You are on page 1of 15

THE STRUCTURAL DESIGN OF TALL AND SPECIAL BUILDINGS Struct. Design Tall Spec. Build.

(2011) Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/tal.699

A comparative study on different analysis approaches for estimating the axial loads on columns and structural walls at tall buildings
Ozgur Kurc*, and Andac Lulec
Structural and Earthquake Engineering Laboratory, Department of Civil Engineering, Middle East Technical University, Ankara, Turkey

SUMMARY Estimating axial loads on columns and structural walls at tall buildings is a complicated task because time dependent deformations of concrete and the way the building is constructed affect the way the gravity loads are carried by them. The accurate computation of axial loads is crucial for determining the size and strengths of columns and structural walls. This study investigates several analysis approaches commonly used during the design of such buildings. Construction sequences, timedependent deformations and longitudinal reinforcement were considered, and their inuences on results were discussed. A simplied model of an actual 37story building was used as a case study. The results indicated that the column and wall axial load might vary up to 45% depending on the type of analysis and effects that were considered. Copyright 2011 John Wiley & Sons, Ltd.
Received 23 August 2010; Revised 27 December 2010; Accepted 23 February 2011
KEY WORDS:

tall buildings; axial loads; construction sequence analysis; creep; shrinkage; timedependent

deformations

1. INTRODUCTION The number of reinforced concrete tall buildings keeps increasing day by day. Especially in large cities of developing countries, there are and will be numerous tall building constructions, mostly reinforced concrete. There are various types of structural systems for such buildings, but at the zones of low seismicity, it is very common to see buildings with at plates and coupled or uncoupled structural wall systems due to speed and ease of construction. If a building is located at a high seismic zone, the structural wall system is usually reinforced with a moment frame system to improve the ductility of the whole structure. Columns and structural walls are one of the most important members of such building systems. Columns, especially the ones at the lower stories, are subjected to very high axial loads, that is, loads very close to their axial strengths. Likewise, structural walls resist considerably high moments and shear forces, and the amount of axial load on the wall inuences the walls exural and shear strength considerably. Thus, accurate estimation of axial loads on such members throughout the buildings service life is crucial. Any underestimation of design loads and member strengths may lead to signicant damages that are very difcult to repair. Accurate estimation of axial loads on columns and walls in a tall building is actually a very complicated task (Taranath, 2005). The way the building is constructed, timedependent effects such as creep, shrinkage and the effect of temperature differences inuence the differential shortening of columns and walls, thus changing the amount of load carried by a particular column or wall (Donping et al., 2009). Kwak and Kim (2006a) presented a detailed study on how construction sequences inuence
* Correspondence to: Ozgur Kurc, Structural and Earthquake Engineering Laboratory, Department of Civil Engineering, Middle East Technical University, Ankara, Turkey Email: kurc@metu.edu.tr Copyright 2011 John Wiley & Sons, Ltd.

O. KURC AND A. LULEC

the member forces of reinforced concrete frames. The shoring system also changes the way the loads are distributed among structural components (Kwak and Kim, 2006b). Kim and Abdelrazaq (2008) investigated the effect of ve different construction schemes on the shortterm and longterm behavior of the at plate systems and presented how the slab deections are inuenced by different shoring systems. Studies on actual tall buildings showed that creep and shrinkage deformations might shorten a column up to 4.6 cm and 6 cm, respectively, at an 80story building (Smith and Coull, 1991). Pfeifer et al. (1971) measured about 1in differential shortening between the column and core wall of a 70story building. They also presented an analytical procedure for the calculation of column and wall deformations that satisfactorily agree with the measured deformations. Pan et al. (1993) measured about 3cm shortening at one of the columns of a 32story building in 2400 days, which induced signicant moments and shears in connecting slabs and beams. Because of this, there are various studies that proposed methods to compensate such differential shortenings to eliminate damage to both structural and nonstructural components (Park, 2003; Taranath, 2005). To estimate the axial loads on vertical members, various methods are usually used by structural engineers such as tributary area, linear static analysis of full 3D model and nite element model of a single story. Tributary area method is generally used during the presizing of columns (Smith and Coull, 1991). If the columns are not subjected to signicant moments due to gravity and lateral loads, which is usually the case in at plate systems, the results of the tributary area method can be used for column design. The more common way to obtain column and wall axial loads is linear static analysis because vertical member design is usually performed by combining different load cases with various load factors and checking the column and wall strengths with interaction diagram software that is embedded into the design modules of the same analysis software (ETABS, 2010; SAP 2000, 2010). For special cases, the 3D nite element models of each different story are created, and the results of such models are combined with the results of full 3D model to obtain the column axial loads at any story. In this case, load distributions among columns are computed as if there are no differential shortenings among them. Unfortunately, neither of these analysis methods, which are frequently used for designing columns and walls, consider the load and deformation history of vertical members and timedependent deformations of concrete. The accurate estimation of column and wall axial loads throughout the lifetime of a building depends on how precisely the relative axial deformations of those members are computed. There is, however, very limited number of studies that focused on the axial load distribution among vertical members of tall buildings. Moreover, neither of these studies examined the effect of longitudinal bars on the axial loads of such vertical members. Thus, the main objective of this paper is to present an in depth examination and comparison of different analysis approaches that are used by structural engineers for computing the axial loads on columns and walls. In addition to this, a simple and practical modeling approach that could be used with any structural analysis software is proposed for considering the effect of reinforcing bars in columns and walls. The analysis approaches considered in the study are (a) tributary area method, (b) the 3D nite element model after the elastic analysis, (c) the 3D model after the construction sequence analysis (CSA) with and without timedependent effects and (d) the nite element oor models of the typical levels. As an example case, a 37story high reinforced concrete building that would be built in Ankara, Turkey, was considered. In this building, the lateral resistance was provided by the structural walls. Columns were designed for carrying the gravity loads only. The accuracy of each analysis approach and their effect on the safety of the design were also addressed in this paper.

2. ANALYSIS APPROACHES The design of a reinforced concrete building is performed in two steps: design against gravity and lateral loadings and the way the building is modeled and analyzed differ according to the loading in consideration. In other words, various structural models are created for obtaining the member design forces that are on the safe side rather than using a single model of the whole building for designing every member. The main reason behind this approach is reinforced concrete being a nonlinear and composite material whose behavior is dictated by the reinforcement detailing and amount of cracking.
Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

The design against gravity loads is usually performed by using simple 2D models where only beams, slabs and top and bottom columns are modeled. ACI 31808 (2008) proposes two different methods, direct design and equivalent frame methods, for the analysis and design of such frames. The column axial loads due to gravity loads can be estimated from the slab tributary areas. Especially in buildings with at plates and structural walls where columns are not the part of the lateral load carrying system, tributary area approach is considered to be easy to use and accurate (Smith and Coull, 1991). The design against lateral loads is usually performed by linear static analysis of 3D nite element model of the whole structure. The lateral loadresisting members such as structural walls, frame beams and columns are designed according to the results of the 3D model. 2.1. Construction sequence analysis Construction sequence analysis is a powerful tool for examining the construction steps such as constructing parts of the structure, applying loads at different times or modifying the boundary conditions (Smith and Coull, 1991; Kwak and Kim, 2006b). CSA is actually a series of nonlinear static analysis where new construction activities are applied to an already deformed and stressed structure. Let Ki and Fi represent the stiffness matrix and force vector of the initial structure composed of structural members that will be constructed and loaded at the same time. The rst step of the CSA is to solve the nonlinear equilibrium equations (Equation (1)) of the initial structure for obtaining the deformed state and corresponding internal forces of the members. Ki Ui Ui Fi 0 (1)

As the analysis of the rst step is nalized, the next part of the structure and additional loads are added to the structural model. For the following step of the analysis, the nonlinear equilibrium equations become Ki Uj Ka Uj Kr Uj Uj Fi Fa Fr 0 (2)

In the above equation, Ki represents the stiffness matrix of the initial structure in the deformed and stressed state, Ka is the stiffness matrix of the newly added structure and Kr is the stiffness matrix of the removed or demolished members. Likewise, Fa and Fr represent the newly added and removed loads, respectively. It is important to note that dening the geometry of the newly added structure is not a straight forward process since the nodes of the new structure must be dened according to the deformed initial structure. The rst approach is to dene the coordinates of the newly added nodes according to the undeformed geometry (Figure 1(a)). This approach is very suitable for modeling tall buildings since during the construction of such buildings, the newly constructed story is rst leveled to its predened elevation and the following story is constructed afterward. The other approach is to dene the

Figure 1. Construction sequence analysismodeling approaches.


Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

O. KURC AND A. LULEC

coordinates of the newly added nodes to a location relative to the deformed adjacent nodes (Figure 1 (b)). Such an approach is frequently used in modeling the bridge construction, especially at segmental deck bridges (Kwak et al., 2000; Malm and Sundquist, 2010). For each predened construction step, the system equation (Equation (2)) is updated and solved. Timedependent effects such as elastic modulus variation, creep, shrinkage and steel relaxation can be incorporated in the analysis by updating the Ki(Uj) and Fa matrices at every analysis step. 2.2. Shrinkage and creep effects The term shrinkage refers to the shortening of concrete during hardening and drying under constant temperature due to loss of a layer of water from the surface of the gel particles (Wight and MacGregor, 2008). In other words, shrinkage is caused by moisture evaporation from the surface and mostly inuenced by the volumetosurface ratio of a concrete member. The shortening caused by creep signicantly depends on the loading history and magnitude of the sustained stress. Creep is very effective at an early age of concrete, and its effect decreases with the age of concrete. To determine the creep and shrinkage effects for normal strength concrete, several models exist. Among those, CEBFIP Model Code (CEBFIP, 1993) was simple and frequently used in practical design application. In this model, the shrinkage strains (cs) are calculated by the following equation: cs t; ts cs0 s t; ts (3)

where cs0 is the notional shrinkage coefcient and s denes the development of shrinkage with time. The notional shrinkage coefcient is calculated by using Equations (4) and (5): cs0 s fcm RH (4)

s fcm 160 10sc 9fcm =fcm0 106

(5)

According to Equation (4), notional shrinkage coefcient, cs0, depends on s( fcm) and relative humidity of the ambient environment ( RH). s( fcm) describes the effect of concrete strength on shrinkage strains and is calculated by Equation (5). In Equation (5), fcm is the mean compressive strength of concrete at the age of 28 days in MPa, and fcm0 is dened as 10 MPa. The value of sc depends on the type of cement and can be taken as 5 for normal hardening cements. The variation of shrinkage strains with respect to time is dened in CEBFIP Model Code as follows: " s tts #0:5 (6)

tts =t1 350h=h0 2 tts =t1

h2

Ac u

(7)

In Equation (6), t is the age of concrete (days) at the moment considered, ts is the age of concrete (days) at the beginning of the analysis. t1 and h0 can be taken as 1 day and 100 mm, respectively. h is
Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

the notional thickness (mm) of the member and calculated by Equation (7). In Equation (7), Ac is the area of concrete crosssection and u is the perimeter of the member, which is in contact with the atmosphere. According to CEBFIP Model Code, creep strains are calculated by Equations (8) and (9): c t 0 t; t 0 Eci

cc t; t 0

(8)

t; t0 0 c tt0

(9)

In Equation (8), co(t0) is the compressive stress on the concrete under sustained loading, Eci is the modulus of elasticity for 28dayold concrete and (t,t0) is the creep coefcient calculated by Equation (9). In Equation (9), 0 is the notional creep coefcient calculated by Equation (10) and c(t t0) represents variation of creep strains with respect to time and is calculated by Equation (11). 0 RH fcm t0 (10)

tt0 =t1 c tt0 H tt0 =t1

!0:3 (11)

The parameters of Equations (10) and (11) are calculated by Equations (12)(15). In Equation (12), reference relative humidity of the ambient environment, RH0 , can be taken as 100%. 1 RH = RH0 0:46h=h0 1=3

RH 1

(12)

f cm

5:3 f cm =f cm0 0:5

(13)

t 0

1 0:1 t0 =t1 0:2

(14)

H 150 1 1:2 RH RH 0

18 )

h 250 h0

(15)

The creep and shrinkage strains are applied as equivalent loads to the frame or shell elements during static analysis. In case of CSA, the creep and shrinkage strains are calculated and
Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

O. KURC AND A. LULEC

converted into equivalent loads after each construction step. The loads that are computed with Equation (16) are applied to the structure as an additional loading. Fa BT EfCS CC gdV

(16)

In the above equation, B is the strain displacement and E is the constitutive relationship matrices. 2.3. Effect of reinforcement on timedependent axial deformations The CEBFIP Model Code denes the timedependent strains on concrete members without considering the amount of reinforcement. In an actual structure, however, when a reinforced concrete column is subjected to sustained compressive stresses, there is a gradual transfer of stresses from concrete to the reinforcements due to the shortening caused by creep and shrinkage effects. Because of this, the effects of longitudinal reinforcement ratio on timedependent deformation of such members should be considered in a detailed analysis. Smith and Coull (1991) presented an analytical method for estimating the shortening caused by creep and shrinkage. The method was mostly based on the studies of Fintel and Khan (1969), but they also considered the inuence of column and wall longitudinal reinforcement on axial shortenings. As a rst step, by using equilibrium and compatibility conditions between steel and concrete, they calculated the changes in both steel and concrete stresses due to ultimate creep and shrinkage. Then, they modied the creep and shrinkage strains by considering additional steel strains that were calculated from steel stresses. Pfeifer et al. (1971) estimated column and wall deformations with a single equation that considered shrinkage, creep and amount of reinforcement. This and similar analytical methods may successfully predict the amount of shortening of a single column but cannot estimate the load transfer amount of vertical members due to relative axial shortening. Moreover, the loading history including the construction sequences and the effect of story leveling cannot be considered in such methods. Thus, to model the longitudinal reinforcement of vertical members in a structural analysis and design software and consider the effect of reinforcement on creep and shrinkage deformations, the modeling approach presented in Figure 2 was used. In this approach, the columns were rst modeled by frame elements whose material properties represent the behavior of concrete. Then, a truss element was dened for the longitudinal reinforcement that connects to the start and end nodes of the frame member (Figure 2(a)). The area of the truss element was dened as being equal to the total area of the corresponding columns longitudinal reinforcements. In a similar manner, the longitudinal reinforcement of walls were modeled with truss elements that were connected to the top and bottom nodes of every mesh point of the shell elements used for modeling shear walls (Figure 2(b)). In this

Figure 2. Modeling of reinforcement in columns and structural walls.


Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

case, the area of each truss element was dened in such a way that it provided existing reinforcement ratio at its corresponding tributary area. This modeling approach assumes that the axial shortening of the concrete and reinforcement at the whole crosssection is the same and there is a perfect bond between the reinforcement and concrete.

3. CASE STUDY Several analyses for estimating the column and wall axial loads were performed on an actual building model that was a part of Guneypark project (Kurc and Ozcebe, 2007). Guneypark project was a mass residential building complex that involves the construction of 11 tall buildings (six 37story and ve 35story buildings) together with parking decks and townhouses. One of the main challenges of this project was the speed. The whole construction had to be nalized in 18 months. Because of this, reinforced concrete at plate system together with shear walls was chosen as the structural system. Steel formworks would be manufactured and used during the construction of the typical levels. The layout of the structural elements, however, was determined in such a way that the formworks were able to slide to the outside of the building from the edges and placed to the upper levels. This constraint limited the column shapes, number, location and size of beams. In addition to this, the highest concrete class that could be produced at the site was C40 ( fc = 40 MPa) that increased the column sizes at the below levels. The architectural drawings of the building considered are presented in Figure 3. The typical oor plan framing and column and wall labels of the 37story building are presented in Figure 4(a). The columns and structural walls were in black, whereas the beams were shown in dark gray. The building is almost square with 32.5m 31.5m plan dimensions with 3m oortooor height. In this building, the lateral load resistance was provided by the structural walls at the center and corners; thus, the moment and shear carried by the columns were negligible. Because of this, the amount of axial load on each column was the key element that governed the size and design of columns. The column and wall sizes are presented in Table 1. The gravity loads were carried by the 22cmdeep reinforced concrete twoway at plate system. The slabs were reinforced with additional secondary beams at various locations to reduce the deections and amount of reinforcements. The maximum allowable depth of the beams was 50 cm; otherwise, it would not be possible to slide the oor formworks to the outside of the building. There were upturn beams at the perimeter of the slabs. For the comparative study, 30 residential stories of the building that have the same oor plans were modeled (Figure 4(b)). This way, the effect of differential shortening of vertical members on axial load distribution was examined more clearly. In this model, structural walls and slabs were modeled with shell elements, but the bending stiffness of the slab and beam elements was reduced to 35% and 50%, respectively, to account for minor cracking due to gravity loads. During the CSA, the structure was loaded with its selfweight only. The analysis of the nite element model of a single story was performed with ETABS (2010) v9.6. SAP 2000 (2010) v14 was used for the linear static analysis and CSA. CSA with timedependent effects is performed in SAP 2000 by rst dening timedependent material properties for concrete. SAP 2000 uses CEBFIP Model Code for dening creep and shrinkage and requires four parameters: relative humidity percentage ( RH), notional thickness (h), shrinkage coefcient (sc) and shrinkage start age (t0). Shrinkage and creep strains for every time step is then computed with these parameters by using the equations as presented in the previous section. For all analyses performed for this study, the humidity percentage, shrinkage coefcient and shrinkage start age were taken as 70% (normal environmental conditions), 5 (normal hardening cement) and 0 day, respectively. Notional thicknesses for every column and wall segment were calculated using Equation (7). Construction sequence analysis was dened as nonlinear staged construction in SAP 2000. It is possible to dene the initial conditions before initiating the analysis. For each construction stage, structural members belonging to the corresponding stage are rst grouped and added to the stage with their construction duration. Loadings for the corresponding stage are also dened at this step. If time dependent effects are considered in the analysis, timedependent material properties should be chosen as nonlinear parameters. Once all the construction stages are dened, the analysis is being performed
Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

O. KURC AND A. LULEC

Figure 3. Architectural drawings.

Figure 4. Typical oor and simplied building model.


Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

Table 1. Column and wall dimensions. Column label


C1, C3, C8 C2, C4, C5 C6, C7

Dimensions
32 160 cm 32 210 cm 32 145 cm

Wall label
W1 W2 W3 W4

Dimensions
30 615 cm 30 460 cm 40 330 cm 32 330 cm 30 915 cm (Xdirection) (Ydirection) (Ydirection) (Ydirection) (Ydirection)

as described in the previous section. During CSA, SAP 2000 uses the rst modeling approach (Figure 1(a)). This way, the leveling of new slabs to their predened elevation was taken into consideration. 3.1. Axial loads on columns and walls Figures 5 and 6 present the loads transferred to columns C1, C3 and walls W1, W2 from every story, respectively. The axial loads for columns were obtained from six different analysis approaches: tributary area method, nite element model of a single story, 3D linear static analysis, CSA without considering any timedependent effects, CSA with timedependent effects just after the completion of the construction and CSA with timedependent effects after 5 years. Timedependent effects involve creep and shrinkage of concrete. During the analyses with timedependent effects, it was assumed that each story would be constructed in 1 week, and CEBFIP Model Code was used for dening shrinkage and creep effects on concrete. The effect of longitudinal reinforcement at columns and walls was not examined at this point. As can be seen from Figure 5, the column loads obtained from the tributary area method and nite element model of a single story were almost similar, 196 kN/story for C1 and 150 kN/story for C3.

Figure 5. Axial loads transferred to columns.


Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

O. KURC AND A. LULEC

Figure 6. Axial loads transferred to structural walls.

There were, however, considerable differences between the axial loads obtained from the linear static analysis and CSA. According to the results of the linear static analysis, the loads transferred to the columns from each story decreased at higher elevations and walls got more axial loads. For example, 191 kN were transferred to C1 from the rst story, but this value reduced to 146 kN at the top story. The results of the CSA signicantly contradicted the results of the linear static analysis; the loads transferred to columns decreased up to the midstory but started to increase up to the top story. At the midlevel, 163 kN is transferred to C1, but this value became equal to192 kN at the top story. Such differences in column loads at each story, even though the oor plans of all stories are the same, were because of the differential shortening of columns and walls and the leveling of each new story. In linear static analysis, it is assumed that the whole structure is loaded with gravity loads at the same time after the completion of the whole construction. Because of this, the effects of differential shortenings are more pronounced; in other words, more loads are carried by the vertical components that deform less. On the other hand, in CSA, the model of a new story is added on top of the previously modeled, analyzed, thus deformed bottom stories. This way, when the new story is loaded with gravity loads, the differential axial shortenings of its vertical members are only because of the additional gravity loading of the new level. In other words, the differential axial shortening of the vertical members of the bottom stories do not cause any load transfer among the vertical member of the new story. That is the main reason of getting signicantly different results from linear static analysis and CSA. When timedependent effects were considered in the analysis, the differential axial shortenings of columns with respect to walls were increased. In this building, the compressive stresses due to the gravity loads on walls were much smaller than the ones on columns since walls had larger cross sections. Because of this, the shortening caused by shrinkage and creep were less signicant in walls. Because columns have shortened continuously due to creep for 5 years, they transferred considerable amount of axial load to the walls. As presented in Figure 5, the axial loads on columns C1 and C3 were signicantly reduced as time passes. According to the results of the CSA without timedependent effects, 163 kN was transferred to column C1 from level 15. When shrinkage and creep deformations
Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

were considered, this value dropped to 128 kN immediately after the completion of the construction and to 89 kN in 5 years time. In other words, the amount of load transferred from level 15 to column C1 was almost halved. The opposite situation was observed at structural walls (Figure 6). The amount of gravity load transferred to W1 at level 15 was equal to 850 kN according to the results of the CSA. With the effect of creep and shrinkage, this value increased to 1046 kN and 1206 kN immediately after construction and in 5 years, respectively. 3.2. Effect of modeling of reinforcement To investigate the effect of modeling the reinforcement of columns and walls on the axial load distribution, various analyses with reinforcements were performed and compared it with the results of the analyses without reinforcements. In these analyses, the longitudinal reinforcements were modeled with the methodology described in Section 2.3. Two models were created. In the rst model, columns had the minimum 1% reinforcement ratio. In the second one, columns had the maximum 4% reinforcement ratio. (ACI 31808 (2008) allows 8% reinforcement ratio for columns but recommends 4% ratio for column bars that are lap spliced.) With these two models, the inuence of reinforcement amount was examined for following two limiting cases: minimum and maximum reinforcement ratios. The structural walls were modeled with 0.25% reinforcement ratio at both models. Figure 7 presents the effect of modeling the reinforcement on column axial loads. These results are for the analysis approaches that ignore the timedependent effects. In this case, there is not any signicant change in the amount of axial loads carried by the columns for both linear static analysis and CSA. This result is expected since modeling the reinforcement only affected the axial stiffness of vertical members. Having the same reinforcement ratio at columns did not change their differential axial shortenings. There is a minor reduction in the relative axial stiffness of walls with respect to columns, but this did not cause any signicant change in the amount of load carried by each wall or column. For the most extreme case when the columns have 4% and walls have 0.25% reinforcement, the difference was around 7% for column C1 and 4% for wall W1.

Figure 7. Effect of reinforcementanalysis without timedependent effects.


Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

O. KURC AND A. LULEC

Signicant differences were obtained in the column axial loads when creep and shrinkage deformations were considered (Figure 8). In this case, as concrete shortens due to creep or shrinkage, the longitudinal reinforcements start to carry additional axial stresses and reduce the amount of shortening. Modeling of reinforcement reduced the shortening of columns relative to the walls since walls had smaller reinforcement ratio. That is the reason why the axial loads of columns have increased but the ones on walls have decreased (Figure 8). Having the minimum reinforcement ratio in columns did not cause much difference with respect to the results of the model without the reinforcing bars. The most signicant change was observed in wall W1 that corresponds to 6.5% reduction of the axial load at level 15 in 5 years. As the reinforcement ratio increases, the differential shortening of columns and walls decreases; thus, the columns get larger axial loads. For the limit case, when columns have 4% reinforcement, the amount of load transferred to column C1 at level 15 in 5 years was increased to 163 kN, which was equal to 105 kN with 1% reinforcement and 89 kN when the effect of reinforcement was ignored. 3.3. Base reactions and deformations Table 2 presents the base reaction of all analyses that considers construction sequences and linear static analysis for columns C1, C3 and walls W1, W2. The axial shortenings of the same members at level 25 for the same analyses are presented in Table 3. The wall deformations are for the left bottom corner of W1 and the top right corner of W2. The base reactions and deformations are only for the selfweight of the building. The timedependent deformations of concrete caused a signicant axial load transfer from columns to structural walls in this building. The base reaction of column C1 decreased from 5213 kN to 3364 kN in 5 years, which corresponds to an about 36% reduction. These loads were carried by the structural walls because timedependent deformations are less effective due to having smaller compressive stresses. The base reaction of wall W1 increased from 23 949 kN to 34 711 kN, which

Figure 8. Effect of reinforcementanalysis with timedependent effects.


Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

Table 2. Base reactionsconstruction sequence analysis (kN). No reinforcement Linear static analysis
C1 C3 W1 W2 4771 3476 25 730 8962

1% longitudinal reinforcement After 5 years


3364 2233 34 711 11 939

4% longitudinal reinforcement After construction


5500 3737 26 223 9090

No time dependent effects


5213 3767 23 949 8287

After construction
4595 3179 28 807 9686

After construction
4798 3370 27 859 9486

After 5 years
3809 2638 32 627 11 466

After 5 years
5335 3450 29 019 10 566

Table 3. Column and wall deections (cm). No reinforcement Linear static analysis
C1 C3 W1 W2 1.60 1.21 0.73 0.66

1% longitudinal reinforcement After 5 years


5.23 4.29 3.11 2.80

4% longitudinal reinforcement After construction


1.10 0.83 0.59 0.52

No time dependent effects


0.57 0.43 0.23 0.21

After construction
1.50 1.16 0.69 0.60

After construction
1.36 1.06 0.66 0.42

After 5 years
4.72 3.90 2.97 2.68

After 5 years
3.79 3.05 2.63 2.42

corresponds to an about 45% difference. The modeling of reinforcement and the reinforcement ratio of columns inuenced the base reactions considerably. When the columns had the minimum reinforcement, the difference in base reactions with respect to the analysis without reinforcement remained below 7% and 15% for columns and walls, respectively. As the amount of reinforcement in columns increase, the amount of shortening of columns decrease, and they carry more axial load. The base reaction of column C1 becomes equal to 5335 kN after 5 years, whereas this value was computed as 3809 kN with 1% reinforcement ratio. The wall reactions, on the other hand, decreased as columns shorten less. The base reaction of wall W1 reduced from 32 627 kN to 29 019 kN when the column reinforcement ratio increased to 4% from 1%. For all cases, linear static analysis signicantly underestimated the wall base reactions when timedependent deformations are considered. Such an error may lead to compression failure under high moments due to lateral loads. The column C1 with 1% reinforcement ratio continued to shorten, and its axial load was decreased to 3809 kN from 4798 kN in 5 years, which corresponds to a 20% reduction. As the reinforcement ratio in columns increases, the longterm differential shortenings become less effective. In this building, the axial load on column C1 changed only 3% in 5 years with 4% reinforcement ratio (55005335 kN). Linear static analysis results for column C1 involve large errors especially when compared with the results of time dependent analysis after 5 years (+25% for 1% and 12% for 4% column reinforcement, respectively). Thus, even though linear static analysis of 3D structural models can be very practical during the design stages of tall buildings, their results can lead to considerable errors in estimating the axial load distribution among vertical members. Accurate estimation of column and wall deformations is crucial for predicting any possible damage to nonstructural elements. The maximum deection limits are usually dened as spandeection ratio. ACI 31808 (2008) sets the deection limit to ln/480 for spans carrying deectionsensitive materials such as masonry walls and ln/240 for other locations where ln represents the clear span length. According to the analysis results when timedependent deformations were not considered, the relative deformation between columns C1 and C3 and C3 and W2 were about 0.2 cm, which was not signicant. When creep and shrinkage effects were considered, the relative deformation between C3 and W2 became equal to 1.22 cm with 1% reinforcement, whereas linear static analysis estimation
Copyright 2011 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

O. KURC AND A. LULEC

was only equal to 0.55 cm. The relative 1.22cm deection corresponds to a deection ratio equal to ln/245, which indicates a possible damage at the exterior masonry wall and thus violates the codes serviceability conditions. The code limit can only be satised by increasing the reinforcement ratio. When 4% reinforcement was used, the relative deformation was reduced to 0.6 cm, which corresponded to a deection equal to ln/500.

4. CONCLUSIONS Several analysis approaches for estimating the axial loads on columns and structural walls were investigated by using a simplied model of an actual 37story reinforced concrete building. The time dependent effects such as creep and shrinkage and the inuence of modeling the longitudinal reinforcement on axial load distribution were also examined. The following conclusions were drawn from this study: The way the building is modeled and the type of analysis performed signicantly inuenced the column/wall axial loads and deformations. Thus, the design engineers should be very cautious while using the analysis results and always question the assumptions of the analysis method that they are using. The tributary area method and nite element model of a single story ignore the differential shortening of columns and walls, which resulted in an overestimation of column axial loads and an underestimation of wall axial loads. Linear static analysis cannot consider oor leveling during construction and timedependent effects and may lead to considerable errors in column and wall axial loads. Shortening caused by creep and shrinkage signicantly affected the amount of axial load carried by columns and walls. Creep was more effective in columns due to high compressive stresses, thus causing a considerable load transfer from columns to walls. The longitudinal reinforcement reduced the amount of differential shortening among vertical members, thus changing the axial loads on these members considerably. Increasing the reinforcement ratio in columns reduced the creep deformations for the long term. The effect of creep and shrinkage deformations must be investigated for the following time periods: immediately after construction and in 5 years time. The longitudinal reinforcement should also be modeled since they signicantly inuence the axial load distribution among vertical members. This way, the upper and lower axial load limits would be obtained. Then, the design of columns and walls should be checked for each case.

REFERENCES
ACI 31808. 2008. Building Code Requirements for Structural Concrete. American Concrete Institute: MI. CEBFIP. 1993. CEBFIP Model Code 1990. Bull. dInf. Thomas Telford Ltd.: London, UK. Donping F, Haifeng XI, Xiaoming W, Chuanmin Z. 2009. Inuences of shrinkage, creep, and temperature on the load distributions in reinforced concrete buildings during construction. Tsinghua Science and Technology 14: 756764. ETABS. 2010. Extended 3D Analysis of Building Systems, v9.6.0, Computers and Structures, Inc: CA. Fintel M, Khan FR. 1969. Effects of column creep and shrinkage in tall structuresprediction of inelastic column shortening. ACI Journal (December): 957967. Kim JY, Abdelrazaq AK. 2008. Construction sequence analysis of the at plate system in a highrise building and its impact on the construction cycle. Structural Design of Tall and Special Buildings, doi:10.1002/tall.443 Kurc O, Ozcebe G. 2007. The Analysis and Design of the Tallest Buildings in Ankara. Structural Engineers World Congress: Bangalore, India. Kwak HG, Kim JK. 2006a. Timedependent analysis of RC frame structures considering construction sequences. Building and Environment 41: 14261434. Kwak HG, Kim JK. 2006b. Determination of efcient shoring system in RC frame structures. Building and Environment 41: 19131923. Kwak HG, Seo YJ, Jung CM. 2000. Effects of the slab casting sequences and drying shrinkage of concrete slabs on the shortterm and longterm behavior of composite steel boz girder bridgesPart I. Engineering Structures 23: 14531466. Malm R, Sundquist H. 2010. Timedependent analyses of segmentally constructed balanced cantilever bridges. Engineering Structures, doi:10.1016/j.engstruct.2009.12.030

Copyright 2011 John Wiley & Sons, Ltd.

Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

AXIAL LOADS ON COLUMNS AND STRUCTURAL WALLS

Pan LB, Liu PC, Bakoss SL. 1993. Longterm shortening of concrete columns in tall buildings. Journal of the Structural Engineering 119: 22582262. Park HS. 2003. Optimal compensation of differential column shortening in highrise buildings Structural Design of Tall and Special Buildings 12: 4966. Pfeifer DW, Magura DD, Russel HG, Corley WG. 1971. Time Dependent Deformations in a 70 Story Structure, Paper SP 277, Special Publication SP27: Designing for Effects of Creep, Shrinkage and Temperature in Concrete Structures, January 1. American Concrete Institute: Detroit, MI. SAP 2000. 2010. Static and Dynamic Finite Element Analysis of Structures v14.1.0. Computers and Structures, Inc: CA. Smith BL, Coull A. 1991. Tall Building Structures: Analysis and Design. John Wiley and Sons, Inc.: NY. Taranath BS. 2005. Wind and Earthquake Resistant Buildings: Structural Analysis and Design. CRC Press, Taylor & Francis Group: FL. Wight JK, MacGregor JG. 2008. Reinforced Concrete: Mechanics and Design, fth edn. Prentice Hall: Upper Saddle River, New Jersey, USA.

AUTHORS BIOGRAPHY
Ozgur Kurc is assistant professor in the Structural and Earthquake Engineering Laboratory at Department of Civil Engineering, Middle East Technical University (METU), Ankara, Turkey. He received his bachelors and masters degree from METU and PhD degree from Georgia Institute of Technology, Atlanta, USA, in 2005. He also worked as structural engineer at BLA Associates, Atlanta, for more than six years. His current research areas related to structural engineering are analysis and design of tall buildings, wind effects on buildings and bridges, and pseudodynamic testing of structures. Andac Lulec is a masters student and research assistant in the Structural and Earthquake Engineering Laboratory at the Department of Civil Engineering, Middle East Technical University (METU), Ankara, Turkey. He received his dual bachelors degree in civil engineering and industrial engineering from METU in 2008 and 2009, respectively. His current research involves comparison of seismic design approaches of building codes and effect of longterm deformations at tall buildings.

Copyright 2011 John Wiley & Sons, Ltd.

Struct. Design Tall Spec. Build. (2011) DOI: 10.1002/tal

You might also like