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 revisits the topic of earthquake recovery in Christchurch City more than a decade after the Canterbury earthquakes. Despite promising visions of a community reconnected and a sustainable and liveable city, significant portions of the city’s core – the Red Zone – remain dilapidated and “eerily empty”. At the same time, new developments in other areas have proven to be alienated or underutilised. Currently, the Canterbury Earthquake Recovery Authority’s plans for the rebuilding highlight the delivery of more residential housing to re-populate the city centre. However, prevalent approaches to housing development in Christchurch are ineffective for building an inclusive and active community. Hence, the central inquiry of the thesis is how the development of housing complexes can revitalise the Red Zone within the Christchurch city centre. The inquiry has been carried out through a research-through-design methodology, recognising the importance of an in-depth investigation that is contextualised and combined with the intuition and embodied knowledge of the designer. The investigation focuses on a neglected site in the Red Zone in the heart of Christchurch city, with significant Victorian and Edwardian Baroque heritage buildings, including Odeon Theatre, Lawrie & Wilson Auctioneers, and Sol Square, owned by The Regional Council Environment Canterbury. The design inquiry argues, develops, and is carried through a place-assemblage lens to housing development for city recovery, which recognizes the significance of socially responsive architecture that explores urban renewal by forging connections within the social network. Therefore, place-assemblage criteria and methods for developing socially active and meaningful housing developments are identified. Firstly, this thesis argues that co-living housing models are more focused on people relations and collective identity than the dominant developer-driven housing rebuilds, as they prioritise conduits for interaction and shared social meaning and practices. Secondly, the adaptive reuse of derelict heritage structures is proposed to reinvigorate the urban fabric, as heritage is seen to be conceived as and from a social assemblage of people. The design is realised by the principles outlined in the ICOMOS charter, which involves incorporating the material histories of existing structures and preserving the intangible heritage of the site by ensuring the continuity of cultural practices. Lastly, design processes and methods are also vital for place-sensitive results, which pay attention to the site’s unique characteristics to engage with local stakeholders and communities. The research explores place-assemblage methods of photographic extraction, the drawing of story maps, precedent studies, assemblage maps, bricolages, and paper models, which show an assembly of layers that piece together the existing heritage, social conduits, urban commons and housing to conceptualise the social network within its place.
A review of the literature showed the lack of a truly effective damage avoidance solution for timber or hybrid timber moment resisting frames (MRFs). Full system damage avoidance selfcentring behaviour is difficult to achieve with existing systems due to damage to the floor slab caused by beam-elongation. A novel gravity rocking, self-centring beam-column joint with inherent and supplemental friction energy dissipation is proposed for low-medium rise buildings in all seismic zones where earthquake actions are greater than wind. Steel columns and timber beams are used in the hybrid MRF such that both the beam and column are continuous thus avoiding beam-elongation altogether. Corbels on the columns support the beams and generate resistance and self-centring through rocking under the influence of gravity. Supplemental friction sliders at the top of the beams resist sliding of the floor whilst dissipating energy as the floor lifts on the corbels and returns. 1:20 scale tests of 3-storey one-by-two bay building based on an earlier iteration of the proposed concept served as proof-of-concept and highlighted areas for improvement. A 1:5 scale 3-storey one-by-one bay building was subsequently designed. Sub-assembly tests of the beam-top asymmetric friction sliders demonstrated repeatable hysteresis. Quasi-static tests of the full building demonstrated a ‘flat bottomed’ flag-shaped hysteresis. Shake table tests to a suite of seven earthquakes scaled for Wellington with site soil type D to the serviceability limit state (SLS), ultimate limit state (ULS) and maximum credible event (MCE) intensity corresponding to an average return period of 25, 500 and 2500 years respectively were conducted. Additional earthquake records from the 22 February 2011 Christchurch earthquakes we included. A peak drift of 0.6%, 2.5% and 3.8% was reached for the worst SLS, ULS and MCE earthquake respectively whereas a peak drift of 4.5% was reached for the worst Christchurch record for tests in the plane of the MRF. Bi-directional tests were also conducted with the building oriented at 45 degrees on the shake table and the excitation factored by 1.41 to maintain the component in the direction of the MRF. Shear walls with friction slider hold-downs which reached similar drifts to the MRF were provided in the orthogonal direction. Similar peak drifts were reached by the MRF in the bi-directional tests, when the excitation was amplified as intended. The building self-centred with a maximum residual drift of 0.06% in the dynamic tests and demonstrated no significant damage. The member actions were magnified by up to 100% due to impact upon return of the floor after uplift when the peak drift reached 4.5%. Nonetheless, all of the members and connections remained essentially linearelastic. The shake table was able to produce a limited peak velocity of 0.275 m/s and this limited the severity of several of the ULS, MCE and Christchurch earthquakes, especially the near-field records with a large velocity pulse. The full earthquakes with uncapped velocity were simulated in a numerical model developed in SAP2000. The corbel supports were modelled with the friction isolator link element and the top sliders were modelled with a multi-linear plastic link element in parallel with a friction spring damper. The friction spring damper simulated the increase in resistance with increasing joint rotation and a near zero return stiffness, as exhibited by the 1:5 scale test building. A good match was achieved between the test quasi-static global force-displacement response and the numerical model, except a less flat unloading curve in the numerical model. The peak drift from the shake table tests also matched well. Simulations were also run for the full velocity earthquakes, including vertical ground acceleration and different floor imposed load scenarios. Excessive drift was predicted by the numerical model for the full velocity near-field earthquakes at the MCE intensity and a rubber stiffener for increasing the post joint-opening stiffness was found to limit the drift to 4.8%. Vertical ground acceleration had little effect on the global response. The system generates most of its lateral resistance from the floor weight, therefore increasing the floor imposed load increased the peak drift, but less than it would if the resistance of the system did not increase due to the additional floor load. A seismic design procedure was discussed under the framework of the existing direct displacement-based design method. An expression for calculating the area-based equivalent viscous damping (EVD) was derived and a conservative correction factor of 0.8 was suggested. A high EVD of up to about 15% can be achieved with the proposed system at high displacement ductility levels if the resistance of the top friction sliders is maximised without compromising reliable return of the floor after uplift. Uniform strength joints with an equal corbel length up the height of the building and similar inter-storey drifts result in minimal relative inter-floor uplift, except between the first floor and ground. Guidelines for detailing the joint for damage avoidance including bi-directional movement were also developed.
