A poster created by Empowered Christchurch to advertise their submission to the CERA Draft Transition Recovery Plan on social media.The poster reads, "Submission, CERA Draft Transition Recovery Plan. Seismic Risk. One thing we can learn from the past is that seismic risk in Canterbury has been underestimated before the earthquakes struck. This is confirmed in a report for EQC in 1991 (paper 2005). It is also the conclusion of the Royal Commission in the CTV report. A number of recommendations have been made but not followed. For example, neither the AS/NZS 1170.5 standard nor the New Zealand Geotechnical Society guidelines have been updated. Yet another recovery instrument is the Earthquake Prone Building Act, which is still to be passed by Parliament. As the emergency response part of the recovery is now behind us, we need to ensure sustainability for what lies ahead. We need a city that is driven by the people that live in it, and enabled by a bureaucracy that accepts and mitigates risks, rather than transferring them to the most vulnerable residents."
This thesis presents the application of data science techniques, especially machine learning, for the development of seismic damage and loss prediction models for residential buildings. Current post-earthquake building damage evaluation forms are developed for a particular country in mind. The lack of consistency hinders the comparison of building damage between different regions. A new paper form has been developed to address the need for a global universal methodology for post-earthquake building damage assessment. The form was successfully trialled in the street ‘La Morena’ in Mexico City following the 2017 Puebla earthquake. Aside from developing a framework for better input data for performance based earthquake engineering, this project also extended current techniques to derive insights from post-earthquake observations. Machine learning (ML) was applied to seismic damage data of residential buildings in Mexico City following the 2017 Puebla earthquake and in Christchurch following the 2010-2011 Canterbury earthquake sequence (CES). The experience showcased that it is readily possible to develop empirical data only driven models that can successfully identify key damage drivers and hidden underlying correlations without prior engineering knowledge. With adequate maintenance, such models have the potential to be rapidly and easily updated to allow improved damage and loss prediction accuracy and greater ability for models to be generalised. For ML models developed for the key events of the CES, the model trained using data from the 22 February 2011 event generalised the best for loss prediction. This is thought to be because of the large number of instances available for this event and the relatively limited class imbalance between the categories of the target attribute. For the CES, ML highlighted the importance of peak ground acceleration (PGA), building age, building size, liquefaction occurrence, and soil conditions as main factors which affected the losses in residential buildings in Christchurch. ML also highlighted the influence of liquefaction on the buildings losses related to the 22 February 2011 event. Further to the ML model development, the application of post-hoc methodologies was shown to be an effective way to derive insights for ML algorithms that are not intrinsically interpretable. Overall, these provide a basis for the development of ‘greybox’ ML models.
This thesis investigates life-safety risk in earthquakes. The first component of the thesis utilises a dataset of earthquake injuries and deaths from recent earthquakes in New Zealand to identify cause, context, and risk factors of injury and death in the 2011 MW6.3 Christchurch earthquake and 2016 MW7.8 Kaikōura earthquake. Results show that nearly all deaths occurred from being hit by structural elements from buildings, while most injuries were caused by falls, strains and being hit by contents or non-structural elements. Statistical analysis of injured cases compared to an uninjured control group found that age, gender, building damage, shaking intensity, and behaviour during shaking were the most significant risk factors for injury during these earthquakes. The second part of the thesis uses the empirical findings from the first section to develop two tools for managing life-safety risk in earthquakes. The first tool is a casualty estimation model for health system and emergency response planning. An existing casualty model used in New Zealand was validated against observed data from the 2011 Christchurch earthquake and found to underestimate moderate and severe injuries by an order of magnitude. The model was then updated to include human behaviour such as protective actions, falls and strain type injuries that are dependent on shaking intensity, as well as injuries and deaths outside buildings. These improvements resulted in a closer fit to observed casualties for the 2011 Christchurch earthquake. The second tool that was developed is a framework to set seismic loading standards for design based on fatality risk targets. The proposed framework extends the risk-targeted hazard method, by moving beyond collapse risk targets, to fatality risk targets for individuals in buildings and societal risk in cities. The framework also includes treatment of epistemic uncertainty in seismic hazard to allow this uncertainty to be used in risk-based decision making. The framework is demonstrated by showing how the current New Zealand loading standards could be revised to achieve uniform life-safety risk across the country and how the introduction of a new loading factor can reduce risk aggregation in cities. Not on Alma, moved and emailed. 1/02/2023 ce
