Analysis of continuous reinforced concrete beams in serviceability limit state

The paper investigates to what extent the effect of actions and deformations of statically indeterminate reinforced concrete beams are modified if, instead of the elastic, uncracked model (which is simple but approximate), an elastic, cracked model is applied (which is more accurate) in the computation of the effect of actions at serviceability check. (Continuous beams were modelled with beams fixed at both ends.) The effects of different load levels and moment-redistribution at design were also analysed. It is assumed that the variable load reached its characteristic value prior to the time considered, thus, the location of cracked zones and the tension stiffening effect is determined for that load level. Instead of the interpolation between the two values of deformation, the tension stiffening effect was taken into account by an effective moment of inertia (I_eff), which is equivalent to the other method.


Introduction
In the past decades, the limitation of stresses, deformation and crack width of reinforced structures under serviceability conditions has gained considerable importance due to the following facts: • Buildings tend to be tali er, structural spans tend to be greater, and structures with hinged connections (e.g.precast concrete structures) are spreading.
• High-strength structural materials have been applied lately, while their elastic moduli have hardly increased (e.g. in case of concrete) or not increased at ali (e.g. in case of steel).
• Social expectations conceming the aesthetics and service of structures have increased; the limitation of deformation, vibration and crack width is based on several requirements.
• Structures used to be analysed applying the method of allowable stresses, that is, under service loads, assuming elastic, cracked state.However, according to the method of limit states appii ed recently, structural dimensions are determined for a different load level, assuming a different stress state.Thus, the latter method does not involve an indirect check of serviceability.
In case of statically determinate structures, the same method is employed at the computation of the effects of actions both at design situations and at the analysis of serviceability limit states: the conditions of equilibrium must be satisfied.However, in case of statically indeterminate structures, the flexural stiffness of the structure influences the effects of actions considerably.
The draft European Standard prEN 1992-1-1 :2003 proposes the following four model s for the computation of the effect of actions if a structure is designed for the ultimate limit state: • A linear elastic model applying the stiffness of the uncracked gross cross-section.(This model is obviously inaccurate, but its application is simple.) • The previous model, assurning a limited redistribution of the effects of actions subsequently.(The effects of actions may be modified, what may be favourable in certain respects.) • A non-linear elastic model, taking the effect of cracking into account.(This model is appii ed mainly for the check of structures, since the reinforcement area at cracked cross-sections is assumed to be known.) • A plastic approach, assuming very ductile structural elements, in which plastic hinges form at ultimate limit state.
There are two model s for the analysis of the effect of actions at serviceability check: • The linear elastic model described above, applying the stiffness of the uncracked gross cross-section.(This simple model is generally employed in engineering practice.) • A non-linear elastic model, which takes the effect of cracking into consideration.
In this paper, the non-linear elastic model was applied for the analysis of statically indeterminate reinforced concrete elements in serviceability limit state.Sections 2 to 5 introduce the assumptions of the analysis.The propo sed algorithm for the analysis is illustrated by Fig. 5 in Section 6. Section 7 presents the numerical analysis of a beam fixed at both ends, which is similar to one span of a continuous beam.

The Influence of Flexural Stiffness on the Effect of Actions
Cross-sections of a structural member are considered to be in state 2 when we apply the non-linear elastic model.Flexural stiffness is not constant in different sections of the element, what influences the effect of actions in case of statically indeterminate elements.However, the variation of stiffness has different effects on specific flexural elements: • Beams normally have quite short uncracked segments (i.e.cross-sections in state 1, Il), which do not cause great changes in the behaviour of the structural element, since the moment at these parts is small.In case the beam was designed applying the linear elastic analysis, the ratio of the maximum and minimum bending moment (and, correspondingly, the reinforcement ratio) is approximately 2. In this case, the effects of actions (i.e.bending moments) computed according to a homogeneous cross-section are assumed to be only slightly modified if the variation of flexural stiffness is taken into account.
• In case of beams designed for plastic distribution of moments or redistribution of moments (when the moment-diagram is shifted either upwards or downwards) (Fig. 6), values of the moment of inertia are different compared to those in the previous case.This results in a shift in the moment-diagram in service state.(See Section 7, Numericai Analysis, for data.)Unlike in the previous case, effects of actions at design and in service state are not proportional, what suggests that highly stressed locations should be checked to make sure that steel stress does not exceed 0.8f\'k.(fyk is the characteristic yield stress.) (Under serviceability conditions, steel stresses that would lead to inelastic deformation must be avoided, since that would involve excessive deformations and crack width (prEN 1992-1-1:2003,7.2(4),(5)).) • FIat slabs normally have both cracked and uncracked zones, and there is significant difference between reinforcement ratios of certain locations (the difference of reinforcement ratios of certain locations may be fivefold, corresponding to moments).Therefore, flat slabs show great variation in stiffness, what is supposed to lead to considerable changes in the effect of actions in service state, as compared to the effect of actions of a homogeneous, elastic slab.

The Effect of Cracking on Flexural Stiffness
For the computation of flexural stiffness values, which are necessary for the calculation of moments and deformations, cracked and uncracked zones (i.e. the crack pattem) have to be at first determined.During the service period of a building, the probability that the rare combination of actions (in case of variable actions of only one type, Fr = Ck + Qd occurs is 40 to 50 percent.Therefore it is advisable to assume that cracks induced by the rare combination of actions form prior to the time being considered (since serviceability requirements have to be fulfilled following crack formation, too) (Fig. 1).
According to provision 7.1 (2) of prEN 1992-1-1 :2003, a concrete section has to be assumed cracked if the concrete stress exceeds the mean value of the tensile strength of concrete (űc > JCIIIl) under the rare combination of actions.However, for serviceability (irreversible) limit state, Annex A of ENV 1991-1 specifies a target reliability index (B) of 1.5 for the design working life, i.e. the failure probability (P l) is 0.7 .10-1.It is advisable to apply this reliability level also for reversible limit states.(The difference between reversible and irreversible limit states lies in the combination of actions to be taken into account (Deák.2000)) [1].Applying the characteristic values of material properties (Ecb JClk) is a good method of obtaining this reliability level (DEÁK et al. 1998 [3]).
Following the determination of the sectional properties (flexural stiffness) of the structural element under the rare combination of actions, deformations shall be computed usually for the quasi-permanent (F qp = Ck + 'jf2Qk) or frequent combination of actions (Ff = C k + 'fil Qú Values of moment of inertia obtained in the previous step can be considered constant, since the tension stiffening effect (i.e. the stiffening effect of the cracked concrete in tension exerted on the reinforcement) determined for the highest expected load in service state will not change if the load decreases (Fig. 1).(lt is disregarded that the concrete may prevent the steel reinforcement from shortening.) Provision 7.4.where a is the deformation (curvature, rotation, deflection etc.), aj is the deformation assuming state l, a2 is the deformation assuming state 2, and where is the factor of interpolation between lj and 12 due to the tension stiffening effect, fi is a factor taking into account the type and duration of load, Mer is the cracking moment.and M is the actual bending moment.
According to Expression (7.20) in section 7.4.3(5), where i fJ is the final creep coefficient of concrete.
The final creep coefficient of concrete, i fJ (00, to), is specified in Fig. 3 in prEN 1992-1-1 :2003.The notional size of the cross-section, the age at loading and atmospheric conditions are taken into consideration here.According to prEN 1992-1-1:2003, 2.3.2.2(3), the effects of creep should be evaluated under the quasi-permanent combination of actions.It implies that 1f12 Q should be considered as a long-term action.
Loads induce the greater concrete creep in a structure the earlier they are applied after concrete casting and the longer they act.If no data are provided, we may assume that the dead load (G) starts to act when the concrete strength reaches its design value and it acts permanently afterwards.The variable part of Assuming that K = M / Ee/eff, this procedure may be simpli-the quasi-permanent combination of actions (1fI2Q) also starts fied by introducing an effective moment of inertia derived from to act at an early age, and it acts in approx.50 percent of the the above formula: working life.Therefore, the total value of these two actions has where is the curvature, is the modulus of elasticity of concrete, and are moments of inertia of the transformed section in state l and state 2, respectively.
been taken into account at the computation of long-term defor-(3) mations due to creep.Since the variable part of the frequent and that of the rare combinations of actions are assumed to start to act later and for a shorter period of time, they were not considered as actions that induce creep.A k factor is introduced to obtain the creep coefficient of concrete for the quasi-permanent combination of actions: where Eck is the characteristic value of the modulus of elasticity of concrete, and where Gk is the characteristic value of the permanent action (self weight), Qk is the characteristic value of the variable action, and 1f12 is a multiplying factor for variable actions at quasi-permanent combination of actions.(6) 4=----------------.: ..... }( (curvature)

Effective Modulus of Elasticity of Concrete
At the computation of deformations, prEN 1992-1-1 :2003 allows to take into account the concrete creep by replacing the When determining the flexural stiffness of a concrete section after taking creep into account, it is not only the modulus of elasticity of concrete that changes but also the moment of inertia of the transformed concrete section.This change is due to the fact that a factor (a) is applied at the computation of the moment of inertia of the transformed section: (7) modulus of elasticity of concrete, Eem, by an effective modulus where Es is the modulus of elasticity of reinforcing steel, not of elasticity, Ee,eff.
taking concrete creep into consideration.However, a will change if creep is considered: (J. = Es/Ee,eff. ( Due to this fact, the moment of inertia of the transformed section will increase.The degree of increase of the moment of inertia is different in case of different reinforcement ratios.Straindiagrams of concrete sections in case of low and high reinforcement ratios can be seen in Fig. 2. Since the stress-diagram is proportional to the strain-diagram, point E: = O represents the neutral axis.The figure illustrates that, in case of different reinforcement ratios, the same degree of creep increases the height of the compression zone (and thus, the moment of inertia) to a different extent.(There is a slight ch ange in the steel stress, and correspondingly in steel strain, due to creep, but this is ignored in the figure .)that the variable action reaches its characteristic value during the first year of service time is only 2 percent.Therefore, it is probable that the main part of creep has already taken place by the time the total load reaches the level of the rare combination of actions.This loading history is illustrated by Fig. 4.  ---------------- The increase of the moment of inertia is greater in case of Taking into account ali the above considerations and assumsmaller reinforcement ratios, therefore, the flexural stiffness de-ing the loading history illustrated by Fig. 1, the algorithm accreases to a lesser extent.It is indicated in Fig. 3.According to cording to (Fig. 5) is propo sed for the computation of deforma-Colonnetti's first theorem, the deformation of a statically inde-tion of flexural RC elements.terminate structure loaded by constant load will increase but the effect of actions will rem ain unchanged, if the creep law of each structural element is the same.Since the reinforcement ratio is not constant along the eleme nt, the creep law is different in each cross-section, what leads to a ch ange in the effect of actions.Fig. 1 illustrates the loading history in which the total load reaches the load level of the rare combination of actions at an early age, what is followed by creep.However, concrete creep x starts as early as loading (including self-weight), and the effects The effect of the factors described in the previous sections on the bending moments and deformations of a flexural RC element was studied in a numerical analysis.The analysis was carried out applying a computer program that divided the beam into finite segments along the span, which have different sectional properties (reinforcement ratio, moment of inertia, etc.) The exarnined structural element was a beam of 8 m, fixed at both ends; its cross-section was 30/50 cm (Fig. 6).The permanent action (Le.self-weight of the structure, g) was taken 20 kN/m.In order to have three different tension reinforcement ratios, the variable actions (q) were chosen to be 9, 18 and 24 kN/m.(The smallest live load is that of a hotel or hospital room, the next one is that of a school hall, and the greatest live load belongs to a department store.)The multiplying factor of the variable load: 1f12 = 0.5.The total load at the quasipermanent combination of actions (Pqp = g + 0.5 q) was 24.5, 29 and 32 kN/m.S 500 steel and C 25/30 concrete was used (Eek = 2780 kN/cm 2 ).The final creep coefficient of concrete, ep, was taken 1.6.Values of the reinforcement ratio varied between 3%c and 17%c.
According to prEN 1992-1-1:2003,5.5(4), the moments calculated using a linear, elastic analysis may be redistributed, provided that the ratio of the redistributed moment to the moment before redistribution is less than or equal to 0.7 (in case of high ductility steel).Therefore, at the design of the reinforcement, four cases were considered: l) a bending mo ment diagram assuming homogeneous cross-section (l = const.),without redistribution, 2) and 3) moment diagrams shifted by 15 percent upwards and downwards at the supports, respectively, and 4) a moment diagram shifted downwards by 30 percent at the supports (Fig. 6).Applying such a redistribution, the stress limitations specified in prEN 1992-1-1 :2003 may be assumed to be satisfied if the stress in the reinforcement is lower than 0.8•f y k under the rare combination of actions, Le. the steel reinforcement will have no inelastic deformations under service loads (prEN 1992-1-1:2003,7.2.(5)).
The numerical analysis was carried out for both types of loading history illustrated by Figs. 1 and 4 above.The same results were obtained both for bending moments and deflections; the different assumptions for loading history led to no practical changes in the final results.
Due to the effect of creep, the effective modulus of elasticity of concrete, Ee.eff, decreased to 38 percent of Eeb while the increase of the moment of inertia of cracked sections, Ieff(Ee.eff)/ IeffCEed, was 180 to 210 percent.The value of the flexural stiffness taking creep into account, Ee. eff Ieff(Ee.eff), is approx.70 to 80 percent of the initial value, EekIeff(Eed.At the analysis of moments at service state, it was found that the distribution of moments was only slightly modified as compared to the diagram obtained assuming constant stiffness (-M max = p .L 2 /12; +M max = P .L 2 /24, where p is the total load, p = g + q).However, the increased stiffness re- sulting from raising the moment diagram at design led to an approx.10 percent increase of the moments at the support, and consequently to a decrease of the midspan moment.In case of lowering the moment diagram by 15 percent at the support, the increased stiffness of the rnidspan resulted in an approx.10 percent increase of the midspan moment, and a decrease of the negative moments.The ratio of moments computed according to the proposed method (see the algorithm in Fig. 5) and according to the traditional method (assuming constant stiffness) is indicated in Fig. 7.
Values of deflection obtained from the proposed method and those deterrnined according to the traditional method were compared in the same way as moments.According to the traditional method, the deflection was computed by the formula 1 Pqp .L 4 amidspan = 384 .E . I e.eff (9) where amidspan is the deflection at midspan, Pqp is total load at the quasi-permanent combination of actions, L is the span of the element.
If the work method is applied for the computation of deflection, a moment diagram corresponding to a unit force is employed (Fig. 6c), and the two moment diagrams are graphically integrated.Since the region of negative moments of the first diagram is integrated with small values of the second diagram, it is the midspan regions of the diagrams that account for a considerable 'part of the result, Le. the deflection.Therefore, in Formula 9 above, hmidspanCEe.eff)and Ic were substituted for I, and the deflection was obtained by interpolating between the results by (.When the moment-diagram was shifted upwards at design, the proposed moment-diagram also moved upwards, what led to a decrease in the midspan moment in service state.This explains why the deflection obtained from the proposed method is smaller than the deflection computed traditionally.Accordingly, shifting the moment-diagram downwards at design led to 2007381 a down ward move of the diagram in service state, i.e. to an increase of the midspan moment as compared to the moments assuming constant stiffness.This resulted in greater deflection than the one ca1culated by the other method.These results are illustrated by Fig. 8 below.While shifting the moment diagram downwards by 15 percent at design led to a deflection (aprop) that is up to 20 percent greater than atrad, shifting the diagram downwards by 30 percent resulted in an increase of the deflection (aprop) of up to 40 percent as compared to atrad.The different values of the live load and the different cases at design res ul ted in several different values of reinforcement area for the same concrete cross section.It provided data for an analysis on what values of the moment of inertia belong to different values of the reinforcement ratio.This may be useful at the analysis of flat slabs, where there are greater differences in the reinforcement ratio than in case of beams.Increasing the reinforcement ratio by 100 percent leads to an increase of the moment of inertia of approx.60 percent.However, as Fig. 9 indicates, the function is not perfectly linear.

Conclusions
The paper discussed how the effects of cracked, elastic state of cross-sections of flexural elements and creep influence the effects of actions and deformations.A method was proposed to take the se effects into consideration.Results of a numerical analysis demonstrated that plastic redistribution of moments at design led to a different moment distribution from the one assuming constant stiffness.In certain cases, this resulted in up to 20 percent greater deformations than those obtained by a traditional method (see Fig. 8).
3(3) of prEN 1992-1-1:2003 takes the tension stiffening effect into account by applying a factor (n to interpolate between va1ues of deformation computed according to Katalin Rákóczy / György Deák cross-sections in state l and state 2:

Fig. 1 .
Fig. 1.Loading history and change of flexura] stiffness of a section of a Re element, in case the load reaches the rare combination of actions prior to creep

Fig. 2 .
Fig. 2. Strain-diagrams of cross-sections with Jow and high reinforcement ratio.p, illustrating the effect of creep on strain
Fig, 4. Loading history and change of flexura!stiffness of a section of a Re eJement. in ca~e creep precedes eracking due to the rare combination of actions

Fig. 3 .
Fig. 3.The modification of flexura!stiffness due to ereep as a function of the reinforcement ratio

Fig. 6 .
Fig. 6. a) The examined structural element; b) Different moment-diagrams considered at design: c) Moment diagram corresponding to a unit force.applied in work method.