You are on page 1of 6

Proceedings of the Ninth Pacific Conference on Earthquake Engineering Building an Earthquake-Resilient Society 14-16 April, 2011, Auckland, New


Response of Out-of-Plumb Structures in Earthquake
T. Masuno, G.A. MacRae & V.K. Sadashiva
Department of Civil and Natural Resources Engineering, University of Canterbury, Christchurch, New Zealand.

A. Wada
Department of Architectural Engineering, Structural Engineering Research Centre, Suzukakedai Campus, Tokyo Institute of Technology, Tokyo, Japan.

ABSTRACT: No building is built perfectly plumb. The out-of-plumb may not just result from construction influences, but it may also occur as a result of permanent displacements after an earthquake. While there are methods to account for out-of-plumb for non-seismic design, such methods do not exist for seismic design, as the effect of different levels of out-of-plumb has not been quantified. It is desirable to know how a new building with a specified out-of-plumb is likely to behave in a earthquake, and how one which is out-of-plumb as a result of a previous earthquake may perform in an aftershock. In this research, the effect of building out-of-plumb is quantified. Simple models of structures with different amounts of out-of-plumb are made. These are designed with different levels of ductility and with different distributions of strength/stiffness over their height. Steel structures with elastic-perfectly plastic hysteresis loops are considered as these are considered to be most susceptible to large displacements. The structures are then subject to a suite of earthquake records using inelastic dynamic time history analysis to quantify the effect of out-of-plumb. Records from both directions are considered. Relationships between out of displacement and increase in peak and permanent displacement response are developed, and recommendations are made for estimation of response due to different out-of-plumb limits. 1 INTRODUCTION Traditional methods for the analysis of multi-storey structures in non-seismic zones are based on the use of elastic analysis. Modifications to the results obtained to account for many effects, such as in– plane P-delta effects, out-of plane bucking and plasticity effects are then used in a second step. Often these methods use effective length factors which are applied differently to the frame braced moments and sway moments (e.g. AISC 2005, NZS3404 1997). This requires that different analyses be conducted to obtain the braced-frame and sway-frame moments as shown in Figure 1. However, for seismic design of steel frames, such a rigorous analysis is not generally conducted. Because of the complexity of the traditional methods for non-seismic design, such as those given in Figure 1, new analysis methods have been developed. One of these is the Direct Analysis method (AISC 2005). Such a method uses higher order analysis to estimate the frame elastic moments directly. It uses the dependable, rather than nominal, member stiffnesses, and member residual stress effects are also included in the member stiffness. Also, frame out-of-plumb is considered explicitly. The out-ofplumb of 0.2% was used following the work of Surovek and White (2004). Neither residual stress effects nor out-of-plumb were considered in the traditional methods. After the moment demands are established, further modifications are still required to check the frame capacity. An even more powerful method of analysing non-seismic steel frames, an extension of the direct analysis method, has recently been developed. It is termed “Extended Direct Analysis (or EDA)” (Lu

Paper Number 061

Because it is conducted using freely available software including the residual stresses associated with the NZ column curves. it was found that. Also.N*1 V1 N*2 V2 N*3 V3 V1 V2 V3 N *1 R1 N *2 N *3 R2 R3 R1 R2 R3 Original Frame = Nonsway Frame For Mfbx (i. If the structure does not collapse then it is not satisfactory and member sizes may be increased. However. They conducted studies to evaluate the effect of various types of vertical irregularity on structural response during seismic excitations. Two classes of the structure stiffness distribution were designed (i) the Constant Stiffness ratio (CS) and (ii) the Constant Interstorey Drift Ratio (CISDR) shown in Figure 2. In the process of the research two structural models . Continuous column stiffness ratio αacci (MacRae et al. EDA considers all the factors in the AISC Direct Analysis method. The flexural beam was pinned at the bottom. Hi = storey height of the ith floor level. 2 RELATED STUDIES No study is known to have been conducted on out-of-plumb in buildings. in the mass irregularity study. et al. for seismic design the effects of residual stress are not considered directly on member stiffness. then it is not satisfactory. 2009). 5. (2009) was used as part of this research so it is discussed below. or due to it being deformed in some way. if the structure fails and collapses under the applied factored forces. (2010) that variation in response due 2 . However. and Koi = initial stiffness of the ith floor level. Graduating undergraduate students from the University of Canterbury. it is used by a number of engineers in practice. A higher order plastic analysis considering moment-axial force interaction is therefore conducted. mass irregularity was characterized as mass increase in top.e. as well as New Zealand engineers. “What is the relationship between out-of-plumb and change in response?” It summarises the work of Masuno (2010). It is shown by Tagawa et al. Ii = moment of inertia at the ith floor level. These served as extreme practical limits for the analysis. Because it also considers the dependable strength of a structure. 9 and 15 storey height structures. This paper aims to answer the question. are aware of this method. A rigid link between shear beam and flexural beam slaves the horizontal displacement of the joined nodes.Shear-Beams (SB) and Shear-Flexural Beams (SFB) were compared. 2004) which represents the flexural beam stiffness relative to shear beam at ith floor is computed using Equation 1 where E = Elastic Modulus. Therefore. and it may be out-of-plumb due to construction tolerances. From this comparison. No frame is perfectly plumb. unrealistically large soft-storey displacements occurred in structures modelled as SB. but it also considers frame plasticity with the dependable member strengths. This is the topic of this paper. mid height and bottom floor of 3. For example. construction errors. The SFB was modelled as shown in Figure 3 below. P-δ) + Sway Frame For Mfsx (i. the effect of frame out-of-plumb is not normally considered. P-∆) Figure 1: Obtaining braced and sway moments. the model reasonably realistically represents the real structure.e. The mass at the level considered was increased as a ratio of the regular structure mass. such as by a previous earthquake. therefore only the SFB model was used in this research. A study is being conducted on the significance of this (Lu 2011). work by Sadashiva et al.

5% with step of 0. Rayleigh damping with a critical damping ratio of 5% was used in the first mode and the mode corresponding to the number of storeys in the structure (Carr 2004). structure height. 2. it can be expected that these structures will also have a tendency for larger displacements in the direction in which they are leaning. The frame of the structure is modelled as shear beams and continuous flexure columns slaved together. design codes limit the amount of construction out-ofplumb.2% (or 1/500) (e.01% are considered. AISC 2005 or NZS3404 1997). 9 and 15. They showed that during earthquake shaking. the moment tended to cause extra deformation in the direction in which the moment was applied. Structures with out-of-plumb ranging from 0. of 20. Specific details on how the members were sized are available in Sadashiva et al. A value representing that of the seismic columns alone was therefore selected. These 3 .000 kg at each floor. 3 MODELLING AND EVALUATION APPROACH The models used in the studies by Sadashiva et al. (2009). each multi-storey structure is assumed to have a constant lumped mass.g. Structures were designed according to the Equivalent Static Method in NZS1170. (2009) are also used in this study. 4 and 6. and that associated with the continuity of all the columns at a level is small. Structures were designed with a target interstorey drift of 1. A study by MacRae and Kawashima (1993) and Yeow (2010) looked at the behaviour of bridge columns subject to axial force and moment before earthquake shaking occurred. m. That is. 6.8% and target design ductility’s (similar to lateral force reduction factors) of 1. For short structures. the allowable out-of-plumb changes with height. h = 3. For taller structures. Because out-of-plumb building structures such as those described in this study also have an eccentric moment. α cci = E Ii H i3 K oi (1) The RUAUMOKO (Carr 2004) computer program was used to run inelastic dynamic time-history analysis subjected to 20 earthquake records with a specified design interstorey drift ratio.Forces Deflection profile Deflection profile (a) Constant Stiffness (CS) (b) Constant Interstorey Drift Ratio (CISDR) Shear beam Flexural beam Figure 2: Two classes of stiffness distribution models Figure 3: SFB model to αacci between that associated with the continuity of the seismic columns alone. The structure is also assumed to have equal storey height.1%-2. The number of stories (N) considered in this study are 3. The basic structure was designed as an ordinary building in Wellington close to the fault on site class C. mass irregularity and structural ductility factor. the maximum out-of-plumb is 0. To minimize problems related to out-of-plumb.5 (2004).5m. Both CISDR and CS models were used as in Figure 3. A bilinear hysteresis loop with bi-linear factor of 1% was used in the study.

The residual response is very dependent on the hysteresis loop characteristics (MacRae and Kawashima 1993). Equations 4 and 5 were developed to estimate the median effect of out-of-plane for the most critical (CISDR) model peak response. the median peak response (MPR) and the median residual response (MRR) was normalized with the MPR or MRR from initially perfectly plumb models. If a structure has a permanent drift at the end of an earthquake of 2%. Normalization was done using Equation 3.2% + 1) = 2. . lateral force reduction factor). The two programmes were automated to run analysis and the desired output values were extracted in the process. From the information obtained. 2002) where xi is the peak or residual storey drift ratio from the ith record and n is the number of earthquake records considered.072 × µ × N × OOP (%) + 1 5 EXAMPLE For example.2% = 2. The twenty SAC (2000) earthquake ground motion records for Los Angeles with probability of exceedance of 10% in 50 years were used.35%.structures were entered into the programme in their deformed configuration before the seismic analysis started. were computed assuming a lognormal distribution according to Equation 2 (Cornell et al. Input files for RUAUMOKO are generated using MATLAB (2009). The predictions are compared to the actual results for the case in Figure 4. Thus the peak demand is increased by less than 10% and the total peak displacement relative to perfectly plumb is likely to be 2. so that directional trends in displacement from individual records did not affect the results. To evaluate the effect of out-of-plumb. NMRR.2% with target ductility of 4 and a target drift of 2% is likely to have a NMPR = 2 %* (0. then in the opposite direction. perfectly plumb models and models with initial out-of-plumb were analysed. The total displacement relative to the perfectly plumb value at (4) (5) (3) 4 . This model was more critical than the CS model. ˆ x= ⎛1 ⎜ ⎜n e⎝ ∑ ln( xi ) ⎟ ⎟ i =1 n ⎞ ⎠ (2) In order to compare the response. One example is shown in Figure 4 for a 9 storey frame with the CISDR model. Pdelta effects were considered.0091*4*10*0.46% if the same earthquake as that above occurs. When they were applied they were run both in one direction. a 10 storey structure with a construction out of plumb of 0. Each record was used for each design ductility (i.15%.e. period and level of out-of-plumb. [MPR or MRR (out-of-plumb)]/ [MPR or MRR (perfectly straight)] 4 EFFECT OF OUT-OF-PLUMB The increase in response due to out-of-plumb (OOP) increased due to ductility (duc or µ). then the NMPR = 3.15% + 0. NMPR. The dynamic inelastic time history computer programme RUAUMOKO (Carr 2004) was used in this project to run the analysis. The median peak and residual interstorey drifts ratio (ISDR). number of stories (N) and increased initial out-of-plumb. The residual displacement response increases faster than the median peak response. NMPR = 0. and residual response. Bi-linear hysteresis loop with bi-linear factor of 1% was used for shear beams and continuous column of the model. Critical damping of 5% was adopted from study of Sadashiva et al.0091 × µ × N × OOP (%) + 1 NMRR = 0. but the residual response is always less than the peak response. (2009).

2011. MacRae G.C.Normalised Peak Response the end of the earthquake is therefore likely to be 2% + 3.2 duc=2 TE µ=2 2 duc=4 µ=6 1. Journal of Structural Engineering.5 OOP (%) 2 2.4 1. Fatemeh J. Inc.2 1 0 0.4 Actual µ=4 duc=1 APPROXIMA 2. Specification for Structural Steel Buildings.F. MacRae G.A. 1993. Department of Civil Engineering. In addition. Ruaumoko 2D Inelastic dynamic time-history analysis program. Cornell C.C.. to be published in the proceedings of SEE6 Tehran Conference (2011).J.6 1. Effect of column stiffness on braces frame seismic behaviour.Y.W. Residual stress effects on the seismic performance of steel frames. d) An example of how peak displacement increase due to out-of-plumb was provided. Ziemian R. 2002.A.5 1 1. 526-533.H. NZ Society of Structural Engineering (SESOC) Journal. 5 . b) Based on the observed response equations were developed to estimate the likely median increase in response for different levels of initial out-of-plumb. & Roeder C. Responses (CISDR..H. N=9) 2. University of Canterbury. Hahn C. structures with a greater number of storeys and those with greater design ductility’s also tend to have greater response. storey number and design ductility. Christchurch. Peng B.. Master’s Dissertation. & Foutch D. Lu A.C. REFERENCES: AISC..46%. The seismic response of bilinear oscillators using Japanese earthquake records. This is likely to be unsatisfactory and the structure may require replacement.8 duc=6 1. 128(4). 30. & Kawashima K. Carr A. 7 ACKNOWLEDGEMENTS This paper is similar to a paper titled “Effect of building out-of-plumb on seismic response”.Y. & Clifton G. Journal of Research of the Public Works Research Institute. New Zealand. Christchurch.A. Hamburger R..A. Kimura Y.46% = 5.. 2004. 2009. University of Canterbury.O. 2004. Probabilistic basis for 2000 SAC FEMA steel moment frame guidelines. Lu A. 22(2). c) The rate of increase in residual interstorey displacement was almost an order of magnitude more sensitive to out-of-plumb than was peak displacement. It was shown that: a) Greater initial out-of-plumb generally causes greater response increases relative to structures with no initial out-of-plumb..A.5 Figure 4: Comparison of actual and approximated peak responses – 9 storey CISDR design 6 CONCLUSIONS A number of analyses were conducted on shear-type structures to evaluate the effect of out-of-plumb on their performance. Extended direct analysis of steel frames. MacRae G. American Institute of Steel Construction. 2005.

Structural Design Actions Part 5: Earthquake actions New Zealand. NZS 3404:1997. AIJ.. Journal of Structural Engineering. Masuno T.continuous solumn effects in steel moment frames in perspective of dynamic stability (Part 2).S. Journal of Structural and Constructional Engineering. University of Canterbury. Transactions of the Architectural Institute of Japan.. Christchurch. steel moment-resisting frame structures.Journal of Structural Engineering. 2000. Continuous column effects of gravity columns in U. Christchurch. 42(4). 2009. The MathWork . Sadashiva V. Alternative Approaches for elastic analysis and design of steel frames.W. Response of multi-storey structures with initial out-of-plumb. Matlab R2009b. Standards New Zealand. Berkeley. Yeow T. 2004. 381-391. MATLAB. NZS 1170. & Lowes. 288-301. Wellington. 130 (3). SAC.L. L. 2010. SEAOC-ATC-CUREE . Inc. New Zealand. 2010. Tagawa H. Undergraduate Report.5:2004.The SAC steel project. I: Overview.A. & Deam B. Determination of structural irregularity limits. Bulletin of The New Zealand Society for Earthquake Engineering. MacRae G. Steel Structures Standard. 2009.E. New Zealand. 75(650).Mass irregularity example. 2010. MacRae. New Zealand. 1997.A. Wellington. G. New Zealand. 6 .K. Undergraduate Report. Ca. & White D. Standards New Zealand. University of Canterbury. Seismic behaviour of cantilever bridge columns. Surovek-Maleck A. 761-770. 130(8). 2004.