Structural engineers rely heavily on predictions. Unlike other engineering products, building structures cannot make use of a prototype where the “bugs” can be ironed out. In a sense, each building is its own prototype. The dominant method in today’s practice for structural analysis is the Finite Element Analysis method (FEA).
All finite element models are a simplification of reality. Multiple factors affect how close the model and the results are to reality.
There are two types of FEA: linear and nonlinear. In today’s practice, the classic design and assessment of structures usually involves the development of a linear elastic model to calculate the global structural response. A second step may be needed for critical components in which nonlinear analysis may be required. Nonlinear analysis can serve as a valuable tool in forensic analysis supporting origin and cause investigations as well as assessing the expected behaviour of retrofitted structures, or in rationally selecting amongst various repair alternatives.
Reasons for caution when using nonlinear analysis
Unlike, say, structural analysis of a statically determinant structure, the application of nonlinear finite element analysis is hardly straightforward. This necessitates that models be developed and results examined with a healthy degree of skepticism and caution.
Sophisticated nonlinear finite element analysis software packages are likely to be used by only a small number of people with a specialized knowledge or interest. Analysis software and models should be validated or calibrated against tested specimens.
Analysts must be mindful of that no one approach performs well over the entire range of varying geometrical and loading conditions encountered in practice. Therefore, analysts should carefully select the appropriate approach for the unique situation.
Engineering judgment is a key in nonlinear finite element analysis. Decisions made with respect to selection of material behaviour models, mesh layout, type of element used, support conditions, method of loading, convergence criteria, and selection of material behaviour model will produce a range of results. Prediction competitions between experienced analysts show a variation of the prediction results. More variation is expected with predictions of deflections and deformations just before failure, compared to failure load predictions. Analysts must be aware of the possible effects of each model parameter. A study of the sensitivity of the results to a certain parameter should be insightful.
Crucially, nonlinear finite element analysis is known to yield large amounts of data. Analysts must be aware of what to look for and how to interpret it. Even with experienced analysts, the possibility for misinterpretation cannot be totally eliminated.
In conclusion, nonlinear finite element analysis is a great tool but should only be used by those who understand its capabilities and limitations and approach it with a healthy degree of skepticism and caution.
VANCOUVER TORONTO OTTAWA