Sunteți pe pagina 1din 12

992

Rational model for calculating deflection of reinforced concrete beams and slabs
Peter H. Bischoff

Abstract: Deflection control is an important performance criterion that needs to be satisfied to ensure serviceability of the structure for its intended use. The extent of cracking and amount of reinforcement affects the flexural rigidity, EI, of a reinforced concrete member and both the Canadian concrete design standard (CSA A23.3-04) and ACI Building Code (ACI 318-05) use an effective moment of inertia, Ie, that was originally proposed by Branson to compute beam deflection. This is an empirically derived equation that works well within a narrow range of limits corresponding to steel-reinforced concrete beams with a reinforcing ratio between 1% and 2%. However, the equation underestimates deflection for steel-reinforced concrete beams and slabs with a reinforcing ratio less than 1% and for most beams reinforced with low-modulus, fibre-reinforced-polymer (FRP) bars. Deflection of slender tilt-up wall panels can also be underestimated with Bransons equation. This paper provides an explanation of why the Branson equation does not always work well in predicting deflection, and presents a rational approach to develop an alternative expression for the effective moment of inertia that works equally well for both steel- and FRP-reinforced concrete at all reinforcing ratios. A rational expression is also introduced for continuous beams that uses an averaged moment of inertia, Ie,avg, to calculate beam deflection. Changes are included in a proposed revision to deflection prediction requirements specified in clause 9.8 of CSA A23.3-04. Key words: reinforced concrete, deflection, effective moment of inertia, serviceability. Rsum : Le contrle des flches est un critre de performance important qui doit tre rencontr pour assurer un comportement en service des structures en fonction de lusage planifi. Le niveau de fissuration et la quantit darmature affectent la rigidit en flexion EI dun lment en bton arm. La norme canadienne de Calcul des ouvrages en bton (CSA A23.3-04) et l ACI Building Code (ACI 318-05) utilisent un moment dinertie efficace Ie , propos initialement par Branson, pour calculer la flche des poutres. Il sagit dune quation drive empiriquement qui fonctionne bien dans une plage troite correspondant aux poutres de bton arm ayant un pourcentage darmature en acier entre 1 et 2 %. Toutefois, lquation sous-estime les flches pour les poutres et les dalles en bton arm ayant un pourcentage darmature en acier infrieur 1 % ainsi que pour la plupart des poutres en bton arm avec des tiges renforces de fibres de polymres PRF faible module. La flche des panneaux muraux minces mis en place par relvement peut tre galement sous-estime par lquation de Branson. Le prsent article expliqure pourquoi lquation de Branson ne prdit pas toujours adquatement la flche et prsente une approche rationnelle qui permet de dvelopper une autre expression pour le calcul du moment dinertie efficace qui fonctionne aussi bien pour le bton arm darmatures en acier ou en PRF, et ce, pour tous les pourcentages darmature. Une expression rationnelle est galement prsente pour les poutres continues pour lesquelles on utilise un moment dinertie moyen, Ie, avg , pour calculer la flche des poutres. Des suggestions pour rviser les exigences du paragraphe 9.8 de la norme CSA A23.3-04 pour la prdiction des flches sont inclus dans larticle. Mots-cls : bton arm, flche, moment dinertie efficace, comportement en service. [Traduit par la Rdaction] Bischoff 1002

Introduction
Reinforced concrete structures are typically designed for strength using factored loads to ensure that the structure is safe from collapse. Member deflection and crack widths are subsequently checked to ensure that the structure remains
Received 21 June 2006. Revision accepted 4 February 2007. Published on the NRC Research Press Web site at cjce.nrc.ca on 7 September 2007. P. Bischoff. Department of Civil Engineering, University of New Brunswick, PO Box 4400, Fredericton, NB E3B 5A3, Canada (e-mail: bischoff@unb.ca). Written discussion of this article is welcomed and will be received by the Editor until 31 December 2007.
Can. J. Civ. Eng. 34: 9921002 (2007)

serviceable under the specified loading conditions. Although deflection was often less of a problem when design was conservatively based on limiting stresses in the concrete and steel at service loads, the adoption of strength design and use of higher-strength materials has led to more slender members that have a greater potential for experiencing deflection problems. Use of advanced composite materials for internal reinforcement can also lead to more flexible members because of their lower stiffness. Hence, serviceability failures related to excessive deflection are more common than one might expect (Gilbert 2001) and can lead to costly litigation and repairs. Deflection problems are avoided by either limiting the span to depth ratio of the member (to give a minimum member thickness) or ensuring that calculated deflection values
2007 NRC Canada

doi:10.1139/L07-020

Bischoff

993 Fig. 1. (a) Flexural member response based on an effective moment of inertia Ie. (b) Flexural member response incorporating tension stiffening.

do not exceed specified limits that are often based on maximum deflection to span ratios found in most codes. Minimum member thickness values are intended for members not supporting or attached to partitions or other construction likely to be damaged by large deflections and do not apply when the partitions are susceptible to cracking. More often than not, minimum thickness values are used regardless of whether damage to the partition is likely (Scanlon and Choi 1999). For this reason, it has been suggested that deflection values always be computed and minimum thickness values be used as a guideline only for preliminary selection of member size. Whereas calculation of deflections may appear to be slightly onerous, it should be remembered that this procedure is typically included in most software packages used for design of reinforced concrete structures, many of which use an effective moment of inertia to compute deflection (Scanlon et al. 2001). Both the Canadian concrete design standard CSA A23.3-04 (CSA 2004) and ACI Building Code ACI 318-05 (ACI Committee 318 2005) use an effective moment of inertia, Ie, that was originally proposed by Branson (1965, 1977) to calculate beam deflection. Bransons approach was quite innovative at the time; his expression for Ie (eq. [1]) represents a gradual transition from the uncracked flexural stiffness, EcIg, to a fully cracked stiffness, EcIcr, as loading on the member increases beyond the cracking point (see Fig. 1a). [1] M I e = I cr + ( I g - I cr ) cr M a M = cr M a
3

3 M I g + 1 - cr M a

I cr I g the cracking moment. The uncracked moment of inertia is typically used when the applied load is less than the cracking moment; however, these members may still crack over time as additional stresses develop from shrinkage and temperature effects. For this reason, the Australian standard AS3600 (AS 1994) imposed an upper limit of 0.6Ig for steel-reinforced beams and slabs with a reinforcing ratio less than 0.5% (Gilbert 2001). This paper provides an explanation of why the Branson equation does not always work well in predicting deflection, and presents a rational approach to develop an alternative expression for the effective moment of inertia that works equally well for both steel- and FRP-reinforced concrete at all reinforcing ratios. The focus is on flexure members reinforced with steel, although some consideration is given to FRP-reinforced members. Revisions to the section on control of deflections in CSA A23.3-04 (clause 9.8) (CSA 2004) are proposed and include changes to expressions for the effective moment of inertia Ie and for the averaged effective moment of inertia Ie,avg of continuous members. Direct calculation of deflection is also considered for continuous members in lieu of using an averaged moment of inertia. Only immediate (short-term) deflection is considered in this study.

In eq. [1], Ig represents the gross (uncracked) moment of inertia and Icr is the cracked transformed moment of inertia. Deflection is calculated for the applied service load moment Ma at the critical section and the cracking moment is given by Mcr. The cubic term in this equation, (M cr / M a )3, accounts for both tension stiffening and the variation of member stiffness along the beam length. Bransons equation is an empirically derived relationship that was calibrated for steel-reinforced concrete beams with a reinforcing ratio between 1% and 2%. This corresponds to beams with an uncracked to cracked stiffness (EcIg/EcIcr = I g / I cr ) ratio between 2 and 3 (Bischoff 2005) and the equation has appeared to work well in the past because most beams fall within this narrow range of limits. However, Bransons approach consistently underestimates short-term deflection of lightly reinforced members (Scanlon et al. 2001; Gilbert 2006) and members reinforced with low-modulus, fibre-reinforced-polymer (FRP) bars (ACI Committee 440 2006). Problems also arise for slender tilt-up wall panels with a central layer of reinforcement (Bischoff 2006). In all cases, this occurs because member stiffness is overestimated with Bransons equation when flexure members have an I g / I cr ratio greater than 3. As a result, eq. [1] begins to underpredict deflection when steel reinforcing ratios are less than approximately 1% for beams, corresponding to an I g / I cr ratio greater than 3 (Bischoff 2005). For very low reinforcing ratios, the service load can be near or below

Modeling tension stiffening in beams


The overall stiffness and corresponding deformation of a reinforced concrete member is affected by tension stiffening
2007 NRC Canada

994

Can. J. Civ. Eng. Vol. 34, 2007

that arises from tension carried by the concrete between cracks. The concrete contribution in beams is modeled with a tension stiffening factor c ts ( = Df / Dfmax ) that decreases with increasing load once the member has cracked (Fig. 1b). The maximum tension stiffening curvature at first cracking is given by Dfmax = M cr (1 - I cr / I g ) / E c I cr . The change in curvature relative to the cracked member response is called the tension stiffening curvature Df ( = c ts Dfmax ) and is used to define a rational expression for the effective moment of inertia Ie. Figures 1a and 1b help show the derivation for Ie, giving [2] Ie = Ma Ma = E c fa E c ( M a / E c I cr - Df) = I cr Ig 1 - c tsh( M cr / M a )

where h = 1 - I cr / I g and fa is the curvature at Ma. Approximating the tension stiffening factor cts with M cr / M a then gives [3] Ie = I cr 1 - h( M cr / M a ) 2 Ig

This approximation assumes that the tension stiffening curvature varies inversely with bar stress at the crack locations. That is, c ts f b,cr / f b , where f b,cr is the stress in the bar at first cracking and fb is the stress in the bar (at a crack) at Ma (Nemirovsky 1949; Rao and Subrahmanyam 1973; Rostasy et al. 1976). The same tension stiffening factor has been adopted by the Comit Euro-International du Bton (CEB) model codes (CEB-FIP 1978, 1993) and CEB-based Eurocode 2 (CEN 2004). Equation [3] represents a rational and unified approach that works well for predicting deflection of both steel- and FRP-reinforced concrete beams without the need for correction factors (Bischoff 2005; Bischoff and Scanlon 2007). This equation is simple, easy to use, and is proposed as a suitable replacement for the Branson equation.

with FRP bars, as these types of beams typically have an I g / I cr ratio between 5 and 25 (Bischoff 2005). Tilt-up wall panels with a central layer of reinforcement also exhibit an unrealistically stiff computed response when using Bransons equation because the I g / I cr ratio for this type of wall typically ranges from 15 to 25 (Bischoff 2006; Bischoff and Scanlon 2007; Lawson 2007). Figure 3 plots the influence of steel reinforcing ratio on I g / I cr for rectangular beams and the corresponding level of service load relative to the cracking moment ( M a / M cr ) at which deflection is calculated. Service load moments ( M a = M r / a load ) are approximately equal to 60% of the nominal moment capacity (Mn) provided that the beam is underreinforced, where Mr is the factored moment resistance calculated using material resistance factors, Mn is the nominal moment capacity without material resistance factors, and a load is an average load factor equal to 1.375 for an assumed dead to live load ratio of 1:1. All calculations are based on the requirements of CSA A23.3-04 (CSA 2004) and are carried out for concrete with a specified compressive strength f c of 35 MPa and steel with a yield strength fy of 400 MPa. Figure 3 shows how the I g / I cr ratio increases significantly at low reinforcing ratios less than about 0.6%. Corresponding service loads in this reinforcing range are typically less than twice the cracking moment, where the influence of tension stiffening is most significant. The concrete contribution in Bransons equation is evaluated by calculating the corresponding tension stiffening factor c ts . This is determined by substituting Ie from eq. [1] into the expression for the tension stiffening factor [4] to get [5] c ts = M a / M cr 1 + [1 - ( M cr / M a ) 3 ] ( I cr / I g ) / ( M cr / M a ) 3 c ts = Df /Dfmax = ( M a / M cr ) (1 - I cr / I e ) (1 - I cr / I g )

A fresh perspective on Bransons equation


Tension stiffening is characterized by the change in curvature Df relative to the cracked member response and should not exceed the change in curvature at first cracking Dfmax if the tensile stress in the concrete is to remain less than the cracking strength fcr (Fig. 1b). Hence, the tension stiffening factor c ts = Df /Dfmax 1, which gives a member response bounded by an upper limit with constant tension stiffening (c ts = 1) and a lower limit with no tension stiffening (c ts = 0). Member responses using both Bransons expression for Ie (eq. [1]) and the unified approach (eq. [3]) are plotted in Fig. 2 for beams with I g / I cr = 2, 3, 5, 8, and 11 (corresponding to steel reinforcing ratios of r = 1.7%, 1%, 0.5%, 0.3%, and 0.2%, respectively). Results clearly show that tension stiffening is overestimated using Bransons approach for beams with I g / I cr greater than 3 (corresponding to r < 1% for steel-reinforced concrete beams), whereas there is little difference between the two approaches for steel-reinforced members at higher reinforcing ratios (greater than 1%). Tension stiffening is grossly overestimated for beams reinforced

Figure 4 compares Bransons tension stiffening factor with the factor c ts = ( M cr / M a ) incorporated into eq. [3], for beams with I g / I cr equal to 2, 3, 5, 8, 25, and 75. These plots show that the tension stiffening factor in Bransons equation tends towards M a / M cr as I g / I cr goes to infinity. In other words, Bransons expression for Ie approaches Ig as the ratio I g / I cr increases. Overestimating tension stiffening leads to a stiffer response and less deflection than expected, which is the problem with Bransons equation.

Comparison with rational proposal


Bransons equation gives a weighted average of the gross and cracked moments of inertia at any given load level. This is analogous to having an uncracked and cracked spring placed in parallel as shown in Fig. 5a, where the equivalent spring stiffness ke = kg + kcr approaches the uncracked stiffness kg of the stiffer spring as the difference between the two spring stiffness values increases. This occurs when the ratio k g / k cr becomes larger as the cracked stiffness kcr decreases for smaller values of Icr. Consequently, the member response is incorrectly pulled towards a weighted value of the un 2007 NRC Canada

Bischoff Fig. 2. Member response comparison for steel reinforced concrete beams (fc = 35 MPa). Fig. 4. Tension stiffening component in Bransons equation.

995

Fig. 3. Variation of Ig/Icr and Ma/Mcr with steel reinforcing ratio r.

cracked response E c I g ( M cr / M a ) 3 as the reinforcing ratio and (or) elastic modulus of the reinforcing bar decreases. This results in an unrealistically high amount of tension stiffening and smaller deflections than expected when Icr drops below one-third of Ig (Bischoff 2005). Tension stiffening is modeled properly by putting the uncracked and cracked springs in series (Fig. 5b) to represent the variation in stiffness of the uncracked and cracked parts of the beam along the member length. Using this approach, the member stiffness is then pulled more realistically towards the response with the lower stiffness (that is, the cracked member response). The equivalent or effective spring stiffness is now determined by taking a weighted average of the inverse stiffness values (1/EI), leading to a subtle change in Bransons original eq. [1] as given below. [6]
m 1 M cr 1 M cr = + 1 - Ie Ma Ig Ma

flection calculations. Equation [3] is thus proposed as a suitable replacement for Bransons equation because it represents a physical model that correctly accounts for tension stiffening in reinforced concrete beams and slabs. Differences between the two methods used to calculate Ie (eqs. [1] and [3]) are shown in Figs. 6a and 6b for different levels of applied load. A comparison of these two approaches is plotted in Fig. 7 and demonstrates that differences start to become evident for steel reinforcing ratios less than about 1% when the ratio of applied service load moment to cracking moment ( M a / M cr ) equals 2. The effective moment of inertia using the proposed approach (eq. [3]) decreases dramatically for low reinforcing ratios less than about 0.5%, especially when the applied moment Ma is close to the cracking load, whereas Bransons expression predicts values of Ie that are significantly closer to Ig. For example, a beam with r = 0.5% that is subjected to a moment just 10% greater than the cracking moment (Ma = 1.1Mcr) has an effective moment of inertia Ie equal to about 0.8Ig (using Bransons eq. [1]) compared with a more realistic value of 0.58Ig obtained with the proposed approach. Computed values of Ie are typically greater than 0.6Ig at low reinforcing ratios when using Bransons equation for moments less than 1.2Mcr. This helps to explain the upper limit of 0.6Ig imposed on Ie by the Australian standard AS3600 (SA 1994) when using Bransons equation to compute deflection of beams and slabs with a reinforcing ratio less than 0.5% (see Introduction). Note how the applied service load is close to the cracking moment at low reinforcing ratios (Fig. 3), where the influence of tension stiffening is most significant.

Evaluation and validation of deflection prediction at service loads


Beam deflections calculated with the proposed expression for Ie (eq. [3]) are compared with Bransons method (eq. [1]) in Fig. 8. Results are compared at the full service load defined in Fig. 3, clearly showing that both the Branson equation and proposed equation give comparable results (within a few percent of each other) for reinforcing ratios between 1% and 2%. Results also compare well with existing beam tests by Washa and Fluck (1952). However, differences between the two approaches become evident once steel reinforcing ratios drop below 1%, with Bransons expression
2007 NRC Canada

1 1 I cr I g

Calibration of this equation with eq. [1] for a beam with I g / I cr = 2.2 (Branson calibrated his original equation using beams with this ratio) gives a power of m = 2 (Bischoff 2005). This is equivalent to using a tension stiffening factor c ts = M cr / M a . Equation [6] with m = 2 is a rearranged form of the proposed eq. [3] and is also the same as the curvature-based equation used in Eurocode 2 (CEN 2004) for de-

996

Can. J. Civ. Eng. Vol. 34, 2007

Fig. 5. (a) Equivalent spring model for Branson expression (eq. [1]). (b) Equivalent spring model for proposed expression (eqs. [3] and [6] with m = 2).

Fig. 6. Variation of effective moment of inertia using (a) Bransons approach and (b) proposed approach.

Fig. 7. Comparison of Branson (eq. [1]) and proposed (eq. [3]) approach.

Fig. 8. Calculated beam deflection at service loads defined in Fig. 3.

underpredicting deflection by up to half of the calculated value using the proposed approach for a beam with r = 0.3%. Experimental support of this behavioural trend is provided by Gilbert (2006) for a number of steel-reinforced slabs having a reinforcing ratio between 0.18% and 0.84%. Figure 9a shows one example comparing the two approaches to Gilberts measured member response for a simply supported slab with a very low reinforcing ratio of 0.18%, which results in an Ig/Icr ratio of 13.3. Bransons expression clearly produces a member response that is much too stiff at this low reinforce-

ment ratio; however, the proposed model (eq. [3]) works quite well. Work by Bischoff and Scanlon (2007) and Lawson (2007) confirm that Bransons expression for Ie underestimates deflection of slender tilt-up wall panels, whereas eq. [3] gives reasonable results at service loads. Figure 9b shows that for glass FRP (GFRP) reinforced beams, Bischoffs expression for Ie compares well with the ACI 440
2007 NRC Canada

Bischoff

997 Fig. 9. (a) Slab response comparison for r = 0.18% and Ig/Icr = 13.3 (after Gilbert 2006). (b) Beam response comparison for GFRP-reinforced concrete.

(ACI Committee 440 2006) approach using a modified form of the Branson equation. As expected, Bransons original expression gives a response that is too stiff because of the high I g / I cr ratio and requires a correction factor to soften the computed member response. Practical examples are given in Table 1 for a rectangular concrete section having steel reinforcement ratios of 0.4%, 1.0%, and 2.0%. A GFRP-reinforced beam with a reinforcing ratio of 1.0% is also included for comparison and the GFRP bars are assumed to have an elastic modulus of 40 GPa. While analysis is carried out for concrete with a compressive strength of 35 MPa, the proposed expression for Ie is expected to be equally valid for higher strength concretes. Deflection is evaluated at full service load for the steel-reinforced concrete sections based on a service load moment Ma = Mr /1.375 (as explained in an earlier section) that is equal to 60% of the nominal moment capacity Mn, which gives a strain in the reinforcement of about 1200 microstrains. The service load for the GFRP section is based on a limiting bar strain of 3000 microstrains that allows for larger crack widths (Bischoff 2005), which gives an applied service load equal to about 25% of the nominal moment capacity for this section (compared with 60% for steel-reinforced concrete). Little difference is seen between Bransons approach (eq. [1]) and the proposed approach (eq. [3]) for the sections with steel reinforcing ratios of 1% and 2%. Note how the effective moment of inertia Ie is within a few percent of Icr for these two reinforcing ratios. This occurs because of the relatively high ratio of Ma/Mcr at full service load when the reinforcing ratio is greater than 1% (see p. 120 of Branson 1977 where Ie Icr when Ma/Mcr > 3). Bransons approach underestimates deflection by 34% (compared with Bischoffs approach) when the steel reinforcing ratio r drops to 0.4%. This occurs because the lower r value increases the I g / I cr ratio beyond 3 to a value of about 5, whereas the service load has dropped down to 1.3Mcr as indicated by Fig. 3. Recall that the influence of tension stiffening is more significant when service loads are less than twice the cracking load. The GFRP beam with a reinforcing ratio of 1.0% has an I g / I cr ratio close to 10 and Ma/Mcr ratio of 1.6. In this case, Bransons equation gives a deflection that is only half the expected value. The solution to dealing with this problem in the past has been to apply a correction factor to Bransons equation for computing deflection of FRP members (ACI Committee 440 2006). For steel-reinforced concrete, CSA A23.3-04 (CSA 2004) uses one-half the cracking moment to compute Ie for two-way slabs that are typically lightly reinforced. The reduced cracking moment compensates for the stiffened bulge evident with Bransons computed member response at low rs (Fig. 2) and also accounts for a decrease in member stiffness resulting from shrinkage restraint cracking.

0.15(Ie1 + Ie2) for two continuous ends and Ie,avg = 0.85Iem + 0.15Iec for one continuous end. Iem is the value of Ie at midspan, Ie1 and Ie2 are the values of Ie at the two end supports, and Iec is the value of Ie at the one continuous end. ACI 318-05 (ACI 2005) suggests using a simple averaged value of Ie. Both approximations were proposed by Branson (1977) and are based on work carried out by ACI 435 (ACI Committee 435 1973). The averaged value for Ie is used in conjunction with a generalized elastic equation for midspan deflection, such as that given by [7] D=K 5 M m L2 48 E c I e,avg

Continuous members
For continuous members, CSA A23.3-04 (CSA 2004) provides an averaged moment of inertia Ie,avg based on a weighted average of the effective moment of inertia taken at midspan and the end supports. This is given by Ie,avg = 0.7Iem +

where the beam deflection coefficient K = 1 + 01 ( M a1 + M a2 ) / . M m for a beam with uniformly distributed load. Mm, Ma1, and Ma2 are the moments at midspan and the two end supports, respectively. The value of K can also be expressed with the equation 1.2 0.2Mo /Mm, where Mo is the total static moment wL2/8. Note that algebraic (negative) values are used for the two end moments. This simplification uses a single averaged value of flexural rigidity to approximate deflection and is evaluated by comparing computed values of deflection to a beam having three segments of constant flexural rigidity corresponding to the
2007 NRC Canada

998 Table 1. Steel- and GFRP-reinforced beam deflection comparison using Ie. Steel, reinforcement ratio (%) 2.0 Ig /bd (10 ) Icr /bd3 (103) Ig /Icr Ma/Mcr Ie,Branson/bd3 (103) Ie,Bischoff /bd3 (103) DBranson/DBischoff
a

Can. J. Civ. Eng. Vol. 34, 2007

GFRP, reinforcement ratio (%) 0.4 114.3 21.8 5.2 1.3 63.6 41.7 0.66 1.0 114.3 12.0 9.6 1.6 39.1 19.0 0.49

1.0 114.3 45.7 2.5 3.1 48.0 48.8 1.01

3 a

114.3 75.2 1.5 5.5 75.5 76.1 1.02

d/h = 0.90.

two negative end span regions and one positive midspan region with each segment defined by the location of the inflection points (ACI Committee 435 1973). Both methods (single value and three-segment solution) are approximate since the variation of Ie within each segment and stiffening from the uncracked parts of the beam span are not considered. Comparisons are made for steel-reinforced concrete sections only; extension to FRP-reinforced concrete requires further evaluation. The approximate method using a single value of Ie,avg gives the same computed value of deflection as the three segment solution only when the rigidity at the end supports is equal to the rigidity of the midspan region; that is, when Ie1/Iem = Ie2/Iem = 1 (except when the beam is simply supported with end support moments equal to zero). Downwards deflection of the beam increases as rigidity of the end supports increases (relative to the midspan rigidity), which implies that the average moment of inertia Ie,avg should decrease and drop below the midspan value I em when 0.5 ( I e1 + I e2 ) / I em > 1. Comparison is made for a given set of support moments and redistribution of moments from changes in rigidity at the end supports is not considered. Hence, deflection is actually underestimated with either the CSA A23.3-04 (CSA 2004) or ACI 318 (ACI Committee 318 2005) approach whenever the stiffness at the end supports is greater than the stiffness at midspan, as the computed value of Ie,avg is greater than Iem using either of these two approaches. Using the midspan value Iem alone in this case (that is, substituting Iem for Ie,avg) would give a better approximation, although deflection will still be underestimated, but not to the same extent. Alternatively, deflection can be computed by the integration of curvature along the beam length. Assuming a parabolic distribution of curvature along the span of a continuous beam subjected to a uniformly distributed load, midspan deflection D is approximated by [8] L2 D = ( f1 + 10fm + f2 ) 96

ports are stiffer than the midspan region and underestimates deflection when the end supports are more flexible. The accuracy of this method is also tempered by the fact that the uncracked parts of the beam are neglected in this approximation. In lieu of using this type of direct approach for calculating deflection, the deflection value from eq. [8] can be equated with the generalized elastic equation for midspan deflection (eq. [7]) to give a rational expression for the averaged moment of inertia. [9] I e , avg = KI em Ig 1 + 01bM1 b I 1 + 01bM 2 b I 2 . .

where L is the span length, f1 and f 2 are the curvatures M a1 / E c I e1 and M a2 / E c I e2 at the two end supports, respectively, and f m is the curvature M m / E c I em at midspan. A detailed explanation of this approximation is given by Ghali (1993). When compared with the three-segment solution, this approach overestimates deflection when the end sup-

with b M1, 2 = M a1, 2 / M m and b I 1, 2 = I em / I e1, 2 . This gives Ie,avg = Iem for a simply supported beam and I e,avg = 0.8I em / (1 - 0. 2I em / I e1 ) for a typical interior span with a positive midspan moment equal to wL2/16 and having the same effective moment of inertia at the two end supports (Ie1 = Ie2). Therefore, the new expression for Ie,avg gives a more flexible member than using the midspan value of Iem whenever the end supports are stiffer than the midspan region. The CSA A23.2-04 approach (CSA 2004) is compared with eq. [9] in Fig. 10 for the interior span conditions just described. Comparison is also made with deflection calculated using a single moment of inertia equal to the midspan value and with the three-segment solution. Redistribution of moments after cracking is not considered. Results show that the existing code approach stiffens the member response when Ie is greater at the ends (Ie1,2 > Iem), as explained previously, whereas the new proposal using eq. [9] gives a more flexible response that results in greater deformation as expected (recall that midspan deflection should increase as the end supports become stiffer). Thus, when member stiffness is greater at the end supports than at midspan, deformation of continuous beams can be underpredicted by up to 50% or more using the present code approach for the average moment of inertia Ie,avg. The comparison in this case is being made with the three-segment approximation. Once again, deflection is more appropriately reflected by taking a weighted average of the inverse stiffness values (1/EcIe) as opposed to the averaged stiffness (EcIe) values stipulated in either A23.3-04 (CSA 2004) or ACI 318-05 (ACI Committee 318 2005). A word of caution, as the proposed approach
2007 NRC Canada

Bischoff Fig. 10. Comparison of average moment of inertia Ie,avg using code equation and proposed approaches.

999

underpredicts deflection significantly when the stiffness at the end supports drops below one-half of the midspan stiffness for this example (Fig. 10). Note that eq. [9] does not apply to fixed-end cantilevers.

Deflection example of continuous beam


To help evaluate the different approaches proposed for computing deflection of continuous members, calculated midspan deflection values under full dead plus live load are compared for both the exterior and interior span of a continuous T beam in a building frame. Only short-term deflections are computed. Details of this example are given in Fig. 11 and Table 2. Effective moments of inertia are based on moment envelopes using the CSA (CSA 2004) moment coefficients, whereas calculated deflections use the bending moment corresponding to the loading case for maximum moment at midspan (typically based on the moment coefficient at midspan). In this example, the average moment at the supports ranges from 0.75 to 1 times the midspan moment, and the effective moment of inertia at the end supports is between 1.5 to 2.25 times the value at midspan. Results are summarized in Table 3 and show that in both spans the use of Ie from midspan alone (Iem) gives the closest approximation to the three-segment solution. Deflection is underestimated by less than 2% even though the average support stiffness is more than twice the midspan stiffness in the end span (see Fig. 10). Deflection using the weighted average from eq. [9] overestimates midspan deflection by about 8%, whereas the CSA (CSA 2004) weighted average underestimates deflection by up to 30%. However, these comparisons need to be put into perspective by realizing that computed deflections using the three-segment solution are overestimated by about 15% in this example, as the variation in Ie for each segment and the uncracked parts of the beam are neglected. Despite this, the weighted average recommended by CSA A23.3-04 (CSA 2004) still underestimates deflection when considering the uncracked parts of the beam, whereas the other methods overestimate deflection. Note how Bransons equation overestimates Ie by about 25% at the supports for the interior span, where the service load moment is only 30% greater than the cracking value. Comparison of results from the example described above suggests that the best and most simple estimate of deflection

for a continuous beam in a frame is obtained by approximating the average moment of inertia with the midspan value Iem. This approach gives a reasonable approximation of midspan deflection provided that the support moments do not exceed twice the midspan moment (corresponding to a beam with fixed end conditions) and the average stiffness at the supports is within 0.5 to 3 times the midspan stiffness. Similar recommendations were made by ACI 435 (ACI Committee 435 1973). Within this range, use of the midspan value for Ie,avg underestimates deflection by less than 12% (when the support stiffness is greater than midspan stiffness) and overestimates deflection by 30% at most when the end supports are more flexible than the midspan stiffness. More accurate methods should be used outside this range. When assessing methods to compute deflection for design, it is important to realize that deflection is a highly variable process even under controlled laboratory conditions. Assumed values of the support and midspan moments in continuous members are also not exact and, to a large extent, depend on the loading pattern and flexural rigidities developed along the beam and in other members framing into the supports. Moments are often based on the gross (uncracked) flexural rigidity and do not account for redistribution of forces when the members crack. Long-term deflection only complicates matters further. Therefore, computed values of deflection are at best an approximation of what is likely to occur and as such should be used to provide only an indication of whether problems with deflection are likely to be encountered. With this in mind, almost any of the expressions discussed in this section would be adequate, but use of the effective moment of inertia at midspan is the simplest approach and appears to provide the most reasonable result for practical applications.

Proposed revisions to clause 9.8 in CSA A23.3-04


This part of the paper presents proposed revisions to clause 9.8 (entitled Control of Deflections) in the CSA concrete design standard CSA A23.3-04 (CSA 2004). Changes are made to the definitions of Ie and Ie,avg in clauses 9.8.2.3 and 9.8.2.4, respectively, and a more direct approach for calculating deflection is included. Additional notation and suggested commentary explaining reasons for the changes are also provided. Other parts of the clause remain unchanged.
9.8. Control of deflections 9.8.2.3. E and I Unless deflections are determined by a more comprehensive analysis, immediate deflection shall be computed using elastic deflection equations, a modulus of elasticity, Ec, for concrete as specified in Clause 8.6.2, and the effective moment of inertia as follows: I e = I cr / [1 - h( M cr / M a )2 ] I g where h = 1 - I cr / I g M cr = frI g / yt (9 - 2) ( 9 - 3) ( 9 - 1)

and fr is as given in Clause 8.6.4. Note: The moment Ma may be due to construction loads.
2007 NRC Canada

1000

Can. J. Civ. Eng. Vol. 34, 2007

Fig. 11. Continuous T-beam example used for deflection comparison: (a) beam elevation and (b) beam cross-section. All dimensions in millimetres.

Table 2. Details for deflection comparison of continuous members (wa = 46.7 kN/m; fc = 35 MPa; fy = 400 MPa). End spana Support 1 Ig /10 (mm ) Icr /107(mm4) Ig /Icr Mcr (kN@m) Moment envelope using coefficients Ma (kN@m) Ma/Mcr Ie/107 (Bischoff) Ie/107 (Branson) Bending moment Ma (kNm)
a b

Interior spanb Midspan 1157.7 332.5 3.48 99.9 wLn2/14 246.7 2.47 376.5 387.3 wLn2/14 246.7 Support 2 1157.7 344.6 3.36 243.9 wLn2/10 345.4 1.42 530.3 630.9 wLn2/22.4 154.2 Support 1 1157.7 329.4 3.51 243.9 wLn2/11 314.0 1.29 579.6 717.6 wLn2/16 215.9 Midspan 1157.7 332.5 3.48 99.9 wLn2/16 215.9 2.16 392.4 414.3 wLn2/16 215.9 Support 2 1157.7 329.4 3.51 243.9 wLn2/11 314.0 1.29 579.6 717.6 wLn2/16 215.9

1157.7 245.2 4.72 243.9 wLn2/16 215.9 0.89 1157.7 1157.7 wLn2/16 215.9

K = 0.85, (Ma1 + Ma2)/2Mm = 0.75, (Ie1 + Ie2)/2Iem = 2.24. K = 0.80, (Ma1 + Ma2)/2Mm = 1.0, (Ie1 + Ie2)/2Iem = 1.48.

Table 3. Results of deflection comparison for continuous members. End span Averaging approach using Ie proposed by Bischoff (eq. [3]) Integrationa Three segmentsb Equations [8] or [9] CSA A23.3-04 (CSA 2004) Iem (at midspan)
a b

Interior span Deflection (mm) 14.9 16.3 17.6 11.7 16.1 Ie,avg /107 (mm4) 363.1 448.6 392.4 Deflection (mm) 11.6 12.9 13.8 11.1 12.7

Ie,avg /107 (mm4) 345.1 516.7 376.5

Integration of moment curvature includes variation of Ie and uncracked parts of beam. Integration of three segments with constant Ie (Ie1, Iem, and Ie2).

Commentary: Bransons equation used for Ie in earlier editions of A23.3 for computing deflection underestimates deflection of steel-reinforced concrete beams with a reinforcing percentage less than 1%. Service load deflection of slender (tilt-up) walls is also underestimated. Equation (9-1) computes reasonable values of deflection for both

steel- and fibre-reinforced-polymer (FRP)- reinforced concrete members at all reinforcing ratios. 9.8.2.4. Moment of inertia for continuous spans For continuous prismatic members, the effective moment of inertia may be based on Ie values obtained from equa 2007 NRC Canada

Bischoff tion (9-1) for the critical positive and negative moment sections. An average value for the effective moment of inertia can be approximated with the midspan value as given below I e, avg = I e,m provided that (a) ( M a1 + M a2 )/ 2 Mm 2 and (b) 0. 5I em ( I e1 + I e2 )/ 2 3I em In lieu of using equation (9-4), deflection may be approximated directly as follows: L2 D = ( f1 + 10 fm + f2 ) 96 ( 9 - 5) ( 9 - 4)

1001

stiffness at the end supports is greater than the stiffness at midspan.

Acknowledgements
Support for this work was provided by the Natural Sciences and Engineering Research Council of Canada and the University of New Brunswick.

References
ACI Committee 435. 1973. Deflections of continuous concrete beams. ACI Journal, 7(12): 781787. ACI Committee 318. 2005. Building code requirements for structural concrete (ACI 318-05) and commentary (ACI 318R-05). American Concrete Institute, Farmington Hills, Mich. ACI Committee 440. 2006. Guide for the design and construction of concrete reinforced with FRP bars. ACI 440.1R-06, American Concrete Institute, Farmington Hills, Mich. Bischoff, P.H. 2005. Re-evaluation of deflection prediction for concrete beams reinforced with steel and fiber reinforced polymer bars. Journal of Structural Engineering, ASCE, 131(5): 752767. Bischoff, P.H. 2006. Closure to Re-evaluation of deflection prediction for concrete beams reinforced with steel and FRP bars by Peter H. Bischoff. Journal of Structural Engineering, ASCE, 132(8): 1330. Bischoff, P.H., and Scanlon, A. 2007. Effective moment of inertia for calculating deflections of concrete members containing steel reinforcement and FRP reinforcement. ACI Structural Journal, 104(1): 6875. Branson, D.E. 1965. Instantaneous and time-dependent deflections of simple and continuous reinforced concrete beams. Alabama Highway Department, Bureau of Public Roads, Ala. HPR Report No. 7, Part 1. Branson, D.E. 1977. Deformation of concrete structures. McGraw-Hill Inc., New York, N.Y. CEB-FIP. 1978. CEB-FIP model code for concrete structures (MC78). 3rd ed. Comit Euro-International du Bton (CEB), Paris, France. CEB-FIP. 1993. CEB-FIP model code (MC-90). Comit Euro-International du Bton (CEB), Thomas Telford Ltd., London, UK. CEN. 2004. Eurocode 2: design of concrete structures Part 1-1: general rules for buildings. European prestandard, DD ENV 1992-1-1: 2004, European Committee for Standardization (CEN), Brussels, Belgium. CSA. 2004. Design of concrete structures. Standard A23.3-04, Canadian Standards Association (CSA), Toronto, Ont. Ghali, A. 1993. Deflection of reinforced concrete members: a critical review. ACI Structural Journal, 90(4): 364373. Gilbert, R.I. 2001. Deflection calculation and control Australian code amendments and improvements. In Code provisions for deflection control in concrete structures. Edited by E.G. Nawy and A. Scanlon. American Concrete Institute, Farmington Hills, Mich. SP-203, pp. 4577. Gilbert, R.I. 2006. Discussion to Re-evaluation of deflection prediction for concrete beams reinforced with steel and FRP bars by Peter H. Bischoff. Journal of Structural Engineering, ASCE, 132(8): 13281330. Lawson, J. 2007. A call for unified design: deflection limits for tilt-up wall serviceability. Structural Engineer, 18(12): 3032. Nemirovsky, J.M. 1949. Rigidity of flexural loaded reinforced concrete members and opening of cracks. Central Research
2007 NRC Canada

for continuous members with a uniformly distributed load, where f1 = Ma1/EcIe1 f2 = Ma2/EcIe2 fm = Mm/EcIem Note: Ma1 and Ma2 are taken as algebraic values (ve value for negative moments). Commentary: Deflection is affected by the effective moment of inertia at the end supports and at midspan, and may be approximated by using an appropriate elastic deflection equation and assuming that the average effective moment of inertia is equal to the value at midspan. An alternative approach for calculating deflection directly is based on integration of an assumed parabolic distribution of curvature. Additional notation: h = cracked stiffness factor (1 Icr /Ig) Ma1,2 = support moments at end 1 (Ma1) and end 2 (Ma2)

Conclusions
The tension stiffening component in Bransons equation increases unrealistically for beams with an Ig/Icr ratio greater than 3 and because of this flexural stiffness is overpredicted for most FRP beams and for steel-reinforced concrete beams and slabs with a reinforcing ratio less than about 1%. Computed deflections are then underestimated. Similar problems are encountered for slender walls with a central layer of reinforcement. A rational alternative to this equation is proposed that correctly accounts for tension stiffening, giving an effective moment of inertia I e = I cr / [1 - h ( M cr / M a ) 2 ] with h =1 - I cr / I g . This equation predicts deflection equally well for both steel- and FRP-reinforced concrete members. A rational expression giving the averaged moment of inertia Ie,avg for continuous beams is developed based on the integration of curvature. The existing approach in CSA A23.3-04 (CSA 2004) averages stiffness at the critical sections and underestimates deflection when the flexural stiffness at the end supports is greater than the stiffness at midspan. Proposed revisions to deflection requirements contained in CSA A23.3-04 (CSA 2004) provide more realistic values of deflection for flexural members with low reinforcement ratios. Changes to the average effective moment of inertia in continuous members reflect the unconservative nature of the existing equation under certain conditions, namely when the

1002 Institute for Building Design (TsNIPS), Research Results for Conventional and Prestressed Reinforced Concrete Construction, Collected Papers, Stroiisdat, State Publication for Construction Literature, Moscow, Russia. [In Russian] Rao, P.S., and Subrahmanyam, B.V. 1973. Trisegmental moment-curvature relations for reinforced concrete members. ACI Journal, 70(5): 346351. Rostasy, F.S., Koch, R., and Leonhardt, F. 1976. Zur Mindestbewehrung fur Zwang von Aussenwanden aus Stahlleichtbeton (Minimum restraint reinforcement for exterior walls of lightweight concrete). Deutscher Ausschuss fr Stahlbeton, Vol. 267. pp. 183. [In German.] SA. 1994. Australian standard for concrete structures. AS3600, Standards Australia (SA), Sydney, Australia. Scanlon, A., and Choi, B.-S. 1999. Evaluation of ACI 318 minimum thickness requirements for one-way slabs. ACI Structural Journal, 96(4): 616621. Scanlon, A., Cagley Orsak, D.R., and Buettner, D.R. 2001. ACI code requirements for deflection control: a critical review. In Code provisions for deflection control in concrete structures. Edited by E.G. Nawy and A. Scanlon. American Concrete Institute, Farmington Hills, Mich. SP-203, pp. 114. Washa, G.W., and Fluck, P.G. 1952. Effect of compressive reinforcement on the plastic flow of reinforced concrete beams. Proceedings American Concrete Institute, 49(8): 89108. Ie1 Ie2 Iec Iem Ig K kcr ke kg L Ln m Ma Ma1 Ma2 Mm Mcr Mn Mo Mr w wa a load bM1,2 bI1,2 D Da Dcr Df Dfmax h r f f1 f2 fa fcr fm cts

Can. J. Civ. Eng. Vol. 34, 2007 effective moment of inertia at first support effective moment of inertia at second support effective moment of inertia at continuous end effective moment of inertia at midspan gross moment of inertia beam deflection coefficient cracked spring stiffness equivalent spring stiffness uncracked spring stiffness beam or slab span length clear span power coefficient (=2 in eq. [6]) applied service load moment support moment at first support support moment at second support midspan moment cracking moment nominal moment capacity total static moment (wL2/8) factored moment resistance uniformly distributed load uniformly distributed service load average load factor for dead plus live load ratio of end support moments (Ma1 and Ma2) to the midspan moment (Mm) ratio of the effective moment of inertia at midspan (Iem) to the effective moment of inertia at the supports (Ie1 and Ie2) deflection service load deflection deflection at first cracking tension stiffening curvature tension stiffening curvature at first cracking stiffness reduction coefficient (1 Icr /Ig) reinforcing ratio curvature curvature at first support using Ie1 curvature at second support using Ie2 curvature at Ma curvature at first cracking (Mcr /EcIg) curvature at midspan using Iem tension stiffening factor (= D f /D f max )

List of symbols
b d E Ec fb fb,cr fcr fc fy h I Icr Ie Ie,avg beam width effective depth of reinforcing steel elastic modulus elastic modulus of concrete bar stress (at a crack) bar stress at first cracking concrete cracking strength compressive strength of concrete yield stress of reinforcing steel beam height moment of inertia cracked transformed moment of inertia effective moment of inertia averaged effective moment of inertia

2007 NRC Canada

S-ar putea să vă placă și