1 Reliability analysis of moment redistribution in RC beams

3 Hassan Baji: PhD candidate at the University of Queensland

4 Hamid Reza Ronagh: Senior lecturer at the University of Queensland

5 Address: School of Civil Engineering, The University of Queensland, St Lucia,

6 Brisbane, QLD, Australia, 4072

7 Telephone: +61733659117

8 Email: h.ronagh@uq.edu.au

9 Date text was written: 10 May 2012

10 Number of words in main text: 5425

11 Number of tables: 1

12 Number of figures: 8

13

14

15

16

17

1
 1 ABSTRACT

2 Design codes allow limited amount of moment redistribution in continuous reinforced

3 concrete beams and often make use of lower bound values in the procedure of estimating the

4 moment redistribution factors. Here, based on the concept of demand and capacity rotation,

5 and by means of Monte Carlo Simulation, a probabilistic model is derived for the evaluation

6 of moment redistribution factors. Results show that in all considered cases, the evaluated

7 mean and nominal values of moment redistribution factor are greater than the values provided

8 by the ACI code. On the other hand, the 5th percentile value of moment redistribution factor

9 could be lower than those specified by the code. Although the reduction of strength limit state

10 reliability index due to uncertainty in moment redistribution factors is not large, it is

11 comparable to the reduction in reliability index resulted from increasing the ratio of live to

12 dead load.

13 Keywords: Beams and girders; Buildings, structures and design; Concrete structures

14 NOTATION:
15 As tensile rebar area
16 b width of the rectangular section
17 c neutral axis depth
18 d effective depth of the rectangular section
19 db rebar diameter
20 Ec secant modulus of concrete
21 Es modulus of steel
22 f c' characteristic concrete compressive strength

23 f cm' mean cylinder compressive concrete strength

24 fy reinforcement yield strength

25 h depth of the rectangular section
26 k neutral axis parameter

2
 1 k1, k2, k3 stress block parameters
2 l beam span length
3 lp equivalent plastic hinge length
4 MD dead load bending moment
5 ML live load bending moment
6 MRn nominal bending capacity
7 Mu ultimate bending moment
8 n ratio of modulus of elasticity of steel to concrete
9 z distance from point of contra-flexure
10 1 , 1 equivalent stress block parameters
11 moment redistribution factor
12 c strain at peak stress of concrete

13 cu extreme fiber concrete ultimate strain

14 t extreme tensile rebar strain

15 y rebar steel yield strain

16 demand demand rotation of RC beam

17 capacity rotational capacity of RC beam

18 curvature ductility

19 tensile rebar percentage

20 b rebar percentage at parameter state

21 u ultimate curvature

22 y yield curvature

23 u ultimate uniformly distributed load

24

25

26

27

3
 1 INTRODUCTION

2 Normally, beams are loaded with different patterns of live load, after which an elastic

3 analysis is performed for each of the chosen live load patterns and then designed for the

4 envelope of these. Therefore, for any combination of live load patterns, certain critical

5 stations along the beam might reach the ultimate strength while other stations hold extra

6 capacity. In elastic analysis, this reserve capacity is not utilized; however, a full inelastic

7 analysis based on hinge formation could take advantage of this reserve capacity. The very

8 common way of dealing with this is to perform the analysis elastically but make use of

9 moment redistribution factors to account for the redistribution. The amount of moment

10 redistribution depends on the ductility of inelastic regions, the geometry of beams and the

11 loading pattern. The moment redistribution in continuous reinforced concrete beams is one of

12 the simplest applications of member ductility in the procedure of design. This prevents the

13 congestion of rebars at critical sections along the beams and allows a more even distribution

14 of rebars along the length. Design codes worldwide permit limited amount of moment

15 redistribution and each present a different formula for it. Usually provisions of codes do not

16 comprehensively consider the effect of all important parameters in the redistribution.

17 Mattock (1959) and Cohn (1964) did the first experimental programs on the subject of

18 moment redistribution. They concluded that cracking and deflection of beams designed for

19 limited moment redistribution are not significantly greater at service loads than for beams

20 designed by the distribution of moments according to the elastic theory. Shakir and

21 Rogowsky (2000) presented a model for calculating the plastic rotation capacity and

22 permissible moment redistribution factor in reinforced concrete beams. Their results showed

23 a good agreement with experimental results and their conclusion was that although the CSA

24 23.3-94 provides a reasonable estimate of moment redistribution factor for unfavorable

4
 1 combinations of important parameters, the code can be very conservative when conditions are

2 favorable for the moment redistribution to occur. Mostofinejad and Farahbod (2007)

3 implemented a parametric study on moment redistribution in continuous RC beams using

4 ductility demand and capacity concept. Their results showed that the permissible moment

5 redistribution in continuous reinforced concrete beams based on the current codes is

6 conservative for majority of cases but is not safe for some.

7 Strength and ductility capacities of reinforced concrete members depend on various

8 geometric and material properties, most of which are of a random nature. Therefore, a level

9 of uncertainty exists in the strength and ductility of reinforced concrete members. There have

10 been numerous studies on the strength of reinforced concrete members, the results of which

11 are already implemented in the design codes (Szerszen and Nowak, 2003, Bartlett et al.,

12 2003). In contrast, limited research can be found on the probabilistic inelastic deformation

13 and ductility. Trezos (1997) calculated the probabilistic parameters of the curvature ductility

14 of confined reinforced concrete columns using Monte Carlo Simulation (MCS). Parametric

15 and sensitivity analyses were carried out and the results were compared with the proposed

16 values of Euro Code No. 8 for structures in seismic zones. Kappos et al. (1999) investigated

17 the uncertainty of strength and ductility of confined reinforced concrete members using MCS

18 and Response Surface Method (RSM) and evaluated the concrete model and curvature

19 ductility provisions of Euro Code No. 8. Lu and Gu (2004) conducted a probabilistic analysis

20 of reinforced concrete member deformation limits for different performance levels. They

21 reported that curvature and drift limits generally follow a normal distribution.

22 In this study, firstly a closed form expression of curvature ductility and moment

23 redistribution is derived using ductility demand and capacity method developed by Silva and

24 Ibell (2008). Then, a probabilistic analysis is performed in order to find the reliability of the

5
 1 nominal and the code specified moment redistribution factors. Finally the effect of

2 considering uncertainty in evaluating the moment redistribution factor associated with the

3 strength limit state is investigated.

4 PROVISIONS OF CODES ON MOMENT REDISTRIBUTION

5 By means of plastic hinge length, the plastic hinge rotation could be related to

6 curvature ductility. Curvature ductility in reinforced concrete beams is directly related to the

7 percentage of tensile rebar area which in turn is correlated to the strain in extreme tensile

8 steel, t and neutral axis parameter ku c / d . Current design codes worldwide have different

9 moment redistribution forms. Some codes like the Canadian (CSA A23.3-04, 2004) and the

10 Australian (AS 3600, 2009) use the neutral axis parameter as the indicator of ductility in high

11 moment sections, i.e., the greater is the value of c / d , the lower are the ductility and the

12 permissible moment redistribution factor. Other design codes such as ACI (ACI 318-08,

13 2008) use t as an indicator of ductility and permissible moment redistribution factors. In the

14 previous ACI 318-99 (1999), the rebar percentage was used to calculate the moment

15 redistribution factors. Fig. 1 shows the curves for evaluating the moment redistribution factor

16 based on different design codes (ACI 318-08, 2008, AS 3600, 2009, CSA A23.3-04, 2004,

17 CEB-FIP, 1990).

18 One of the concerns in moment redistribution of reinforced concrete beams is the

19 serviceability limit state. Allowing large amounts of moment redistribution, which happens in

20 highly ductile sections with low rebar percentages, can create excessive deflection along the

21 beam span. In order to limit these deflections, one should prevent redistribution under service

22 loads. Shakir (2006) has proposed some equations to evaluate the maximum allowable

6
 1 moment redistribution factor considering the serviceability.

2 MOMENT REDISTRIBUTION REQUIREMENTS IN RC BEAMS

3 The basic idea for moment redistribution in continuous RC beams is that the demand

4 rotation required for the development of plastic hinges at ends and middle of the spans should

5 be lower than the rotational capacity of plastic hinge or hinges that yield first. The rotational

6 capacity in members could easily be transferred to section curvature capacity using the

7 concept of plastic hinge length.

8 Fig. 2 shows a typical continuous reinforced concrete beam with equal span length

9 under a uniformly distributed load. In this study, for the sake of simplicity, a beam fixed at

10 both ends, which can approximately represent an interior span of a multi-span beam is

11 considered. While the representation is not exact, the results can be trusted to be adequately

12 accurate. It is assumed that the RC beam has a constant stiffness, EI, along its length and

13 plastic hinges are first formed at the ends of the beam. The end hinges should show adequate

14 ductility and deform adequately to allow the formation of another hinge at the middle span of

15 the beam.

16 The idealized stress-strain curve for steel and that of moment-curvature curve at the

17 critical sections are shown in Fig. 3. In this study, the post yield rigidity has been neglected.

18 Usually for conventional reinforced concrete beams with normal reinforcement, the slope of

19 the post-yield part of the moment-curvature curve could be neglected. This always brings

20 about more conservative results. As is seen in Fig. 3, the width of the equivalent stress block
' '
21 is given by the product k1k3 f c . The term f c here represents the real concrete compressive

22 strength rather than the characteristic value. The k3 factor takes into account the difference

7
 1 between the in-situ compressive strength of concrete and the strength determined from

2 standard cylindrical tests as well as load duration effect. The k2 factor represents the stress

3 block depth factor.

4 The ultimate deformation capacity is expressed through ultimate curvature of the

5 section. The ultimate curvature is a state at which either of the specific ultimate compressive

6 strain in the concrete or the specific ultimate strength of extreme tensile rebar are reached.

7 Usually, for unconfined concrete, which is assumed for reinforced concrete beams, reaching

8 the ultimate compressive strain in concrete governs the ultimate curvature; because the

9 ultimate concrete compressive strain of unconfined concrete is relatively low and the rebar

10 steel even for high strength concrete has adequate ductility prior to rupture. In this study, it is

11 assumed that the ultimate curvature is governed by crushing of the extreme fiber of

12 reinforced concrete beam section. In addition the effect of the compressive rebar is neglected.

13 Here, firstly, the capacity of curvature ductility for critical section is derived using

14 basic mechanics of reinforced concrete. The curvature ductility is defined as the ratio of

15 ultimate to yield curvature (Park and Paulay, 1975):

u
(1)
y

16 According to Fig. 3, the ultimate curvature is defined as the gradient of strain over the

17 section height. Using geometry, compatibility and equilibrium, the yield and ultimate

18 curvature can be easily obtained:

cu f c' 1
u k1k2 k3 cu (2)
c f y d

8
 f y / Es f y / Es
y
d (1 k ) 2 (3)
1 n 1 1 d
n

1 In the case of yielding steel, the stress in the extreme compressive fiber of concrete

2 could be much lower than the cylinder strength f c' . The stress-strain curve for concrete is

3 approximately linear up to 0.70 f c' . Therefore, by using the elastic theory and assuming that

4 concrete stress does not exceed this value when the extreme steel yields, the neutral axis

5 parameter at yield is calculated as shown in Eq. (3).

6 In Eq. (3), is the tensile rebar percentage and n is the ratio of modulus of elasticity

7 of steel to concrete. Substituting Eqs. (2) and (3) in Eq. (1), the final expression for curvature

8 ductility in singly reinforced rectangular beam section can then be derived as:

2
1 n 1 1
f '
n (4)
k1k2 k3 Es cu
c

f y
2

9 In Eq. (4), the k1k2 k3 factor represents the equivalent stress block parameters,

f c'
10 Es cu factor is related to material properties and the last multiplier represents the cross
f y2

11 sectional dimensions.

12 The demand ductility (rotational or curvature) depends on the geometry of the

13 reinforced concrete beam, type of loading and plastic hinge length in the critical regions.

14 Referring to Fig. 2 and using the moment-area method, the demand rotation for the formation

15 of plastic hinges at the ends and middle of the beam can be calculated as shown in Eq. (5).

l u l 2 l
demand Mu Me Mu (5)
2 EI 12 2 EI

9
 1 Using the concept of equivalent plastic hinge length, the rotational capacity of end

2 hinges can also be calculated as shown in Eq. (6) (Park and Paulay, 1975).

capacity (u y )l p (6)

3 In Eq. (6), y and u are the yield and ultimate curvature at the end sections of the

4 beam while l p is the equivalent length of plastic hinge. The relationship in between the

5 parameters of ultimate uniform load u , moment redistribution factor , ultimate moment at

6 beam ends M u and elastic moment M e can simply be written as Eq. (7).

u l 2
Mu (1 ) M e (1 ) (7)
12

7 Equating Eq. (5) and (6) and substituting Eq. (7), results in Eq. (8) for demand

8 curvature ductility at the end critical sections.

1 l
1 (8)
2 1 lp

9 Finally, rearranging parameters in Eq. (8) results in Eq. (9) in which the permissible

10 moment redistribution factor can be found.

1
1
l
1 2 p 1
(9)
l

11 In Eq. (9), is the curvature ductility of the section which is calculated using Eq. (4).

12 Applying different end conditions for the reinforced concrete beam would result in different

13 values for moment redistribution factors. In the considered case of this study, which is a fixed

ended beam, the bending moment is 0.083u l . For instance, in three-bay continuous
2
14

reinforced concrete beams the maximum bending moment is 0.10u l that is less than that of
2
15

10
 1 a fixed ended beam. The smaller values of maximum bending moments at supports result in

2 lower plastic rotation and consequently lower curvature ductility demands, so they are more

3 conservative when it comes to the investigation of moment redistribution factors (Silva and

4 Ibell, 2008).

5 BASIC RANDOM VARIABLES

6 Important parameters for the estimation of moment redistribution factors are

7 presented in Eqs. (7) and (12). Eq. (12) shows that the plastic hinge length and the curvature

8 ductility at the critical section are the main random variables in the probabilistic analysis of

9 the moment redistribution factor. In this section, all random variables are reviewed and the

10 nominal, mean and standard deviation values as well as best fit probability density function

11 for each variable are selected based on the available literature.

12 Dimensions

13 The uncertainties in dimensions are relatively small and do not have significant effect

14 on the results of reliability analysis. Therefore for the sake of simplicity, in this study, the

15 statistical model used by Szerszen et al. (2005) is applied. According to this study, the bias

16 factor and coefficient of variation of 1.0 and 0.015 are used for the rebar area and 1.0 and

17 0.04 for the effective depth and width of RC sections.

18 Concrete properties

19 In this study, both strength and ductility are of concern. Therefore, a statistical model

20 that includes concrete compressive strength and all parameters of equivalent stress block is

21 required. Attard and Stewart (1998) in their probabilistic analysis of concrete stress block

22 used the lognormal distribution as the probability density function for the concrete

11
 1 compressive strength based on other studies (Tabsh and Aswad, 1995, Diniz and Frangopol,

2 1997, Pham, 1985). They used the mean value of ( f c' 7.5MPa) and the standard deviation

3 of 6MPa in their analysis. As in the current study, the statistical model of equivalent stress

4 block are derived from Attard and Stewart research, their model for concrete compressive

5 strength is also being used.

6 Attard and Stewart based on the mentioned probabilistic model for the concrete

7 compressive strength and the results of Setunge (1993), proposed models for the modulus of

8 elasticity Ec , and strain of concrete at peak stress c , for a wide range of concrete

9 compressive strength. They used the stress-strain model proposed by Attard and Setunge

10 (1996) to find the ultimate strain and equivalent stress block parameter variation. Their

11 results for the equivalent stress block parameters included both the dog bone and the

12 sustained load value for k3 . In this research, the main focus in the probabilistic analysis is

13 directed towards dog bone results. For all of these parameters, Attard and Stewart proposed

14 the use of normal probability functions.

15 Rebar steel

16 Based on Mirza nad McGregor (1979), yield strength of rebar steel can be modeled by

17 a Beta distribution. In their study a four-parameter Beta function was used for the probability

18 density function fitting. According to their study, the mean value and coefficient of variation

19 of 337MPa and 0.107 and 490MPa and 0.093 obtained for G40 (280MPa yield stress) and

20 G60 (420MPa yield stress), respectively.

21 Recently, Bournonville et al. (2004) gathered extensive data for various steel types for

22 the United States and Canada. Their study included different types of steel and a wide range

12
 1 of rebar sizes. Bournoville et al. (2004) used different types of steel produced according to

2 the ASTM standard. In the current study, grade 615 ASTM steel is considered in the three

3 different types of G40 (280 MPa), G60 (420 MPa) and G75 (500 MPa) and the statistical data

4 are taken from Bournonville et al. (2004). According to Bournonville et al. (2004) study, G40

5 and G60 steels follow Beta distribution while for G75, either normal or Beta distributions

6 could be used as the best fit probability functions. It is assumed that the modulus of elasticity

7 follows lognormal distribution with a mean equal to 201GPA and a coefficient of variation of

8 0.033.

9 Plastic hinge length

10 Various formulations have been suggested for the determination of equivalent plastic

11 hinge length. There is high uncertainty involved in the plastic hinge length as it is affected by

12 many uncertain factors. Lu and Gu (2004) combined the experimental data from previous

13 studies (Priestley and Park, 1987, Sheikh et al., 1994, Bayrak, 1999) and plotted the plastic

14 hinge length versus z (the distance from the point of contra-flexure) and db (the longitudinal

15 rebar diameter). Using linear regression analysis, they proposed the following equation for

16 the estimation of plastic hinge length:

l p 0.077 z 8.16db (10)

17 In order to include uncertainty in the Eq. (10), a model uncertainty factor has to be

18 associated to it. Based on the experimental results and a linear regression, (Lu and Gu, 2004)

19 found that the model uncertainty factor can be presented with normal distribution with a

20 mean equal to unity and coefficient of variation of 0.198. In this study, the same model is

21 used for the plastic hinge random variable.

13
 1 A summary of the probabilistic model for plastic hinge length is presented in Table 1

2 in which f c' represents the specified or nominal concrete compressive strength and f cm' shows

3 the mean concrete compressive strength of concrete.

4 PROBABINISTIC ANALYSIS OF MOMENT REDISTRIBUTION FACTOR

5 In order to consider the effect of variation of concrete compressive strength, three

6 different strengths of 20, 40 and 60 MPa are selected for this study. The nominal values of all

7 random variables that are summarized in Table 1 are taken from ACI 318-08 which signifies

8 the assumption that the code specified values represent the nominal values. In this study, two

9 different spans to depth ratios of 10 and 20 are used and the effective depth of reinforced

10 concrete beams is assumed to be 500 mm. In an ideal fixed ended beam, the distance from the

11 point of contra-flexure to the section at which the maximum moment occurs, z, is about one

12 fifth of the whole span length. Consequently, using Eq. (10), assuming 25 mm rebar diameter

13 and having 500 mm effective depth, a plastic hinge length of 281 mm and 358 mm are

14 calculated, respectively. In these cases, the ratios of plastic hinge to span length are about

15 0.0562 and 0.0358, respectively. The assumptions for span length, effective depth and bar

16 diameter are presented as an example here. Any other values that provide ratios close to the

17 code considered lower bound or other limits could be used.

18 Design codes basically use the 5th percentile value (95% chance of being exceeded) as

19 the nominal value of resistance related parameters. Moment redistribution factor could be

20 treated in a similar manner. In evaluating the 5th percentile values, simulated data are used to

21 find the probability density function of moment redistribution factor. Fig. 4 shows the

22 theoretical 5th percentile moment redistribution value and ACI 318-08 specified value of

23 moment redistribution factor as a function of steel strain for different types of steel. In all

14
 1 considered cases, the curve proposed by ACI 318-08 falls below the evaluated mean value.

2 Furthermore, as the steel rebar yield strength increases, the mean and nominal curves merge

3 closer. Almost in all cases, mean moment redistribution curves lay between the nominal and

4 the code curves. For the large span to depth ratio of 20, the code specified moment

5 redistribution factors are smaller than the 5th percentile values; while, for the smaller span to

6 depth ratio of 10, the code specified moment redistribution factors are slightly larger. The 5th

7 percentile values are much smaller than the nominal and the mean values and this indicates

8 the high uncertainties in the evaluation of moment redistribution factors.

9 Fig. 5 shows the probability of exceedance as a function of steel strain for different

10 types of steel. The results in Fig. 5 are based on the span to depth ratio of 20. The code

11 specified nominal and mean values are treated as deterministic values and the probability of

12 exceeding these values is calculated using the simulated data. The probability of exceedance

13 gives estimation for safety of the calculated moment redistribution factors and those specified

14 by the code. Results in Fig. 5 show that the code specified values for moment redistribution

15 provide higher safety margins of above 75% comparing to mean and nominal values. For the

16 Mean and the Nominal values, the chance of exceedance is about 50%. The chance of

17 exceedance is not uniform for the code specified moment redistribution factors, whereas the

18 nominal and the mean values show almost uniform probability against steel strain. The 50%

19 chance of being exceeded indicates that the probability density function of moment

20 redistribution factor is close to a symmetric distribution such as the normal distribution.

21 In order to investigate the effect of different types of concrete and rebar steel on the

22 reliability of code specified values, Fig. 6 presents the chance of exceedance for ACI 318-08

23 specified values of moment redistribution factor considering different concrete and steel rebar

24 strengths. For the range of 0.0075-0.020 steel strain, the chance of exceeding code specified

15
 1 moment redistribution factors is above 75% for all of the considered cases. The results for

2 span to depth ratio of 10 are much higher than those for span to depth ratio of 20 and close to

3 95%.

4 The reason for the values shown in Figs 5 and 6 for strains below 0.0075 being too

5 different from the values above this level of strain is that the ACI code does not allow any

6 redistribution of moment in this region. (i.e. moment redistribution factor =0). Furthermore,

7 as the moment redistribution factor is kept at 0.2 for strains above 0.02, the rate of increase in

8 the chance of exceedance increases more rapidly for strains beyond 0.02.

9 STRENGTH RELIABILITY BY CONSIDERING UNCERTAINTY IN

10 MOMENT REDISTRIBUTION FACTOR

11 The reliability of reinforced concrete beams under dead and live loads is a classic

12 problem in structural reliability and has been investigated by many researchers. Here, the

13 reliability of reinforced concrete beam is evaluated taking into consideration the uncertainty

14 of moment redistribution factor.

15 The strength limit state function of moment resistance of any reinforced concrete

16 beam can be stated as shown in Eq. (11). In this limit state, both the dead load and the live

17 load are considered.

MR
g MD ML (11)
1

18 In Eq. (11), MR, MD, and ML represent the moment resistance of RC section, dead load

19 effect and live load effect, respectively. The factor stands for the moment redistribution

20 factor. The moment redistribution factor can be calculated using Eq. (9). Resistance of a

21 reinforced concrete section depends on its dimensions, and the material properties of concrete

16
 1 and steel. According to Fig. 3, the moment resistance of a singly reinforced rectangular

2 section is calculated as shown in Eq. (12).

fy
M R bd 2 f y 1 ' (12)
2k1k3 f c

3 All variables of Eq. (12) were defined in the previous sections. These variables are

4 random in nature.

5 Dead load is treated as a normal variable with the mean of 1.05 times its nominal

6 value and the coefficient of variation of 0.10 according to Ellingwood et al. (1980). Live load

7 in Eq. (14) is the maximum life time live load and is modeled by extreme type I distribution

8 with bias factor (mean to nominal) and coefficient of variation of 1.0 and 0.23, respectively

9 (Ellingwood et al., 1980). For the design to suit, the demand must be less than the capacity.

10 Eq. (13) shows the governing equation relating the load effects (demand) to the capacity

11 (moment resistance).

M Rn
D M Dn L M Ln (13)
1 n
12

13 In Eq. (13), subscript n denotes the nominal value of the variable. The factor represents the

14 load factor. In this study, the nominal value of moment redistribution factor, , is calculated

15 either using Eq. (9) by putting the nominal values of all variables or based on ACI 318-08

16 design code. Monte Carlo Simulation (MCS) is utilized to find the probability of failure and

17 the reliability index of the strength limit state shown in Eq. (11). Three cases are considered

18 in the reliability analysis; in the first case, the effect of moment redistribution is not

19 considered i.e. elastic analysis and design is considered; while in the other two cases, the

20 moment redistribution factor is calculated using Eq. (9) i.e. based on the section capacity and

17
 1 the ACI 318-08 design code.

2 Fig. 7 shows the results of reliability analysis for two different concrete compressive

3 strengths and the live to dead load ratio of 1.0. The reliability indices for the elastic analysis

4 are almost constant and do not depend on the rebar percentage and the concrete properties.

5 As expected, considering the uncertainty in the moment redistribution factor causes a

6 reduction in the reliability index (in case where the section capacity is used to derive the

7 moment redistribution factor). The amount of reduction for high steel strain (low rebar

8 percentage) is more than that of low steel strain. The reason behind this result is that the

9 moment redistribution factors for high steel stain are higher and moment redistribution factor

10 has a proportionally larger contribution in the design. On the other hand, when low steel

11 strain is used in the design, lower ductility is produced and consequently the role of moment

12 redistribution uncertainly is decreased. In case of =0, this case converges to the elastic case.

13 Fig. 7 shows that if the code specified moment redistribution factors are used in the

14 design, the reliability indices will be even higher than those of an elastic analysis. As was

15 shown in the previous sections, the code specified values are more conservative than the

16 calculated ones based on section capacity. Therefore, it is obvious that if the code specified

17 moment redistribution factors are used in a design, higher reliability indices will be resulted

18 for the strength limit states.

19 In order to investigate the effect of different live to dead load ratio, several reliability

20 analyses with various live to dead ratio were conducted. The results in Fig. 8 are based on

21 moment redistribution factor resulting from a section capacity analysis (nominal moment

22 redistribution factor is used in design). As is seen, although the reduction of reliability index

23 due to uncertainty in moment redistribution factors is not large, it is comparable to the

18
 1 reduction in reliability index resulted from increasing the ratio of live to dead load.

2 CONCLUSIONS

3 A reliability based analysis of moment redistribution was performed using MCS

4 method. The statistical properties and distributions of all random variables were derived from

5 the currently available literature. According to the results of reliability analysis, the evaluated

6 mean and nominal values of moment redistribution factor are greater than the values provided

7 by ACI 318-08 code in all of the considered cases. The results show that in some cases, the

8 5th percentile values of moment redistribution factor could be smaller than those specified by

9 the code. However, due to the lack of adequate statistical models for parameters related to the

10 moment redistribution, no specific judgment can be made on the code specified values.

11 Generally, the probability of exceedance of moment redistribution factors specified by ACI

12 318-08 are above 75 percent, and increasing with the increase in the steel strain. On the other

13 hand, the probability of exccedance for the mean values of moment redistribution is about 50

14 percent.

15 As expected, considering the uncertainty in the moment redistribution factor causes a

16 reduction in the reliability index of the strength limit state. Effect of considering the

17 uncertainty of moment redistribution factor in the moment resistance reliability index is in the

18 order of about 0.5 in the reliability index scale. Although the reduction of reliability index

19 due to uncertainty in moment redistribution factors is not large, it is comparable to the

20 reduction in reliability index resulted from increasing the ratio of live to dead load.

21 Lack of probabilistic models for important parameters such as plastic hinge, ultimate

22 strain of concrete, concrete stress block parameters and other important variables complicates

23 the judgment on reliable values for moment redistribution factors. The results presented here

19
 1 are highly dependent on the statistical models chosen for the main random variables.

2 Nevertheless, they show the importance of probabilistic evaluation of moment redistribution

3 factors.

4 REFERENCES

5 ACI 318-08 2008. Building Code Requirements for Structural Concrete and
6 Commentary. Farmington Hills, MI, USA: American Concrete Institute.

7 ACI 318-99 1999. Building Code Requirements for Structural Concrete. Farmington
8 Hills, MI, USA: American Concrete Institute.

9 AS 3600 2009. Australian concrete structures standard. Standards Australia. Sydney,

10 Australia.

11 ATTARD, M. & SETUNGE, S. 1996. Stress-strain relationship of confined and

12 unconfined concrete. ACI Materials Journal, 93.

13 ATTARD, M. M. & STEWART, M. G. 1998. Two parameter stress block for high-
14 strength concrete. ACI Structural Journal, 95, 305-317.

15 BARTLETT, F. M., HONG, H. P. & ZHOU, W. 2003. Load factor calibration for the
16 proposed 2005 edition of the National Building Code of Canada: Companion-action load
17 combinations. Canadian Journal of Civil Engineering, 30, 440-448.

18 BAYRAK, O. 1999. Seismic performance of rectilinearly confined high strength

19 concrete columns. PhD, University of Toronto.

20 BOURNONVILLE, M., DAHNKE, J. & DARWIN, D. 2004. Statistical Analysis of

21 the Mechanical Properties and Weight of Reinforcing Bars. Kansas: The University of
22 Kansas, Structural Engineering and Materials Laboratory.

23 CEB-FIP 1990. CEB-FIP model code 1990. Lausanne, Switzerland.

24 COHN, M. Rotation Compatibility in the Limit Design of Reinforced Concrete

25 Continuous Beams. International Symposium on the Flexural Mechanics of Reinforced
26 Concrete, 1964 Miami. ASCE-ACI, 359-381.

27 CSA A23.3-04 2004. Design of concrete structures. Canadian Standards Association.

28 DINIZ, S. M. C. & FRANGOPOL, D. M. 1997. Reliability assessment of high-

29 strength concrete columns. International Conference on Structural Safety and Reliability.
30 Kyoto, Japan.

31 ELLINGWOOD, B., GALAMBOS, T. V., MACGREGOR, J. G. & CORNELL, C. A.

32 1980. Development of a probability-based load criterion for American National Standard
33 A58, Washington, D.C., National Bureau of Standards, NBS Special Publication 577.

20
 1 KAPPOS, A. J., CHRYSSANTHOPOULOS, M. K. & DYMIOTIS, C. 1999.
2 Uncertainty analysis of strength and ductility of confined reinforced concrete members.
3 Engineering Structures, 21, 195-208.

4 LU, Y. & GU, X. 2004. Probability analysis of RC member deformation limits for
5 different performance levels and reliability of their deterministic calculations. Structural
6 Safety, 26, 367-389.

7 MATTOCK, A. H. Redistribution of design bending moments in reinforced concrete

8 continuous beams. ICE Proceedings, 1959 1959 London, England. Institution of Civil
9 Engineers (ICE), 35-46.

10 MIRZA, S. A. & MACGREGOR, J. G. 1979. Variability of mechanical properties of

11 reinforcing bars. Journal of the Structural Division, 105, 921-937.

12 MOSTOFINEJAD, D. & FARAHBOD, F. 2007. Parametric study on moment

13 redistribution in continuous RC beams using ductility demand and ductility capacity concept.
14 Iranian Journal of Science and Technology, Transaction B: Engineering, 31, 459-471.

15 PARK, R. & PAULAY, T. 1975. Reinforced concrete structures, John Wiley & Sons
16 Inc.

17 PHAM, L. 1985. Reliability analyses of reinforced concrete and composite column

18 sections under concentric loads. Institution of Engineers, Australia, Civil Engineering
19 Transactions, CE27, 68-72.

20 PRIESTLEY, M. & PARK, R. 1987. Strength and ductility of concrete bridge

21 columns under seismic loading. ACI Structural Journal, 84, 61-76.

22 SETUNGE, S. 1993. Structural properties of very high strength concrete. PhD thesis,
23 Monash University.

24 SHAKIR, A. 2006. Moment redistribution in reinforced concrete structures. PhD

25 Thesis, University of Alberta.

26 SHAKIR, A. & ROGOWSKY, D. M. 2000. Evaluation of ductility and allowable

27 moment redistribution in reinforced concrete structures. Canadian Journal of Civil
28 Engineering, 27, 1286-1299.

29 SHEIKH, S. A., SHAH, D. V. & KHOURY, S. S. 1994. Confinement of high-

30 strength concrete columns. ACI Structural Journal, 91, 100-100.

31 SILVA, P. F. & IBELL, T. J. 2008. Evaluation of moment distribution in continuous

32 fiber-reinforced polymer-strengthened concrete beams. ACI Structural Journal, 105, 729-39.

33 SZERSZEN, M. M. & NOWAK, A. S. 2003. Calibration of design code for buildings

34 (ACI 318): Part 2 - Reliability analysis and resistance factors. ACI Structural Journal, 100,
35 383-391.

36 SZERSZEN, M. M., SZWED, A. & NOWAK, A. S. 2005. Reliability analysis for

37 eccentrically loaded columns. ACI Structural Journal, 102, 676-688.

21
1 TABSH, S. & ASWAD, A. 1995. Statistical properties of plant-produced high
2 strength concrete in compression. PCI journal, 40, 72-76.

3 TREZOS, C. G. 1997. Reliability considerations on the confinement of RC columns

4 for ductility. Soil Dynamics and Earthquake Engineering, 16, 1-8.

22
1

2 Fig. 1 Permissible moment redistribution factor versus c / d based on different

3 design codes

5 Fig. 2 Reinforced concrete beam geometry

23
1

2 Fig. 3 Basic assumptions on mechanical model for reinforced concrete

24
 0.60 0.60
Moment Redistriution Factor () Nominal Nominal

Moment Redistriution Factor ()

0.50 Mean 0.50 Mean
5th Percentile 5th Percentile
0.40 ACI 318-08 0.40 ACI 318-08

0.30 0.30

0.20 0.20

0.10 0.10

0.00 0.00
0.000 0.005 0.010 0.015 0.020 0.025 0.000 0.005 0.010 0.015 0.020 0.025
Steel Strain, t Steel Strain, t

(a) G60 steel rebar and l/d=20 (b) ) G60 steel rebar and l/d=10

0.60 0.60
Nominal Nominal
Moment Redistriution Factor ()

Moment Redistriution Factor ()

0.50 Mean 0.50 Mean
5th Percentile 5th Percentile
0.40 ACI 318-08 0.40 ACI 318-08

0.30 0.30

0.20 0.20

0.10 0.10

0.00 0.00
0.000 0.005 0.010 0.015 0.020 0.025 0.000 0.005 0.010 0.015 0.020 0.025
Steel Strain, t Steel Strain, t

(c) ) G75 steel rebar and l/d=20 (d) ) G75 steel rebar and l/d=10

1 Fig. 4 Mean, Nominal and 5th percentile values of moment redistribution factor

2 ( f c' = 40MPa)

25
 1.00 1.00
Chance of Being Exceeded

Chance of Being Exceeded

0.75 0.75

0.50 0.50

0.25 Nominal 0.25 Nominal

Mean Mean
ACI 318-08 ACI 318-08
0.00 0.00
0.000 0.005 0.010 0.015 0.020 0.025 0.000 0.005 0.010 0.015 0.020 0.025
Steel Strain, t Steel Strain, t

(a) G60 steel rebar (b) G75 steel rebar

1 Fig. 5 Reliability of code specified, Mean and Nominal values of moment redistribution

2 factor ( f c' = 40MPa)

(a) l/d=20 (b) l/d=10

5 Fig. 6 Reliability of code specified values of moment redistribution factors for different

6 types of concrete and steel rebar

26
 4.50 4.50

3.75 3.75
Reliability Index

Reliability Index
3.00 3.00

2.25 2.25

1.50 1.50
ACI 318-08 ACI 318-08
0.75 Section Capacity 0.75 Section Capacity
Elastic Elastic
0.00 0.00
0 0.005 0.01 0.015 0.02 0.025 0.000 0.005 0.010 0.015 0.020 0.025
Steel Strain, t Steel Strain, t

(a) fc' 20MPa (b) fc' 60MPa

1 Fig. 7 Reliability indices of strength limit state for different cases (G60 steel, l/d=20 and

2 LL/DL =1.0)

4.00 4.00

3.50 3.50

3.00 3.00
Reliability Index

Reliability Index

2.50 2.50

2.00 2.00

1.50 LL/DL=0.5 1.50 LL/DL=0.5

1.00 LL/DL=1.0 1.00 LL/DL=1.0

LL/DL=2.0 LL/DL=2.0
0.50 0.50
LL/DL=5.0 LL/DL=5.0
0.00 0.00
0.000 0.005 0.010 0.015 0.020 0.025 0.000 0.005 0.010 0.015 0.020 0.025
Steel Strain, t Steel Strain, t

(a) G60 steel rebar (b) G75 steel rebar

4 Fig. 8 Effect of live to dead load ratios on strength limit state reliability indices

5 ( f c' = 40MPa and l/d=20)

27
1 Table 1 Summary of statistical models of random variables

COV2/Standard
Variable Bias1/Mean PDF3 Reference
Deviation
(Szerszen et al.,
b 1.00 0.04 Normal
2005)
Dimensions

(Szerszen et al.,
d 1.00 0.04 Normal
2005)
(Szerszen et al.,
As 1.00 0.015 Normal
2005)
(Attard and
f c' Nominal+7.5MPa 6.0MPa Lognormal
Stewart, 1998)

Ec 1.01 4370.3 fcm' 0.5164 0.15 Normal

(Attard and
Stewart, 1998)
4.11 f cm' (Attard and
c 0.18 0.001 fc' Normal
Concrete

Ec 4
f ' Stewart, 1998)
cm

c 2.8133 fcm'
0.2093 (Attard and
cu 0.19 Normal
Stewart, 1998)

1.2932 f c'
.0998 (Attard and
(k1k3 ) 0.09 Normal
Stewart, 1998)

1.0948 f c'
0.091 (Attard and
k2 0.03 Normal
Stewart, 1998)

Es 1.00 0.033 Lognormal (Lu and Gu, 2004)

Steel

(Bournonville et
G60 1.22 0.068 Beta
al., 2004)
(Bournonville et
G75 1.16 0.048 Beta
al., 2004)
Plastic
Hinge

lp 0.077 z 8.16db 0.198 Normal (Lu and Gu, 2004)

1. Mean to Nominal value

2. Coefficient of Variation
3. Probability Density Function
2

28