Recurrent liquefaction in Christchurch during the 2010-2011 Canterbury earthquake sequence created a wealth of shallow subsurface intrusions with geometries and orientations governed by (1) strong ground motion severity and duration, and (2) intrinsic site characteristics including liquefaction susceptibility, lateral spreading severity, geomorphic setting, host sediment heterogeneity, and anthropogenic soil modifications. We present a suite of case studies that demonstrate how each of these characteristics influenced the geologic expressions of contemporary liquefaction in the shallow subsurface. We compare contemporary features with paleo-features to show how geologic investigations of recurrent liquefaction can provide novel insights into the shaking characteristics of modern and paleo-earthquakes, the influence of geomorphology on liquefaction vulnerability, and the possible controls of anthropogenic activity on the geologic record. We conclude that (a) sites of paleo-liquefaction in the last 1000-2000 years corresponded with most severe liquefaction during the Canterbury earthquake sequence, (b) less vulnerable sites that only liquefied in the strongest and most proximal contemporary earthquakes are unlikely to have liquefied in the last 1000-2000 years or more, (c) proximal strong earthquakes with large vertical accelerations favoured sill formation at some locations, (d) contemporary liquefaction was more severe than paleoliquefaction at all study sites, and (e) stratigraphic records of successive dike formation were more complete at sites with severe lateral spreading, (f) anthropogenic fill suppressed surface liquefaction features and altered subsurface liquefaction architecture.
In this paper, the characteristics of near-fault ground motions recorded during the Mw7.1 Darfield and Mw 6.2 Christchurch earthquakes are examined and compared with existing empirical models. The characteristics of forward-directivity effects are first examined using a wavelet-based pulse-classification algorithm. This is followed by an assessment of the adequacy of empirical models which aim to capture the effect of directivity effects on amplifying the acceleration response spectra; and the period and peak velocity of the forward-directivity pulse. It is illustrated that broadband directivity models developed by Somerville et al. (1997) and Abrahamson (2000) generally under-predict the observed amplification of response spectral ordinates at longer vibration periods. In contrast, a recently developed narrowband model by Shahi and Baker (2011) provides significantly improved predictions by amplifying the response spectra within a small range of periods surrounding the directivity pulse period. Although the empirical predictions of the pulse period are generally favourable for the Christchurch earthquake, the observations from the Darfield earthquake are significantly under-predicted. The elongation in observed pulse periods is inferred as being a result of the soft sedimentary soils of the Canterbury basin. However, empirical predictions of the observed peak velocity associated with the directivity pulse are generally adequate for both events.
Cats all over the world hunt wild animals and can contribute to the extinction of threatened species. In New Zealand, around half of all households have at least one cat. When cats live close to a natural area, such as a wetland, they may have impacts on native species. A previous study on the movements and hunting behaviour of domestic (house) cats around the Travis Wetland, Christchurch, New Zealand during 2000-2001 raised concerns about the effects of cats on the local skink population, as skinks were a frequent prey item. My study is a comparison to the prior study, to determine if impacts have changed alongside changes in human populations in the area post-earthquake. The domestic cat population in the area was estimated by a household survey in March-April 2018. For a 6 month period from March-September 2018, 26 households recorded prey brought home by their 41 cats. During April-July 2018, 14 cats wore Global Positioning System (GPS) devices for 7 days each to track their movements. Skink abundance was measured with pitfall trapping over 20 days in February 2018. There were more households in the area in 2018 than there were in 2000, but the numbers of cats had decreased. In the 196 ha study area around Travis Wetland, the domestic cat population was estimated at 429 cats, down from the previous 494. Most of the cats were free roaming, but the majority had been desexed and many were mostly seen at home. A total of 42 prey items were reported from 26 households and 41 cats over 6 months. Of these, 62% were rodents, 26% were exotic birds, and 12% were native birds. There were no native skinks, other mammals, or other vertebrates such as fish and amphibians (invertebrates were not included in this study). Eight male and six female cats were tracked by GPS. Home range sizes for the 100% minimum convex polygons (MCPs) ranged from 1.34 to 9.68 ha (mean 4.09 ha, median 3.54 ha). There were 9/14 (64%) cats that entered the edge of the wetland. Males had significantly larger home range areas at night and in general compared with females. However, age and distance of the cat’s household to the wetland did not have a significant effect on home range size and there was no significant correlation between home range size and prey retrieved. In 20 days of skink trapping, 11 Oligosoma polychroma were caught. The estimated catch rate was not significantly different from an earlier study on skink abundance in Travis Wetland. The apparently low abundance of skinks may have been the result of successful wetland restoration creating less suitable skink habitat, or of other predators other than cats. In the future, increased education should be provided to the public about ways to increase wildlife in their area. This includes creating lizard friendly habitat in their gardens and increasing management for cats. Generally, only a few cats bring home prey often, and these select cats should be identified in initial surveys and included in further studies. In New Zealand, until management programmes can target all predators in urban areas, domestic cats could stay out at night and inside during the day to help decrease the abundance of rodents at night and reduce the number of birds and lizards caught during the day.
On 4 September 2010 the Magnitude 7.1 'Darfield' Earthquake marked the beginning of the Canterbury earthquake sequence. The Darfield earthquake produced strong ground shaking throughout the centralCanterbury Plains, affecting rural areas, small towns and the city of Christchurch. The event produced a 29km long surface rupture through intensive farmland, causing localised flooding and liquefaction. The central Canterbury plains were subjected to a sustained period of thousands of aftershocks in the months after the Darfield earthquake. The primary sector is a major component of the in New Zealand economy. Business units are predominantly small family-run farm organisations, though there are increasing levels of corporate farming. The agribusiness sector contributes 20 per cent of real GDP and 47 per cent of total exports for New Zealand. Of the approximately 2,000 farms that are located in the Canterbury Plains, the most common farming sectors in the region are Mixed farming (mostly comprised of sheep and/or beef farming), Dairy farming, and Arable farming (cropping). Many farms on the Canterbury Plains require some form of irrigation and are increasingly capital intensive, reliant on built infrastructure, technology and critical services. Farms are of great significance to their local rural economies, with many rural non-farming organisations dependent on the health of local farming organisations. Despite the economic significance of the sector, there have been few, if any studies analysing how modern intensive farms are affected by earthquakes. The aim of this report is to (1) summarise the impacts the Darfield earthquake had on farming organisations and outline in general terms how farms are vulnerable to the effects of an earthquake; (2) identify what factors helped mitigate earthquake-related impacts. Data for this paper was collected through two surveys of farming and rural non-farming organisations following the earthquake and contextual interviews with affected organisations. In total, 78 organisations participated in the study (Figure 1). Farming organisations represented 72% (N=56) of the sample.
A zone of active tectonism occurs in mid and north Canterbury, from the Rakaia to the Waipara Rivers, which coincides with seismicity concentrations and several Quaternary surface anomalies and is here defined as the Porters Pass Tectonic Zone. Although parallel to the Marlborough faults to the north, the lack of regional definition suggests this zone is much younger in its inception reflecting a southward movement of the plate rotation vector. The objectives of this study were to map the structures associated with this zone in the segment between the Rakaia and Waimakariri Rivers with detailed analysis concentrated in the upper Kawai Valley. Quaternary offsets on the main lineament of the Porters Pass Fault were traced through the area and evidence for the rate of movement, probable magnitudes and return periods of related seismic events was sought. The basement was found to be complicated by pre-existing deformation structures in Torlesse Group rocks which have been subsequently been re-activated or rotated by recent fault movement probably beginning in the Pleistocene. This phase is dominantly thrusting and uplift has lead to the erosion of most of the overlying sedimentary cover. Remnants of the Cret-Tertiary sediments still remain as fault-bounded packets. Evidence suggests that a change to development of a regional lateral shear associated with the Porters Pass Tectonic Zone transects the thrust system with complex interaction between the older reverse and new strike-slip faults. Offset rates along the segments of the Porters Pass Fault are not well constrained but are believed to be approximately in the range of 11-13 mm/year for at least the last 130,000 years. This rate is similar to other large faults in the Marlborough region. Two earthquake events have been identified and dated at 600 and 2000 years ago, with a magnitude of greater than 6.5. Evidence suggests characteristic earthquakes along the Porters Pass Fault are greater than Magnitude 7. This result has some major ramifications for the expected seismic hazards for nearby Christchurch.
When disasters and crises, both man-made and natural, occur, resilient higher education institutions adapt in order to continue teaching and research. This may necessitate the closure of the whole institution, a building and/or other essential infrastructure. In disasters of large scale the impact can be felt for many years. There is an increasing recognition of the need for disaster planning to restructure educational institutions so that they become more resilient to challenges including natural disasters (Seville, Hawker, & Lyttle, 2012).The University of Canterbury (UC) was affected by seismic events that resulted in the closure of the University in September 2010 for 10 days and two weeks at the start of the 2011 academic year This case study research describes ways in which e-learning was deployed and developed by the University to continue and even to improve learning and teaching in the aftermath of a series of earthquakes in 2010 and 2011. A qualitative intrinsic embedded/nested single case study design was chosen for the study. The population was the management, support staff and educators at the University of Canterbury. Participants were recruited with purposive sampling using a snowball strategy where the early key participants were encouraged to recommend further participants. Four sources of data were identified: (1) documents such as policy, reports and guidelines; (2) emails from leaders of the colleges and academics; (3) communications from senior management team posted on the university website during and after the seismic activity of 2010 and 2011; and (4) semi-structured interviews of academics, support staff and members of senior management team. A series of inductive descriptive content analyses identified a number of themes in the data. The Technology Acceptance Model 2 (Venkatesh & Davis, 2000) and the Indicator of Resilience Model (Resilient Organisations, 2012) were used for additional analyses of each of the three cases. Within the University case, the cases of two contrasting Colleges were embedded to produce a total of three case studies describing e-learning from 2000 - 2014. One contrast was the extent of e-learning deployment at the colleges: The College of Education was a leader in the field, while the College of Business and Law had relatively little e-learning at the time of the first earthquake in September 2010. The following six themes emerged from the analyses: Communication about crises, IT infrastructure, Availability of e-learning technologies, Support in the use of e-learning technologies, Timing of crises in academic year and Strategic planning for e-learning. One of the findings confirmed earlier research that communication to members of an organisation and the general public about crises and the recovery from crises is important. The use of communication channels, which students were familiar with and already using, aided the dissemination of the information that UC would be using e-learning as one of the options to complete the academic year. It was also found that e-learning tools were invaluable during the crises and facilitated teaching and learning whilst freeing limited campus space for essential activities and that IT infrastructure was essential to e-learning. The range of e-learning tools and their deployment evolved over the years influenced by repeated crises and facilitated by the availability of centrally located support from the e-Learning support team for a limited set of tools, as well as more localised support and collaboration with colleagues. Furthermore, the reasons and/or rate of e-learning adoption in an educational institution during crises varied with the time of the academic year and the needs of the institution at the time. The duration of the crises also affected the adoption of e-learning. Finally, UC’s lack of an explicit e-learning strategy influenced the two colleges to develop college-specific e-learning plans and those College plans complemented the incorporation of e-learning for the first time in the University’s teaching and learning strategy in 2013. Twelve out of the 13 indicators of the Indicators of Resilience Model were found in the data collected for the study and could be explained using the model; it revealed that UC has become more resilient with e-learning in the aftermath of the seismic activities in 2010 and 2011. The interpretation of the results using TAM2 demonstrated that the adoption of technologies during crises aided in overcoming barriers to learning at the time of the crisis. The recommendations from this study are that in times of crises, educational institutions take advantage of Cloud computing to communicate with members of the institution and stakeholders. Also, that the architecture of a university’s IT infrastructure be made more resilient by increasing redundancy, backup and security, centralisation and Cloud computing. In addition, when under stress it is recommended that new tools are only introduced when they are essential.
The Porters Pass fault (PPF) is a prominent element of the Porters Pass-Amberley Fault Zone (PPAFZ) which forms a broad zone of active earth deformation ca 100 km long, 60-90 km west and north of Christchurch. For a distance of ca 40 km the PPF is defined by a series of discontinuous Holocene active traces between the Rakaia and Waimakariri Rivers. The amount of slip/event and the timing of paleoearthquakes are crucial components needed to estimate the earthquake potential of a fault. Movement was assumed to be, coseismic and was quantified by measuring displaced geomorphic features using either tape measure or surveying equipment. Clustering of offset data suggests that four to five earthquakes occurred on the PPF during the Holocene and these range between ca 5-7 m/event. Timing information was obtained from four trenches excavated across the fault and an auger adjacent to the fault. Organic samples from these sites were radiocarbon dated and used in conjunction with data from previous studies to identify the occurrence of at least four earthquakes at 8500 ± 200, 5300 ± 700, 2500 ± 200 and 1000 ± 100 years B.P. Evidence suggests that an additional event is also possible at 6200 ± 500 years B.P. The ~1000, 5300 and 6200 years B.P. paleoearthquakes were previously unrecognised, while the 500 year event previously inferred from rock-avalanche data has been discarded. The present data set produces recurrence intervals of ~2000-2500 years for the Holocene. The identification of only one Holocene PPF rupture to the west of Red Lakes indicates the presence of a segment boundary that prevents the propagation of rupture beyond this point. This is consistent with displacement data and results in slip rates of 0.5-0.7 mm/yr and 2.5-3.4 mm/yr to the west and east of Red Lakes respectively. It is possible that the nearby extensional Red Hill Fault influences PPF rupture propagation. The combination of geometric, slip rate and timing data has enabled the magnitude of prehistoric earthquakes on the PPF to be estimated. These magnitudes range from an average of between 6.9 for a fault rupture from Waimakariri River to Red Lakes, to a maximum of 7.4 that ruptures the entire length of the PPAFZ, including the full length of the PPF. These estimates are approximately consistent with previous magnitude estimates along the full length of the PPAFZ of between 7.0 and 7.5.
To this extent, modern buildings generally demonstrated good resistance to collapse during the recent earthquakes in New Zealand. However, damage to non-structural elements (NSE) has been persistent during these events. NSEs include secondary systems or components attached to the floors, roofs, and walls of a building or industrial facility that are not explicitly designed to participate in the main vertical or lateral load-bearing mechanism of the structure. They play a major role in the operational and functional aspects of buildings and contribute a major portion of the building’s overall cost. Therefore, they are expected to accommodate the effects of seismic actions such as drifts and accelerations. Typical examples of NSEs include internal non-loadbearing partitions, suspended ceilings, sprinkler piping systems, architectural claddings, building contents, mechanical/electrical equipment, and furnishings. The main focus of this thesis is the drift sensitive NSEs: precast concrete cladding panels and internal partition walls. Even though most precast concrete cladding panels performed well from a life-safety point of view during recent earthquakes in NZ, some collapsed panels posed a significant threat to life safety. It is, therefore, important that the design and detailing of the panel-to-structure connections ensure that their strength and displacement capacity are adequate to meet the corresponding seismic demands, at least during design level earthquakes. In contrast, the partition wall is likely to get damaged and lose serviceability at a low inter-story drift unless designed to accommodate the relative deformations between them and the structure. Partition walls suffered wide-ranging damage such as screw failures, diagonal cracking, detachments to the gypsum linings, and anchorage failures during the 2011 Canterbury Earthquake Sequence in NZ. Therefore, the thesis is divided into two parts. Part I of the thesis focuses on developing novel low-damage precast concrete cladding panel connections, i.e. “rocking” connection details comprising vertically slotted steel embeds and weld plates. The low-damage seismic performance of novel “rocking” connection details is verified through experimental tests comprising uni-directional, bi-directional, and multi-storey scaled quasi-static cyclic tests. Comparison with the seismic performance of traditional panel connections reported in the literature demonstrated the system’s significantly improved seismic resilience. Furthermore, the finite element models of panel connections and sealants are developed in ABAQUS. The force-drift responses of the “rocking” panel system modelled in SAP2000 is compared with the experimental results to evaluate their accuracy and validity. Part II of the thesis focuses on a) understanding the seismic performance of traditional rigid timber-framed partition wall, b) development and verification of low-damage connections (i.e. “rocking” connection details comprising of dual-slot tracks), and c) seismic evaluation of partition walls with a novel “bracketed and slotted” connections (comprising of innovative fastener and plastic bracket named Flexibracket) under uni-directional and bidirectional quasi-static cyclic loadings. Moreover, parametric investigation of the partition walls was conducted through several experimental tests to understand better the pros and cons of the rocking connection details. The experimental results have confirmed that the implementation of the proposed low damage solutions of precast cladding panels and internal partition walls can significantly reduce their damage in a building.
The Porter's Pass-Amberley Fault Zone (PPAFZ) is a complex zone of anastomosing faults and folds bounding the south-eastern edge of the transition from subducting Pacific Plate to continental collision on the Australia Plate boundary. This study combines mapping of a 2000 km2 zone from the Southern Alps northeast to the coast near Amberley, 40 km north of metropolitan Christchurch, with an analysis of seismicity and a revision of regional seismic hazard. Three structural styles: 1) a western strike-slip, and 2) a more easterly thrust and reverse domain, pass into 3) a northwest verging fold belt on the northern Canterbury Plains, reflecting the structural levels exposed and the evolving west to east propagation. Basal remnants of a Late Cretaceous-Cenozoic, largely marine sedimentary cover sequence are preserved as outliers that unconformably overlie Mesozoic basement (greywacke and argillite of the Torlesse terrain) in the mountains of the PPAFZ and are underlain by a deeply leached zone which is widely preserved. Structure contouring of the unconformity surface indicates maximum, differential uplift of c.2600 m in the southwest, decreasing to c.1200 m in the coastal fold belt to the northeast. Much lower rates (or reversal) of uplift are evident a few kilometres southeast of the PPAFZ range-front escarpment. The youngest elements of the cover sequence are basement-derived conglomerates of Plio-Pleistocene age preserved on the SE margin. The source is more distant than the intervening mountains of the PPAFZ, probably from the Southern Alps, to the west and northwest. The absence of another regional unconformity on Mesozoic basement, older than Pleistocene, indicates that this uplift is post-Pliocene. Late Pleistocene(<100 kyr) differential uplift rates of c.0.5-2.7 m/kyr from uplifted marine terraces at the east coast, and rates of 2.5-3.3 m/kyr for tectonically-induced river-down cutting further west, suggest that uplift commenced locally during the last 1 Ma, and possibly within the last 0.5 Ma, if average rates are assumed to be uniform over time. Analysis of seismicity, recorded during a 10 week regional survey of micro earthquakes in 1990, identified two seismic zones beneath North Canterbury: 1) a sub-horizontal zone of activity restricted to the upper crust (≤12 km); and 2) a seismic zone in the lower crust (below a ceiling of ≤17 km), that broadens vertically to the north and northwest to a depth of c.40 km, with a bottom edge which dips 10°N and 15°NW, respectively. No events were recorded at depths between 12 km and 17 km, which is interpreted as a relatively aseismic, mid-crustal ductile layer. Marked differences (up to 60°) in the trend of strain axes for events above and below the inferred ductile layer are observed only north of the PPAFZ. A fundamental, north-to-south increase in the Wave-length of major geological structures occurs across the PPAFZ, and is interpreted as evidence that the upper crust beneath the Canterbury Plains is coupled to the lower crust, whereas the upper crust further north is not. Most of the recorded micro earthquakes <12 km deep beneath the PPAFZ have strike-slip mechanisms. It is probable that faults splay upward into the thrusts and folds at the surface as an evolving transpression zone in response to deep shear in basement. There have been no historic surface ruptures of the PPAFZ, but the zone has been characterised historically by frequent small earthquakes. Paleoseismic data (dated landslides and surface ruptures) compiled in this study, indicate a return period of 1500-1900 years between the last two M>7-7.5 earthquakes, and 500-700 years have elapsed since the last. The magnitudes of these events are estimated at c.M7.5, which represents a probable maximum magnitude for the PPAFZ. There are insufficient data to determine whether or not the frequency of large earthquakes conforms to a recognised model of behaviour, but comparison of the paleoseismic data with the historic record of smaller earthquakes, suggests that the magnitudes of the largest earthquakes in this zone are not exponentially distributed. A seismicity model for the PPAFZ (Elder et al., 1991) is reviewed, and a b-value of 1.0 is found to be consistent with the newly acquired paleoseismic data. This b-value reduces the predicted frequency of large earthquakes (M≥7.0) in this zone by a factor of 3.5, while retaining a conservative margin that allows for temporal variations in the frequency of large events and the possibility that the geological database is incomplete, suggesting grounds for revising the hazard model for Christchurch.
Non-structural elements (NSEs) have frequently proven to contribute to significant losses sustained from earthquakes in the form of damage, downtime, injury and death. In New Zealand (NZ), the 2010 and 2011 Canterbury Earthquake Sequence (CES), the 2013 Seddon and Cook Strait earthquake sequence and the 2016 Kaikoura earthquake were major milestones in this regard as significant damage to building NSEs both highlighted and further reinforced the importance of NSE seismic performance to the resilience of urban centres. Extensive damage in suspended ceilings, partition walls, façades and building services following the CES was reported to be partly due to erroneous seismic design or installation or caused by intervening elements. Moreover, the low-damage solutions developed for structural systems sometimes allow for relatively large inter-story drifts -compared to conventional designs- which may not have been considered in the seismic design of NSEs. Having observed these shortcomings, this study on suspended ceilings was carried out with five main goals: i) Understanding the seismic performance of the system commonly used in NZ; ii) Understanding the transfer of seismic design actions through different suspended ceiling components, iii) Investigating potential low-damage solutions; iii) Evaluating the compatibility of the current ceiling system with other low-damage NSEs; and iv) Investigating the application of numerical analysis to simulate the response of ceiling systems. The first phase of the study followed a joint research work between the University of Canterbury (UC) in NZ, and the Politecnico Di Milano, in Italy. The experimental ceiling component fragility curves obtained in this existing study were employed to produce analytical fragility curves for a perimeter-fixed ceiling of a given size and weight, with grid acceleration as the intensity measure. The validity of the method was proven through comparisons between this proposed analytical approach with the recommended procedures in proprietary products design guidelines, as well as experimental fragility curves from other studies. For application to engineering design practice, and using fragility curves for a range of ceiling lengths and weights, design curves were produced for estimating the allowable grid lengths for a given demand level. In the second phase of this study, three specimens of perimeter-fixed ceilings were tested on a shake table under both sinusoidal and random floor motion input. The experiments considered the relationship between the floor acceleration, acceleration of the ceiling grid, the axial force induced in the grid members, and the effect of boundary conditions on the transfer of these axial forces. A direct correlation was observed between the axial force (recorded via load cells) and the horizontal acceleration measured on the ceiling grid. Moreover, the amplification of floor acceleration, as transferred through ceiling components, was examined and found (in several tests) to be greater than the recommended factor for the design of ceilings provided in the NZ earthquake loadings standard NZS1170.5. However, this amplification was found to be influenced by the pounding interactions between the ceiling grid members and the tiles, and this amplification diminished considerably when the high frequency content was filtered out from the output time histories. The experiments ended with damage in the ceiling grid connection at an axial force similar to the capacity of these joints previously measured through static tests in phase one. The observation of common forms of damage in ceilings in earthquakes triggered the monotonic experiments carried out in the third phase of this research with the objective of investigating a simple and easily applicable mitigation strategy for existing or new suspended ceilings. The tests focused on the possibility of using proprietary cross-shaped clip elements ordinarily used to provide seismic gap as a strengthening solution for the weak components of a ceiling. The results showed that the solution was effective under both tension and compression loads through increasing load bearing capacity and ductility in grid connections. The feasibility of a novel type of suspended ceiling called fully-floating ceiling system was investigated through shaking table tests in the next phase of this study with the main goal of isolating the ceiling from the surrounding structure; thereby arresting the transfer of associated seismic forces from the structure to the ceiling. The fully-floating ceiling specimen was freely hung from the floor above lacking any lateral bracing and connections with the perimeter. Throughout different tests, a satisfactory agreement between the fully-floating ceiling response and simple pendulum theory was demonstrated. The addition of isolation material in perimeter gaps was found effective in inducing extra damping and protecting the ceiling from pounding impact; resulting in much reduced ceiling displacements and accelerations. The only form of damage observed throughout the random floor motion tests and the sinusoidal tests was a panel dislodgement observed in a test due to successive poundings between the ceiling specimen and the surrounding beams at resonant frequencies. Partition walls as the first effective NSE in direct interaction with ceilings were the topic of the final experimental phase. Low-damage drywall partitions proposed in a previous study in the UC were tested with two common forms of suspended ceiling: braced and perimeter-fixed. The experiments investigated the in-plane and out-of-plane performance of the low-damage drywall partitions, as well as displacement compatibility between these walls and the suspended ceilings. In the braced ceiling experiment, where no connection was made between ceiling grids and surrounding walls no damage in the grid system or partitions was observed. However, at high drift values panel dislodgement was observed on corners of the ceiling where the free ends of grids were not restrained against spreading. This could be prevented by framing the grid ends using a perimeter angle that is riveted only to the grid members while keeping sufficient clearance from the perimeter walls. In the next set of tests with the perimeter-fixed ceiling, no damage was observed in the ceiling system or the drywalls. Based on the results of the experiments it was concluded that the tested ceiling had enough flexibility to accommodate the relative displacement between two perpendicular walls up to the inter-storey drifts achieved. The experiments on perimeter-fixed ceilings were followed by numerical simulations of the performance of these ceilings in a finite element model developed in the structural analysis software, SAP2000. This model was relatively simple and easy to develop and was able to replicate the experimental results to a reasonable degree. Filtering was applied to the experimental output to exclude the effect of high frequency noise and tile-grid impact. The developed model generally simulated the acceleration responses well but underestimated the peak ceiling grid accelerations. This was possibly because the peak values in time histories were affected by impact occurring at very short periods. The model overestimated the axial forces in ceiling grids which was assumed to be caused by the initial assumptions made about the tributary area or constant acceleration associated with each grid line in the direction of excitation. Otherwise, the overall success of the numerical modelling in replicating the experimental results implies that numerical modelling using conventional structural analysis software could be used in engineering practice to analyse alternative ceiling geometries proposed for application to varying structural systems. This however, needs to be confirmed through similar analyses on other ceiling examples from existing instrumented buildings during real earthquakes. As the concluding part of this research the final phase addressed the issues raised following the review of existing ceiling standards and guidelines. The applicability of the research findings to current practice and their implications were discussed. Finally, an example was provided for the design of a suspended ceiling utilising the new knowledge acquired in this research.
