Post-earthquake collapse prognostication of structural systems using sparse response data

Open Access
Kolli, Purna Chand
Graduate Program:
Civil Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 09, 2015
Committee Members:
  • Gordon Patrick Warn, Thesis Advisor
  • Collapse prognostication
  • sparse response data
  • genetic algorithms
  • parameter estimation
  • nonlinear structural systems
In order to reduce the impact of major earthquakes on communities, a rapid assessment of the state of structural systems for the purpose of re-occupancy of safe buildings is necessary. However, this determination is typically carried out through time-consuming visual inspections. Structural health monitoring is a promising alternative but existing techniques require a dense array of instrumentation on the structure rendering them impractical for many reasons including prohibitive costs of installation, maintenance, and data management and processing. A methodology is proposed in this study that uses sparse response data, for example the accelerations at a few floor levels, to identify the state of nonlinear numerical models that are further used for prognosticating collapse under future seismic hazards. The utility of the proposed methodology is demonstrated using the data from the shake table tests performed on a 4-story scaled model. The methodology is able to successfully predict the observed collapse of the physical structure. Furthermore, it has been observed that the vulnerability to collapse following an earthquake is greatly increased even to lower intensity earthquakes due to the accumulation of residual deformations and stresses in the structure.