What we do
Estimating Expected Loss and Maximum Foreseeable Loss
Central to many of our engagements is the task of estimating the scale of potential economic and environmental losses that may be suffered over time due to bushfires. This is important for those responsible for managing fire in a state or region, or enterprises with exposure to fires that are either caused by it or by third parties in areas such as forestry zones or electricity network footprints. We can be interested in expected losses, the probability distribution of annual losses over time, and/or maximum foreseeable losses.
For wildfire our go-to method is numerical simulation, usually through use of the Spark Wildfire Toolkit that we deploy for massive computational experiments on cloud and/or high performance computing resources. Our largest single computational experiment to date is a 2.1 million simulation ensemble executed for Citipower-Powercor in 2018 in collaboration with ENEA Australia, and one of our most recent is a 1.3 million simulation ensemble for SA Power Networks. In these experiments for power companies the simulated fire ignition points are selected at no more than 500m intervals along electricity network corridors, and several dozen simulations each with a different representative weather pattern are executed for every ignition point. The results are processed and aggregated into a comprehensive picture of fire consequence which then serves as input to quantitative risk analyses supporting multiple functions including infrastructure investment decision support, regulatory development, regulatory exemption submissions, safe and risk-reducing operational procedure development, and insurance negotiations.
Risk is the product of likelihood and consequence, and we have many years of experience in data science directed at the task of estimating the rate of fire ignitions in regions due either to multiple causes (arson, lightning, machinery, etc.) or to specific groups of causes (most often, electrically-ignited fires). Much of our effort in this vein during the past decade has been in estimating the rates of faults and fire ignitions on High Voltage (HV) and Medium Voltage (MV) electricity distribution systems.
Quantitative Assessment of New Technologies
Our ignition likelihood estimation work has included multi-year engagements in Victoria, where we developed the technology-dependent ignition likelihood models that have underpinned electrical safety regulatory changes (specifically the 2016 Amendments to the Victorian Electrical Safety Act 2013) and decision support tools for industry. The 2016 Amendments require electricity distribution companies to progressively install Rapid Earth Fault Current Limiter (REFCL) technologies so as to reduce bushfire ignition rates by 50% to 60%. The quantification of the estimated ignition rate reduction due to REFCL is one of the key outcomes of our ongoing work for the Victorian Powerline Bushfire Safety Program (PBSP). This was carried out in conjunction with Facio Pty. Ltd.
Firefighting First Attack Success
Another important aspect of likelihood is the estimation of firefighting first attack success probabilities. In all places were large-scale forest fire and/or grass fire is a threat, firefighting is our most important defence. Firefighting routinely contains 90% or more of fire starts during first attack even on the most dangerous of fire days. Any quantitative assessment of fire risk must account faithfully for this critically important function. We have developed and applied first attack success probability models for South Australia, Victoria, and the forestry areas of Chile.
Data analysis applied to firefighting can also provide great insight into the effectiveness of firefighting tactics, equipment types, force sizes and relevant operational KPIs. In collaboration with CSIRO Chile and forestry industry partner Arauco we have undertaken studies in response to the major fire events that occurred in Chile in early 2017. We have informed and developed updated tactics, resourcing levels and first attack firefighting sortie compositions that have contributed to dramatic decreases (by 80% to 90%) in fire-related forestry plantation losses since 2018.
Analytics for Risk Reduction and Mitigation Programs
Analyses of fire consequence, ignition likelihood and suppression effectiveness come together to inform decisions in risk reduction initiatives and risk mitigation programs. Our flagship series of projects in this sense is with the Victorian Powerline Bushfire Safety Program. In these projects we have provisioned PBSP with fire risk analytics for targeting mitigation investments (incl. powerline undergrounding and zone substation selection for REFCL deployment), parameterizing the “F Factor” fire ignition reduction incentive scheme, estimating and reporting on annual fire risk reduction progress, and analytics supporting regulatory change. We have done this with partners including Facio Pty Ltd and Mondo.
For individual electricity distribution businesses we have provided fire risk assessments that: guide investment in asset renewal and fire mitigation actions; the formulation of network operational procedures including de-energization of lines; inform insurance negotiations; and constitute part of distribution business’ submissions to the Australian Energy Regulator justifying capital investments and operational expenditure for fire safety initiatives. We have executed projects and partnerships with more than half of all electricity distributors in Australia over the period 2012-2020.
Regulatory Exemption Studies
Regulations such as the Victorian Electrical Safety Act (2013/2016) bring about positive change in the fire risk profiles of electricity businesses, for example through requiring the installation of equipment that will de-energize powerlines before an electrical discharge can ignite a fire. However, there are cases where the mandated actions and investments are not cost or risk optimal due to special circumstances in particular geographies. Our team has partnered with electricity distribution businesses to build quantitative and semi-quantitative risk assessments of alternative mitigations such as powerline undergrounding, network re-configuration, vegetation management, and covered conductor systems. In doing so we have helped to save businesses and consumers at least $60M in avoided capital costs. The alternative solutions have also simultaneously improved electricity system reliability and bushfire safety (i.e., extra fire risk reduction).
Risk Mitigation and Long-Term Planning in Forestry
How and where fire risk will increase under future climates is of intense interest to the finance sector, citizens and businesses on the land, fire agencies and land managers, and exposed industries including forestry. Predicting fire occurrence and fire outcomes/losses under climate change is an ongoing science challenge, but there are decisions that must be made today using the best available information. As a key example, our risk analytics expertise and models have been applied within optimisation (Operation Research) models co-developed with forestry industry players. These inform multi-decadal decisions around plantation species, species/product regionalisation and diversification, and forestry supply chain planning. Fire activity can be highly damaging for forestry supply chains over many years and multiple crop generations. Its direct consideration in long-term forestry planning is necessary in any region where forest fires may have a non-trivial impact.
Combustible Cladding Removal Prioritization
Recently we have expanded our focus from bushfire/wildfire risk into combustible cladding, i.e, building façade systems which are flammable and pose what is now considered to be an unacceptable risk to life and property. The world has witnessed events such as the Grenfell Tower disaster in June 2017, and in multiple places worldwide this has prompted actions to mitigate these risks. These actions include the removal of the combustible cladding materials from buildings and their replacement with safer alternatives. Victoria is a world leader in this regard: the state government has established an approximately $0.5B fund for the rectification of residential buildings with combustible cladding. This fund is now administered by Cladding Safety Victoria (CSV) and the scope of CSV includes the development of a building risk prioritization approach which can be used to prioritize buildings for rectification so that the highest risks are retired earlier. CSV engaged the Data61 Fire Risk Analytics team in 2020 to develop the prioritization methods. The methods are based on a hybridization of expert judgements and risk-based deductive logic approaches. In early 2021 the prioritization methods were applied to the fleet of buildings in CSV’s remit and the prioritized lists of buildings can now be used as inputs in decision making through the remainder of the program’s timeline.