Development of a Seismic Damage Prediction Model Using Machine Learning

Research papers, University of Canterbury Library

Natural catastrophes are increasing worldwide. They are becoming more frequent but also more severe and impactful on our built environment leading to extensive damage and losses. Earthquake events account for the smallest part of natural events; nevertheless seismic damage led to the most fatalities and significant losses over the period 1981-2016 (Munich Re). Damage prediction is helpful for emergency management and the development of earthquake risk mitigation projects. Recent design efforts focused on the application of performance-based design engineering where damage estimation methodologies use fragility and vulnerability functions. However, the approach does not explicitly specify the essential criteria leading to economic losses. There is thus a need for an improved methodology that finds the critical building elements related to significant losses. The here presented methodology uses data science techniques to identify key building features that contribute to the bulk of losses. It uses empirical data collected on site during earthquake reconnaissance mission to train a machine learning model that can further be used for the estimation of building damage post-earthquake. The first model is developed for Christchurch. Empirical building damage data from the 2010-2011 earthquake events is analysed to find the building features that contributed the most to damage. Once processed, the data is used to train a machine-learning model that can be applied to estimate losses in future earthquake events.

Validation of ground motion simulations via response history analysis of s…

Research papers, University of Canterbury Library

Validation is an essential step to assess the applicability of simulated ground motions for utilization in engineering practice, and a comprehensive analysis should include both simple intensity measures (PGA, SA, etc), as well as the seismic response of a range of complex systems obtained by response history analysis. In order to enable a spectrum of complex structural systems to be considered in systematic validation of ground motion simulations in a routine fashion, an automated workflow was developed. Such a workflow enables validation of simulated ground motions in terms of different complex model responses by considering various ground motion sets and different ground motion simulation methods. The automated workflow converts the complex validation process into a routine one by providing a platform to perform the validation process promptly as a built-in process of simulation post-processing. As a case study, validation of simulated ground motions was investigated via the automated workflow by comparing the dynamic responses of three steel special moment frame (SMRF) subjected to the 40 observed and 40 simulated ground motions of 22 February 2011 Christchurch earthquake. The seismic responses of the structures are principally quantified via the peak floor acceleration and maximum inter-storey drift ratio. Overall, the results indicate a general agreement in seismic demands obtained using the recorded and simulated ensembles of ground motions and provide further evidence that simulated ground motions can be used in code-based structural performance assessments in-place of, or in combination with, ensembles of recorded ground motions.