Professional Documents
Culture Documents
Abstract: Following the collapse of the World Trade Center towers in September 2001, there has been heightened interest among
building owners and government entities in evaluating the progressive collapse potential of existing buildings and in designing new
buildings to resist this type of collapse. The General Services Administration and Department of Defense have issued general guidelines
for evaluating a building’s progressive collapse potential. However, little detailed information is available to enable engineers to confi-
Downloaded from ascelibrary.org by GUELPH UNIVERSITY on 06/07/13. Copyright ASCE. For personal use only; all rights reserved.
dently perform a systematic progressive collapse analysis satisfying these guidelines. In this paper, we present four successively more
sophisticated analysis procedures for evaluating the progressive collapse hazard: linear-elastic static; nonlinear static; linear-elastic dy-
namic; and nonlinear dynamic. We discuss the advantages and disadvantages of each method. We conclude that the most effective analysis
procedure for progressive collapse evaluation incorporates the advantageous parts of all four procedures by systematically applying
increasingly comprehensive analysis procedures to confirm that the possibility of progressive collapse is high.
DOI: 10.1061/共ASCE兲0887-3828共2004兲18:2共79兲
CE Database subject headings: Structural failures; Collapse; Structural analysis; Nonlinear analysis; Dynamic analysis; New York;
New York City; Terrorism.
Our main objective is to formulate an easy-to-follow, comprehen- initiating damage; it includes all normal service loads, such as
sive analysis procedure that will, in most cases, yield reliable and dead and live loads. Therefore the load sequence is the key to
accurate results to estimate the likelihood of progressive collapse. accurately capturing the structural response in the analysis. It is
To attain this objective, we perform the following tasks: approximated as follows: at first, the building is loaded with dead
1. Description of progressive collapse phenomena; and live loads 共normal service loads兲 and then, suddenly, struc-
2. Review of current guidelines and state of the practice; tural bearing elements are removed. This could be represented
3. Evaluate and compare various analysis procedures; and mathematically as a sudden change in the stiffness matrix of the
4. Formulate the preferred method of analysis. structure. However, in this paper the ‘‘at rest’’ equilibrium of the
Major topics covered in this paper include: structure is represented by applying the internal reaction forces of
• Selection of the analysis procedure, the lost element to the supported structure. Initiating damage is
• Verification and validation of the results, and simulated by suddenly removing this internal force 共reaction兲 as-
• Evaluation of the results. sociated with the damaged element.
Dynamic Behavior
Description of the Phenomena
Sudden removal of an element in a structure causes an immediate
Progressive collapse includes two types of loadings: the load that geometric change in the structure, resulting in release of potential
causes the structural element to fail 共primary load兲 and the loads energy and rapid alteration of internal static dynamic forces, in-
that are generated due to the structural motions caused by sudden cluding inertia forces. Usually this energy is confined to the im-
collapse of the element 共secondary loads兲. External abnormal mediate vicinity of the damage. Since this behavior is localized,
loads, such as blast pressures due to explosive attacks, could care must be exercised when performing dynamic analysis; in
cause primary loads, while secondary loads result from internal particular, all high modes of vibration should be included when
static and dynamic loads and are caused by sudden changes in the using modal superposition or Ritz vector analysis methods. Direct
load path through the structure geometry. Although estimation of step-by-step integration methods are preferable, since such algo-
primary loads is important, this paper deals with the effects of the rithms account for all possible vibration modes associated with
secondary loads. the given finite element mesh and analysis time step 共Clough and
There are several approaches to evaluating progressive col- Penzien 1993; Wilson 2002兲.
lapse potential. The simplest is the indirect method, where general
design upgrades are implemented to enhance the overall robust- Nonlinear Behavior
ness of the structure 共Corley 2002兲. Direct methods include the
alternate load path method and the direct design method. For the True structural response to any loading, including internal dy-
alternate load path method, the loss of a primary structural com- namic loads, is nonlinear, and thus nonlinear behavior must be
ponent is posited and the ability of the surrounding structure to considered whenever feasible, as well as when required by exist-
remain intact is evaluated. For the direct design method, the ac- ing guidelines. However, nonlinear dynamic evaluation can be
tual loads that initiate collapse of a primary structural member are very time-consuming, even using today’s fastest computers.
used to evaluate whether collapse will be initiated. The alternate
load path evaluation is performed by removing one or several Structural Stability
major structural bearing elements 共i.e., introducing an initiating
damage兲 and analyzing the remaining structure to determine if Structural stability analysis is required to determine kinematical
this initiating damage propagates. We use the alternate load path stability of the damaged structure after the occurrence of the ini-
method in this paper. An advantage of this method is that it is tiating damage, as well as after the remaining structure is stabi-
independent of the initiating load, so that the solution is valid for lized. This analysis ensures that the structure maintains its kine-
any type of hazard that causes member loss. matical stability during and after the event.
Analysis methods used to evaluate the possibility of progres-
sive collapse vary widely, ranging from the simple two-
dimensional linear elastic static procedure to complex three- Review of Existing Guidelines
dimensional nonlinear time history analysis. The loss of the
bearing elements 共initiating damage兲 is modeled as a sudden dy- Two major guidelines available in the United States are the GSA’s
namic or quasistatic removal, depending on the method used. Progressive collapse analysis and design guidelines 共GSA 2000兲
Generally, an advanced structural analysis computer program and the DoD’s ‘‘Minimum antiterrorism standards for buildings’’
• Includes dynamic behavior, and include building a computer model and performing static and sta-
• Includes material nonlinear behavior. bility analysis. By exploiting similarities in the analysis proce-
The disadvantages are: dures and taking into consideration that engineering resources are
• Can be very time consuming, used as exhaustively and as efficiently as possible, we can formu-
• Requires extensive verification and validation of findings, late a preferred analysis method which includes the most infor-
• Hard to evaluate the results. In most of the cases the results of mative parts of all analysis methods and which should escalate
nonlinear dynamic analysis have to be verified and validated from simple to increasingly complex analysis processes as a vali-
independently. Independent peer review analysis, alternate dation of the results. As a result, we have an analysis procedure
modeling, and sensitivity studies could validate the accuracy which progresses from simple linear elastic static analysis to com-
of the analyses, plex nonlinear time history analysis, which we call the progres-
• High complexity, and sive analysis method.
• Incorrect assumptions or incorrect modeling may lead to erro- The progressive analysis method requires evaluating a struc-
neous results. ture for its vulnerability to progressive collapse against increas-
The limitation is: Nonlinear time history analysis can be very ingly realistic performance criteria through increasingly precise
time consuming, which may limit the number of nonlinearities in analysis. The process commences with a basic linear-elastic static
order to reduce the model and subsequently the computation time. analysis. If the building passes this step, whose evaluation re-
The steps required for performing the analysis are: quirements are the most conservative, then the analysis is com-
1. Build computer model; plete; if the building fails, we proceed in turn to increasingly
2. Determine ‘‘at rest’’ force distribution; complex linear and nonlinear analyses. At each step, we test the
3. Perform stability analysis; building’s performance against increasingly less conservative cri-
4. Estimate load time step; teria. The structural evaluation process is stopped after the build-
5. Estimate element capacities and force-displacement relation- ing meets evaluation criteria set forth for that particular analysis
ship; procedure, provided these simpler methods of analysis are appli-
6. Perform nonlinear time history analysis; cable to the complexity of the structure 共i.e., are allowed under
7. Perform validation and various sensitivity studies to verify the applicable FEMA/GSA/DoD Selection of Analysis Proce-
and validate the results; and dures Guidelines兲. This approach ensures that the computer model
8. Evaluate the results. is stable and the results are validated at every step of the analysis
The performance evaluation criteria used: Most realistic. by comparing and evaluating the results with the analysis results
obtained in previous steps. It ensures also that the results progress
in accordance with commonly used engineering practice 共i.e., me-
Analysis Procedures Summary Table
chanics of materials, structural kinematics, statics, and dynamics兲.
A summary of advantages and disadvantages of all four progres- Two major advantages of the progressive analysis procedure in-
sive collapse analysis procedures is presented in Table 1. clude:
• Progressive analysis methodology inherits the simplicity of
linear elastic static analysis methods, as well as the robustness
Preferred Method for Progressive Analysis of advanced analysis methods since it encompasses all avail-
able analysis procedures, and
As is implied by the preceding, more complicated and compre- • Since progressive collapse analysis evolves from simpler
hensive methods of analysis provide more accurate results. This analysis methods to a more complex methodology, the results
accuracy, however, comes with costs: meticulous engineering and can be easily validated by comparing the results of simpler
enormous computer resources, waiting sometimes weeks before analysis methods with more complicated analysis methods at
examining the output. For all methods, the greatest challenge is every step of the analysis process.
verifying and validating the results. In general, more complex
analysis methods require a more comprehensive verification and Conclusions and Recommendations
validation process. This may include alternate modeling and vari-
ous sensitivity studies. Simpler analysis methods, however, re- Progressive collapse is caused by a series of structural element
quire, at most, simplistic validation methods, such as hand calcu- failures due to large internal loads that exceed the elements’ bear-
lations and visual inspections of deflected shapes and force ing capacities. The internal loads are generated by sudden geom-
diagrams. Thus simpler analysis methods are preferred from a etry change of the structure and include inertia and damping
computational point of view. forces.
static 2. Calculations 2. Does not consider material nonlinearity cannot be evaluated with confidence 2. Perform static analysis
analysis done quickly 2. Limited to simple structures with 3. Perform stability analysis
3. Easy to perform predictable behavior 4. Verify, validate, and evaluate
4. Easy to evaluate the results
and validate results
Nonlinear 1. Includes material 1. Does not consider dynamic effects 1. Limited to relatively simple 1. Build computer model Very conservative
static nonlinear behavior 2. Relative complexity structures with predictable behavior 2. Perform stability analysis
analysis 3. Could be time consuming 2. Cannot be effectively used for 3. Estimate element capacities
4. Leads to overly conservative results progressive collapse analysis and force-displacement relationship
4. Perform nonlinear static analysis
5. Verify, validate, and evaluate the results
Linear-elastic 1. Includes 1. Does not account for material 1. Limited to structures that do not 1. Build computer model Conservative for structures
dynamic dynamic nonlinear behavior exhibit large plastic deformations 2. Perform static analysis to determine that exhibit nearly elastic
analysis behavior 2. Could be time consuming for internal forces behavior and could become
large computer models 3. Determine ‘‘at rest’’ force distribution nonconservative for structures
3. Moderate complexity 4. Estimate load time step that exhibit large plastic
4. Requires additional calculations 5. Perform time history analysis deformations
to obtain time-step and internal forces 6. Verify, validate, and evaluate
5. Dynamic amplification, inertia the results
and damping forces may be incorrectly
calculated for structures that exhibit large
plastic deformations
Nonlinear 1. Provides most 1. Could be very time consuming 1. Nonlinear time history analysis can 1. Build computer model Most realistic
dynamic realistic results 2. Requires extensive verification be very time consuming, which may 2. Determine ‘‘at rest’’ force distribution
analysis 2. Includes dynamic and validation of findings limit the number of nonlinearities in 3. Perform stability analysis
behavior 3. Hard to evaluate the results. order to reduce the model and 4. Estimate load time step
3. Includes material In most of the cases the results of subsequently the computation time 5. Estimate element capacities
nonlinear behavior nonlinear dynamic analysis have to be and force-displacement relationship
verified and validated independently. 6. Perform nonlinear time history analysis
Independent peer review analysis, 7. Verify, validate, and evaluate the results
alternate modeling, and sensitivity studies 8. Perform validation and various
could validate the accuracy of the analyses. sensitivity studies to verify the results
4. High complexity
5. Incorrect assumptions or incorrect
modeling may lead to erroneous results