1
Insurance loss drivers and mitigation for Australian housing in severe wind events Daniel J. Smith1, David J. Henderson1, John D. Ginger1 Cyclone Testing Station, College of Science, Technology, and Engineering, James Cook University, Townsville, Australia email:
[email protected],
[email protected],
[email protected]
1
ABSTRACT: Damage investigations conducted by the Cyclone Testing Station (CTS) following severe wind events have typically shown that Australian homes built prior to the mid-1980s do not offer the same level of performance and protection during severe wind events as homes constructed to contemporary building standards. A direct relationship between observed damage modes and societal cost is needed to inform cost-benefit analysis of retrofit mitigation solutions. Policy data from one insurer in the North Queensland region of Australia during Cyclone Yasi (2011) were analyzed to identify correlations between claim value, typical damage modes, and construction age. Preliminary results suggest that ancillary damages (i.e. fabric shade coverings, roofing vents, wind-borne debris impacts on wall cladding, etc.) contribute significantly to overall loss for all ages of housing construction. In addition, the frequency of severe structural and water ingress damages is shown to be higher for legacy housing in a region subjected to wind speeds below design level. Structural retrofitting details exist for some forms of legacy housing in Australia but the utility of these details is limited. Hence, the issues of retrofitting legacy Australian housing, including practicality, economy, etc. must also be analyzed. A recently conducted survey of building industry representatives in Australia indicates that the utility of current retrofitting guidelines is inhibited by cost and facultative status. The details of the survey are discussed herein. KEY WORDS: ICWE14; Resilience; Retrofitting; Legacy housing; Standards Australia; Insurance; Claims Analysis; Structural upgrades; Wind Resistance; Cyclone Yasi. 1
INTRODUCTION
Damage investigations carried out by the Cyclone Testing Station (CTS) following severe wind storms have typically shown that Australian houses built prior to the mid-1980s do not offer the same level of performance and protection during windstorms as houses constructed to contemporary building standards. Structural retrofitting details exist for some forms of legacy housing but the uptake of these details is limited. There is also evidence that retrofitting details are not being included into houses requiring major repairs following severe storm events, thus missing the ideal opportunity to improve resilience of the house and community. Hence, the issues of retrofitting legacy housing, including feasibility and hindrances on take-up, etc., must be analyzed. The primary objective of this research is to develop cost-effective strategies for mitigating damage to housing from severe windstorms across Australia. Strategies will be (a) tailored to aid policy formulation and decision making in government and industry, and (b) provide guidelines detailing various options and benefits to homeowners and the building community for retrofitting typical at-risk houses in Australian communities. Tropical Cyclone Tracy resulted in extreme damage to housing in December 1974, especially in the Northern suburbs of Darwin [1]. Changes to design and building standards of houses were implemented during the reconstruction. The Queensland Home Building Code (HBC) was introduced as legislation in 1982 with realization of the need to provide adequate strength in housing. By 1984 it is reasonable to presume that houses in the cyclonic region of Queensland were being fully designed and built to its requirements [2-4]. Damage investigations of housing, conducted by the Cyclone Testing Station (CTS) in Northern Australia from cyclones over the past fifteen years have suggested that the majority of houses designed and constructed to current building regulations have performed well structurally by resisting wind loads and remaining intact [5-12]. However, these reports also detail failures of contemporary construction at wind speeds below design requirements, in particular for water-ingress related issues. The poor performance of these structures resulted from design and construction failings, poor connections (i.e. batten/rafter, rafter/top plate) (Figure 1 and Figure 2), or from degradation of construction elements (i.e., corroded screws, nails and straps, and decayed or insect-attacked timber). Hence, the development of retrofit solutions for structural vulnerabilities are critical to the performance longevity of all ages of housing. Damage surveys invariably reveal some failures due to loss of integrity of building components from aging or durability issues (i.e., corrosion, dry rot, insect attack, etc.). The CTS conducted a detailed inspection of houses built in the 1970s and 1980s in the tropics [13]. Although the majority of surveyed houses appeared in an overall sound condition, the majority had potential issues like decay of timber members, corrosion at a connections, missing/removed structural elements, etc. The damage survey after 14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015
2
Cyclone Yasi showed substantial corrosion of roof elements in houses less than 10 year old [6]. This study confirmed that retrofitting for improved wind resistance is only part of the process. Ongoing maintenance is also an important part of improving community resilience in severe weather. Considering the prevalence of roofing failures due to inadequate upgrading techniques, current building industry literature for upgrading the wind (and water-ingress) resistance of existing Australian housing were reviewed. In parallel, a brief internet-based questionnaire was distributed to a wide range of Australian building industry constituents in order to identify specific limitations of current upgrading guidelines. The results of the survey are discussed in the current paper. A recently conducted analysis of insurance claims data from Cyclones Yasi (2011) and Larry (2006) in the North Queensland coastal region of Australia is also discussed. The primary objective was to extract insights on typical vulnerabilities in Australia residential housing during cyclonic wind events. Preliminary findings of the analysis are discussed herein, emphasizing comparisons to typical damage modes observed from post-event damage assessments.
Figure 1. Wind-induced failure of a new (< 1 year) metal cladding roof on an old house at the rafter to top-plate connection during Cyclone Marcia (2015) in Yeppoon, Australia.
Figure 2. Wind-induced failure of a metal cladding roof (left) at the batten to rafter connection (right) during Cyclone Marcia (2015) in Yeppoon, Australia. 2
POST-EVENT DAMAGE OBSERVATIONS
Following Cyclone Yasi in 2011, Boughton et al [6] showed that homes correctly designed and constructed to the Australian building standards introduced in the 1980s generally performed well under wind load actions. Damage survey results indicated that in the most highly affected areas, ~3% of post-1980s homes experienced significant roof damage, in contrast to ~15% for pre1980s homes. Greater than 20% of the pre-1980s housing experienced significant roof loss in some areas. The relatively low incidence of roofing damage to post-1980s buildings indicates that modern building practices are able to deliver higher performance for the roofing structure in severe wind event conditions. A damage survey following Cyclone Larry [7] showed that although wind-induced structural damage was minimal for 95% of contemporary housing, many contemporary houses experienced water ingress damage from wind-driven rain. A survey conducted by Melita [10], details building envelope failures during Cyclone Larry. Approximately 75% of post-1985 homes experienced
14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015
3
water ingress through breaches in the building envelope (i.e. broken windows, punctured cladding, failed fascia or guttering, etc.). In many cases replacement of internal building components and owner contents were required. These observations are similar to those of other other post-event damage assessments in Australia (e.g. Cyclone Winifred [11], Cyclone Vance [12], Cyclone Ingrid [8], and Cyclone George [5]). Consistent findings include: • In general, contemporary construction performance for single family residential housing was adequate under wind loading • Significant structural damage to legacy housing was typically associated with loss of roof cladding and/or roof structure. There were many examples of legacy housing with relatively new roof cladding installed to contemporary standards (i.e. screwed fixing as opposed to nailed) but lacking upgrades to batten/rafter or rafter/top-plate connections, resulting in loss of roof cladding with battens attached • Corrosion or degradation of connections and framing elements initiated failures • Where wind-induced structural failures were observed for contemporary housing, they were often associated with either poor construction practice or the application of an incorrect site design wind speed • Breaches in the building envelope (i.e. failed doors and windows, debris impact, etc.) exacerbated failure potential from increased internal pressures • Extensive water ingress damages were observed for structures with and without indication of exterior building damage These observations suggest the majority of contemporary houses remained structurally sound, protecting occupants and therefore meeting the life safety objective of Australia’s National Construction Code (NCC). However, contemporary homes did experience water ingress (resulting in loss of amenity) and component failures (i.e. doors, soffits, guttering, etc.) with the potential for damage progression to other buildings, thus failing to meet specific objectives and performance requirements of the NCC. 3
INSURANCE CLAIMS ANALYSIS
A direct relationship between observed damage modes and societal cost is needed to inform cost-benefit analysis of retrofit mitigation solutions. Policy data from one insurer in the North Queensland region of Australia during Cyclone Yasi (2011) were analyzed to identify correlations between claim value, typical damage modes, and construction age. This was achieved by extracting qualitative and quantitative insights from aggregated insurance policy data from one insurer at the time of the event. The aggregated data included information on policies both with and without a claim for Cyclone Yasi. A more detailed analysis that addresses topographic effects, etc. on the wind field at policy level will be completed as a next step (current results are preliminary). Loss Ratios The claims data was subdivided by loss ratio (i.e. claim value / insured value) into five bins (Table 1). Bins were specified based on a general range of damage modes observed for a corresponding range of loss ratios. These relationships were determined by a review of 180 assessor’s reports for claims of varying loss ratio. The loss ratio bins are used broadly to discuss correlation between damage severity and construction age of housing in the Townsville region. Table 1. Loss ratio bins for Cyclone Yasi insurance claims analysis (one insurer) with typical damage modes extracted from assessor’s reports (note: typical damages for higher loss ratio bins also include all damages from lower ratio bins). Loss Ratio Bin Description Typical Damage 0 No claim filed N/A Minor roofing issues and water ingress, minor 0 – 0.09 Minor damage debris impact damage, fencing, fabric shade coverings, roofing vents, etc. 0.1 – 0.49
Moderate damage
0.5 – 0.99
Severe damage
> 1.0
Severe damage /underinsured
Moderate roofing and water ingress, ceiling damage, broken fenestration, exterior cladding, etc. Severe roofing failures, extensive water ingress damages, interior components damage and broken fenestration, etc. Severe roofing failures, extensive water ingress damages, interior components damage and broken fenestration, etc.
The “No claim filed” bin refers to the number of policies that did not file a claim following Cyclone Yasi and hence are assumed to have avoided damage. The “Minor damage” bin refers to claims with small loss ratio (< 10%), which was generally associated with ancillary damages (i.e. fabric shade coverings, fencing, roof vents, etc.) and minor water ingress damages. The “Moderate damage” bin refers to claims with moderate damages typically to roof cladding, the building interior, etc. due to wind load and water ingress. The “Severe damage” bin refers to claims with relatively large loss ratios associated with damages typically including severe roofing failures, large-scale water ingress damage to building interior, etc. The “Severe damage/underinsured” bin refers to claims with loss ratios greater than one (i.e. claim exceeds policy limit). This bin was created to characterize the number proportion of housing that suffered complete failure. Damages observed for claims in this bin were typically extreme
14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015
4
cases of the damage modes described for the “Severe damage” bin. Figure 3 shows the spatial distribution of properties falling within the four non-zero loss ratio bins for the North Queensland coastal region and the estimated wind field near the point of landfall for Cyclone Yasi.
Figure 3. North Queensland coastal region impacted by Cyclone Yasi (2011) with distribution of claims subdivided by four loss ratio bins (claim value/insured value) and wind field estimation [6]. Loss ratios increase directly with wind speed. The frequency of large loss ratios (i.e. large claims) was greatest, in geographic locations nearest the point of landfall (i.e. Cardwell, Mission Beach, etc.) where wind speed estimates were 210-240 km/h [6]. However, claim frequency was significant for the entire North Queensland region, even in areas of relatively low wind speed estimates (i.e. ~135 km/h in Townsville, ~90 km/h in Cairns). A total of 26% (14,282) of policies in the area affected by Cyclone Yasi (North Queensland coastal region) filed a claim. Approximately 86% (12,296) of those claims were within the “Minor damage” loss ratio bin. Claims in this loss ratio bin represent 29% of the total claims payout cost accrued by one insurer for Cyclone Yasi in the North Queensland coastal region. Approximately 12% (1,665) of claims in the North Queensland coastal region were the “Moderate damage” loss ratio bin, contributing 44% to the total payout cost for one insurer. The majority of claims this size were filed in the areas subjected the highest wind speeds (Figure 3). Severe damage claims (including underinsured bin) represented at total of 27% of the total claim-related losses for one insurer in the North Queensland coastal region. Typical damage modes for claims of this size generally more extreme versions of those described for the “Moderate damage” bin (i.e. severe roofing failures, interior damage from water ingress, fenestration damage, etc.). The contributions to overall cost from moderate to severe claims typically associated with roofing and building envelope failures, indicate that a mitigation program emphasizing structural roofing and opening protection upgrades may significantly reduce the frequency of relatively large claims and thus decrease societal cost. Townsville Analysis Region To isolate a relatively high population of housing subjected to a similar range of wind field characteristics (i.e. velocity, direction, and duration) and rain fall intensity, preliminary analysis emphasized the Townsville region. Peak 3-second gust wind speed measured at the Townsville airport weather station (10 m) was 135 km/h during Cyclone Yasi. Figure 4 shows the spatial distribution of construction ages for policies in the city centre of the Townsville region. For clarity, the greater Townsville area of modern housing suburbs to the North, West, and South are omitted. A total of 23,878 policies from the North Queensland coastal region were included in the Townsville region, 30% of which filed a claim associated with Cyclone Yasi.
14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015
5
Figure 4. Spatial distribution of residential property construction age in the city centre area of the Townsville region for policy holders from one insurer during Cyclone Yasi (2011) . The inland moving housing development of the central Townsville area over time is clearly shown. In general, for the main part of the city, the housing stock near the coastline is of an older construction age. This is an important consideration for risk modelling. Wind conditions near the coast will generally be more severe than inland areas during cyclonic events. Figure 5 shows the loss ratio for claims across the central Townsville area. Considering wind speeds during Cyclone Yasi were just above 50% design level for the region, there was a high amount of claims distributed throughout the city and across all age groups. “Minor damage” claims were dominant and occurred uniformly throughout the region. Their occurrence appears to be independent of housing age and proximity to the coast. However, moderate and severe loss ratio claims are more prevalent in areas near the coastline where older housing is also more prevalent.
Figure 5. Spatial distribution of claims subdivided by four loss ratio bins (claim value/insured value) in the city centre area of the Townsville region for policy holders from one insurer during Cyclone Yasi (2011).
14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015
6
Table 2 shows the relative contributions of claims in each loss ratio bin as a proportion of the total number of claims filed in the Townsville region. As suggested by Figure 5, the majority of claims (~94%) were within the “Minor damage” loss ratio bin (00.09). The sum of claim values from Cyclone Yasi for the Townsville region was $AUD 63.5 million (one insurer). Minor claims (e.g., fencing, fabric shade coverings, garage doors, minor impact damages to wall cladding, etc.) comprised 60% of the total claims cost of for the Townsville Region. For “Moderate damages” (0.1-0.49) in the Townsville Region, 390 claims were filed (i.e. 5% of the Townsville regions claims contributed 32% of the total claims cost). These claims were generally associated with moderate damage to the roofing structure, water ingress damages to the building interior, etc. and occurred in an area where wind speed estimates were significantly less than design level (240 km/h). This supports the need for improved building standards for both upgrading of older housing and maintenance and certification for newer construction. Table 2. Frequency and cost statistics for the four non-zero damage levels and claims from one insurer in the Townsville Region during Cyclone Yasi (2011). Loss Ratio
% Total Claim Cost
# Claims
% Claims
0-0.09
60%
6851
94%
0.1-.49
32%
390
5%
0.5-.99
6%
27
= 1.0
2%
5
= 1.0
25.6%
50%
70.8%
73.4%
0% 1.0” bin proportions are
