Engineering Structures 30 (2008) 36193628

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Investigation of progressive collapse resistance and inelastic response for an earthquake-resistant RC building subjected to column failure
Meng-Hao Tsai , Bing-Hui Lin
Department of Civil Engineering, National Pingtung University of Science and Technology, No.1 Hseuh-Fu Road, Neipu, Pingtung 912, Taiwan

article

info

a b s t r a c t
Following the linear static analysis procedure recommended by the US General Service Administration (GSA), the potential of an earthquake-resistant RC building for progressive collapse is evaluated in this study. Nonlinear static and nonlinear dynamic analyses are conducted to estimate the progressive collapse resistance of the building subjected to column failure. Under an approximate deflection demand, different collapse resistances are obtained. It indicates that different criteria for estimating the collapse resistance may be adopted for these two nonlinear analysis methods. The nonlinear static approach leads to a conservative estimation for the collapse resistance if 2.0 is used as the dynamic amplification factor (DAF). As the column-removed building is loaded into a significantly yielding phase, different assessed results are obtained by the linear static method and the nonlinear acceptance criterion suggested by the GSA guidelines. A DAF considering the inelastic dynamic effect may be needed in the GSA linear procedure. The capacity curve constructed from the nonlinear static analysis is shown to be capable of predicting the progressive collapse resistance and the DAF of a column-removed RC building. 2008 Elsevier Ltd. All rights reserved.

Article history: Received 20 July 2007 Received in revised form 28 May 2008 Accepted 30 May 2008 Available online 21 July 2008 Keywords: Progressive collapse Nonlinear analysis method Dynamic amplification factor GSA guidelines

1. Introduction Many practicing engineers and academic researchers have been engaged in the prevention of progressive collapse since the partial collapse of the Ronan Point apartment building in 1968. Especially after the malevolent bombing of the Murrah Federal Building in 1995, several changes to the philosophy and practice of design for important buildings have been made in the last decade. Resistance of building structures to progressive collapse has been an important task for the development of structural design codes. Some study results, code approaches, and design strategies or standards have been reviewed, discussed, and/or compared in the literatures [17]. Generally speaking, the investigated issues include abnormal loading events, assessment of loading, analysis methods, and design philosophy. In recent years, the development of analysis methods for evaluating the progressive collapse potential of an existing or new building has been an imperative subject. Linear static, nonlinear static, linear dynamic, and nonlinear dynamic methods are four basic approaches for the progressive collapse analysis. Advantages and disadvantages of these approaches have been discussed by Marjanishvili and Agnew [8,9]. Detailed descriptions of a step-bystep, linear static procedure for progressive collapse analysis have

been issued by the US General Service Administration (GSA) [10] and Department of Defense (DoD) [11]. The GSA linear static analysis approach has been applied to evaluate the potential of a steel moment frame and a RC frame for progressive collapse [8,12]. Terrorist events are quite few in the history of Taiwan. Even so, the potential hazard of terrorist attacks always exists because of the trend of globalization. Since Taiwan is located in an earthquake-hazardous region, most of the RC buildings are detailed according to the seismic design code. Some studies indicated that seismic design detailing might help to enhance the resistance of buildings against progressive collapse [1315]. Hence, seismically designed RC buildings are expected to have low potential for progressive collapse. In this paper, the progressive collapse potential of an earthquake-resistant RC building under four threat-independent, column-removed conditions is evaluated by using the GSA linear static analysis procedure. Nonlinear static and dynamic analyses are carried out to verify the linear analysis result and estimate the progressive collapse resistance for each column-removed condition. The catenary effect is neglected and only the flexural failure mode is considered herein. Dynamic effect on the assessed results obtained from the linear or nonlinear static method is discussed. Application of the nonlinear static method to the estimation of the progressive collapse resistance and the dynamic amplification factor (DAF) of a column-removed building is proposed and demonstrated.

Corresponding author. Tel.: +886 8 7703202x7193. E-mail address: mhtsai@mail.npust.edu.tw (M.-H. Tsai).

0141-0296/$ see front matter 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.engstruct.2008.05.031

3620 Table 1 Dimensions of RC member sections (cm) Floor 1F 2F11F Column 70 100 70 90 Beam 50 90 50 75

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

Joist 30 65, 20 50 30 65, 20 50

2. Descriptions and modeling of the RC building 2.1. Descriptions The RC building is an 11-storey, moment-resisting frame structure with a 2-storey basement. Its first storey is an open space for the public. The center-to-center plan dimensions are 17.75 m in length and 12.25 m in width from the ground floor to the roof, as shown in Fig. 1. There are three bays with center-to-center span length arranged as 5.6 m, 6.55 m, and 5.6 m in the longitudinal (westeast) direction, and two bays with a 6.6 m and a 5.75 m span in the transverse (northsouth) direction. The storey height is 3.8 m for the first storey and 3.2 m for the others. In addition to the self weight, a dead load (DL) of 0.98 kN/m2 is applied to the roof and 0.245 kN/m2 to other floors. The service live load (LL) is 4.91 kN/m2 for the roof and 1.96 kN/m2 for other floors. Conventionally, the structural design consultants in Taiwan use larger imposed DL and LL on the roof to account for the loading of special waterproof roofing and some accessory facilities (e.g. water reservoir, ventilation system, etc.), respectively. Table 1 presents the section dimensions of the RC members for the building. A compressive strength equal to 27 500 kN/m2 is used for the concrete. The design yield strength is 412 000 kN/m2 for the main reinforcements and 275 000 kN/m2 for the stirrups. The building is located at a soft soil site and its design spectral response acceleration, SaD , is equal to 0.47g estimated at the fundamental period. All the beams and columns are designed and detailed according to seismic code requirements. The beam members have at least three continuous #10 steel bars for the top and bottom reinforcement. As required by the seismic demand, more #10 top and bottom bars are provided and continuous through the column lines at the beamcolumn joints. The positive moment strength at each joint face is larger than one half of the negative moment strength at that face of the joint. Also, sum of the nominal flexural strengths of the columns framing into a joint is at least 1.2 times larger than that of the beams framing into the joint. Hence, a strong column-weak beam mechanism may be ensured. If any interior column on the ground floor is removed, the two-span beam will redistribute loads to adjacent columns. Flexural hinges may form at the two ends of the beams when they cannot resist the instantaneous loading in an elastic manner. If the plastic hinge strength is insufficient to sustain the loading, the beam deflection will further increase to mobilize catenary tensile action, which is the final protection against collapse. 2.2. Structural modeling A beamcolumn frame model is constructed for the RC building using the SAP2000 commercial program [16]. The model is assumed to be fixed on the ground. Self weight of the exterior walls is distributed to the spandrel beams. Also, self weight of the interior walls and partitions is estimated and applied to the floor slab as a distributed load. Thereafter, according to the tributary area, self weight of the slab and all the dead loads and live load on it are distributed to the beam elements for each floor. The fundamental period of the building model is equal to 1.35 and 1.34 s in the longitudinal and transverse direction, respectively. The reinforcement disposition of each member section is simulated based on the design drawings of the RC building. There

are twenty types of reinforcement disposition and nine different spacing of shear stirrup for all beam sections. The nominal moment and shear strength vary from 620 kN m to 1460 kN m and 730 kN to 920 kN, respectively, for the beam members. Flexural plastic hinges are assigned to both ends of beam elements. Default moment-hinge properties based on the FEMA-273 guidelines [17] are adopted for the hinge model. Different performance levels are represented by circular symbols with different shadows, as shown in Fig. 2. Although, as recommended by the GSA guidelines, strength increase factors for material properties may be used in the analysis, they are not considered in this study. Preliminary studies [18] indicated that collapse of the RC building under column-removed conditions is governed by the flexural failure mode of beam elements. Also, the column members remain elastic even when the ultimate moment capacities of the connected beam sections have been developed. Hence, shear failure is not considered and the column members are assumed to be elastic in this study. 3. Progressive collapse potential 3.1. Loading and criterion A downward loading combination Pst = 2(DL + 0.25LL) (1)

recommended by the GSA guidelines is adopted for the linear static analysis. DL includes the structural weight and additional dead loads. Pst is defined as the GSA loading herein. In the linear static analysis, the GSA loading is applied to the RC building subjected to column failure, and the demand-to-capacity ratio (DCR) of flexural moment is calculated to assess the progressive collapse potential. Since the building has a typical structural configuration, the acceptance criterion for the primary structural components is DCR 2.0. When the DCR value is larger than 2.0, a hinge has to be inserted to the member end for releasing the moment. For dynamic analysis, the DAF, 2, in Eq. (1) is omitted and the downward loading is changed to Pdy = (DL + 0.25LL). 3.2. Collapse potential Four threat-independent, column-removed conditions, designated as Case 1B, Case 2A, Case 1A, and Case 2B, are considered for the building. According to the bay line numbers shown in Fig. 1, the removed column of the first storey is 1B, 2A, 1A, and 2B for Case 1B, 2A, 1A, and 2B, respectively. Linear static, nonlinear static and nonlinear dynamic analyses are carried out to investigate the column failure responses of the building. 5% inherent damping ratio is assumed for the dynamic analysis. Similar to the results observed by Sucuoglu et al. [19], most of the downward loading originally sustained by the failed column is transferred to the plane frames intersecting at the line of the failed column. Therefore, the DCR values, plastic hinge distribution, and deflection of those intersected frames are the major concerns in this paper. Fig. 3a shows the DCR values and the plastic hinges obtained from the GSA linear static and nonlinear static analysis for Case 1B, respectively. Force-controlled method with force magnitude equal to the GSA loading (Pst ) is used for the nonlinear static analyses. Moment distribution of the linear static analysis is presented in Fig. 3b. Because of different local axis definitions, the column moments are not displayed for the BB frame. It is seen that there is no DCR value larger than 2. The column-removed building has low potential for progressive collapse and no moment-released (2)

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

3621

Fig. 1. Plan dimensions of the building.

Fig. 2. The plastic hinge model.

hinge has to be inserted to any member. However, the nonlinear static analysis results show several plastic hinges occurred at the member ends. Similar result is observed from the nonlinear dynamic analysis, as shown in Fig. 3c. Based on the performance levels in Fig. 2, the plastic hinges in Figs. 3a and 3c are classified as immediate occupancy under the GSA loading. It is observed that a DCR value close to or larger than 1.0 will lead to the formation of a plastic hinge. This implies that the GSA guidelines permit certain flexural ductility to be developed. Similar results are obtained for Case 2A, 1A, and 2B, which are not shown here. Table 2 summarizes the maximum deflection at the column-removed point for those four conditions. Approximate deflections confirm their insignificantly yielding behavior under the GSA loading.

Table 2 Maximum deflection under the GSA loading (cm) Method Elastic static Nonlinear static Nonlinear dynamic Case 1B 1.88 2.15 1.94 Case 2A 1.61 1.64 1.56 Case 1A 2.11 2.25 1.83 Case 2B 1.73 1.82 1.82

4. Progressive collapse resistance As expected, only minor damage is induced under the GSA static loading for the earthquake-resistant RC building. The progressive collapse resistance, which is defined as the ultimate downward loading capacity of the column-removed building, is further

3622

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

Fig. 3a. DCR values and nonlinear static plastic hinges for Case 1B.

the GSA loading are also marked with symbols in the figure. If expressed as a multiplier of 2(DL + 0.25LL), the maximum loading capacity is equal to 1.25Pst , 1.62Pst , 1.39Pst , and 1.39Pst for Case 1B, 2A, 1A, and 2B, respectively. It is realized that Case 1B has the smallest progressive collapse resistance as compared to others. However, since the dynamic effect is not considered in the nonlinear static analysis, incremental dynamic analysis is carried out to verify the ultimate loading capacities. 4.2. Nonlinear dynamic method A series of nonlinear time history analyses under different dynamic loadings are conducted. A step function multiplied by (DL + 0.25LL) is used to simulate the dynamic loading applied to the column-removed building. The magnitude of the step function is increased gradually till extremely large deflection occurs at the column-removed point. P effect and large displacement are considered in the nonlinear dynamic analysis. Typical displacement time histories under (DL + 0.25LL) and 2(DL + 0.25LL) for Case 1B are shown in Fig. 5. The peak displacement response of each time history is collected to construct the loaddisplacement envelopes for the incremental dynamic analysis, as shown in Fig. 6. It is seen that the ultimate loading capacities are developed at the deflections approximated to that obtained from the nonlinear static analysis. This is due to the fact that identical plastic hinge models are adopted for the nonlinear static and nonlinear dynamic analyses. The estimated collapse resistance is about 2.15Pdy , 2.75Pdy , 2.4Pdy , and 2.4Pdy , respectively for Case 1B, 2A, 1A, and 2B. Table 3 summarizes the progressive collapse resistance obtained from the nonlinear static and nonlinear dynamic analyses. 4.3. Dynamic amplification factor From Table 3, it is seen that larger loading capacities are obtained from the nonlinear static analysis than from the nonlinear dynamic analysis. This difference may be attributed to the

Fig. 3b. Elastic moment distribution for Case 1B.

estimated using the nonlinear static and the incremental dynamic analysis method. 4.1. Nonlinear static method Displacement-controlled procedure with a maximum deflection of 20 cm is used to investigate the collapse resistance of the column-removed building. Fig. 4 shows the loaddisplacement responses of those four column-removed conditions. The ordinate represents the exerted loading in terms of the percentage of (DL + 0.25LL), while the abscissa represents the deflection at the column-removed point. The loaddisplacement responses under

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

3623

Fig. 3c. DCR values and nonlinear dynamic plastic hinges for Case 1B. Table 3 Loaddisplacement response and DAF for collapse resistance Method Nonlinear static Nonlinear dynamic DAF
a

Case 1B (2.50, 12.84) (2.15, 12.33) 1.16

a

Case 2A (3.24, 12.92) (2.75, 12.30) 1.18

Case 1A (2.78, 12.13) (2.4, 12.61) 1.16

Case 2B (2.78, 12.82) (2.4, 12.51) 1.16

Loading multiplier of (DL + 0.25LL), deflection in cm.

Fig. 4. Loaddisplacement curves from the nonlinear static analyses.

Fig. 5. Typical displacement time histories for Case 1B.

dynamic effect. The DAF is usually defined as the ratio of the dynamic displacement response (dy ) of an elastic single-degreeof-freedom (SDOF) system to its static displacement response (st ) under an equal applied load. This displacement-based definition results in a larger DAF than 2.0 for elasticplastic systems [2022]. Consider st = Pst /kst and dy = Pdy /kdy , st = dy leads to DAF = kst /kdy = Pst /Pdy (3)

where Pst and Pdy are, respectively, the required static and dynamic force under the same deflection. kst and kdy represent the equivalent static and dynamic stiffness of the SDOF system,

respectively. Hence, the DAF may be expressed as the ratio of the static force response to the dynamic force response under an equal displacement demand, as explained by Fig. 7. Based on this force-based definition, the DAFs for those four column-removed conditions are obtained as shown in Fig. 8. It is seen that the DAF decreases with increasing displacement of the column-removed point. As shown in Table 3, under the displacement demand of the ultimate loading, the DAF may decrease to 1.16, which is similar to that indicated by Ruth et al. [23]. This reflects that, based on a DAF of 2, the displacement-controlled, nonlinear static analysis method may lead to a conservative estimation of progressive collapse resistance for a column-removed building.

3624

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

the ability of the GSA linear static procedure for capturing the nonlinear static behavior in the significantly yielding phase is examined. Also, application of the DAF 2.0 to the GSA linear static analysis for a significantly yielding building is evaluated. 5.1. Nonlinear static maximum loading The GSA linear static analyses are performed for the columnremoved building subjected to the maximum loadings estimated by the nonlinear static method. According to the GSA guidelines, moments of the member ends with DCR > 2.0 are released and the linear static analysis is repeated until no DCR value is larger than 2. Fig. 9a shows the final DCR values and static plastic hinges for Case 1B, where only three DCR values are larger than 2. Performance levels of the plastic hinges are presented by the shading of the hinge circles. There are three moment-released hinges at the final run of the linear static analysis. However, a significantly yielding mechanism is observed from the nonlinear static analysis. The maximum deflection obtained from the GSA linear procedure is 2.6 cm, which is less than 12.84 cm from the nonlinear static analysis. Similar phenomena are revealed by other column-removed conditions, as shown in Figs. 9b9d. The linear static approach fails to simulate the nonlinear static response in the significantly yielding phase. 5.2. Nonlinear dynamic maximum loading A hinge rotation angle equal to 6 is recommended as the nonlinear acceptance criterion for RC beams in the GSA guidelines. According to the hinge model adopted in this study, all the plastic hinge rotation angles are far less than the allowable value under the nonlinear dynamic maximum loadings. Thus, based on the GSA nonlinear criterion, progressive collapse may be limited for the four column-removed conditions. Based on a DAF of 2, the dynamically determined maximum loadings are doubled to conduct the GSA linear static analyses. This means that the multiplier of the loading combination (DL + 0.25LL) is increased to 4.3, 5.5, 4.8, and 4.8, respectively for Case 1B, 2A, 1A, and 2B. Contrary to the result assessed by the GSA nonlinear acceptance criterion, progressive collapse occurs for all four conditions. According to the criterion of removing failed members (three-hinges failure mechanism), all those adjacent bays interconnected at the column-removed line collapse progressively. Fig. 10 shows the final DCR values and the plastic hinges, respectively, obtained from the GSA linear static and the nonlinear dynamic analysis for Case 1B. The plastic hinges are classified as life safety by the hinge model. Therefore, a discrepancy between the results assessed by the GSA linear static method and by its nonlinear acceptance criterion may arise under an extreme downward loading. The DAF of 2 appears to be invalid as the RC building undergoes significantly yielding behavior. 6. Prediction of the collapse resistance
Fig. 8. Variation of the DAF for the column-removed conditions.

Fig. 6. Loaddisplacement curves from the incremental dynamic analyses.

Fig. 7. Diagrammatic explanation of the dynamic amplification factor.

5. Comparison of inelastic responses In the GSA linear static procedure, beam sections with DCRs larger than 2.0 are replaced with inserted hinges to simulate the inelastic response of the column-removed building under vertical downward loadings. This implies that elasticperfectly plastic models are assumed for the inserted hinges. In this section,

Nonlinear incremental dynamic analysis is usually a promising method for estimating the progressive collapse resistance. Nevertheless, it is time-consuming to perform the incremental dynamic analysis for the maximum loading capacity. As a consequence, an efficient approach may be needed to predict the progressive collapse resistance precisely. The area below the nonlinear static loaddisplacement curve represents the absorbed energy of the building subjected to the downward loading. A capacity curve may be constructed

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

3625

Fig. 9a. DCR values and static plastic hinges for Case 1B under 2.5(DL + 0.25LL).

Fig. 9b. DCR values and static plastic hinges for Case 2A under 3.24(DL + 0.25LL).

by dividing the accumulated stored energy by its corresponding displacement [12]. It is mathematically expressed as PCC (ud ) = 1 ud
0 ud

PNS (u)du

(4)

where PCC (u) and PNS (u) are, respectively, the capacity function and the nonlinear static loading estimated at the displacement demand u. Hence, the capacity curve of the column-removed building is numerically obtained from dividing the accumulated area under the nonlinear static loaddisplacement curve by

the corresponding displacement of the column-removed point. Figs. 11a11d show the capacity curves of those four conditions along with the associated nonlinear static and incremental dynamic loaddisplacement curves. It is observed that the capacity curve approximately coincides with the variation of the incremental dynamic curve up to the maximum loading capacity. Therefore, the capacity curve may be used to predict the progressive collapse resistance of the column-removed building. The predicted collapse resistances are determined by the maximum value of the capacity curve and summarized in Table 4.

3626

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

Fig. 9c. DCR values and static plastic hinges for Case 1A under 2.78(DL + 0.25LL).

Fig. 9d. DCR values and static plastic hinges for Case 2B under 2.78(DL + 0.25LL).

It is seen that the estimated values agree well with those obtained from the incremental dynamic analyses. Meanwhile, the DAF may be estimated from dividing the nonlinear static loaddisplacement curve by the corresponding capacity curve up to the maximum loading, as shown in Fig. 12. 7. Conclusions The vulnerability of an earthquake-resistant RC building to progressive collapse is evaluated in this study. According to the

Table 4 Predicted maximum loadings Case 1B 2A 1A 2B Multiplier of (DL + 0.25LL) 2.16 2.78 2.38 2.40

GSA linear static analysis approach, the building has a low potential for progressive collapse. Nonlinear analysis results reveal that

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628

3627

Fig. 10. DCR values and dynamic plastic hinges for Case 1B under Pdy 2.15(DL + 0.25LL). The dash lines represent the collapsed members.

Fig. 11a. The capacity curve of Case 1B. Fig. 11c. The capacity curve of Case 1A.

Fig. 11b. The capacity curve of Case 2A. Fig. 11d. The capacity curve of Case 2B.

the flexural ductility of beam members may be developed to some extent in the GSA linear static analysis procedure. Moreover,

different progressive collapse resistances are obtained by the nonlinear static and the incremental dynamic method. Different

3628

M.-H. Tsai, B.-H. Lin / Engineering Structures 30 (2008) 36193628 [3] Ellingwood BR. Mitigating risk from abnormal loads and progressive collapse. J Perform Constr Fac 2006;20(4):31523. [4] Dusenberry DO. Review of existing guidelines and provisions related to progressive collapse. In: Workshop on prevention of progressive collapse. Washington (DC): National Institute of Building Sciences; 2002. [5] Ellingwood BR, Smilowitz R, Dusenberry DO, Duthinh D, Lew HS, Carino NJ. Best practices for reducing the potential for progressive collapse in buildings. Report no. NISTIR-7396. National Institute of Standards and Technology, Technology Administration, US Department of Commerce; 2007. [6] Starossek U. Progressive collapse of structures: Nomenclature and procedures. IABSE, Structural Engineering International 2006;16(2):1137. [7] Starossek U, Wolff M. Progressive collapse: Design strategies. In: IABSE symposium, structures and extreme events. 2005. [8] Marjanishvili SM. Progressive analysis procedure for progressive collapse. J Perform Constr Fac ASCE 2004;18(2):7985. [9] Marjanishvili S, Agnew E. Comparison of various procedures for progressive collapse analysis. J Perform Constr Fac ASCE 2006;20(4):36574. [10] GSA, Progressive collapse analysis and design guidelines for new federal office buildings and major modernization projects. General Service Administration. 2003. [11] Department of Defense (DoD) Unified facilities criteria (UFC): Design of buildings to resist progressive collapse, UFC 4-023-03, U. S. DoD. 2005. [12] Abruzzo J, Matta A, Panariello G. Study of mitigation strategies for progressive collapse of a reinforced concrete commercial building. J Perform Constr Fac ASCE 2006;20(4):38490. [13] Corley WG. Applicability of seismic design in mitigating progressive collapse. In: Proceedings of workshop on prevention of progressive collapse. Washington (DC): National Institute of Building Sciences; 2002. [14] Corley WG, Mlakar Sr PF, Sozen MA, Thornton CH. The Oklahoma city bombing: Summary and recommendations for multihazard mitigation. J Perform Constr Fac 1998;12(3):10012. [15] Hayes Jr JR, Woodson SC, Pekelnicky RG, Poland CD, Corley WG, Sozen M. Can strengthening for earthquake improve blast and progressive collapse resistance?. J Struct Eng, ASCE 2005;131(8):115777. [16] SAP2000 r Version 8.0. Linear and nonlinear static and dynamic analysis and design of three-dimensional structures. Berkeley (CA, USA): Computers and Structures Inc.; 2002. [17] FEMA. NEHRP guidelines for the seismic rehabilitation of buildings (FEMA Publication 273). Washington (DC): Federal Emergency Management Agency; 1997. [18] Lin BH. Progressive collapse analysis and evaluation of an earthquake-resistant RC building. Master thesis. Taiwan: Department of Civil Engineering, National Pingtung University of Science and Technology: 2007. [19] Sucuoglu H, Citipitioglu E, Altin S. Resistance mechanisms in rc building frames subjected to column failure. J Struct Eng ASCE 1994;120(3):76582. [20] Kaewkulchai G, Williamson EB. Beam element formulation and solution procedure for dynamic progressive collapse analysis. Comput Struct 2004;82: 63951. [21] Kaewkulchai G, Williamson EB. Dynamic behavior of planar frame during progressive collapse. In: 16th ASCE engineering mechanics conference. 2003. [22] Pretlove AJ, Ramsden M, Atkins AG. Dynamic effects in progressive failure of structures. Int J Impact Eng 1991;11(4):53946. [23] Ruth P, Marchand KA, Williamson EB. Static equivalency in progressive collapse alternate path analysis: Reducing conservatism while retaining structural integrity. J Perform Constr Fac 2006;20(4):34964.

Fig. 12. Estimated DAF from the capacity curves.

acceptance criteria for determining the collapse resistance may be adopted for these two methods. The nonlinear static analysis gives a conservative estimation for the collapse resistance if the dynamic amplification factor (DAF) is equal to 2.0. Based on a force-based definition, the DAF is less than 2.0 as the flexural ductility is developed. Also, the linear static analysis procedure and the nonlinear acceptance criterion suggested by the GSA guidelines may have different evaluation results as the columnremoved building is loaded into significantly yielding phase. Under this condition, a different DAF from 2.0 is needed for the GSA loading combination to account for the inelastic dynamic effect. The capacity curve constructed from the nonlinear static analysis is shown to be capable of predicting the progressive collapse resistance and the DAF for a RC building subjected to column failure. Acknowledgments The writers wish to express their sincere appreciation to the reviewers for very constructive comments. The study presented in this paper was supported by the National Science Council of Taiwan under Grants NSC 95-2221-E-020-035. The support is gratefully acknowledged. References
[1] Mohamed OA. Progressive collapse of structures: Annotated bibliography and comparison of codes and standards. J Perform Constr Fac 2006;20(4):41825. [2] Nair RS. Preventing disproportionate collapse. J Perform Constr Fac 2006; 20(4):30914.