Unreinforced masonry (URM) structures comprise a majority of the global built heritage. The masonry heritage of New Zealand is comparatively younger to its European counterparts. In a country facing frequent earthquakes, the URM buildings are prone to extensive damage and collapse. The Canterbury earthquake sequence proved the same, causing damage to over _% buildings. The ability to assess the severity of building damage is essential for emergency response and recovery. Following the Canterbury earthquakes, the damaged buildings were categorized into various damage states using the EMS-98 scale. This article investigates machine learning techniques such as k-nearest neighbors, decision trees, and random forests, to rapidly assess earthquake-induced building damage. The damage data from the Canterbury earthquake sequence is used to obtain the forecast model, and the performance of each machine learning technique is evaluated using the remaining (test) data. On getting a high accuracy the model is then run for building database collected for Dunedin to predict expected damage during the rupture of the Akatore fault.
Recent global tsunami events have highlighted the importance of effective tsunami risk management strategies (including land-use planning, structural and natural defences, warning systems, education and evacuation measures). However, the rarity of tsunami means that empirical data concerning reactions to tsunami warnings and tsunami evacuation behaviour is rare when compared to findings about evacuations to avoid other sources of hazard. To date empirical research into tsunami evacuations has focused on evacuation rates, rather than other aspects of the evacuation process. More knowledge is required about responses to warnings, pre-evacuation actions, evacuation dynamics and the return home after evacuations. Tsunami evacuation modelling has the potential to inform evidence-based tsunami risk planning and response. However to date tsunami evacuation models have largely focused on timings of evacuations, rather than evacuation behaviours. This Masters research uses a New Zealand case study to reduce both of these knowledge gaps. Qualitative survey data was gathered from populations across coastal communities in Banks Peninsula and Christchurch, New Zealand, required to evacuate due to the tsunami generated by the November 14th 2016 Kaikōura Earthquake. Survey questions asked about reactions to tsunami warnings, actions taken prior to evacuating and movements during the 2016 tsunami evacuation. This data was analysed to characterise trends and identify factors that influenced evacuation actions and behaviour. Finally, it was used to develop an evacuation model for Banks Peninsula. Where appropriate, the modelling inputs were informed by the survey data. Three key findings were identified from the results of the evacuation behaviour survey. Although 38% of the total survey respondents identified the earthquake shaking as a natural cue for the tsunami, most relied on receiving official warnings, including sirens, to prompt evacuations. Respondents sought further official information to inform their evacuation decisions, with 39% of respondents delaying their evacuation in order to do so. Finally, 96% of total respondents evacuated by car. This led to congestion, particularly in more densely populated Christchurch city suburbs. Prior to this research, evacuation modelling had not been completed for Banks Peninsula. The results of the modelling showed that if evacuees know how to respond to tsunami warnings and where and how to evacuate, there are no issues. However, if there are poor conditions, including if people do not evacuate immediately, if there are issues with the roading network, or if people do not know where or how to evacuate, evacuation times increase with there being more bottlenecks leading out of the evacuation zones. The results of this thesis highlight the importance of effective tsunami education and evacuation planning. Reducing exposure to tsunami risk through prompt evacuation relies on knowledge of how to interpret tsunami warnings, and when, where and how to evacuate. Recommendations from this research outline the need for public education and engagement, and the incorporation of evacuation signage, information boards and evacuation drills. Overall these findings provide more comprehensive picture of tsunami evacuation behaviour and decision making based on empirical data from a recent evacuation, which can be used to improve tsunami risk management strategies. This empirical data can also be used to inform evacuation modelling to improve the accuracy and realism of the evacuation models.
Post-earthquake cordons have been used after seismic events around the world. However, there is limited understanding of cordons and how contextual information of place such as geography, socio-cultural characteristics, economy, institutional and governance structure etc. affect decisions, operational procedures as well as spatial and temporal attributes of cordon establishment. This research aims to fill that gap through a qualitative comparative case study of two cities: Christchurch, New Zealand (Mw 6.2 earthquake, February 2011) and L’Aquila, Italy (Mw 6.3 earthquake, 2009). Both cities suffered comprehensive damage to its city centre and had cordons established for extended period. Data collection was done through purposive and snowball sampling methods whereby 23 key informants were interviewed in total. The interviewee varied in their roles and responsibilities i.e. council members, emergency managers, politicians, business/insurance representatives etc. We found that cordons were established to ensure safety of people and to maintain security of place in both the sites. In both cities, the extended cordon was met with resistance and protests. The extent and duration of establishment of cordon was affected by recovery approach taken in the two cities i.e. in Christchurch demolition was widely done to support recovery allowing for faster removal of cordons where as in L’Aquila, due to its historical importance, the approach to recovery was based on saving all the buildings which extended the duration of cordon. Thus, cordons are affected by site specific needs. It should be removed as soon as practicable which could be made easier with preplanning of cordons.
Rapid, accurate structural health monitoring (SHM) assesses damage to optimise decision-making. Many SHM methods are designed to track nonlinear stiffness changes as damage. However, highly nonlinear pinched hysteretic systems are problematic in SHM. Model-based SHM often fails as any mismatch between model and measured response dynamics leads to significant error. Thus, modelfree methods of hysteresis loop tracking methods have emerged. This study compares the robustness and accuracy in the presence of significant measurement noise of the proven hysteresis loop analysis (HLA) SHM method with 3 emerging model-free methods and 2 further novel adaptations of these methods using a highly nonlinear, 6-story numerical structure to provide a known ground-truth. Mean absolute errors in identifying a known nonlinear stiffness trajectory assessed at four points over two successive ground motion inputs from September 2010 and February 2011 in Christchurch range from 1.71-10.52%. However, the variability is far wider with maximum errors ranging from 3.90-49.72%, where the second largest maximum absolute error was still 19.74%. The lowest mean and maximum absolute errors were for the HLA method. The next best method had mean absolute error of 2.92% and a maximum of 10.51%. These results show the clear superiority of the HLA method over all current emerging model-free methods designed to manage the highly nonlinear pinching responses common in reinforced concrete structures. These results, combined with high robustness and accuracy in scaled and fullscale experimental studies, provide further validation for using HLA for practical implementation.
High-quality ground motion records are required for engineering applications including response history analysis, seismic hazard development, and validation of physics-based ground motion simulations. However, the determination of whether a ground motion record is high-quality is poorly handled by automation with mathematical functions and can become prohibitive if done manually. Machine learning applications are well-suited to this problem, and a previous feed-forward neural network was developed (Bellagamba et al. 2019) to determine high-quality records from small crustal events in the Canterbury and Wellington regions for simulation validation. This prior work was however limited by the omission of moderate-to-large magnitude events and those from other tectonic environments, as well as a lack of explicit determination of the minimum usable frequency of the ground motion. To address these shortcomings, an updated neural network was developed to predict the quality of ground motion records for all magnitudes and all tectonic sources—active shallow crustal, subduction intraslab, and subduction interface—in New Zealand. The predictive performance of the previous feed-forward neural network was matched by the neural network in the domain of small crustal records, and this level of predictive performance is now extended to all source magnitudes and types in New Zealand making the neural network applicable to global ground motion databases. Furthermore, the neural network provides quality and minimum usable frequency predictions for each of the three orthogonal components of a record which may then be mapped into a binary quality decision or otherwise applied as desired. This framework provides flexibility for the end user to predict high-quality records with various acceptability thresholds allowing for this neural network to be used in a range of applications.
