Lessons learned from forensic FEA of failed RC structures24 Oct 2010
One of the things I’ve enjoyed the most about my time in graduate school at Georgia Tech is the opportunity I’ve had to support the consulting activities of several professors by conducting numerical simulations. The presentation below, given at the ACI Fall 2010 Convention in Pittsburgh, is a summary of several key lessons learned from forensic FEA of failed RC structures that we feel would directly benefit practicing design engineers.
If you were there at the talk today, thanks for stopping by.
You can download my slides here:
Here is the full abstract:
The application of the finite element method in the design of reinforced concrete systems has gained signifcant momentum in engineering practice. While various finite element based design techniques have been widely implemented in general purpose structural engineering software, few practical guidelines have been published to equip engineers to navigate practical analysis issues. Additionally, as practicing engineers face pressures of time and economy, simplifying assumptions are often employed in practice which may lead to glaring omissions in accurately predicting structural behavior. In this presentation, three case studies from the authors’ forensic consulting experience are presented which highlight various ways in which simplifying assumptions in finite element analysis of reinforced concrete structures during the design process failed to capture key behavioral phenomena, resulting in significant damage, poor serviceability, and/or collapse of the constructed structure. Case studies presented include: (a) a composite RC/steel pedestrian bridge which collapsed during placement of concrete, (b) a reinforced concrete building structure which suffered severe cracking due to shrinkage, as well as (c) a reinforced concrete industrial structure which detected excessively due to differential support settlement. For each case, practical techniques using general purpose linear elastic finite element analysis software are demonstrated which could have easily predicted each damaged state at little increase in computational effort. Techniques are presented to characterize the implications of simplifying assumptions in analysis of RC structures as well as to equip engineers to determine when various structural phenomena must be explicitly evaluated in structural models of reinforced concrete systems.