Multiple infrastructure failures and restoration estimates from an Alpine Fault earthquake: Capturing modelling information for MERIT
T. R. Robinson R. Buxton T. M. Wilson W. J. Cousins A. M. Christophersen ERI Research Report 2015/04 November 2015
ECONOMICS of RESILIENT INFRASTRUCTURE
e Economics of Resilient Infrastructure programme is a collaborative New Zealand Government funded research programme between the following people and organisations:
DISCLAIMER
This report has been prepared by the Economics of Resilient Infrastructure (ERI) Research Programme as part of a collaborative research programme funded by the New Zealand Government.
Unless otherwise agreed in writing by ERI, the ERI collaborators accept no responsibility for any use of, or reliance on any contents of this Report by any person or organisation and shall not be liable to any person or organisation, on any ground, for any loss, damage or expense arising from such use or reliance.
Contact organisation for inquiries and correspondence is GNS Science Ltd, 1 Fairway Drive, Avalon, PO Box 30368, Lower Hutt 5040.
BIBLIOGRAPHIC REFERENCE
Robinson, T. R; Buxton, R.; Wilson, T. M.; Cousins, W. J.; Christophersen, A. M. 2015. Multiple infrastructure failures and restoration estimates from an Alpine Fault earthquake: Capturing modelling information for MERIT, ERI Research Report 2015/04. 80 p.
T. R. Robinson, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand R. Buxton, GNS Science, PO Box 30368, Lower Hutt, 5040, New Zealand T. M. Wilson, University of Canterbury, Private Bag 4800, Christchurch, New Zealand W. J. Cousins, GNS Science, PO Box 30368, Lower Hutt, 5040, New Zealand A. M. Christophersen, GNS Science, PO Box 30368, Lower Hutt, 5040, New Zealand
© Institute of Geological and Nuclear Sciences Limited, 2015
ISSN 2382-2325 (Print) ISSN 2382-2287 (Online) ISBN 978-0-908349-47-0 (Print) ISBN 978-0-908349-48-7 (Online)
CONTENTS ABSTRACT ...... V KEYWORDS ...... V 1.0 INTRODUCTION ...... 1 2.0 MODELLING THE ECONOMICS OF RESILIENT INFRASTRUCTURE TOOL ...... 3 3.0 METHODS ...... 4
3.1 DISASTER SCENARIO DEVELOPMENT ...... 4 3.1.1 Disaster Event ...... 4 3.1.2 Scenario Modelling ...... 4 3.2 LOSS AND RESTORATION MODELLING ...... 8 3.2.1 Critical Infrastructure Networks ...... 8 3.2.2 Exposure Analysis ...... 13 3.2.3 Expert Elicitation ...... 14 3.2.4 Network Analysis ...... 18 4.0 INFRASTRUCTURE LOSSES AND RESTORATION ...... 19
4.1 STATE HIGHWAYS ...... 19 4.1.1 Pre-Earthquake ...... 19 4.1.2 T=0 ...... 19 4.1.3 Restoration Strategy ...... 22 4.1.4 T=3 days ...... 27 4.1.5 T=14 days ...... 32 4.1.6 T=30 days ...... 36 4.1.7 T=90 days ...... 40 4.1.8 Aftershock Impacts ...... 43 4.1.9 Mitigation ...... 47 4.1.10 Adaptation ...... 48 4.2 RAIL ...... 48 4.2.1 Pre-Earthquake ...... 48 4.2.2 T=0 ...... 49 4.2.3 Restoration Strategy ...... 49 4.2.4 T=1 day...... 50 4.2.5 T=3 days ...... 51 4.2.6 T=25 days ...... 51 4.2.7 T=100 days ...... 51 4.2.8 T=186 days ...... 52 4.2.9 Aftershock Impacts ...... 53 4.3 HEP TRANSMISSION ...... 54 4.3.1 Pre-Earthquake ...... 54 4.3.2 T=0 ...... 54 4.3.3 Restoration Strategy ...... 56 4.3.4 T=2 days ...... 56
ERI Research Report 2015/04 i
4.3.5 T=30 days ...... 56 4.3.6 T=80 days ...... 56 4.3.7 Aftershock Impacts ...... 56 4.3.8 Mitigation ...... 58 4.3.9 Adaptation ...... 58 4.4 PIPED SERVICES-THE APPROACH USED...... 58 4.5 SUPPORTING SYSTEMS ...... 62 4.6 WATER SUPPLY ...... 65 4.7 SEWERS ...... 66 4.8 STORMWATER ...... 66 4.9 THE OVERALL REPAIR TIMES ...... 67 4.10 AFTERSHOCK IMPACTS ...... 67 5.0 DISCUSSION AND CONCLUSIONS ...... 68 6.0 REFERENCES ...... 70
FIGURES
Figure 1 Active faults in the South Island of New Zealand showing the location of the plate boundary Alpine Fault and Marlborough Faults Zone...... 5 Figure 2 Top: Isoseismals resulting from an M8 rupture of the Alpine Fault between Milford Sound and the Ahaura River...... 6 Figure 3a) Location map of the South Island State Highways. Green triangles show significant place locations; ...... 9 Figure 3b) Location map of the South Island rail network. Green triangles show significant place locations; ...... 10 Figure 3c) Location map of the Hydroelectric Power (HEP) Transmission network; ...... 11 Figure 3d) Township locations for the Wastewater, Stormwater and Water study...... 12 Figure 4 Horizon lines defined using the built in Arc GIS ‘Skyline’ Tool and used to identify the area of landslide hazard surrounding a point on a network to estimate coseismic landslide exposure...... 14 Figure 5 Landslide, surface rupture, and bridge losses on the State Highway network immediately after an Alpine Fault earthquake (T=0)...... 21 Figure 6 Restoration strategy for the State Highway network following an Alpine Fault earthquake...... 26 Figure 7 Example of an aftershock intensity map...... 43 Figure 8 State highways most impacted by aftershocks...... 46 Figure 9 Landslide, surface rupture, and bridge losses to the South Island Rail network immediately after an Alpine Fault earthquake (T=0)...... 50 Figure 10 South Island rail network and aftershocks...... 53 Figure 11 Landslide losses to the HEP Transmission network immediately following an Alpine Fault earthquake (T=0)...... 55 Figure 12 The South Island HEP network and the Alpine Fault aftershock zone...... 57 Figure 13 The calculated repair rates per km for ground shaking intensities for some locations in the study...... 62
ii ERI Research Report 2015/04
TABLES
Table 1 STEP model aftershock rate from generic parameters in different magnitude and time intervals...... 7 Table 2 Functionality state descriptions for each of the critical infrastructure networks designated by the corresponding network providers...... 17 Table 3 Impedance values added to functionality states between F=0 and F=1 for various infrastructure networks...... 18 Table 4 Travel time in minutes between key South Island nodes pre-earthquake (times derived from the network model)...... 20 Table 5a Change in travel time in minutes between key South Island nodes at T=0 compared to pre-earthquake (Table 3) for various vehicle types...... 23 Table 5b Change in travel time in minutes between key South Island nodes at T=0 compared to pre-earthquake (Table 3) for various vehicle types...... 24 Table 5c Change in travel time in minutes between key South Island nodes at T=0 compared to pre-earthquake (Table 3) for various vehicle types...... 25 Table 6a Change in travel time in minutes between key South Island nodes at T=3 days compared to pre-earthquake (Table 3) for various vehicle types...... 29 Table 6b Change in travel time in minutes between key South Island nodes at T=3 days compared to pre-earthquake (Table 3) for various vehicle types...... 30 Table 6c Change in travel time in minutes between key South Island nodes at T=3 days compared to pre-earthquake (Table 3) for various vehicle types...... 31 Table 7a Change in travel time in minutes between key South Island nodes at T=14 days compared to pre-earthquake (Table 3) for various vehicle types...... 33 Table 7b Change in travel time in minutes between key South Island nodes at T=14 days compared to pre-earthquake (Table 3) for various vehicle types...... 34 Table 7c Change in travel time in minutes between key South Island nodes at T=14 days compared to pre-earthquake (Table 3) for various vehicle types...... 35 Table 8a Change in travel time in minutes between key South Island nodes at T=30 days compared to pre-earthquake (Table 3) for various vehicle types...... 37 Table 8b Change in travel time in minutes between key South Island nodes at T=30 days compared to pre-earthquake (Table 3) for various vehicle types...... 38 Table 8c Change in travel time in minutes between key South Island nodes at T=30 days compared to pre-earthquake (Table 3) for various vehicle types...... 39 Table 9a Change in travel time in minutes between key South Island nodes at T=90 days compared to pre-earthquake (Table 3) for various vehicle types...... 41 Table 9b Change in travel time in minutes between key South Island nodes at T=90 days compared to pre-earthquake (Table 3) for various vehicle types...... 42 Table 10 Expected Mw5-6 aftershocks within 5km of highways, within 90 days of the main shock for the South Island by road section (each highway has many component parts)...... 44 Table 11 Total of potentially damaging aftershocks (Mw5-Mw8) by State Highway...... 45 Table 12 Change in accessibility for nodes on the rail network at T=0...... 51 Table 13 Change in accessibility for nodes on the rail network at T=1 day compared to T=0 (Table 12)...... 52 Table 14 Change in accessibility for nodes on the rail network at T=3 days compared to T=0 (Table 12)...... 52 Table 15 Locations with and without electrical power due to HEP Transmission network losses immediately following an Alpine Fault earthquake (T=0)...... 55
ERI Research Report 2015/04 iii
Table 16 The South Island townships included in the study...... 58 Table 17 Estimated ground class distributions for the locations in the pipe services study...... 59 Table 18 The liquefaction factors and multiplier estimates for the pipe services locations...... 60 Table 19 The total multiplication factors applied to the base rate for all the locations in the piped services study...... 60 Table 20 The base case pipe breakage rates...... 61 Table 21 Support elements for infrastructure services included...... 62 Table 22 MMI to acceleration conversions used...... 63 Table 23 Damage classes and consequences...... 63 Table 24 Water treatment plant damage states...... 63 Table 25 Sewage treatment plant damage states...... 63 Table 26 Water storage tank damage states...... 63 Table 27 Pumps (water, sewage lift stations, stormwater pumps)...... 64 Table 28 Supporting infrastructure component damage based on Figure 9 isoseismals...... 64 Table 29 The supporting plant repair rates (days per item, described above) assumed...... 64 Table 30 Restoration time estimates for support infrastructure...... 65 Table 31 The South Island townships water supply pipe repair times using Te Ripahapa (Figure 9) isoseismals...... 65 Table 32 The South Island townships sewer pipe repair times...... 66 Table 33 Total days restoration times...... 67 Table 34 The modelled aftershock numbers (Mw5–6)...... 67
APPENDICES APPENDIX 1: STATE HIGHWAY INITIAL DAMAGE AND RESTORATION TABLES ...... 74 APPENDIX 2: RAIL NETWORK INITIAL DAMAGE AND RESTORATION TABLES ...... 79
APPENDIX TABLES
Table A1 Initial damage and subsequent restoration times (in days) for surface rupture affecting State Highways...... 74 Table A2 Initial damage and subsequent restoration times (in days) for sections of State Highway affected by landslides...... 74 Table A3 Initial damage and subsequent restoration times (in days) for major State Highway bridges...... 76 Table A4 Initial damage and subsequent restoration times (in days) for major rail bridges...... 79 Table A5 Initial damage and subsequent restoration times (in days) for surface rupture affected rail lines. aMID – Midland Line...... 79 Table A6 Initial damage and subsequent restoration times (in days) for landslide affected rail lines...... 80
iv ERI Research Report 2015/04
ABSTRACT
The Economics of Resilient Infrastructure (ERI) research program aims to develop techniques and tools that allow the modelling and analysis of the economic consequences of infrastructure outages. The development of a Modelling the Economic Resilience of Infrastructure Tool (MERIT) has been a cornerstone deliverable for the ERI project. The underpinning concept of MERIT is that of a System Dynamic economic model with some aspects of Computable General Equilibrium (CGE). MERIT is, in fact, three different economic tools that are best suited to small, medium and large scale infrastructure outages.
To aid in the MERIT development process four outage scenarios of varying levels of severity and complexity have been produced. Previous work has concentrated on scenarios based around a single impacted infrastructure type. These scenarios and the MERIT modelling undertaken on them provided some insight into the possible regional and national economic implications of a water supply outage or an electricity outage in Auckland.
This work addresses an Alpine Fault Earthquake scenario. This is a larger and more complex scenario than those previously considered in the ERI program and which impacts multiple infrastructure types over a large area of the South Island and, especially when aftershocks are considered, has a large temporal footprint. The largest impacts are to the transportation networks (roads and rail) which could isolate many small communities on the West Coast.
KEYWORDS
Alpine Fault, economics, modelling
ERI Research Report 2015/04 v
1.0 INTRODUCTION
The Economics of Resilient Infrastructure (ERI) programme aims to develop a new tool (MERIT = Modelling the Economics of Resilient Infrastructure Tool) to quantify the spatial and temporal economic impacts resulting from critical infrastructure failures during disasters and to examine post-event recovery strategies. Existing economic models have a limited capability for estimating the economic impacts of major disruption and infrastructure losses. This is because they are unable to capture complex systems behaviour resulting from interdependencies, business adaption, and shifts in spatial patterns. Thus, there is the requirement for a tool that allows better understanding of economic behaviours, both spatially and temporally, during a disaster/disruption scenario and its consequent recovery phase. The intended outputs are expected to be able to inform government (both local and central) policies and guide future investment decisions. As a result of its importance to disaster resilience in New Zealand, the ERI research programme has been funded by the New Zealand Government. Funding of NZ$2.8 million over a period of four years was granted by the New Zealand Ministry of Business, Innovation and Employment from their 2012 funding round.
The project has four key research tasks: 1. Infrastructure disruption scenarios a. Single infrastructure disruption; b. Multiple infrastructure disruption; 2. Business behaviours in response to infrastructure disruption; 3. Modelling of economic impacts of infrastructure disruption; and 4. Informing policy and investment through stakeholder engagement.
This report focuses on Research Aim 1b, which is principally concerned with understanding the loss of service to multiple critical infrastructure networks from a single disruption scenario. The purpose of this report is to detail the spatial and temporal losses of service and restoration to critical infrastructure following an Alpine Fault earthquake in New Zealand’s South Island.
Single infrastructure disruption scenarios were completed in 2014 (Buxton et al., 2014) in order to develop confidence in the procedures and techniques being developed for MERIT. Single infrastructure disruption scenarios represent failures resulting from a specific problem within a network, such as lightning strikes causing power outages. Multiple infrastructure disruption scenarios are more likely to result instead from major geophysical or meteorological disasters, such as earthquakes, volcanic eruptions and cyclones (Preuss and Godfrey, 2006). This report intends to build upon the initial confidence building work of the single infrastructure scenarios by providing MERIT with a realistic earthquake disaster scenario resulting in multiple infrastructure losses.
The sectors considered in this work are roads (State Highways), Hydroelectric Power Transmission, Wastewater and Water. Telecommunication were not included because of time constraints and confidentiality issues.
ERI Research Report 2015/04 1
The focus of this report is to: 1. Detail the initial losses to State Highways, Rail, Hydroelectric Power Transmission, Wastewater, and Water services likely to result from an anticipated Alpine Fault earthquake; 2. Discuss the likely temporal and spatial restoration of these networks post-event.
Where possible, each network has been assessed in terms of: 1. ‘Business as usual’ (the impacts and restoration that would occur under current response plan procedures); 2. ‘Mitigation’ (where a mitigation measure, hypothetical or real, is introduced before the event); 3. “Adaptation’ (where different adaption measures are considered during the response).
The focus of this report is on the initial losses and subsequent recovery of infrastructure networks, not the earthquake disaster itself. No analysis has been undertaken as to the likelihood of this precise scenario; however an Alpine Fault earthquake has been chosen as it is considered to be highly likely in the near future (30% chance in the next 50 years; Biasi et al., 2015). The Alpine Fault scenario provides a different type of outage scenario to the single outage scenarios and the Auckland Volcanic Field Scenario considered in the ERI program in that it results in very widespread impact area. Other scenarios in ERI are far more concentrated. As such the Alpine Fault scenario presents a set of different challenges.
The purpose of the scenario is to provide an opportunity for the economic modellers to refine their modelling approaches. The information needs of the modellers have evolved throughout the scenario modelling process. This report contains details of the information arising from the scenario at the time of writing, and consequently the final input data used by the ERI economic modellers may differ slightly from that included here.
Initially, aftershocks were not considered due to time and scoping limitations. However, given the likely highly disruptive nature of the aftershocks resulting from an Alpine Fault event some aftershock modelling and a discussion of the possible additional delays in restoration times caused by these is included in each relevant infrastructure section. Additional impacts from the damming of rivers and streams are not included here.
Various infrastructure providers have participated in an expert elicitation process to develop the data included herein and have allowed the details of this elicitation process to be published. Given the complexities and necessary assumptions involved, the results are considered order of magnitude estimates only and do not account for a number of potential sources of uncertainty.
2 ERI Research Report 2015/04
2.0 MODELLING THE ECONOMICS OF RESILIENT INFRASTRUCTURE TOOL
A detailed description of the MERIT model is provided in Smith and McDonald (2014) as well as a brief summary in Buxton et al. (2014). Herein, a short overview of the model and its required elements is included in order to highlight the necessary data collection undertaken by this study. Readers should refer to Smith and McDonald (2014) for a more detailed description of MERIT.
MERIT is an adaptation of economic computable general equilibrium (CGE) models that attempt to deal with issues such as time-path trajectories and out-of-equilibrium dynamics, which traditional CGE models are unable to incorporate (Barker, 2004; Grassini, 2004; Scrieciu, 2007). Instead of identifying steady states of economic equilibrium, as per CGE, MERIT uses a systems dynamics approach to model the economy as a complex system. In this way, MERIT better reflects reality by modelling an economy that is constantly changing and adapting to external factors and is rarely considered to be in steady state equilibrium.
One of the key aspects of modelling economies as complex dynamic systems is explicit modelling of vital supply and demand relationships. This includes modelling a business’s ability to achieve required production levels as well as its ability to adequately distribute goods to where they are in demand.
For MERIT to effectively model the economic effects of major disasters, explicit data on the initial loss of service and subsequent restoration of critical infrastructure are required. In order to inform post-disaster recovery options, MERIT must also consider potential alternative situations which may change either the initial losses or the subsequent restoration. This information is critical for effective modelling of both the spatial and temporal evolution of the key supply and demand relationships post-disaster. Consequently, the data collected herein will enable time-path trajectories vital to the supply-demand chain to be evaluated at any time after a given disaster in order to assess the overall impact of the event to the economy.
In order to do this effectively, this study focuses on detailing the functionality of infrastructure on a 0 to 1 scale, with 0 representing complete loss of functionality and 1 representing no loss in functionality (i.e. pre-earthquake functionality). These functionality states are updated at various time-steps following a disaster as restoration progresses. Understanding the likely functionality of infrastructure following a disaster is vital for understanding the capacity of a network and for determining both the immediate and long-term recovery post-disaster (Whitman et al., 1997). The precise information needs of the MERIT tool are still being developed and the information presented herein is, in some cases, likely only part of that required to successfully undertake the MERIT analysis.
ERI Research Report 2015/04 3
3.0 METHODS
This study has involved two key stages: 1. Development of a suitably detailed disaster scenario from which infrastructure losses of service will occur; and 2. Estimates of the consequent infrastructure losses of service and the times for their subsequent restoration.
Below is a description of the methods undertaken during each of these stages.
3.1 DISASTER SCENARIO DEVELOPMENT
3.1.1 Disaster Event
New Zealand is one of the most seismically active countries in the world, and the South Island in particular has experienced multiple damaging earthquakes since European settlement in the 1840’s (Robinson and Davies, 2013). Following the 2010–11 Canterbury earthquake sequence, there has been a strong drive by the New Zealand public and government at all levels to increase earthquake resilience in general. This is primarily a result of the infrastructural damage that resulted from the Canterbury earthquake sequence (see Giovinazzi et al., 2011) and has led the ERI project to investigate the economic effects likely to result from a large future earthquake scenario. One of the largest and most anticipated earthquake scenarios for New Zealand is a major rupture of the plate boundary Alpine Fault, which runs for >600 km through the South Island (Figure 1). Recent studies have shown that this fault has regular earthquakes with an average recurrence interval of 329±68 years, with 298 years since the last known rupture (Berryman et al., 2012). Consequently, there is currently thought to be ~30% probability of an Alpine Fault earthquake in the next 50 years (Biasi et al., 2015). This study has therefore focussed on developing detailed loss and restoration estimates resulting from a rupture of the Alpine Fault.
3.1.2 Scenario Modelling
As a consequence of the 2010–11 Canterbury earthquake sequence and the high likelihood of an Alpine Fault rupture, the South Island Civil Defence and Emergency Management (CDEM) Groups held an Island-wide emergency response exercise in 2013 focussing on an Alpine Fault earthquake. This exercise was developed to represent a maximum-credible event occurring in late May (the time of the exercise). The scenario used was one of the most detailed and realistic scenarios developed for a CDEM exercise in New Zealand, and included the most up-to-date scientific knowledge at the time. This study therefore uses the same scenario, with some additions in areas where scientific understanding has increased since the exercise. The methods to develop the exercise and final scenario itself are described in detail in Robinson et al., (2014) and thus only a short summary is provided herein.
4 ERI Research Report 2015/04
3.1.2.1 Seismic Scenario
To develop their earthquake scenario, Robinson et al. (2014) focussed on a review of the pertinent literature. Given the likelihood of an Alpine Fault earthquake, numerous studies have investigated the fault in an attempt to better understand its seismic potential. Consequently, there is a significant amount of data available upon which a sufficiently detailed Alpine Fault earthquake scenario could be developed.
Figure 1 Active faults in the South Island of New Zealand showing the location of the plate boundary Alpine Fault and Marlborough Faults Zone. Inset: Tectonic setting of New Zealand with plate motions (mm/a).
As well as showing that the Alpine Fault experiences earthquakes regularly, Berryman et al. (2012) also showed that the fault is likely uni-modal, with the majority of known pre-historic earthquakes inferred to have had rupture lengths >400 km. Studies of offset geomorphic features on the central section of the fault have shown that the last three earthquakes involved lateral displacements of ~7.5 m (De Pascale et al., 2014). On the southern section of the fault, lateral displacement of moraines from 270 ka have been measured at ~8,000 m, corresponding to single event displacements of ~9.5 m assuming an earthquake recurrence of 329 years (Barth et al., 2014). This apparent difference in offsets can be attributed to the change in motion between the central and southern segments of the fault, from dextral-
ERI Research Report 2015/04 5
oblique to purely dextral respectively. Combined with studies of dune formations, rockfall/landslide deposits, and other off-fault evidence (see Adams, 1980; Sutherland, 1994; Wells et al., 1999; Wells and Goff, 2007) this suggests that previous earthquakes have all been M~8. The scenario earthquake included in the CDEM exercise and herein therefore replicates a M8 earthquake involving ~400 km rupture between Milford Sound and the Ahaura River (Figure 2).
Figure 2 Top: Isoseismals resulting from an M8 rupture of the Alpine Fault between Milford Sound and the Ahaura River.
6 ERI Research Report 2015/04
3.1.2.2 Modelling aftershocks
To model the aftershock sequence arising from the scenario main shock on the Alpine fault the Short-Term Earthquake Probability (STEP) model was used. The STEP model (Gerstenberger, 2003; Gerstenberger et al., 2005) is an aftershock model based on the idea of superimposed earthquake sequences, each following the Omori-Utsu law for aftershock decay (Utsu et al., 1995). The model typically involves a mix of a background seismicity model and a time-dependent clustering model. Here only the time-dependent component is used since we want to model only the aftershocks. The expected number of aftershocks within 0–90 days, 0–365, and 365–720 days of the mainshock are calculated. The aftershocks are distributed spatially along a two-segment idealised fault with the coordinates -45.0416°, 166.9833°, -42.8933°, 171.15°, and -41.7967°, 172.8783°. Aftershock rates are calculated in 0.05° times 0.05° grid cells, and in magnitude bins from 5.0–8.0 in 0.1 magnitude steps. Spatially, the aftershock rates taper off according to the inverse square of the distance from the fault. This spatial distribution for the aftershocks was determined from the analysis of numerous aftershock sequences in California (Gerstenberger, 2003). In the scenario modelling, there is no opportunity for the STEP model to apply its standard updating procedures for the Reasenberg and Jones parameters, or to allow for spatial heterogeneities within the sequence. Therefore the forecast rates are for somewhat idealised aftershock sequences. More heterogeneous sequences would be expected in practice.
Three sets of parameters are applied to estimate the expected number: the generic STEP model parameters determined by Pollock (Pollock, 2007), and earlier version of the same parameters by (Eberhart-Phillips, 1998), and a version with a slightly different formulation of aftershock productivity (Christophersen and Gerstenberger, 2008). The largest earthquakes that contributed to the derivation of model parameters were in the order of M7.1, and there is always a risk in extrapolating a model beyond the range of data for which it was derived. For a magnitude M8.2 within the first year, the number of aftershocks of M5 and greater are 428, 397 and 482, respectively for the above mentioned three models. We prepared rate files for the generic models. Table 1 lists aftershock of different sizes and within different forecast time.
Table 1 STEP model aftershock rate from generic parameters in different magnitude and time intervals.
0–90 0–365 365–730 Magnitude days days days
M5+ 380 428 23
M6+ 35 40 2
M7+ 3 4 0.2
3.1.2.3 Secondary hazards
Robinson and Davies (2013) showed that an Alpine Fault earthquake is also likely to involve substantial secondary hazards, primarily in the form of landsliding, liquefaction, debris flows etc. At the time the CDEM scenario was developed it was not possible to model the resulting landslide hazard, which Robinson and Davies (2013) suggested to be one of the most widespread and potentially catastrophic secondary effects. Bird and Bommer (2004) have shown that as well as ground shaking, landsliding during earthquakes is one of the most important hazards to consider in terms of infrastructure losses. Modelling the consequent landslide hazard resulting from the seismic scenario is therefore vital for accurate loss and
ERI Research Report 2015/04 7
restoration estimates. Herein, the resulting landslide hazard has been taken from Robinson (2014) who modelled the hazard using the method of Kritikos et al. (2015). The method and results are briefly described here.
To model landslide hazard from scenario earthquakes in regions with no detailed geotechnical data or accurate historic landslide inventories (e.g. New Zealand), Kritikos et al. (2015) used historic inventories from multiple events globally to identify common pre- disposing factors. By identifying these factors from multiple different locations, it can be assumed that these factors may also have a similar influence on landsliding in scenario earthquakes elsewhere. Robinson (2014) showed the method successfully modelled the spatial distribution of landslides from the 2003 M7.2 and 2009 M7.8 Fiordland earthquakes. He thus applied this method to the seismic scenario described above.
By assessing the exposure of critical infrastructure to the ground shaking, surface rupture, and landsliding described, a detailed understanding of the potential initial losses that require restoration can be estimated. Other hazards such as liquefaction, debris flows etc. are also likely (Robinson and Davies, 2013), however Bird and Bommer (2004) demonstrated that the majority of infrastructure losses can be attributed to ground shaking and landsliding, and thus other hazards are not considered in detail herein.
3.2 LOSS AND RESTORATION MODELLING
3.2.1 Critical Infrastructure Networks
The critical infrastructure networks included in this study are the: • State Highways; • Rail; • Hydroelectric Power (HEP) Transmission; • Wastewater; and • Water networks (Figure 3d).
These networks have been included as, firstly, there is substantial evidence from the Canterbury earthquake sequence (e.g. Giovinazzi et al., 2011) that these are some of the most crucial networks in terms of affecting emergency response and long-term recovery. As well as this, they are among the most vital networks for the South Island regional economy. Mining, agriculture, and tourism for instance are three of New Zealand’s largest industries, particularly in West Coast Region where they account for ~43% of total annual GDP (Infometrics, 2012). Each of these industries relies heavily on these infrastructure networks. In the West Coast Region the agricultural industry is dominated by dairy farming, which requires: 1. A reliable source of electricity for milking and storing milk; 2. A transport network to distribute product; 3. Wastewater systems to safely clean/remove stock effluent; and 4. Water systems for irrigation.
8 ERI Research Report 2015/04
Secondly, these networks are not constrained by commercially sensitive information to the same degree as other types of infrastructure networks (e.g. telecommunications), allowing the network providers to contribute vital data on the vulnerability of their respective networks and current restoration plans. Other networks such as telecommunications are constrained by commercially sensitive data, which affects the amount of relevant data the individual providers can share, particularly within the time constraints of the project. These networks have therefore not been considered at this stage of the project.
Figure 3a) Location map of the South Island State Highways. Green triangles show significant place locations;
ERI Research Report 2015/04 9
Figure 3b) Location map of the South Island rail network. Green triangles show significant place locations;
10 ERI Research Report 2015/04
Figure 3c) Location map of the Hydroelectric Power (HEP) Transmission network;
ERI Research Report 2015/04 11
Figure 3d) Township locations for the Wastewater, Stormwater and Water study.
12 ERI Research Report 2015/04
3.2.2 Exposure Analysis
Loss and restoration modelling was undertaken via a combination of exposure analysis and expert elicitation. Exposure analysis was first undertaken in order to identify which sections of the various networks were exposed to the various hazards in the disaster scenario (Figure 2). These exposure maps were then used to assign functionality states to the corresponding sections of network via an expert elicitation process.
Exposure mapping was primarily undertaken using Geographic Information Systems (GIS) by comparing the spatial distribution of each hazard (ground shaking, surface rupture, and landsliding) to the locations of each of the networks. For the State Highway and Rail networks this also required identifying the locations of major bridges, which will behave differently (and thus experience different functionality states) compared to the rest of the network, particularly with regards to ground shaking. Major bridges were identified as being those sections of State Highway and Rail network that cross rivers mapped on the Land Information New Zealand (LINZ) 1:50,000 topographical maps. It is likely that this does not contain all the bridges/viaducts on these networks, however it does contain those that cross large rivers and may therefore be the most significant impediment to the network should they fail.
Surface rupture exposure analysis identified the locations at which the various networks crossed the fault rupture and anticipated the displacement along that section of the fault corresponding to the fault slip measurements in Norris and Cooper (2001). This was not undertaken for the HEP Transmission network as this network comprises high tension cables strung between wooden- or steel-supports. Thus cables are suspended above ground where they cross the fault, and are only affected by becoming stretched or slack due to displacement of supports. In the Christchurch earthquake these effects had minimal impact (Giovinazzi et al., 2011).
Exposure analysis for landsliding was undertaken using a simplified method to that detailed by Robinson (2014). This involved calculating the average landslide hazard in a defined region surrounding a given location on the network (Figure 4). To avoid incorporating slopes in opposing valleys, Robinson (2014) suggested defining the surrounding region using the ‘Skyline’ function in GIS which identifies the horizon at a given distance from a given point.
ERI Research Report 2015/04 13
Figure 4 Horizon lines defined using the built in Arc GIS ‘Skyline’ Tool and used to identify the area of landslide hazard surrounding a point on a network to estimate coseismic landslide exposure.
3.2.3 Expert Elicitation
In order to establish appropriate functionality states, the initial infrastructure losses and the subsequent restoration timelines, an expert elicitation analysis was undertake for each of the networks. Expert elicitation has been shown to be an effective method for solving complex but well-defined problems, with Cooke and Goosens (2004) showing there was no appreciable loss in detail or realism compared to other methods. Herein, this involved a series of structured workshops with the relevant infrastructure providers discussing the hazard scenario, exposure analysis, network provisions and vulnerability, current restoration strategies and inferred timelines, as well as potential mitigation and adaption techniques and their likely effects. The infrastructure providers involved were: • New Zealand Transport Agency (NZTA) – State Highway network; • KiwiRail – Rail network; • Transpower – HEP Transmission network; and • Fulton Hogan Rebuild – Water and Wastewater networks
The individuals who participated in the workshops comprised national/regional network managers and senior structural engineers with significant experience and knowledge of their corresponding network as a whole. They each participated on a professional basis, providing their expert knowledge/opinion as opposed to their personal opinion. The data and results included herein represent each infrastructure provider’s best and most current understanding of their network, its vulnerability to a future Alpine Fault earthquake, and the current most likely restoration strategies and timelines.
14 ERI Research Report 2015/04
The structured workshops were undertaken by the authors individually with each network provider. Discussions began with the authors introducing the project and its aims and objectives, followed by a short, detailed summary of the earthquake scenario and the evidence used to derive it, including its likelihood and any uncertainties. The methods and results of the exposure analysis were then discussed, with the infrastructure providers able to contribute their knowledge on its accuracy, highlighting any inconsistencies in the modelling compared to expert/historic knowledge. Following this, appropriate functionality states for each network were discussed and decided upon based on the requirements and restoration strategies of each network. For instance, when initially constructing, restoring or upgrading sections of State Highway, NZTA aims to minimise the impact of such works by initially opening a single lane carriageway with weight and speed restrictions, before progressing to a double lane carriageway with similar speed and weight restrictions, and finally a fully functioning (i.e. no restrictions) double lane carriageway. Following an Alpine Fault earthquake, the same restoration approach would be undertaken in order to minimise impacts by reducing the out-of-service time by providing first a single lane carriageway followed by a double lane carriageway. The resulting functionality states for each network are shown in Table 2.
To accurately estimate the timeline associated with post-earthquake restoration the workshop discussed the restoration strategy (i.e. the restoration priority for each section of the network) options available for each provider. This primarily involved identifying which sections of the network were most critical and thus required rapid restoration. Each of the network providers already had some level of planning for restoration following an Alpine Fault earthquake, demonstrating the level of awareness and planning that is already in place for such an event. In all instances the scenario presented did not require a dramatic change in the current restoration plans for each provider, further confirming the high levels of planning and preparedness that have already been undertaken by New Zealand’s critical infrastructure providers. Nevertheless, the scale of infrastructure losses, particularly with regards to landsliding, which had not previously been modelled accurately, was higher than the providers had considered.
Following discussions on the restoration strategy, first order timescales for the restoration of affected sections of the networks were discussed. Where required, this included providing estimates of total restoration (i.e. functionality, F=1) time, as well as each intermediate functionality state (see Table 2). The discussions focussed on identifying the relative time required to restore each individual section based on the level of damage it had sustained. Total restoration time required combining this information with the restoration strategy in order to gauge how long after T=0 the restoration could begin for a given damaged section of the network.
On-going hazards such as aftershocks and re-activated landslides were not considered in detail as these are difficult to predict. It should be noted that these will have a substantial effect on the restoration strategy and times however. In some instances, on-going hazards were considered in a general sense only: for instance, NZTA suggested that regardless of the precise aftershock sequence, landslide re-activation etc., it was unlikely to be safe enough to send construction teams into particular areas within the first 6 months. This information was captured and included when there was sufficient confidence in such assumptions from all parties involved in the workshops.
ERI Research Report 2015/04 15
Recent work by GNS (Christophersen and Gerstenberger, 2008) on modelling the aftershock sequences of the Wellington Fault and other faults potentially damaging to Wellington city has suggested that there could be many hundreds of aftershocks of between magnitude M5 and M7 in the first year following an Alpine Fault earthquake. Based on the Wellington work the spatial distribution is likely to be such that most of the aftershocks are clustered around the Alpine Fault itself with the clustering getting less dense as the distance from the fault increases. The impacts that the aftershocks can make to restoration times, particularly in the months immediately following an event, are difficult to quantify. The potential additions to the restoration times that the aftershock sequence may incur are discussed at the end of each relevant section.
For the water supply, sewer and storm water networks a different approach was taken. These networks are not distributed as widely as the others considered in this study and their influences on the economic losses are not expected to rival the other networks. Further, these networks are maintained by local District Councils with maintenance, restoration, improvements etc. contracted to various construction companies that differ for different locations. Consequently, an in-house study was performed by GNS Science using information readily available and procedures and analyses techniques used previously to estimate the damage to piped systems in Wellington following a Wellington Fault earthquake (Moull, 2012).
Although the results for water supply, sewers and stormwater are considered separately in subsequent sections the general approach used to ascertain the damage and recovery times is generic across these sections and this is discussed here.
In order to estimate the breakages occurring in a piped service network and consequently the recovery times needed to restore the service, a number of details about the network need to established. Firstly an estimate of the total length of pipe in the network needs to be ascertained, and secondly the type of ground needs to be determined. The latter is vital as certain types of ground can liquefy, substantially increasing the risk of damage to piped systems. Additionally the type of material the pipe is made of needs to be taken into account as well as an estimate of the severity of the shaking intensity.
The procedure for estimating the impacts to the water supply has several steps: 1. Compile a list of towns and villages likely to be affected by levels of shaking high enough to cause damage to the piped supply network. 2. Estimate the total lengths of pipe for the towns. 3. Estimate the MMI shaking levels at those locations. 4. Estimate the ground class at the locations. 5. Estimate the pipe material. 6. Using fragility functions developed for the Wellington pipe damage study (Cousins et al., 2013) estimate the number of breaks/km, multiply by estimated pipe lengths to get total breaks. 7. Assume 1 day (water) or 2.5 days (sewers and stormwater) to repair 1 break for the water network.
Finally, once the ‘business as usual’ impacts and restoration had been considered, alternative scenarios involving pre-event mitigation and post-event adaption were examined. These could involve actual examples that were being considered or are scheduled to be
16 ERI Research Report 2015/04
implemented by the provider, or ‘blue sky’ examples that may not represent feasible options and are not currently being considered. The effects of the mitigation and adaption on the restoration strategy and/or timeline were then discussed along with its corresponding feasibility.
Table 2 Functionality state descriptions for each of the critical infrastructure networks designated by the corresponding network providers. aThe values for each functionality state do not represent network capacities (i.e. 0.5 ≠ 50% capacity), these are illustrative only. bOnly bridges experiencing MM7 or higher were considered vulnerable to damage requiring restoration. cNZTA suggested there was little benefit in restoring these bridges to double lane sections without restrictions, and noted that many of these bridges are already (i.e. pre-earthquake) single lane with speed and weight restrictions corresponding to the 0.6 functionality state herein.
Network/Infrastructure Functionality Statea State Highways (NZTA)
0 – No Access
0.2 – No public access Emergency/Construction/Military vehicles only, single lane, one vehicle at a time, 10 km/hr speed restriction, 3.5 tonnes maximum weight restriction Bridgesb, c 0.4 – Single lane access for all light (<3.5 tonnes) vehicles, 10 km/hr speed restriction
0.6 – Single lane access for all mid-weight (<6 tonnes) vehicles, 10 km/hr speed restriction
0 – No Access
0.25 – No public access, Emergency/Construction/Military vehicles only
0.5 – Single lane access for all light (<3.5 tonnes) vehicles, 30 km/hr speed Roads restrictions
0.75 – Double lane access for all mid-weight (<6 tonnes) vehicles, 60 km/hr speed restrictions
1 – Full access, no restrictions (100 km/hr) Railways (KiwiRail)
0 – No Access Rails/Bridges 1 – Full Access
HEP Transmission (Transpower)
0 – No Transmission Steel towers/Wooden Poles 1 – Full Transmission Wastewater
Piped systems 0 – No service
1 – Full service Water
Piped systems 0 – No service
1 – Full service Storm Water
Piped systems 0 – No service
1 – Full service
ERI Research Report 2015/04 17
3.2.4 Network Analysis
Following the workshops, a series of network analyses were undertaken for each network at various selected time-steps following T=0. This was undertaken in order to demonstrate the decrease in connectivity between key nodes (towns/cities) throughout the South Island and how this changed temporally. The nodes selected were based upon the distribution of the network considered (Figure 3), and were chosen to allow both inter- and intra-regional connections. Such network analyses are able to effectively demonstrate the initial disruption resulting from the disaster, and the subsequent effect of infrastructure restoration.
This was undertaken using an Origin-Destination (OD) Cost Matrix in GIS (see ESRI, 2015) with network barriers and impedance set relating to the functionality states at the necessary time-steps. The time-steps examined are illustrative only, and are selected to specifically demonstrate the timing of substantial increases in network connectivity. Consequently, selected time-steps are unique to the network being considered. While illustrative time-steps have been selected, the data herein is deliberately presented in such a way as to allow a user to evaluate network connectivity at any desired time after T=0, as is necessary for MERIT.
For the State Highway network, which has functionality states between F=0 and F=1 (Table 2), impedance values were required to represent the increase in time as a result of traversing that section of the network. For points on the network such as bridges and surface ruptures, this impedance is given as an absolute value (i.e. minutes required to traverse), while for line segments such as those affected by landslides, this is given as a scaled factor (i.e. double the F=1 time to traverse). These impedance values were selected to be representative of the corresponding functionality state, considering conditions such as speed restrictions (Table 3). When a network section has F=0 it is considered a network barrier and is impassable (i.e. impedance=∞).
Table 3 Impedance values added to functionality states between F=0 and F=1 for various infrastructure networks. These values are added to the initial (i.e. pre-earthquake) network connectivity when traversed.
Infrastructure Functionality State Impedance Value
0.2 4 minutes State Highway Bridges and Surface 0.4 3 minutes Rupture points 0.6 2 minutes
0.25 x4
State Highway landslide sections 0.5 x3
0.75 x2
Of the infrastructure systems included in this scenario the transportation sector, particularly the road network, has proven to be the most complex from a MERIT modelling point of view. The roads are different from the other services in that their function is to convey different types of goods (impacting different economic sectors) from one location to another. So the impact to the network (functionality state) is only a part, albeit a significant one, of the information needed to model the economic impacts. The information needs of the MERIT tool have evolved throughout the course of the work described in this report and the additional information requirements will be the subject of ongoing work not included within the scope of this report.
18 ERI Research Report 2015/04
4.0 INFRASTRUCTURE LOSSES AND RESTORATION
4.1 STATE HIGHWAYS
4.1.1 Pre-Earthquake
The South Island State Highway network comprises long, thin corridor routes with little redundancy; there are only three routes that cross the Southern Alps connecting the east and west coasts (Figure 3). Speed limits on all highways are 100 km/hr, although vehicles generally travel slower in the steep, narrow, winding Alpine passes. Highways comprise double lane carriageways with a single lane in either direction, although there are multiple single lane bridges throughout the network. Pre-event travel times have been calculated between each of the key nodes in order to compare post-earthquake consequences for network connectivity (Table 4). Sections 4.1.2 to 4.1.7 discuss the recovery timeline in the absence of aftershocks.
4.1.2 T=0
Initial losses to the State Highway network are the result of surface rupture displacing the road at nine different sites, landsliding in the Alpine passes and along SH6, and damage to bridges primarily in West Coast Region (Figure 5). In general, bridges were considered likely to perform well, with only five of the 55 bridges evaluated (9%) thought likely to suffer catastrophic loss (F=0; Figure 5). A further 13 bridges would be sufficiently damaged to require imposing strict weight restrictions for vehicles >3.5 tonnes (F=0.4) but would not fail catastrophically. All other bridges either remain at or are reduced to a single lane with 10 km/hr speed restrictions and maximum weight restrictions of 6 tonnes.
Surface rupture across SH6 at seven sites between Hokitika and Haast displaces the road by 8 m horizontally and 2 m vertically. Where SH6 crosses the fault near Haast, fault motion has primarily transitioned to pure strike-slip (i.e. lateral displacement), and the total horizontal offset increases as a result to 10 m with no vertical offset. SH73 (Arthur’s Pass) also crosses the fault rupture, however the smaller slip rate along this section results in 5 m horizontal displacement and 1 m vertical. All of these locations are immediately rendered impassable to all vehicle types (F=0).
Landsliding primarily effects SH6, SH7 (Lewis Pass), SH73 (Arthur’s Pass), and SH94. There are multiple sites affected, with some having multiple landslides extending over several kilometres. The worst affected locations that are completely blocked and impassable (F=0) are: • Mt Hercules (SH6); • Lake Mapourika (SH6); • Fox Hills (SH6); • Paringa River (SH6); • Lake Moeraki (SH6); • Knights Point (SH6); • Maruia River (SH7); • Lewis Pass (SH7); • Arthur’s Pass/Otira (SH73); and • Hollyford River (SH94).
ERI Research Report 2015/04 19
Table 4 Travel time in minutes between key South Island nodes pre-earthquake (times derived from the network model).
Queenstown Franz Josef Franz Greymouth Invercargill Aoraki/Mt Aoraki/Mt Westport Dunedin Hokitika Reefton Milford Nelson Sound Picton Haast Cook
Christchurch 323 375 282 447 283 548 615 289 407 328 392 338 355
Dunedin 439 475 337 476 303 465 225 600 521 282 531 548
Franz Josef 125 101 92 429 495 332 335 373 272 181 198
Greymouth 226 33 554 620 441 210 248 397 56 73
Haast 193 327 393 230 435 474 170 282 299
Hokitika 521 587 424 243 281 364 89 107
Invercargill 262 332 764 746 171 610 627
Milford Sound 398 830 813 237 676 693
Aoraki/Mt Cook 566 487 175 497 514
Nelson 178 607 154 161
Picton 590 192 199
Queenstown 453 470
Reefton 55
20 ERI Research Report 2015/04
Figure 5 Landslide, surface rupture, and bridge losses on the State Highway network immediately after an Alpine Fault earthquake (T=0). Only bridges experiencing MM7+ shaking have been evaluated.
ERI Research Report 2015/04 21
As well as these, there are multiple locations where landsliding partially blocks the road making it unsafe for all but vital emergency/military/construction vehicles (F=0.25). These locations are primarily near to sections that are completely impassable (see above), however notable sites that are elsewhere include: • Punakaiki (SH6); and • Diana Falls/Haast Pass (SH6).
Finally, there are several locations where minor rockfalls affect the highway, reducing it to a single lane but remaining accessible to all vehicles <3.5 tonnes, albeit with a 30 km/hr speed restriction (F=0.5). The corresponding sections are generally short (<2 km) and are close to sections more heavily damaged, however notable sections include: • Upper Buller Gorge (SH6); • Dobson (SH7); • Wallace River (SH65); and • Inangahua River Bridge (SH69).
The effect of this is to isolate almost the entire West Coast Region from inter- and intra- regional travel (Table 5). Emergency vehicles can gain inter-region access to northern West Coast Region (Westport, Reefton, Greymouth and Hokitika) via routes through the Buller Gorge, however this is not possible for public vehicles. All West Coast communities south of Hokitika are isolated from both inter- and intra-regional connections (Table 5). Retaining intra-regional emergency vehicle connection between Greymouth and Hokitika is vital, as the main hospital for the region is located in Greymouth while many of the doctors and nurses live in Hokitika, which is also where the main airport is located.
4.1.3 Restoration Strategy
Restoration of the State Highway network would initially focus on trying to gain emergency access to the central and southern West Coast regions, which are anticipated to be worst affected by the earthquake. This would involve large construction teams from Blenheim or Nelson first making safe the route to Greymouth via SH6-SH69-SH7 that is initially only viable for emergency vehicles (Figure 6). The largest impedance along this route is the catastrophic failure of the Ahaura River bridge, which is estimated to require 8 weeks to return to F=1 (see Appendix 1). An alternative route on the opposite bank of the Grey River avoids this bridge (Figure 6) without increasing travel time, and is thought unlikely to suffer any substantial disruption. This alternative route will likely be reserved for emergency/ construction/military vehicles to enable rapid access to the West Coast Region, meaning public access must wait for suitable restoration of the Ahaura River bridge.
Once at Greymouth, construction teams will work south along SH6 towards Franz Josef (Figure 6). The primary disruption along this route is the landslide blockages at Mt Hercules. It was inferred that while construction crews from the north were working to clear Mt Hercules, simultaneously the local population and construction crews in the region at the time of the earthquake would be clearing the landslide blockages around Lake Mapourika using earth moving equipment available in the area on most farms. Consequently, once an emergency route across Mt Hercules is completed, access for emergency vehicles would be possible as far south as Franz Josef.
While continuing hazards such as aftershocks and re-activated landslides have not been examined explicitly, the topography and location of the road in the Fox Hills between Franz Josef and Fox Glacier was considered unlikely to allow safe access for construction crews for at least six months after the earthquake. Consequently no road access, emergency or otherwise, is restored south of Franz Josef in this scenario.
22 ERI Research Report 2015/04
Table 5a Change in travel time in minutes between key South Island nodes at T=0 compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes.
Emergency/Construction/Military Vehicles
Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 292 340 0 0 67 0 0 150 140
Dunedin 292 340 0 0 67 0 0 150 140
Franz Josef
Greymouth 16 438 292 30 30 439 78 16
Haast
Hokitika 520 358 46 46 521 94 32
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
ERI Research Report 2015/04 23
Table 5b Change in travel time in minutes between key South Island nodes at T=0 compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes.
Vehicles up to 3.5 tonnes Aoraki/Mt Cook
Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 0 0 67 0 0 150 140
Dunedin 0 0 67 0 0 150 140
Franz Josef
Greymouth 11
Haast
Hokitika
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
24 ERI Research Report 2015/04
Table 5c Change in travel time in minutes between key South Island nodes at T=0 compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes.
Vehicles between 3.5 and 6 tonnes
Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Greymouth Invercargill Westport Dunedin Hoki Reefton Nelson Picton Haast tika
Christchurch 0 0 0 67 0 0
Dunedin 0 0 67 0 0
Franz Josef
Greymouth
Haast
Hokitika
Invercargill 0 128 0 0
Milford Sound
Aoraki/Mt Cook 67 0 0
Nelson 0 129
Picton 0
Queenstown
Reefton
ERI Research Report 2015/04 25
Figure 6 Restoration strategy for the State Highway network following an Alpine Fault earthquake.
26 ERI Research Report 2015/04
Simultaneous to construction crews working south along SH6, further crews will be tasked with restoring Lewis Pass (SH7). This would involve crews from Christchurch working west towards Springs Junction, while crews from Blenheim/Nelson would access Springs Junction via SH65 and work east and west from there (Figure 6). Arthur’s Pass (SH73) was considered unlikely to be safe for construction crews to access for at least six months after the disaster for similar reasons to the Fox Hills on SH6 (Figure 6). No access of any kind is restored through this section in this scenario.
Restoring access along SH6 and SH7 comprise the major restoration efforts. While these are continuing, smaller efforts are given to clearing landslide debris at Punakaiki (SH6). Crews are sent from Westport in the north and Greymouth in the south to restore this section as rapidly as possible. Similarly, crews are sent from Te Anau to clear the landslide blockages on SH94 stopping access to Milford Sound (Figure 6).
The possibility of clearing Haast Pass and Knights Point on SH6 were also considered. Like the Fox Hills and Arthur’s Pass however, the level of damage and the surrounding topography would make restoring these sections unsafe for at least the first six months, regardless of the aftershock sequence. No access is restored to these sections in this scenario (Figure 6).
Despite not being able to gain road access to the region between Franz Josef and Knights Point, construction crews are sent via sea and air to Bruce Bay (Figure 6), from which they could begin to restore surface ruptures and small landslides, and repair damaged bridges between Fox Glacier and Knights Point. This would be undertaken so as to speed the restoration once (if) Haast Pass, Knights Point, and Fox Hills are cleared after at least six months. All affected sections would be fully restored within this time, but the precise timings are not considered further as this will not affect the network analysis due to the long-term (>6 months) blockages that prohibit road access to the region. Access by sea would likely come from the South Island’s deepwater ports at Lyttleton (Christchurch), Port Chalmers (Dunedin), Picton, and Nelson. Air access would likely be gained from Christchurch, where the South Island’s largest airport is located (Figure 6).
4.1.4 T=3 days
Three days after the disaster, most progress has been made on clearing minor landslide/rockfall disruptions and the accessible sections of highway experiencing surface rupture. All locations between Greymouth and Franz Josef where surface rupture disrupted the roads are returned to F=1. Similar restoration speeds occurred for roads displaced during the 4 September 2010 Darfield earthquake, and thus are deemed feasible for this scenario.
Minor landslide/rockfall losses (i.e. F=0.5 at T=0) are restored to F=1 at the following locations: • Upper Buller Gorge (SH6); • Punakaiki (SH6); • Ten Mile Creek (SH6); • Mt Hercules (SH6); • Whataroa River (SH6); • Okarito River (SH6); • Dobson (SH7);
ERI Research Report 2015/04 27
• Old Christchurch Road Junction (SH73); • Warwick River (SH65); • Inangahua River Bridge (SH69); and • Hollyford River (SH94).
Single lane access (F=0.5) is restored to more heavily landslide damaged sections (i.e. F=0 or F=0.25 at T=0) at: • Punakaiki (SH6); • Okarito River (SH6); • Lake Mapourika (SH6); and • Old Christchurch Road Junction (SH73).
Bridge repairs are also progressing, with river fords being set up at: • Makarora River (SH6); • Camerons Creek (SH6); • Stony Creek (SH7); and • Wainihinihi River (SH73)
The fords restore these sections to F=0.6 (the maximum for bridges). A Bailey Bridge is partially installed at Papakeri Creek (SH6) providing single lane access for all light (<3.5 tonnes) vehicles, while repairs are completed on the Taipo River and Griffen Creek bridges on SH73.
The primary result of this is the ability to grant emergency vehicles access to Franz Josef for the first time since the earthquake as well as reduce their travel times to and from Greymouth and Hokitika (Table 5). Vehicles <3.5 t are now able to access Greymouth and Hokitika for the first time, but travel times to all other locations remain unchanged. Vehicles between 3.5 and 6 t have access to Westport and Reefton restored. At this stage, any access by road is still not possible to Milford Sound or Haast (Table 6).
28 ERI Research Report 2015/04
Table 6a Change in travel time in minutes between key South Island nodes at T=3 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=0 (Table 4).
Emergency/Construction/Military Vehicles
Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 386 286 334 0 0 67 0 0 150 140
Dunedin 515 286 334 0 0 67 0 0 150 140
Franz Josef 68 52 750 588 92 92 751 92 84
Greymouth 16 432 286 24 24 433 24 16
Haast
Hokitika 514 352 40 40 515 40 32
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
ERI Research Report 2015/04 29
Table 6b Change in travel time in minutes between key South Island nodes at T=3 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=0 (Table 4).
Vehicles up to 3.5 tonnes Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika R Nelson Picton Haast eefton
Christchurch 0 284 327 0 0 67 0 0 150 140
Dunedin 284 327 0 0 67 0 0 150 140
Franz Josef
Greymouth 11 430 284 22 22 431 70 8
Haast
Hokitika 507 345 33 33 508 81 19
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
30 ERI Research Report 2015/04
Table 6c Change in travel time in minutes between key South Island nodes at T=3 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=0 (Table 4).
Vehicles between 3.5 and 6 tonnes Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 0 0 67 0 0 150 140
Dunedin 0 0 67 0 0 150 140
Franz Josef
Greymouth
Haast
Hokitika
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
ERI Research Report 2015/04 31
4.1.5 T=14 days
Two weeks after the earthquake all landslide affected sections on SH94 have been restored for emergency access only (F=0.25), allowing emergency vehicles to reach Milford Sound by road for the first time since the earthquake.
The landslide blockages (F=0 at T=0) at Okarito River and Mt Hercules on SH6 are fully restored, while emergency access is gained along the blocked section (F=0 at T=0) of SH7 along the Hope River. The following sections which previously had emergency access only are fully restored: • Punakaiki (SH6); • Okarito River (SH6); • Lake Mapourika (SH6); • Maruia River (SH7); and • Old Christchurch Road Junction (SH73).
All landslide losses on SH6 north of Franz Josef are fully restored by this stage.
Restoration on three bridges has progressed, with the Ahaura River on SH7 (which catastrophically failed, F=0 at T=0) restored to F=0.4, the New River bridge on SH6 fully restored, and the Bailey Bridge at Papakeri Creek on SH6 completed. This leaves just six bridges on the network with on-going restoration: • Waimea Creek (SH6); • Mikonui River (SH6); • Waitaha River (SH6); • Mahitahi River (SH6); • Taramakau River (SH6); and • Ahaura River (SH7).
This allows emergency access to Milford Sound, while emergency vehicles travel times to Greymouth, Hokitika, and Franz Josef all improve compared to T=3 days (Table 7). Haast is the only remaining node to which basic emergency access is still not possible, and this remains so for at least six months. Vehicles weighing <3.5 t are now able to access Franz Josef for the first time, and travel times to Greymouth and Hokitika improve (Table 7). The only change from T=3 days for vehicles between 3.5 and 6 t is that access to Greymouth returns (Table 7).
32 ERI Research Report 2015/04
Table 7a Change in travel time in minutes between key South Island nodes at T=14 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=3 days (Table 6).
Emergency/Construction/Military Vehicles
Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 348 277 320 0 8 0 67 0 0 150 140
Dunedin 477 277 320 0 8 0 67 0 0 150 140
Franz Josef 39 28 712 721 550 54 54 713 54 47
Greymouth 11 423 432 277 15 15 424 15 8
Haast
Hokitika 500 509 338 26 26 501 26 19
Invercargill 8 0 128 0 0 296 286
Milford Sound 8 137 8 8 305 295
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
ERI Research Report 2015/04 33
Table 7b Change in travel time in minutes between key South Island nodes at T=14 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=3 days (Table 6).
Vehicles up to 3.5 tonnes
Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Greymouth Invercargill Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 348 277 320 0 0 67 0 0 150 140
Dunedin 477 277 320 0 0 67 0 0 150 140
Franz Josef 39 28 712 550 54 54 713 54 47
Greymouth 11 423 277 15 15 424 15 8
Haast
Hokitika 500 338 26 26 501 26 19
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
34 ERI Research Report 2015/04
Table 7c Change in travel time in minutes between key South Island nodes at T=14 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=3 days (Table 6).
Vehicles between 3.5 and 6 tonnes Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 284 0 0 67 0 0 150 140
Dunedin 284 0 0 67 0 0 150 140
Franz Josef
Greymouth 430 284 22 22 431 70 8
Haast
Hokitika
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
ERI Research Report 2015/04 35
4.1.6 T=30 days
Thirty days after the earthquake, restoration of all remaining bridges is nearing completion. Total restoration is completed on the following bridges: • Waimea Creek (SH6); • Mikonui River (SH6); • Waitaha River (SH6); and • Mahitahi River (SH6).
This leaves only the Taramakau River (SH6) and Ahaura River bridges still in the process of being restored. Both remain open to vehicles <3.5 tonnes on a single lane basis with speed restrictions on 10 km/hr (i.e. F=0.4).
Landslide restoration is now focussed on SH7 and SH94. Remaining minor blockages (F=0.5 at T=14 days) on SH7 at Reefton, Lewis River, and Maruia River, as well as on SH94 at Homer Tunnel and Milford Sound are all fully restored.
The section of SH7 at Lewis Pass with only emergency access (F=0.25 at T=14 days) is upgraded to single lane access for vehicles <3.5 tonnes. The completely blocked sections (F=0 at T=14 days) on SH7 at Lewis Pass, Maruia River and Hope River are restored to double lane carriageways with speed and weight restrictions, as are the Hollyford River sections of SH94.
Nonetheless, landslide restoration remains on going along the Hollyford River section of SH94 and along 27 km of SH7 between the Hope River and Lewis Pass.
The main effect is to decrease travel times for emergency vehicles across the majority of the network, although many routes still have increased travel times compared to pre-earthquake (Table 8). Travel times also decrease for vehicles <3.5 tonnes and these are now able to access Milford Sound for the first time (Table 8). The only change for vehicles between 3.5 and 6 tonnes is access between Hokitika and Franz Josef is now possible. These vehicles still cannot access Greymouth, Hokitika, Franz Josef, and Milford Sound from most other locations however Table 8). Haast remains inaccessible to all vehicles.
36 ERI Research Report 2015/04
Table 8a Change in travel time in minutes between key South Island nodes at T=30 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=14 days (Table 7).
Emergency/Construction/Military Vehicles
Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 217 150 191 0 8 0 2 0 0 23 13
Dunedin 346 150 191 0 8 0 2 0 0 23 13
Franz Josef 35 26 581 590 419 50 50 582 50 43
Greymouth 9 296 305 150 15 15 297 15 8
Haast
Hokitika 371 380 209 24 24 372 24 17
Invercargill 8 0 63 0 0 169 159
Milford Sound 8 72 8 8 178 168
Aoraki/Mt Cook 2 0 0 23 13
Nelson 0 64 0 0
Picton 0 0 0
Queenstown 170 160
Reefton 0
ERI Research Report 2015/04 37
Table 8b Change in travel time in minutes between key South Island nodes at T=30 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=14 days (Table 7).
Vehicles up to 3.5 tonnes Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 217 150 191 0 8 0 2 0 0 23 13
Dunedin 346 150 191 0 8 0 2 0 0 23 13
Franz Josef 35 26 581 590 419 50 50 582 50 43
Greymouth 9 296 305 150 15 15 297 15 8
Haast
Hokitika 371 380 209 24 24 372 24 17
Invercargill 8 0 63 0 0 169 159
Milford Sound 8 72 8 8 178 168
Aoraki/Mt Cook 2 0 0 23 13
Nelson 0 64 0 0
Picton 0 0 0
Queenstown 170 160
Reefton 0
38 ERI Research Report 2015/04
Table 8c Change in travel time in minutes between key South Island nodes at T=30 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=14 days (Table 7).
Vehicles between 3.5 and 6 tonnes Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 284 0 0 67 0 0 150 140
Dunedin 284 0 0 67 0 0 150 140
Franz Josef 26
Greymouth 430 284 22 22 431 70 8
Haast
Hokitika
Invercargill 0 128 0 0 296 286
Milford Sound
Aoraki/Mt Cook 67 0 0 150 140
Nelson 0 129 0 0
Picton 0 0 0
Queenstown 297 287
Reefton 0
ERI Research Report 2015/04 39
4.1.7 T=90 days
By 90 days after the earthquake, all restoration other than the sections too dangerous to access, has been completed. It will be at least another 90 days before these sections will be considered safe enough for restoration to be considered (i.e. T=180 days).
Restoration of the Taramakau and Ahaura bridges is completed, allowing access for all vehicles <6 tonnes via a single lane and with a 10 km/hr speed restriction. Landslide restorations in Lewis Pass (SH7) and on SH94 are completed, returning all sections to pre- earthquake functionality.
Consequently, Haast remains the only location inaccessible to all vehicle types. All other nodes are accessible by all vehicles <3.5 tonnes, although the reduced functionality of some bridges means that travel times between many nodes remain increased compared to pre- earthquake (Table 8). Travel time between Christchurch and Franz Josef for instance, is still 215 minutes (3 hours 35 minutes) longer than pre-earthquake. The most notable changes however are for vehicles between 3.5 and 6 tonnes, which are now able to access all nodes except Haast, albeit with significantly increased travel times across most routes compared to pre-earthquake (Table 9).
40 ERI Research Report 2015/04
Table 9a Change in travel time in minutes between key South Island nodes at T=90 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=30 days (Table 8).
Emergency/Construction/Military and Public Vehicles <3.5 tonnes Aoraki/Mt Aoraki/Mt Milford Sound Queenstown Franz Josef Franz Greymouth Invercargill Westport Dunedin Hokitika Reefton Nelson Picton Haast
Cook
Christchurch 0 215 149 189 0 8 0 2 0 0 23 13
Dunedin 344 149 189 0 8 0 2 0 0 23 13
Franz Josef 34 26 579 588 417 48 48 580 48 42
Greymouth 8 295 304 149 14 14 296 14 8
Haast
Hokitika 369 378 207 22 22 370 22 16
Invercargill 8 0 63 0 0 169 159
Milford Sound 8 72 8 8 178 168
Aoraki/Mt Cook 2 0 0 23 13
Nelson 0 64 0 0
Picton 0 0 0
Queenstown 170 160
Reefton 0
ERI Research Report 2015/04 41
Table 9b Change in travel time in minutes between key South Island nodes at T=90 days compared to pre-earthquake (Table 3) for various vehicle types. Black boxes show inaccessible routes, yellow boxes show decreased travel times compared to T=30 days (Table 8).
Vehicles between 3.5 and 6 tonnes Aoraki/Mt Cook Milford Sound Queenstown Franz Josef Franz Invercargill Greymouth Westport Dunedin Hokitika Reefton Nelson Picton Haast
Christchurch 0 215 149 189 0 8 0 2 0 0 23 13
Dunedin 344 149 189 0 8 0 2 0 0 23 13
Franz Josef 34 26 579 588 417 48 48 580 48 42
Greymouth 8 295 304 149 14 14 296 14 8
Haast
Hokitika 369 378 207 22 22 370 22 16
Invercargill 8 0 63 0 0 169 159
Milford Sound 8 72 8 8 178 168
Aoraki/Mt Cook 2 0 0 23 13
Nelson 0 64 0 0
Picton 0 0 0
Queenstown 170 160
Reefton 0
42 ERI Research Report 2015/04
4.1.8 Aftershock Impacts
The STEP aftershock model produces probabilistic estimates of the expected aftershock numbers and magnitudes within a predefined area surrounding a fault. With the South Island highway network the exposure of different elements of the network to the aftershocks depends on the position and orientation of the particular road section in relation to the aftershock zone. Figure 7 shows a typical output from the STEP model in GIS format with roads overlain.
Figure 7 Example of an aftershock intensity map. Aftershock rates vary from 1.14 eq/pixel (red) to negligible values (yellow).
ERI Research Report 2015/04 43
Table 10 lists the expected number of Mw5–6 aftershocks occurring within 5km of a highway section within the first 90 days following an Alpine Fault earthquake. In Table 10 the middle five columns contain statistics relating to the earthquake rate calculated by the aftershock model. The rate is the number of earthquakes expected per pixel within the time period being considered (90 days for Table 10).
Table 10 Expected Mw5-6 aftershocks within 5km of highways, within 90 days of the main shock for the South Island by road section (each highway has many component parts).
Number of Max Standard Highway Min Rate Range Mean earthquakes Rate deviation expected
SH6 0.021 1.145 1.123 0.380 0.415 20.50
SH6 0.012 1.145 1.133 0.352 0.437 16.90
SH73 0.018 1.145 1.127 0.371 0.444 9.27
SH6 0.002 1.145 1.143 0.124 0.297 7.07
SH65 0.009 1.145 1.136 0.270 0.399 7.03
SH6 0.150 1.145 0.995 0.727 0.422 5.09
SH7 0.009 1.145 1.136 0.229 0.344 4.82
SH63 0.028 1.145 1.117 0.260 0.351 2.34
SH63 0.028 1.145 1.117 0.253 0.331 2.28
SH94 0.002 0.165 0.163 0.014 0.027 0.70
SH7 0.003 0.052 0.049 0.013 0.012 0.38
SH7 0.008 0.021 0.013 0.012 0.003 0.31
SH6 0.005 0.021 0.016 0.010 0.004 0.22
SH73 0.015 0.075 0.060 0.034 0.020 0.20
SH73 0.001 0.018 0.017 0.004 0.004 0.19
SH80 0.002 0.019 0.017 0.006 0.004 0.16
SH6 0.010 0.023 0.013 0.014 0.004 0.14
SH6 0.009 0.019 0.010 0.013 0.003 0.13
SH6 0.002 0.007 0.005 0.003 0.001 0.12
44 ERI Research Report 2015/04
SH6 is the most impacted mainly because it is close to and runs parallel to the zone of high aftershock activity.
The model predicts fewer numbers of larger events, however, the events can generate damage from a greater distance hence Mw6–7 and Mw7–8 within 10km and 20km respectively have been considered. These are all considered potentially damaging aftershocks. Table 11 lists the number of all the potentially damaging (Mw5–Mw8) aftershocks by State Highway number.
Table 11 Total of potentially damaging aftershocks (Mw5-Mw8) by State Highway.
Number of Highway earthquakes expected
SH6 59.35
SH73 11.23
SH65 8.00
SH7 6.86
SH63 5.49
SH94 1.26
SH80 0.16
Figure 8 shows the highways most impacted.
ERI Research Report 2015/04 45
Figure 8 State highways most impacted by aftershocks.
46 ERI Research Report 2015/04
Some proportion of the estimated aftershocks can be assumed to be damaging but all would probably incur some delay to restoration processes through safety inspections and other procedures. For the purposes of this study it has been assumed that half of the estimated aftershocks would cause some additional damage or re-damage work previously repaired. The rest would incur some small time penalty representing safety checks needed before repair work can recommence. A large part of the additional damage within this first 90 days occurs on sections that NZTA would not consider attempting to repair for 6 months following an Alpine Fault earthquake. Some damage, would, however be incurred on the highways farther north specifically earmarked as routes that would be repaired first to enable repair teams to gradually move southwards and westwards relieving trapped communities on the isolated West Coast region.
Considering the first 90 days after the main shock, SH6 into and out of Haast would incur the most additional damage possibly adding several months to the eventual restoration times. SH6 between Hokitika and Franz Josef would possibly be impacted by 8 or 9 further damaging aftershocks and a similar number of lesser but still disruptive shocks adding possibly two months to the recovery times. The western most section of SH73 would likely incur the next most damage being impacted by possibly 4 or 5 further damaging aftershocks adding about a month to the eventual restoration times. The section of SH6 between Haast and Queenstown would possibly be impacted by 3 or 4 damaging aftershocks and a similar number of lesser events adding about 1 month to the restoration times for that section of SH6. The modelling indicates that SH65 would also be impacted by 3 or 4 damaging aftershocks and lesser events adding a month to the restoration times for that section. The western most section of SH7 would possibly be impacted by 2 or 3 damaging aftershocks and a similar number of lesser aftershocks adding possibly 20 days to restoration times for that section. SH63 would possibly be impacted by 2 or 3 damaging aftershocks and a similar number of lesser aftershocks adding possibly 20 days to the restoration times for that section of the highway.
The overall conclusions are that recovery would be hampered by additional damage particularly to SH63 and SH65 which would limit the ability of repair crews to work from Picton (in particular) south. The priority repair route indicated as SH6-SH69-SH7 to Greymouth could also be impacted but not hugely. The suggestion is that the 90 day restoration time currently indicated in Table 8 would be increased by an additional value which is between 1 and about 20 days.
4.1.9 Mitigation
NZTA currently have plans in place to upgrade the Taramakau River Bridge (SH6) in the next five years and the Ahaura River Bridge (SH7) within the next decade. If these improvements were completed before an Alpine Fault earthquake was to occur, it was inferred that neither bridge would sustain any damage.
Maintaining full functionality (i.e. a single lane with some speed restrictions) of the Ahaura River Bridge, would allow all vehicles up to 6 tonnes to retain access to Greymouth throughout the post-earthquake recovery phase. Currently, vehicles up to 3.5 tonnes cannot gain access to Greymouth until T=3 days when the landslide blockages on SH6 between Westport and Greymouth are cleared sufficiently. Vehicles between 3.5 and 6 tonnes do not have access until T=14 days, again gaining access when further landslide debris is cleared at Punakaiki. As well as retaining access post-earthquake, maintaining the route via the Ahaura River Bridge would also allow vehicles to reach Greymouth faster, by not forcing them to access Greymouth via Westport and Punakaiki.
ERI Research Report 2015/04 47
Maintaining full functionality of the Taramakau River bridge would allow vehicles up to 6 tonnes to retain access to Hokitika after 14 days when the New River Bridge is sufficiently restored, and Franz Josef after 28 days when the Waitaha and Mikonui River bridges are sufficiently restored. The only effect for vehicles <3.5 tonnes would be to reduce travel time by 2 minutes between Greymouth and Hokitika and Franz Josef, as the current Taramakau River bridge remains usable by these vehicles in the current scenario.
4.1.10 Adaptation
The considered adaption was to avoid clearance of landslides in the Fox Hills and Haast Pass and permanently use sea and/or air access to reach Bruce Bay and the area between Knights Point and the Fox Hills. This was not considered a realistic option as it would be extremely expensive to maintain such access given the small population (<600 people).
NZTA considered it more likely that access would be gained either by constructing a completely new road route between Franz Josef and Fox Glacier, or by trying to restore Haast Pass and Knights Point. Both options were likely to be extremely expensive, with restoration likely lasting for several years. During this time temporary access via sea and/or air would be required if the local population were to remain, although this would incur further costs. Temporary relocation of these locals to avoid these costs could be considered. A possible alternative to this would be to not restore the roads between Franz Josef and Haast Pass, and instead maintain a dry-weather only gravel road which would likely be sufficient for the local population, but not the large number of tourists that use the road.
4.2 RAIL
4.2.1 Pre-Earthquake
Like the State Highway network, the South Island rail network comprises long, thin corridor routes with little redundancy. Only one route connects the mining and dairy production centres in West Coast Region with the main distribution centre at Lyttleton Port, Christchurch, while only one other route connects the major cities on the east coast (Figure 3). Tracks generally consist of a single line in either direction, with occasional sidings to allow trains to pass, and all trains are diesel powered as opposed to being electrified. Very few passenger routes are available with the majority of rail journeys being freight distribution, although there are infrequent tourist routes between Christchurch and Greymouth, and Christchurch and Picton. Sections 4.2.1 to 4.2.8 discuss the recovery timeline in the absence of aftershocks.
KiwiRail elected to classify their network into two simple functionality states: full service, or no service (Table 2). The reasoning for this was that, unlike for roads, partial service is not a feasible option for rail. For instance, it is not possible to stagger line openings in the same way that roads open single lanes before double lanes. For a train to be able to pass, the full width of the line including both tracks must be available. Similarly, weight restrictions are not feasible because of the large weights involved: restoring a bridge to sustain reduced weights is inefficient as the reduced limit would need to be a substantial proportion of the total limit. Consequently, network analysis for rail only considers whether access to a node can be gained, not the change in travel time. Pre-earthquake all considered nodes are connected to one another allowing both inter- and intra-regional travel.
48 ERI Research Report 2015/04
4.2.2 T=0
The primary losses to the network are from landsliding. All bridges on the network perform well, with none-sustaining any disruption inducing damage. Nevertheless, all bridges require inspection following the earthquake in order to confirm this before trains can use them. Consequently, on the day of the earthquake all rail bridges are closed prior to precautionary inspections.
Landslides primarily affect the Midland Line where it passes through the Southern Alps (Figure 9) at: • Cass; • Waimakariri River to Otira Tunnel (southern portal); • Otira Tunnel (northern portal) to Otira Township; • Otira Township to Taramakau River; • Taramakau River; • Lake Poerua; and • Lake Brunner.
Similar to the State Highways, the worst affected sections (primarily around the Otira Tunnel) have numerous landslides blocking or partially blocking the tracks. Three locations on the Stillwater-Westport Industrial Line are also affected by landslides at: • Dobson; • Reefton; and • Inangahua River Bridge.
Most of these landslides are minor single landslide/rockfall blockages which require little remediation compared to those on the Midland Line.
Surface rupture only affects the rail network at one location at Lake Poerua (Figure 9), displacing the Midland Line 5 m horizontally and 1 m vertically. Similar damage was seen on the Main South Line following the 2010 Darfield earthquake.
The primary effect of these disruptions is to close the network west of Cass on the Midland Line (Table 12). Travel along both the Main North and Main South Lines are possible, connecting the major east coast towns and ports, however both inter- and intra-regional travel west of Christchurch is disrupted (Table 12).
4.2.3 Restoration Strategy
The immediate priority is to inspect all bridges on the network west of Christchurch, and particularly those in West Coast Region. These initial bridge inspections are likely to last require one full day, assuming access to the necessary locations can be gained. Once these have been completed the priority is to restore the Midland Line by remediating the landslide losses. Despite passing alongside the SH73, which is considered too unsafe to restore by NZTA (see above), the Otira Tunnel provides an ~8 km unaffected section of the rail network, allowing KiwiRail to consider restoration of the network a plausible option.
ERI Research Report 2015/04 49
Simultaneous to restoration of the Midland Line, smaller construction crews will be tasked with restoring the Stillwater-Westport Industrial Line where landslides have affected the route. However, given the lack of connection to Christchurch and the east coast via the Midland Line, the Stillwater-Westport Line is assigned a lower priority.
Figure 9 Landslide, surface rupture, and bridge losses to the South Island Rail network immediately after an Alpine Fault earthquake (T=0). Only bridges experiencing MM7+ shaking have been evaluated.
4.2.4 T=1 day
The first day after the earthquake, precautionary bridge inspections are completed and all bridges are confirmed to be undamaged, returning to fully functioning. The small rockfalls at Dobson and the Inangahua River Bridge on the Stillwater-Westport Industrial Line are subsequently cleared, however the rockfalls near Reefton as well as all landslides on the Midland Line remain in place. The surface rupture on the Midland Line at Lake Poerua has been accessed and restoration has begun but at this stage it remains non-functional.
The result is to restore some intra-regional travel between Greymouth and Hokitika, and Westport and Reefton (Table 13). However inter-regional travel to West Coast Region is still unavailable and connections between Westport and Reefton, and Greymouth and Hokitika remain severed until the landslides south of Reefton can be cleared.
50 ERI Research Report 2015/04
Table 12 Change in accessibility for nodes on the rail network at T=0. Black boxes show inaccessible routes. For place locations refer to Figure 3 and Figure 4. Dunedin Greymouth Hokitika Invercargill Picton Reefton Westport
Christchurch
Dunedin
Greymouth
Hokitika
Invercargill
Picton
Reefton
4.2.5 T=3 days
Three days after the earthquake, the surface rupture at Lake Poerua is fully restored, as well as the rockfall south of Reefton, returning full access to the Stillwater-Westport Industrial Line. The small landslides/rockfalls at Cass are remediated on the Midland Line, allowing the first access to the blockages between the Waimakariri and Mingha Rivers. Just three days after the earthquake, the Midland Line is the only line still requiring restoration.
Despite the full restoration of the Stillwater-Westport Industrial Line, the only effect on network connectivity is to fully re-establish intra-regional connections in West Coast Region; connections between the east and west coasts remain severed however (Table 13).
4.2.6 T=25 days
Further restoration is not completed on the Midland Line until 25 days after the earthquake. At this time the landsliding between the Waimakariri and Mingha Rivers is cleared, as are the landslides along the Taramakau River, allowing the first access since the earthquake to both portals of the Otira Tunnel. The scale of landslides around the tunnel portals however is much larger than the landslide cleared to date, and requires substantial restoration work in order to fully complete the network restoration.
Consequently, there is no change to the network connectivity.
4.2.7 T=100 days
Further notable progress is not made until 100 days after the earthquake, when the landslides blocking the southern portal of the Otira Tunnel are cleared. This allows rail access between the major east coast cities and Arthur’s Pass township for the first time since the earthquake. Nevertheless, the most severe blockages at the northern portal near Otira still require remediation, restricting rail connections between the east and west coasts.
ERI Research Report 2015/04 51
Table 13 Change in accessibility for nodes on the rail network at T=1 day compared to T=0 (Table 12). Black boxes show inaccessible routes; yellow boxes show changes from T=0. For place locations refer to Figure 3 and Figure 4. Dunedin Greymouth Hokitika Invercargill Picton Reefton Westport
Christchurch
Dunedin
Greymouth
Hokitika
Invercargill
Picton
Reefton
Table 14 Change in accessibility for nodes on the rail network at T=3 days compared to T=0 (Table 12). Black boxes show inaccessible routes; yellow boxes show changes from T=1 day (Table 13). For place locations refer to Figure 3 and Figure 4. Dunedin Greymouth Hokitika Invercargill Picton Reefton Westport
Christchurch
Dunedin
Greymouth
Hokitika
Invercargill
Picton
Reefton
4.2.8 T=186 days
Almost six months after the earthquake, the final landslide blockages on the Midland Line at the northern portal of the Otira Tunnel are finally cleared. These are the last remaining restoration works on the rail network, restoring the connection between the east and west coasts and returning the network connectivity back to pre-earthquake levels. Normal rail services can now be resumed for the first time since the earthquake.
52 ERI Research Report 2015/04
4.2.9 Aftershock Impacts
Concentrating on the first 90 days after the main event the Midland Line would possibly be impacted by 4 or 5 damaging aftershocks and a similar number of lesser events causing disruption due to safety inspections. It is possible that these events could add a month to the restoration times for the Midland Line resulting in a total time to restoration of about 9 months. Figure 10 shows the main aftershock zone superimposed on the South Island rail network.
Figure 10 South Island rail network and aftershocks. Aftershock rates as per Figure 7.
ERI Research Report 2015/04 53
4.3 HEP TRANSMISSION
4.3.1 Pre-Earthquake
Like the State Highway and Rail networks, the HEP Transmission network comprises long, thin corridor routes with little redundancy (Figure 3). The network comprises of high voltage cables suspended between steel pylons and wooden poles. There are four different voltages carried by the network: 60 kV, 110 kV, 220 kV, and 350 kV. The smaller 60 kV lines are the oldest and are slowly being phased out; the 350 kV line comprises the Inter-Island link, connecting the North and South Island power supplies via Haywards in Wellington (North Island) and Benmore in the Waitaki Valley (South Island). Sections 4.3.1 to 4.3.6 discuss the recovery timeline in the absence of aftershocks.
East of the Southern Alps, transmission cables are primarily supported by steel pylons, while in West Coast Region they are supported by wooden poles. The location of all steel pylons is known and thus the exposure of each individual pylon has been undertaken. For wooden pole supported sections however, only the location of the cables is known. Consequently, the exposure of the cables has been undertaken assuming that wooden poles are equally spaced along the cables every 20 m.
The network transmits electricity from the hydro-lake generation sites primarily in the Waitaki Valley to substations (known as Grid Exit Points or GXPs) for local distribution.1 This study has not evaluated the losses to generation sites or GXPs nor local distribution networks, only the transmission cables and their supports.
4.3.2 T=0
During the Canterbury earthquake sequence, steel pylons and wooden pole supports were observed to perform well despite experiencing strong ground shaking (Giovinazzi et al., 2011). Consequently, ground shaking is considered unlikely to cause significant losses to the HEP Transmission network.
No steel pylons are located directly on the fault rupture, and it is thought that no wooden pole supports do either. Furthermore, sufficient amounts of slack cable are available between supports to accommodate the displacement between the supports. Consequently, surface rupture does not result in any direct losses to the network.
Losses therefore primarily result from landslides. In Arthur’s Pass region, 26 steel pylons carrying 60 kV cables are damaged as a result (Figure 11). In total, 52 km of cables supported by wooden poles are also damaged. Assuming a pole spacing of 20 m this equates to 2,600 wooden poles lost. This primarily affects the 60 kV lines through Arthur’s Pass and in West Coast Region south of Hokitika, although a short (~100 m) section of 110 kV cables are also lost just north of Greymouth as a result of some minor rockfalls (Figure 11).
These disruptions result in power loss in Greymouth, Hokitika, and Franz Josef, while power to the rest of the South Island remains (Table 15).
1 In electrical power networks, transmission refers to the transportation of high voltage electricity over large distances i.e. between generation sites and GXPs. Electricity is then transported locally over short distances with substantially decreased voltages, referred to as distribution.
54 ERI Research Report 2015/04
Figure 11 Landslide losses to the HEP Transmission network immediately following an Alpine Fault earthquake (T=0). Ground shaking and surface rupture were considered not to cause significant losses.
Table 15 Locations with and without electrical power due to HEP Transmission network losses immediately following an Alpine Fault earthquake (T=0).
Location Electricity Available? Christchurch Dunedin Franz Josef X Greymouth X Hokitika X Invercargill Aoraki/Mt Cook Nelson Picton Queenstown Reefton Westport
ERI Research Report 2015/04 55
4.3.3 Restoration Strategy
The priority is to return electrical power to Greymouth, Hokitika, and Franz Josef as quickly as possible. The primary focus is to restore the 110 kV section north of Greymouth and then work south along the west coast restoring the 60 kV section as far as Franz Josef. Similar to the State Highways, it is unlikely to be safe enough for construction crews to access the network between Franz Josef and Fox Glacier and through Arthur’s Pass for at least the first six months after the earthquake. The loss of the 60 kV lines through Arthur’s Pass only presents a local issue however, as most electrical load for West Coast Region uses the 110 and 220 kV lines through the Buller Gorge. Restoring the 110 kV lines is therefore the main priority.
It is estimated that repair time for each wooden pole is six hours compared to a steel pylon, which takes two days. Transpower have up to 30 spare steel pylons in storage at all times. As this is less than the number of pylons lost (26) restoration would therefore not need to include acquisition of further pylon supplies, and can therefore begin immediately.
4.3.4 T=2 days
Two days after the earthquake, restoration of the five damaged wooden poles on the 110 kV line near Greymouth is completed. Simultaneously, teams have worked south from Hokitika restoring a further 8 poles (160 m) damaged at Mt Hercules.
Consequently, electrical power from the national grid has been restored to Greymouth and Hokitika but remains out at Franz Josef.
4.3.5 T=30 days
One month after the earthquake, restoration teams working south along the west coast have managed to restore a further 120 wooden poles (2,400 m), completely restoring the lost section at Mt Hercules. Teams are now focussing on restoring the 200 poles (4 km) section lost near Lake Mapourika (just north of Franz Josef). Consequently, electrical power from the national grid remains out in Franz Josef.
4.3.6 T=80 days
All wooden pole supports on the west coast between Greymouth and Franz Josef are finally restored 80 days after the earthquake. At this stage it is still likely to be unsafe for teams to attempt restoration in the Fox Hills and Arthur’s Pass for at least a further ~100 days (~3.5 months). Consequently, restoration of the HEP Transmission network ceases, as power is restored to Franz Josef. Given the limited electrical load carried by the line through Arthur’s Pass, decision makers at this stage may decide not to restore the network here.
4.3.7 Aftershock Impacts
The aftershock modelling indicates that the 50/60kV network will possibly be impacted by 23 or 24 damaging (Mw5+, within 5km) aftershocks in the first 90 days following the main shock. Most of these would likely occur on the section of line southwest of Hokitika and between Kumara power station and the Otira substation which coincides largely with the pattern of damage shown in Figure 8. Additionally the 220kV line from Kikiwa substation south is modelled as being impacted by possibly 2 damaging aftershocks. No substations or powerstations are expected to be directly impacted. Figure 12 shows the aftershock zone superimposed over the HEP network.
56 ERI Research Report 2015/04
It is possible that the aftershocks will result in 20–30 days of additional repairs and inspections to the 50/60kV network, in particular affecting the eventual restoration time for Franz Josef which may now take approximately 100 days.
Figure 12 The South Island HEP network and the Alpine Fault aftershock zone. Aftershock rates as per Figure 7.
ERI Research Report 2015/04 57
4.3.8 Mitigation
A considered mitigation was to move the 110 kV lines in the Grey River valley north of Greymouth away from the Paparoa Ranges and closer to SH7. This would result in avoiding the loss of five wooden pole supports just north of Greymouth as the highway in this region does not experience any losses as a result of landsliding. The effect would be that Greymouth and Hokitika would remain connected to the HEP Transmission network at all times after the earthquake, leaving Franz Josef as the only disconnected node. In terms of total network restoration time, this would have little effect as crews from Hokitika simultaneously restore sections south of Hokitika as crews from Greymouth restore the losses there. However, this would have the effect of preserving electrical power to ~30,000 West Coast residents in the Greymouth and Hokitika areas who would otherwise lose power for the first two days after the earthquake.
4.3.9 Adaptation
A considered adaptation would be to construct a new 110 or 220 kV cable across the Southern Alps following SH7 through Lewis Pass, providing an alternative route to Arthur’s Pass that was able to carry a large electrical load. This would decrease the reliance on only the 110 and 220 kV cables through the Upper Buller Gorge. This was not considered a plausible option however, due to the amount of landsliding along SH7 in this region. Implementing this adaptation would have no effect on network connectivity post-earthquake and would have the undesired effect of increasing the total level of restoration required.
4.4 PIPED SERVICES-THE APPROACH USED
Work to estimate the towns and villages to include and the shaking intensity estimations for the locations were based on the MMI intensities from Exercise Te Ripahapa (Figure 2).
Table 16 The South Island townships included in the study.
Estimated length of Name MMI (Figure 2) pipe (km)
Greymouth MM8 35
Westport MM7 34
Hokitika MM8 45
Queenstown MM7 60
Franz Josef MM9 5
Mt Cook MM9 3
Fox Glacier MM10 2
Haast MM9 2
Whataroa MM9 9
Harihari MM9 3
Kumara MM9 7
58 ERI Research Report 2015/04
Several of the locations are also subjected to shaking by one or more of the aftershocks listed in the Te Ripahapa scenario. The breakages and repair times for these are calculated separately and the total breakages and repair times are listed.
The overall approach applied to the modelling of piped systems is based on that described in Moull (2012). All relationships, factors and multipliers are based on this work.
The breakage rates for the pipes at each location are calculated based upon a “standard” base rate modified by factors relating to the pipe size and material and the local ground class. The “standard” base rate curve was developed empirically. Modification factors were applied for pipe size and material. All pipes in this study are assumed to be < 400mm diameter.
The contribution to the pipe breakages was determined by assigning NZS1170 Ground Class percentages (% of classes B (Weak Rock), C (Shallow Soil), D (Deep/Soft Soil) and E (Very Soft Soil). Table 17 lists the estimated ground class percentages and liquefaction susceptibility factors for each location used in the calculation of the breakage rates.
Table 17 Estimated ground class distributions for the locations in the pipe services study.
Ground Ground Ground Name Class B/C Class D Class E
Greymouth 10% 85% 5%
Westport 0% 80% 20%
Hokitika 0% 100% 0%
Queenstown 70% 30% 0%
Franz Josef 0% 100% 0%
Mt Cook 100% 0% 0%
Fox Glacier 0% 100% 0%
Haast 0% 100% 0%
Whataroa 0% 100% 0%
Harihari 0% 100% 0%
Kumara 0% 100% 0%
Based upon the ground class estimates, a set of liquefaction susceptibility factors were estimated and an overall liquefaction multiplication factors was calculated. The estimates and the calculated factors for each location are listed in Table 18.
In Table 18 the weighted liquefaction multipliers are listed in the column labelled “Multiplier”. Each multiplier value is the weighted sum of the none/low/medium, high and very high contributing components, for example for Greymouth;
(0.6 * 2) + (0.35*9) + (0.05*27) = 5.7
ERI Research Report 2015/04 59
Table 18 The liquefaction factors and multiplier estimates for the pipe services locations.
Liqf Liqf Mult Liqf Mult Liqf Mult Name Liqf H Liqf VH Multiplier N/L/M N/L/M H VH
Greymouth 60% 35% 5% 2 9 27 5.7
Westport 60% 35% 5% 2 9 27 5.7
Hokitika 80% 15% 5% 2 9 27 4.3
Queenstown 90% 10% 0% 1 9 27 1.8
Franz Josef 100% 0% 0% 1 9 27 1
Mt Cook 100% 0% 0% 1 9 27 1
Fox Glacier 100% 0% 0% 1 9 27 1
Haast 0% 100% 0% 2 9 27 9
Whataroa 80% 20% 0% 2 9 27 3.4
Harihari 80% 20% 0% 2 9 27 3.4
Kumara 80% 20% 0% 2 9 27 3.4
Note in Table 18: Liqf = liquefaction; N/L/M = none/low/medium; H = high; VH = very high; Mult = multiplier.
A weighted average liquefaction multiplier was calculated for each location. The multiplication factors are listed in Table 19.
Table 19 The total multiplication factors applied to the base rate for all the locations in the piped services study.
Water Sewers & Stormwater Name Size Total Size Total Multiplier Others Multiplier Others Mult Factor Mult Factor
Greymouth 5.7 4 1 22.8 5.7 4 1.5 34.2
Westport 5.7 4 1 22.8 5.7 4 1.5 34.2
Hokitika 4.3 4 1 17.2 4.3 4 1.5 25.8
Queenstown 1.8 4 1 7.2 1.8 4 1.5 10.8
Franz Josef 1 4 1 4 1 4 1.5 6
Mt Cook 1 4 1 4 1 4 1.5 6
Fox Glacier 1 4 1 4 1 4 1.5 6
Haast 9 4 1 36 9 4 1.5 54
Whataroa 3.4 4 1 13.6 3.4 4 1.5 20.4
Harihari 3.4 4 1 13.6 3.4 4 1.5 20.4
Kumara 3.4 4 1 13.6 3.4 4 1.5 20.4
From Table 19 it can be seen the difference in the total multiplication factor for water pipes and sewers and stormwater pipes arises from a factor of 1.5 (versus 1) in “Others” for sewers and stormwater pipes. This difference relates to the assumption in the modelling that all water pipes are ductile whereas it is assumed that 50% of sewers (and stormwater pipes) are brittle.
60 ERI Research Report 2015/04
The base case breakage rates are listed in Table 20.
Table 20 The base case pipe breakage rates.
MM Zone MMI Breaks/km
6.1 0.0003
6.3 0.0006
MM6 6.5 0.0011
6.7 0.0018
6.9 0.0029
7.1 0.0045
7.3 0.0069
MM7 7.5 0.0102
7.7 0.0149
7.9 0.0212
8.1 0.0297
8.3 0.0409
MM8 8.5 0.0553
8.7 0.0738
8.9 0.0970
9.1 0.1260
9.3 0.1616
MM9 9.5 0.2050
9.7 0.2574
9.9 0.3200
10.1 0.3941
10.3 0.4814
MM10 10.5 0.5831
10.7 0.7011
10.9 0.8369
11.1 0.9923
The base rates are combined with the multiplication factors from Table 19 to arrive at break rate functions customized for each location. Examples of the curves are shown in Figure 12.
ERI Research Report 2015/04 61
100 Haast 10 Greymouth 1 Hokitika
0.1 Whataroa
0.01 Queenstown
Franz Josef Repairs per Kilometre per Repairs 0.001
Base Case 0.0001 6.0 7.0 8.0 9.0 10.0 11.0 Ground Shaking Intensity (MMI)
Figure 13 The calculated repair rates per km for ground shaking intensities for some locations in the study.
4.5 SUPPORTING SYSTEMS
Most, if not all piped infrastructure services require some sort of supporting systems. Because the level of modelling undertaken on the piped systems is at a coarse level, these have not been considered in detail. However, it is important that some consideration be given to these types of support components since in some cases the importance of the repairs to these elements may outweigh the repairs to the pipes.
For the purposes of this study it is assumed that each location has the elements listed below (Table 21). The size or number of elements is scaled based on the population the systems serve.
Table 21 Support elements for infrastructure services included.
Service Items Assumptions
Water • Purification system • Number of pumps < 1000 population = 2 • Storage facility • Increases by 1 for every 3000 afterwards • Pumps
Sewers • Treatment facility • Number of pumps < 1000 population = 2 • Pumps • Increases by 1 for every 3000 afterwards
Stormwater • Pumps • Number of pumps < 1000 population = 2 • Increases by 1 for every 3000 afterwards
Some assumptions have also had to be made concerning the repair times of the various pieces of plant. These are also scaled according to the size of the population they are expected to serve.
For the calculation of the damage to the supporting equipment, fragility functions based on the work undertaken as part of the Syner-g project (http://www.vce.at/SYNER- G/files/project/proj-overview.html) have been used. The fragility functions provide a level of damage based upon ground acceleration values (g) rather than felt (MMI) effects.
62 ERI Research Report 2015/04
MMI to acceleration approximations have been use to arrive at approximate damage states. The conversions are taken from the work of Abbott (2008) and are shown in Table 22.
Table 22 MMI to acceleration conversions used.
MMI6 MMI7 MMI8 MMI9 MMI10
0.09(g)–0.18(g) 0.18(g)–0.34(g) 0.34(g)–0.65(g) 0.65(g)–1.24(g) >1.24(g)
The damage classes and corresponding effects used in the fragility functions are listed in Table 23.
Table 23 Damage classes and consequences.
Damage(%) Functional description Effect Minor 15–30 Operational after minor repairs Normal capacity
Moderate 30–50 Operational after repairs Reduced capacity
Extensive 50–75 Partially operational after extensive repairs Reduced capacity
Complete 75–100 Not repairable No capacity
The damages arising from the MMI shaking for the supporting infrastructure components are shown in Table 24 to Table 27.
Table 24 Water treatment plant damage states.
Water treatment plant Acc(g) MMI approximation
Minor 0.15 MMI6
Moderate 0.3 MMI7
Extensive 0.55 MMI8
Complete 0.9 MMI9 & MMI10
Table 25 Sewage treatment plant damage states.
Sewage treatment plant Acc(g) MMI approximation Minor 0.15 MMI6
Moderate 0.3 MMI7
Extensive 0.45 MMI8
Complete 1 MMI9 & MMI10
Table 26 Water storage tank damage states.
Water storage tanks
Acc(g) MMI approximation Minor 0.25 MMI7
Moderate 0.52 MMI8
Extensive 0.95 MMI9
Complete 1.64 MMI10
ERI Research Report 2015/04 63
Table 27 Pumps (water, sewage lift stations, stormwater pumps).
All pumps Acc(g) MMI approximation
Minor 0.15 MMI6
Moderate 0.3 MMI7 & MMI8
Extensive 1.1 MMI9
Complete 2.1 MMI10
The resulting damage estimated at each of the locations in study is listed in Table 28.
Table 28 Supporting infrastructure component damage based on Figure 9 isoseismals.
Main Water Sewage Town Population Pumps Storage shaking treatment treatment
Greymouth 10000 MM8 Extensive Moderate Moderate Extensive
Westport 5000 MM7 Moderate Moderate Minor Moderate
Hokitika 4000 MM8 Extensive Moderate Moderate Extensive
Queenstown 12500 MM7 Moderate Moderate Minor Moderate
Franz Josef 300 MM9 Complete Extensive Extensive Complete
Mt Cook 250 MM9 Complete Extensive Extensive Complete
Fox Glacier 240 MM10 Complete Complete Complete Complete
Haast 80 MM9 Complete Extensive Extensive Complete
Whataroa 150 MM9 Complete Extensive Extensive Complete
Harihari 300 MM9 Complete Extensive Extensive Complete
Kumara 320 MM9 Complete Extensive Extensive Complete
In order to arrive at some rough estimates of possible supporting plant repair times the following rates have been assumed. For the treatment plants and storage the rates are days/1000 population served, while for the pumps the rates are per pump. These rates are listed in Table 29.
Table 29 The supporting plant repair rates (days per item, described above) assumed.
Water Sewage Pumps Storage treatment treatment Minor 1.5 4 2 1.5
Moderate 3 7 4 3
Extensive 6 10 8 6
Complete 10 16 12 10
The resulting estimates of the repair/rebuild and restoration are listed in Table 30.
64 ERI Research Report 2015/04
Table 30 Restoration time estimates for support infrastructure.
Restoration times (days) Water Sewage Pumps Storage treatment treatment Greymouth 60 35 40 60
Westport 15 21 10 15
Hokitika 24 21 16 24
Queenstown 37.5 42 25 37.5
Franz Josef 3 20 2.4 3
Mt Cook 2.5 20 2 2.5
Fox Glacier 2.4 32 2.88 2.4
Haast 0.8 20 0.64 0.8
Whataroa 1.5 20 1.2 1.5
Harihari 3 20 2.4 3
Kumara 3.2 20 2.56 3.2
4.6 WATER SUPPLY
Using the procedure discussed the estimated breakages and repair times for each location are listed in Table 31.
Table 31 The South Island townships water supply pipe repair times using Te Ripahapa (Figure 9) isoseismals.
Estimated length Breakages Repair time Name MMI (Figure 9) of pipe (km) (total) (days)
Greymouth MM8 35 44 44
Westport MM7 34 8 8
Hokitika MM8 45 43 43
Queenstown MM7 60 4 4
Franz Josef MM9 5 4 4
Mt Cook MM9 3 2 2
Fox Glacier MM10 2 5 5
Haast MM9 2 15 15
Whataroa MM9 9 25 25
Harihari MM9 3 9 9
Kumara MM9 7 20 20
ERI Research Report 2015/04 65
4.7 SEWERS
The same procedure used in estimating the water pipe breakages was used in estimating the sewer breakages and repair times. With sewers, however, some additional assumptions were made: 1. Sewers are generally larger diameter pipes, and are deeper. 2. Sewers are generally made of more brittle materials. 3. Although some may be pumped, sewers are more prone to problems due to a general reliance on gravity feeding. 4. Because of the increased depth and size of the sewer pipes, repairs are assumed to take 2.5 days rather than 1 day.
Table 32 The South Island townships sewer pipe repair times.
Estimated length Breakages Repair time Name MMI (Figure 2) of pipe (km) (total) (days)
Greymouth MM8 35 66 166
Westport MM7 34 12 30
Hokitika MM8 45 64 161
Queenstown MM7 60 7 17
Franz Josef MM9 5 6 15
Mt Cook MM9 3 4 9
Fox Glacier MM10 2 7 17
Haast MM9 2 22 55
Whataroa MM9 9 38 94
Harihari MM9 3 13 31
Kumara MM9 7 29 73
4.8 STORMWATER
Although it is considered as a separate system the stormwater system is treated in exactly the same way as the sewers. The repair times for the stormwater drains is assumed to be as listed in Table 32.
66 ERI Research Report 2015/04
4.9 THE OVERALL REPAIR TIMES
Taking into account the restoration times estimated for pipes and supporting infrastructure components and taking the longest times the overall estimates for the restoration times for each location for water supply, sewers and stormwater are listed in Table 33.
Table 33 Total days restoration times.
Water Sewers Stormwater Greymouth 60 166 166
Westport 15 30 30
Hokitika 44 164 164
Queenstown 38 38 24
Franz Josef 16 20 20
Mt Cook 16 16 16
Fox Glacier 20 20 20
Haast 16 58 58
Whataroa 16 99 99
Harihari 16 33 33
Kumara 20 73 73
4.10 AFTERSHOCK IMPACTS
Table 34 shows the modelled number of Mw5–Mw6 aftershocks, in the first 90 days, within 5km of the towns included in this part of the study.
Table 34 The modelled aftershock numbers (Mw5–6).
Name Min Max Mean Sum
Franz Josef 0.311 1.145 0.813 2.438
Hokitika 0.009 0.012 0.011 0.021
Greymouth 0.006 0.007 0.006 0.019
Westport 0.001 0.002 0.002 0.005
Queenstown 0.001 0.002 0.002 0.005
Mt Cook 0.015 0.019 0.017 0.034
Fox Glacier 0.145 0.821 0.483 0.966
Haast 0.047 0.068 0.057 0.115
Whataroa 0.096 0.346 0.201 0.604
Harihari 0.102 0.388 0.245 0.490
Kumara 0.014 0.018 0.016 0.033
Only Franz Josef and Fox Glacier have expected rates approaching or exceeding 1 so the repair and restoration times have not been modified.
ERI Research Report 2015/04 67
5.0 DISCUSSION AND CONCLUSIONS
This study has assessed the possible losses and disruption to critical infrastructure resulting from a major Alpine Fault earthquake in the South Island of New Zealand. It has been completed as part of the government-funded Economics of Resilient Infrastructure (ERI) research project. The Alpine Fault earthquake scenario presented is part of an on-going Research Aim within ERI to model a variety of different disruption scenarios to aid in the development of MERIT, a key output of the ERI program.
This study extends work undertaken for the 2013 Civil Defence Emergency Management Te Ripahapa Exercise by including new modelling of both the hazards expected to occur and the subsequent impacts to lifelines. The extensions to the initial 2013 scenario have focussed primarily on better quantifying the direct losses to infrastructure and developing first order estimates of the potential restoration times required. This has been accomplished via a collaborative workshop approach that has elicited crucial data from key personnel within the network providers involved. This process was undertaken in order to gather the best information available to date, but also to foster stronger and more widespread working relationships between network providers and hazard researchers.
The scenario developed is, in the author’s opinion, the most complete study of the consequences of an Alpine Fault earthquake as yet undertaken in New Zealand. Previous studies, such as McCahon et al. (2006), have successfully detailed the initial impacts and potential restoration times for an Alpine Fault earthquake, however the level of detail included herein is advanced of these previous scenarios. The increased level of detail in this scenario is considered a result of improved landslide modelling and the explicit and detailed involvement of the various network providers. All of the providers involved in this study had some level of pre-planning for a future Alpine Fault earthquake. This study has enabled the providers to integrate the most up-to-date scientific knowledge on a future earthquake into their planning, and has simultaneously provided the first detailed assessments of the likely disruption duration for multiple infrastructure in a single study. This information is likely to prove vital for exposed businesses, local populations, emergency managers, central and local government, scientists/researchers, and infrastructure providers. It is hoped that as well as providing key data for MERIT, this study can also foster stronger institutional relationships between researchers and infrastructure providers. Such relationships will enable further more robust studies into potential impacts of major disasters as well as provide a mechanism for ensuring the most up-to-date scientific knowledge is disseminated to network providers. Such activities may increase general disaster resilience in New Zealand.
Nevertheless, the practicalities of undertaking the modelling on a challenging timescale have meant that the number of systems included as part of the study have been limited to those mentioned above. A large part of the work has concentrated on the impacts to the transportation networks, namely state highways and rail. This is justified since these elements, particularly roads, are expected to have a major impact on the economic results due to the spatially diverse nature of the South Island communities and the dependence of key industries (mining, dairy, tourism) on reliable transport networks. Also modelled in this study are the impacts to the Hydroelectric Power network that also plays a vital role in key industry practices. A lesser part of the work has been devoted to the water, sewer and stormwater networks of the small townships directly impacted by the shaking.
68 ERI Research Report 2015/04
Despite the large amount of data gathered and presented in this study, several aspects could still be improved. Some estimation of the impacts of an expected aftershock sequence have been included. However, incorporating the extra impacts from an aftershock sequence in more detail would be a major piece of work in itself, requiring a new approach beyond that demonstrated here and is certainly well outside the scope of the research aim within which this work has been undertaken. The modelling of the piped systems (water, sewers and stormwater) and their support systems has been undertaken at a coarse level compared with the elaborate approach used for the State Highways. This is justified since the expected economic impacts arising from the damage to these systems is not likely to be significant compared to the transportation networks.
The modelling of the state highway network indicated that the West Coast region will be extremely isolated with no direct ground access from the eastern cities until many days after the main shock. Parts of the state highway network will be too unsafe to even start restoration until more than six months after the main earthquake. This is because there is expected to be significant aftershock activity following the main shock as well as heavy rainstorms (common in West Coast Region) which will potentially result in further landsliding. The modelling applied to the rail network has indicated that the Midland Line will suffer the worst damage, and with the expected aftershock activity, will face a prolonged restoration process. For the piped network systems an approach based on previous work undertaken to model the restoration of critical services in Wellington following a Wellington Fault earthquake was utilised.
ERI Research Report 2015/04 69
6.0 REFERENCES Abbott, P. L. (2008). Natural disasters. McGraw-Hill. ISBN 978-0-07-305034-8. 510p. Adams, J. (1980). Paleoseismicity of the Alpine Fault seismic gap, New Zealand. Geology, 8, 72-76. Barker, T. (2004). Economic theory and the transition to sustainability: A comparison of general equilibrium and space-time-economic approaches (Tyndall Centre for Climatic Change Research Working Paper 62). UK: University of Cambridge, Tyndall Centre for Climatic Change Research and Applied Economics. Barth, N. C. (2014). The Cascade rock avalanche: Implications of a very large Alpine Fault-triggered failure, New Zealand. Landslides, 11(3), 327-341. Berryman, K., Cochran, U. A., Clark, K. J., Biasi, G., Langridge, R. M., & Villamor, P. (2012). Major earthquakes occur regularly on an isolated plate boundary fault. Science, 336, 1690-1693. Biasi, G. P., Langridge, R. M., Berryman, K. R., Clark, K. J., & Cochran, U. A. (2015). Maximum- likelihood recurrence parameters and conditional probability of a ground-rupturing earthquake on the Southern Alpine Fault, South Island, New Zealand. Bulletin of the Seismological Society of America, 105(1), 94-106. Bird, J. F., & Bommer, J. J. (2004). Earthquake losses due to ground failure. Engineering Geology, 75, 147-179. Buxton, R., Wright, K. C., Daly, M. C., Timar, L., & Mieler, D. (2014). Single infrastructure failures: Capturing outage information for MERIT Modelling Economics of Resilient Infrastructure Tool. GNS Science Report 2014/12, 65p. Christophersen, A., & Gerstenberger, M. C. (2008). Forecasting aftershocks when catalog completeness is high, in Evison Symposium on Seismogenesis and Earthquake Forecasting, edited, Wellington, New Zealand. Cooke, R. M., & Goosens, L. H. J. (2004). Expert judgement elicitation for risk assessments of critical infrastructures. Journal of Risk Research, 7(6), 643-656. Moull, R. (2012). Lifeline utilities restoration times for Metropolitan Wellington following a Wellington Fault Earthquake, Report to the Wellington CDEM Group Joint Committee from the Wellington Lifelines Group. De Pascale, G., Quigley, M., & Davies, T. R. H. (2014). Lidar reveals uniform Alpine Fault offsets and bimodal plate boundary rupture behaviour, New Zealand. Geology, 42(5), 411-414, doi: 10.1130/G35100.1. Eberhart-Phillips, D. (1998). Aftershock sequence parameters in New Zealand, Bull. Seism. Soc. Am., 88, 1095-1097. ESRI. (2015). ArcGIS Desktop Help 10 – How Skyline Works. Retrieved 17 July 2015, from: http://resources.arcgis.com/en/help/main.10.1/index.html //00q90000008t000000.
Gerstenberger, M. C. (2003). Earthquake clustering and time-dependent⌗ probabilistic hazard analysis for California (PhD thesis). Gerstenberger, M. C., S. Wiemer, L. M. Jones, & P. A. Reasenberg (2005). Real-time forecasts of tomorrow's earthquakes in California. Nature, 435(7040), 328-331. Giovinazzi, S., Wilson, T. M., Davis, C., Bristow, D., Gallagher, M., Schofield, A., Villemure, M., Eidinger, J., & Tang, A. (2011). Lifelines performance and management following the 22 February 2011 Christchurch earthquake, New Zealand: highlights of resilience. Bulletin of New Zealand Society for Earthquake Engineering, 4(4), 402-417.
70 ERI Research Report 2015/04
Grassini, M. (2004, September). Rowing along the general equilibrium modelling mainstream. Paper presented at Ecocide conference on “Input-Output and General Equilibrium: Data, Modelling and Policy Analysis”, Brussels. Infometrics. (2012). West Coast labour market and economic profile. Retrieved 27 July 2014, from: http://www.infometrics.co.nz/reports/regional/TEC/WestCoastRevised – Jun2012.pdf. Kritikos, T., Robinson, T. R., & Davies, T. R. H. (2015). Regional coseismic landslide hazard assessment without historical landslide inventories: A new approach. Journal of Geophysical Research: Earth Surface, 120(4), 711-729. McCahon, I., Elms, D., & Dewhirst, R. (2006). Alpine Fault earthquake scenario (Technical Report). West Coast Engineering Lifelines Group. 204p. Norris, R., & Cooper, A. (2001). Late Quaternary slip rates and slip partitioning on the Alpine Fault, New Zealand. Journal of Structural Geology, 23, 507-520. Pollock, D. (2007). Aspects of short-term and long-term seismic hazard assessment in New Zealand (Master Thesis, ETH Zurich). Preuss, J., & Godfrey, J. (2006). Guidelines for developing an earthquake scenario. Oakland, CA.: Earthquake Engineering Research Institute. Robinson, T. R. (2014). Assessment of coseismic landsliding from an Alpine Fault earthquake scenario, New Zealand (PhD Thesis, University of Canterbury, 258p.). Robinson, T. R., & Davies, T. R. H. (2013). Review Article: Potential geomorphic consequences of a future great (Mw = 8.0+) Alpine Fault earthquake, South Island, New Zealand. Natural Hazards and Earth System Sciences, 13, 2279-2299, do:10.5194/nhess-13-2279-2013. Robinson, T. R., Wilson, T. M., Davies, T. R. H., Orchiston, C., & Thompson, J. R. (2014). Design and development of realistic exercise scenarios: A case study of the 2013 Civil Defence Exercise Te Ripahapa. GNS Science Miscellaneous Series 69, 122p. Scrieciu, S. S. (2007). The inherent dangers of using computable general equilibrium models as a single integrated modelling framework for sustainability impact assessment. A critical note on Bohringer and Loschel (2006). Ecological Economics, 60, 678-684. Smith, N. J., & McDonald G. W. (2014). Towards a model of dynamic general equilibrium-seeking economy, (in press). Sutherland, R. (1994). Displacement since the Pliocene along the southern section of the Alpine Fault, New Zealand. Geology, 22(4), 327-330. Utsu, T., Ogata, Y., & Matsu'ura, R. S. (1995). The centenary of the Omori formula for a decay law of aftershock activity. Journal of the Physics of the Earth, 43, 1-33. Wells, A., & Goff, J. (2007). Coastal dunes in Westland, New Zealand, provide a record of paleoseismic activity on the Alpine Fault. Geology, 35(8), 731-734, do:10.1130/G23554A.1. Wells, A., Yetton, M. D., Duncan, R. P., & Stewart, G. H. (1999). Prehistoric dates of the most recent Alpine Fault earthquakes, New Zealand. Geology, 27(11), 995-998. Whitman, R. V., Anagnos, T., Kircher, C.A., Lagorio, H.J., Lawson, R.S., & Schneider, P. (1997). Development of a national earthquake loss estimation methodology. Earthquake Spectra, 13(4), 643-661.
ERI Research Report 2015/04 71
This page is intentionally left blank.
APPENDICES
APPENDIX 1: STATE HIGHWAY INITIAL DAMAGE AND RESTORATION TABLES
Table A1 Initial damage and subsequent restoration times (in days) for surface rupture affecting State Highways. All times are absolute from T=0 accounting for the time required to gain access. The values ‘>180’ represent those sections considered too dangerous to restore within the first 6 months. Not including impacts from aftershocks.
Time to new condition Displacement (H – Horizontal; T0 Infrastructure Link (days) V – Vertical) Condition 0.25 0.5 0.75 1
SH6
East of Haast 10m H 0 >180 >180 >180 >180
Paringa River Bridge 8m H, 2m V 0 1 1 2 2
Karangarua River Bridge 8m H, 2m V (bridge affected) 0 1 1 2 2
Cook River Bridge 8m H, 2m V 0 >180 >180 >180 >180
North of Fox Township 8m H, 2m V 0 >180 >180 >180 >180
Between Waikukupa and 8m H, 2m V 0 >180 >180 >180 >180 Omoeroa Rivers
Either side of Waiho bailey 8m H, 2m V (bridge affected) 0 1 1 2 2 bridge
Either side of Wahataroa 8m H, 2m V (bridge affected) 0 1 1 2 2 bridge
SH73
Rocky point 5m H, 1m V 0 1 1 2 2
Table A2 Initial damage and subsequent restoration times (in days) for sections of State Highway affected by landslides. All times are absolute from T=0 accounting for the time required to gain access. The values ‘ >180’ represent those sections considered too dangerous to restore within the first 6 months.
Time to new condition Debris Debris per T0 Infrastructure Link (days) volume (m3) km Condition 0.25 0.5 0.75 1
SH6
Diana Falls to Gates of Haast 250,000 27,778 0.25 >180 >180
Thomas River 85,000 42,500 0.25 >180 >180 >180
Thomas River 100,000 33,333 0.5 >180 >180
Knights Point Area 650,000 118,182 0 >180 >180 >180 >180
Knights Point Area 230,000 41,818 0.25 >180 >180 >180
Lake Moeraki to Lake Paringa 250,000 125,000 0 >180 >180 >180 >180
Lake Moeraki to Lake Paringa 250,000 41,667 0.25 >180 >180 >180
Lake Moeraki to Lake Paringa 30,000 20,000 0.5 >180 >180
Paringa River 50,000 100,000 0 >180 >180 >180 >180
Paringa River 40,000 20,000 0.5 >180 >180
74 ERI Research Report 2015/04
Time to new condition Debris Debris per T0 Infrastructure Link (days) volume (m3) km Condition 0.25 0.5 0.75 1
Hunts Beach 10,000 10,000 0.5 >180 >180
Karangarua River 50,000 50,000 0.25 >180 >180 >180
Cook River 125,000 62,500 0.25 >180 >180 >180
Fox River 35,000 35,000 0.25 >180 >180 >180
Fox River 20,000 20,000 0.5 >180 >180
Fox Township to Franz Josef 2,750,000 152,778 0 >180 >180 >180 >180 Township
Lake Mapourika 40,000 40,000 0.25 2 3 4
Okarito River 100,000 100,000 0 2 3 4 7
Okarito River 150,000 50,000 0.25 2 3 4
Okarito River 20,000 20,000 0.5 1 2
Whataroa River 100,000 25,000 0.5 1 2
Mt Hercules 250,000 125,000 0 3 4 7 10
Mt Hercules 150,000 50,000 0.5 1 2
Ten Mile Creek 15,000 15,000 0.5 1 2
Punakaiki 40,000 40,000 0.25 2 3 5
Punakaiki 20,000 13,333 0.5 1 2
Upper Buller Gorge 10,000 20,000 0.5 1 2 (Iron Bridge/Lyell)
SH73
Old Christchurch Rd to Taipo 50,000 33,333 0.25 >180 >180 >180 River
Old Christchurch Rd to Taipo 30,000 10,000 0.5 >180 >180 River
Rocky Point to Otira River 700,000 42,424 0.5 >180 >180
Otira to Waimakariri River 2,000,000 125,000 0 >180 >180 >180 >180
Otira to Waimakariri River 500,000 83,333 0.25 >180 >180 >180
Bealey 30,000 30,000 0.25 >180 >180 >180
Cave Stream 10,000 20,000 0.5 >180 >180
SH7
Dobson 10,000 6,667 0.5 1 2
Reefton 50,000 9,091 0.5 1 2
Immediately West of Springs 650,000 130,000 0 17 22 29 43 Junction
Immediately West of Springs 500,000 62,500 0.25 10 14 21 Junction
Along Maruia River 600,000 109,091 0 15 18 24 36
ERI Research Report 2014/04 75
Time to new condition Debris Debris per T0 Infrastructure Link (days) volume (m3) km Condition 0.25 0.5 0.75 1
Along Maruia River 500,000 76,923 0.25 13 17 26
Along Maruia River 25,000 12,500 0.5 3 4
Along Lewis River 1,000,000 83,333 0.25 14 19 28
Along Lewis River 50,000 12,500 0.5 3 4
Hope River 100,000 100,000 0 13 17 22 33
Hope River 250,000 55,556 0.25 9 12 19
Hope River 50,000 12,500 0.5 3 4
SH65
Warwick River 10,000 10,000 0.5 1 2
SH69
Inangahua River Bridge 10,000 20,000 0.5 1 2
SH94
Hollyford Rd to Homer Tunnel 100,000 100,000 0 13 17 22 33
Hollyford Rd to Homer Tunnel 150,000 60,000 0.25 10 13 20
Hollyford Rd to Homer Tunnel 100,000 22,222 0.5 1 2
Hollyford Rd to Milford Sound 200,000 14,286 0.5 1 2
Table A3 Initial damage and subsequent restoration times (in days) for major State Highway bridges. All times are absolute from T=0 accounting for the time required to gain access. Comments column shows if alternative bridge structure (e.g. Ford or Bailey) will be used. Not including aftershocks.
Time to new condition T0 Comments Infrastructure Link MMI (days) Condition 0.2 0.4 0.6 0.8
SH6
Camerons Creek 7 0.4 2/28 Ford/Bailey
Makarora River 7 0.4 2/28 Ford/Bailey
Gates of Haast 7 0.6
Haast River (Haast Pass) 7 0.6
Haast River (Haast 8 0.6 Township)
Waita River 8 0.6
Ship Creek 8 0.6
Whakapohai River 8 0.6
Moeraki River 8 0.6
Paringa River 8 0.6
Mahitahi River 8 0.4 28
Papakeri Creek 9 0 2 2 7 Bailey
76 ERI Research Report 2015/04
Time to new condition T0 Comments Infrastructure Link MMI (days) Condition 0.2 0.4 0.6 0.8
Makawhio River 8 0.6
Manakaiaua River 8 0.4 7 Bailey
Karangarua River 8 0 3 7 21 Surface rupture (internal); Bailey
Cook River 8 0 3 7 21 Surface rupture (internal); Bailey
Fox River 8 0.4 7 Repairs/Ford
Waikukupa River 8 0.6
Omoeroa River 8 0.6
Docherty Creek 8 0.6
Waiho River 8 0.6
Tartare Stream 8 0.6
Potters Creek 8 0.6
Waitangitanoa River 8 0.6
Matainui Creek 8 0.6
Whataroa River (South) 8 0.6
Whataroa River (North) 8 0.6
Poerua River 8 0.6
Wanganui River 8 0.6
Waitaha River 8 0.4 28
Kakapotahi River 8 0.6
Mikonui River 8 0.4 28
Totara River 8 0.6
Hokitika River 8 0.6
Arahura River 8 0.6
Waimea Creek 8 0.4 21
Taramakau River 7 0.4 70
New River 7 0.4 14
Grey River 7 0.6
Ten Mile Creek 7 0.6
Punakaiki River 7 0.6
Porarari River 7 0.6
SH73
Wainihinihi River 8 0.4 2 Ford
Griffen Creek 8 0.4 2
Taipo River 8 0.4 2
ERI Research Report 2014/04 77
Time to new condition T0 Comments Infrastructure Link MMI (days) Condition 0.2 0.4 0.6 0.8
Otira Viaduct 7 0.6
Waimakariri River 7 0.6
SH7
Arnold River 7 0.6
Red Jacks Creek 7 0.6
Ahaura River 7 0 - 7 56
Upper Grey River 7 0.6
Little Grey River (South) 7 0.6
Stony Creek 7 0 1 1 1 Ford
Little Grey River (North) 7 0.6
SH65
Maruia River 7 0.6
78 ERI Research Report 2015/04
APPENDIX 2: RAIL NETWORK INITIAL DAMAGE AND RESTORATION TABLES
Table A4 Initial damage and subsequent restoration times (in days) for major rail bridges. Not including aftershocks.
Time to new T0 Infrastructure Link MMI condition (days) Condition 1 Hokitikia Industrial
Arahura River 8 0 1
Waimea Creek 7 0 1
Taramakau River 7 0 1
New River 8 0 1
Midland
Arnold River 8 0 1
Crooked River 8 0 1
Taramakau River 8 0 1
Bealey River (North) 7 0 1
Bealey River (South) 7 0 1
Waimakariri River 7 0 1 Stillwater-Westport
Arnold River 7 0 1
Red Jacks Creek 7 0 1
Ahaura River 7 0 1
Upper Grey River 7 0 1
Little Grey River (South) 7 0 1
Stony Creek 7 0 1
Little Grey River (North) 7 0 1
Inangahua River 7 0 1
Table A5 Initial damage and subsequent restoration times (in days) for surface rupture affected rail lines. aMID – Midland Line. Not including aftershocks.
Time to new T0 Infrastructure Link Displacement condition (days) Condition 1 Midland
Lake Poerua 5m H, 1m V 0 3
ERI Research Report 2014/04 79
Table A6 Initial damage and subsequent restoration times (in days) for landslide affected rail lines. All times are absolute from T=0 accounting for the time required to gain access. aMID – Midland Line; SWL – Stillwater-Westport Line. Not including aftershocks.
Time to new Debris Debris T0 Infrastructure Link condition (days) Volume per km Condition 1 Midland
Cass (South of Waimakariri River) 30,000 30,000 0 3
Waimakariri River to Mingha River 250,000 71,429 0 21
Mingha River to Otira Tunnel 500,000 125,000 0 58 (Arthurs Pass)
Otira Tunnel (North) to Otira 300,000 60,000 0 25
Otira Tunnel (North) to Otira 500,000 166,667 0 58
Taramakau River 200,000 30,769 0 17
Lake Poerua 40,000 40,000 0 3
Lake Brunner 30,000 30,000 0 3 Stillwater-Westport
Dobson 10,000 6,667 0 1
South of Reefton 20,000 10,000 0 2
Inangahua River Bridge 10,000 10,000 0 1
80 ERI Research Report 2015/04
Contact details: GNS Science 1 Fairway Drive, Avalon Lower Hutt 5010 PO Box 30 368 Lower Hutt 5040 New Zealand T +64-4-570 1444 F +64-4-570 4600