kL
Compare the values of qp from parts A and B of the analy-
where EI is the cracked flexural stiffness of the beam and k sis. The solution is correct if the two values match. If not, a
is the rotational stiffness of the hinge. Moments and deflec- new value of qu is assumed and the entire procedure for
tion curves at the onset of yielding and in the elastic-plastic parts A and B of the analysis is repeated until convergence.
phase are given in Figs. 9c and 9d, respectively. For the
given M–qp relationship, we can write
Design example
Mu - My
[18] qp = The plastic rotation capacity is calculated for the interior
k
support of a uniformly loaded two-span continuous beam,
Equating eqs. [17] and [18], the stiffness of rotational shown in Fig. 9a. Geometry, material properties, and rein-
spring, k, is given as forcement detail for the beam are given in Table 4.

© 2000 NRC Canada


1294 Can. J. Civ. Eng. Vol. 27, 2000

Table 4. Geometry, material properties, and rein- Fig. 10. Example problem: (a) steel stress variation, (b) steel
forcement detail for the design example beam. strain variation, and (c) curvature variation at ultimate.
L = 12 m Ec = 27 700 MPa
b = 300 mm Es = 200 000 MPa
d = 750 Esh = 2270 MPa
L/d = 16 Bar size = No. 20
fy = 400 MPa No. of bars = 6 at interior support
fu = 540 MPa Compression steel = none
fc¢ = 30 MPa Stirrups No. 10 @ 150 c/c
e cu = 0.005 Steel type: CSA G30.18-M92
e su = 0.1 Grade 400R
e sy = 0.002

The tension region is divided into a number of cracked el-


ements. Crack locations are assumed to match the stirrup lo-
cations, i.e., sm = 150 mm. For beams with very wide or
very closely spaced stirrups this might not be appropriate
and one could use the available empirical relations, such as
the one given in the CEB Manual for Cracking and Defor-
mations (1985). The bond stress is given as

t b1 = 0.6 fc¢ 2 / 3 = 0.6 ´ 302/3 = 5.8 MPa

t b2 = 0.3 fc¢ 2/3 = 2.9 MPa

In the first part, qp is computed from the elongation of steel.


The yielding moment and the corresponding load are calcu-
lated as My = 486 kN·m and qy = 27.0 kN/m. Assuming a
value of ultimate load qu greater than qy, say, qu =
51.3 kN/m, it is found that e s = 0.0414, e sm = 0.0246, and
Mu = 570 kN·m. Figures 10a and 10b show the stress and
strain distributions, respectively, within the cracked ele-
ments. The plastic hinge length is limited to the region in
which s s at cracks is greater than or equal to fy and can be
obtained from Fig. 10a. The curvature distribution is shown
in Fig. 10c. The total rotation and the yield rotation are cal-
culated as qu = 0.02618 rad and qy = 0.0039 rad, and the
plastic rotation capacity is obtained as qp = 0.0223 rad.
In the second part, qp is computed from structure geome-
try and loading. The rotational spring stiffness is computed
as k = 3.76 × 109 N·mm/mm and plastic rotation is com-
puted, using either eq. [17] or eq. [18], as qp = 0.0223 rad.
Since the two values of qp from parts one and two of the
analysis match, a correct solution has been obtained. Thus,
for this example, qp = 0.0223 rad.
Figure 11 shows the variation of qp with mechanical rein-
forcement ratio (w = As fy/bd fc¢ ). The straight, slightly rising,
portion of the curve at the left indicates failure governed by
the rupture of steel. The decreasing hyperbolic portion of the
curve at the right indicates the failure governed by crushing
of concrete. For steels with low ductility, the straight portion
plots lower, as shown in Fig. 11 by the dashed line for e su =
0.04. For ordinary Canadian reinforcing bars, plastic rotation Comparison with experimental results
capacity will generally be governed by crushing of the con-
crete. For prestressing steel and perhaps welded wire fabric, The Mattock (1964) and Bosco and Debernardi (1993)
plastic rotation capacity could well be governed by the steel tests were used to validate the proposed model. The Mattock
ductility. test series consisted of 37 beams, and the Bosco and

© 2000 NRC Canada


Shakir and Rogowsky 1295

Fig. 11. Plastic rotation capacity as a function of reinforcement Fig. 12. Comparison of calculated and measured rotation capac-
index. ity (Mattock 1964): (a) including all test beams and (b) includ-
ing only beams with e cu £ 0.01.

Debernardi test series consisted of 44 beams. The following


sections provide a comparison of the experimental and rectangular reinforced concrete beams. Figure 14 shows the
model results. comparison of calculated and measured plastic rotation ca-
pacity. For most beams the model is conservative.
Mattock test
Mattock carried out a series of tests to investigate the Comparison with CSA A23.3-94
moment–rotation characteristics of rectangular reinforced
concrete beams in the support region of a continuous beam. Figure 15 compares the variation of b, the maximum al-
Figure 12 shows the comparison of calculated and measured lowable percentage of moment redistribution, with c/d as ob-
plastic rotation capacity. The 45° line represents the case tained from the proposed model and A23.3-94. The lower
where the calculated rotation capacity is equal to the mea- and upper limits of the theoretical curve are generated by
sured rotation capacity. All the points plot above the 45° considering the practical extreme values of the two main in-
line, which shows that the model is conservative. The rela- fluencing parameters. These parameters are the slenderness
tively high deviation of some plotted points from the 45° ratio (L/d) of the beam and the volumetric ratio of stirrups,
line in Fig. 12 is due to the fact that very high values of ulti- rv = 2(bs + hs)Avs/bsohsos, where Avs is the cross-section area
mate concrete strains (close to or higher than 0.02) were re- of one leg of stirrup, s is the spacing of stirrups, bs is the
ported for some of the tests. centre to centre width of stirrup, bso is the width of stirrup
Figure 13 illustrates the influence of effective reinforce- measured from outside to outside of stirrup, hs is the centre
ment index, (w – w¢), on plastic rotation capacity. The beams to centre height of stirrup, and hso is the height of stirrup
are grouped by slenderness ratio (L/d). The qp versus (w – w¢) measured from outside to outside of stirrup. Two-legged stir-
curves were generated for these beam groups for two values rups are assumed. The lower bound curve is obtained for a
of e cu , 0.005 and 0.008, and are shown in Fig. 13. Since the beam with a slenderness ratio of 21 and a volumetric ratio of
test values of e cu were in the vicinity of 0.01, the model stirrups of 0.0038. The upper bound curve is obtained for a
curves with e cu = 0.008 predict test results better than those beam with a slenderness ratio of 6 and a volumetric ratio of
with e cu = 0.005. stirrups of 0.0295. The value of rv of 0.0038 corresponds to
In general, the model predicts the trend of the test results No. 10, two-legged stirrups at a spacing of 300 mm and the
very well. In particular, the correlation is very good when value of rv of 0.0295 corresponds to No. 15, two-legged stir-
the values of e cu used in the model are close to the values rups at a spacing of 75 mm. It is found that deeper beams
obtained in the test. The choice of e cu has a significant effect with closely spaced stirrups can allow significantly more
on qp. For flexural members, the use of e cu = 0.005 is con- moment redistribution than slender beams with widely
servative. spaced stirrups. For c/d greater than 0.2, the code limit is
similar to the lower limit of the model. For smaller values of
Bosco and Debernardi tests c/d, the code is conservative even for the lower limit of the
Bosco and Debernardi (1993) carried out a series of tests model. The code is very conservative when conditions are
to study the influence of steel ductility, section size, and favourable for moment redistribution (upper limit of the
shape of moment diagram on plastic rotation capacity of model). It should be noted that if one does not wish to have

© 2000 NRC Canada


1296 Can. J. Civ. Eng. Vol. 27, 2000

Fig. 13. Variation of q p with w: (a) group 1, L/d = 5.5; Fig. 14. Comparison of calculated and measured rotation capac-
(b) group 2, L/d = 11; and (c) group 3, L/d = 22. ity (Bosco and Debernardi 1994).

Fig. 15. Comparison of model and code (A23.3-94) limits for al-
lowable moment redistribution.

Equation [20] gives a value of b max = 20% for zero live load
and b max = 33% for zero dead load. The actual limits for
practical cases would lie in between the extremes. The code
redistribution under service loads, then the upper limit on b restricts the allowable moment redistribution to 20% and
depends on the ratios of live load and dead load moments thus corresponds to the lower limit of eq. [20]. Although the
and can be determined from the following equation: code cut off limit for b is certainly conservative, it would be
more rational to calculate b max from eq. [20], using the ac-
(1.25M D + 1.5M L ) - (M D + M L )
[20] b max = ´ 100% tual dead load and live load moments obtained from the
1.25M D + 1.5M L analysis.

© 2000 NRC Canada


Shakir and Rogowsky 1297

Summary and conclusion BS 8110-85. 1987. Structural use of concrete. Handbook to British
Standards. Viewpoint Publications, London, England.
Design codes allow limited redistribution of elastic mo- CEB. 1985. CEB design manual on cracking and deformations.
ments. Significant differences exist among the design codes Ecole Polytechnique Federale de Lausanne, Switzerland,
on the permissible limits for moment redistribution. The pp. 2.30–2.32.
CSA A23.3-94 provision for permissible moment redistribu- CEB–FIP 1990. Model code 1990 for concrete structures. Comite
tion is assessed and found to be deficient in incorporating Euro-International du Beton, Bulletin d’Information n. 213/214,
the key parameters affecting b. A model is presented for Lausanne, May 1993.
evaluating the plastic rotation capacity and permissible mo- Chan, W.W.L. 1955. The ultimate strength and deformation of
ment redistribution of reinforced concrete members. The plastic hinges in reinforced concrete frameworks. Magazine of
model accounts for the effects of shear cracking, bond–slip, Concrete Research, 7(21): 121–132.
and the variability of curvature along the plastic hinge Cohn, M.Z. 1964. Rotation compatibility in the limit design of re-
length. The proposed model gives good agreement with ex- inforced concrete continuous beams. Proceedings of the Interna-
perimental results. tional Symposium on Flexural Mechanics of Reinforced
The model allows one to establish rational limits on the Concrete, ACI, SP-12, pp. 359–381.
amount of moment redistribution. The upper and lower Corley, G.W. 1966. Rotational capacity of reinforced concrete beams.
bound curves of c/d vs. b were generated for practical ex- ASCE Journal of the Structural Division, 92(ST5): 121–146.
tremes of the influencing parameters. It is found that the dif- CSA. 1994. Design of concrete structures for buildings. Standard
ference in the amount of allowable moment redistribution A23.3-94, Canadian Standards Association, Rexdale, Ont.
for the same c/d ratio can be as high as 200%, depending Dilger, W. 1966. Veranderlichkeit der Biege-und Schubsteifigkeit bei
upon the combination of influencing parameters. Compari- Stahlbetontragwerken und ihr EinfluB auf Schnittkraftverteilung
son with the code (CSA A23.3-94) limit for moment redis- und Trablast bei statisch unbestimmter Lagerung. Ddeutscher
tribution shows that the code limit is conservative. Although Ausschuss fur Stahlbeton, Heft 179, Berlin, Germany, pp. 101.
the code provides a reasonable estimate of b for unfavour- DIN. 1978. DIN 1045-78, German Standards.
able combinations of parameters, it can be very conservative DS. 1986. Structural use of concrete. DS 411-86, Danish Stan-
for more favourable combinations of parameters. Deeper dards.
beams with closely spaced stirrups allow significantly more Ernst, G.C. 1956. A brief for limit design. ASCE Transactions,
moment redistribution than that predicted by the code. For 121(605).
such beams, it is recommended that a more fundamental FIP. 1984. Practical design of reinforced and prestressed concrete
analysis be considered. Although a 20% maximum limit on structures (based on the CEB–FIP model code (MC78)).
b is certainly conservative, it is more rational to base this Thomas Telford Limited, London, England.
limit on avoiding moment redistribution under service loads. JSCE. 1986. Standard specifications for design and construction of
Equation [20] shows that up to 33% moment redistribution concrete structures. Part 1: Design. JSCE-86, Japanese Society
of Civil Engineers, Tokyo, Japan.
can be appropriate in some cases.
Langer, P. 1987. Vedrehfahigkeit plasticizer Tragwerksbereiche im
Stahlbetonbau. Ph.D. thesis, University of Stuttgart, Institut fur
Acknowledgements Werkstoffe in Bauwesen.
Mattock, A.H. 1964. Rotational capacity of hinging regions in re-
The work presented here is part of the doctoral research inforced concrete beams. Proceedings of the International Sym-
program of the first author. That effort was supported in part posium on Flexural Mechanics of Reinforced Concrete, ACI,
by the Natural Sciences and Engineering Research Council SP-12, pp. 143–182.
of Canada, the University of Alberta, and Concrete Canada. Mattock, A.H. 1983. Secondary moments and moment redistribu-
Professors J.G. MacGregor, P. Marti, and W. Dilger provided tion in ACI 318-77 code. Proceedings of the International Sym-
valuable comments and suggestions throughout the course of posium on Nonlinearity and Continuity in Prestressed Concrete.
the work. Their assistance is gratefully acknowledged. Preliminary publication, 3, University of Waterloo, Waterloo,
Ont., pp. 27–48.
Naaman, A.E. et al. 1986. Analysis of ductility in partially pre-
References stressed concrete flexural members. Prestressed Concrete Jour-
ACI 318-95. 1995. Building code for reinforced concrete. Ameri- nal, 31(3): 64–87.
can Concrete Institute, Detroit, Mich. Park, R., Priestly, M.J.N., and Gill, W.D. 1982. Ductility of
Bachmann, H. 1970. Influence of shear and bond on rotation ca- squared-confined concrete columns. ASCE Journal of the Struc-
pacity of reinforced concrete beams. IABSE Publications, Zu- tural Division, 108(ST4): 929–950.
rich, Switzerland, Vol. 30-II. Riva, P., and Cohn, M.Z. 1994. Rotation capacity of structural con-
Baker, A.L.L. 1959. Ultimate theory for concrete frame analysis. crete members. Magazine of Concrete Research, 46(168): 223–
ASCE Journal of the Structural Division, 85(ST9): 1–30. 234.
Baker, A.L.L., and Amarakone, A.M.N. 1964. Inelastic hyperstatic Sawyer, H.A. 1964. Design of concrete frames for two failure
frame analysis. Proceedings of the International Symposium on stages. Proceedings of the International Symposium on Flexural
Flexural Mechanics of Reinforced Concrete, ACI, SP-12, Mechanics of Reinforced Concrete, ACI, SP 12, pp. 405–438.
pp. 85–142. Sigrist, V., and Marti, P. 1994. Ductility of structural concrete, a con-
Bosco, C., and Debernardi, P.G. 1993. Influence of some basic pa- tribution. Workshop on Development of EN 1992 in Relation to
rameters on the plastic rotation of reinforced concrete elements. New Research Results and the CEB–FIP Model Code 1990, Czech
CEB Bulletin d’Information No. 218, pp. 25–44. Technical University, Prague, Czechoslovakia, pp. 211–223.

© 2000 NRC Canada


1298 Can. J. Civ. Eng. Vol. 27, 2000

List of symbols es strain in steel


e sm mean steel strain in the first cracked element
As area of tension steel e su ultimate steel strain
As¢ area of compression steel e sy steel strain at yield
Avs cross-section area of one leg of stirrup qp plastic rotation capacity of hinge
a depth of compression block q preqd required plastic rotation at hinge
b width of the beam qu total rotation of plastic hinge
bs centre-to-centre width of stirrup qy yield rotation of plastic hinge
bso width of stirrup measured from outside to outside of rv volumetric ratio of stirrups
stirrup ss stress in steel
bw width of the beam web s scrL steel stress at left face of a cracked element
c depth of neutral axis from the extreme compression fi- s scrR steel stress at right face of a cracked element
ber s s(min ) minimum steel stress within the cracked element
cy depth of neutral axis at first yield tb bond stress
d effective depth of the member fu curvature at ultimate
db diameter of reinforcing bar fu curvature at ultimate
dv distance between the centroid of the tension steel and fu plastic curvature
the centroid of the compression stress block fy curvature at yield
Ec modulus of elasticity of concrete w reinforcement index (As fy /bd fc¢ )
Es modulus of elasticity of steel w¢ compression reinforcement index (As¢ fy/bd fc¢ )
Esh strain hardening modulus of elasticity of steel wb balanced reinforcement index
EI cracked flexural stiffness of beam w eff effective compression reinforcement index (w – w¢ )
fc effective compressive strength of concrete wp pre-stressing reinforcement index
fc¢ specified compressive strength of concrete
fu ultimate strength of steel
fy yield strength of steel Appendix
hs centre-to-centre height of stirrup
hso height of stirrup measured from outside to outside of The following equations provide values of xL1, xL2, xR1,
stirrup xR2, and s s(min) , for different combinations of steel stresses at
K1 effective hinge length factor the crack locations, required for determining steel stress and
k stiffness of rotational spring strain distribution within cracked elements.
L span length
Lp length of plastic hinge to one side of the section Case 1: ␴scrL > fy; ␴scrR > fy; ␴s(min) < fy
Me elastic bending moment
(s scrL - fy)d b
Mu nominal moment of resistance of the section [A1a] x L2 =
Mx moment at a section, located at a distance x from the 4t b2
support
My moment corresponding to yielding of steel (s scrR - fy)d b
[A1b] x R2 =
q uniformly distributed load 4t b2
qu uniformly distributed ultimate load
qy uniformly distributed load corresponding to yielding of sm - (x L2 + x R2 )
steel [A1c] x L1 =
s spacing of stirrups 2
sm crack spacing
Tx tension force in the steel [A1d] xR1 = xL1
Vx shear at a section, located at a distance x from the support
4t b1 4t
wf uniformly distributed factored load [A1e] s s(min) = fy - x L1 = fy - b1 x R1
wfd uniformly distributed factored dead load db db
wfL uniformly distributed factored live load
xL1 left-hand-side region of a cracked element in which
Case 2: ␴scrL > fy; ␴scrR > fy; ␴s(min) > fy
t b = t b1
xL2 left-hand-side region of a cracked element in which (s scrL - s scrR ) d b + 4t b2 sm
t b = t b2 [A2a] x L2 =
8t b2
xR1 right-hand-side region of a cracked element in which
t b = t b1
xR2 right-hand-side region of a cracked element in which [A2b] xR2 = sm – xL2
t b = t b2
Z distance from the contraflexure point to the critical section [A2c] xL1 = xR1 = 0
b moment redistribution factor expressed as a percentage
4t b2 4t
of elastic moment (DM/Me) × 100% [A2d] s s(min) = s scrL - x L2 = s scrR - b2 x R2
e cu ultimate concrete strain at extreme compression fiber db db

© 2000 NRC Canada


Shakir and Rogowsky 1299

Case 3: ␴scrL > fy; ␴scrR £ fy; ␴s(min) < fy 4t b1 4t


[A3e] s s(min) = fy - x L1 = s scrR - b1 x R1
db db
s scrL - fy
[A3a] X L2 = db
4t b2 Case 4: ␴scrL < fy; ␴scrR < fy; ␴s(min) < fy
[A4a] xL2 = xR2 = 0
[A3b] xR2 = 0 (s scrL - s scrR ) d b + 4t b1 sm
[A4b] x L1 =
8t b1
( fy - s scrR ) d b + 4t b1( sm - x L2 )
[A3c] x L1 =
8t b1 [A4c] xR1 = sm – xL1
4t b1 4t
[A4d] s s(min) = s scrL - x L1 = s scrR - b1 x R1
[A3d] xR1 = sm – (xL1 + xL2) db db

© 2000 NRC Canada

You might also like