Risk, Reliability and Societal Safety Aven & Vinnem (eds) 2007 Taylor & Francis Group, London, ISBN 978-0-415-44786-7 A study of hurricane mitigation cost effectiveness in Florida J.-P. Pinelli, C.S. Subramanian & F. Garcia Florida Institute of Technology, Melbourne, Florida, USA K. Gurley University of Florida, Gainesville, Florida, USA ABSTRACT: The mitigation studies described here were carried on by a team of engineers from different institutions in Florida with the sponsorship of the Florida Sea Grant Consortium. The authors used the Florida Public Hurricane Loss Projection Model for a state of the art comprehensive study of mitigation cost effectiveness in the State of Florida. The work included engineering-based cost benefit evaluation of various mitigation measures. The aim is to articulate the relationship between property owners, insurance companies, financial institutions, and regulatory agencies, to devise the most effective mitigation strategies. A direct outcome from this research is maps of (benefits minus costs) for the entire State of Florida, for different mitigation measures, applied to different building types. These maps are presented and discussed. Conclusions regarding the more cost effective mitigation strategies for Florida are presented. The methods presented can be extrapolated to other regions of the world and to other perils. 1 INTRODUCTION Historically, hurricanes have presented a constant threat to the State of Florida. The cost of the wind induced destruction affects individual home owners, the insurance industry and the State in general. Mitigation of hurricane damage is therefore a critical issue in the State, and given the limited resources available, it is important to define which mitigation measures would be more cost effective for home owners. Consequently, the Florida Sea Grant Consortium has funded research at Florida Tech and the University of Florida to investigate the cost effectiveness of different mitigation measures for residential structures. This paper presents some result of this work. The authors analyzed the cost effectiveness of various combinations of mitigation measures for different types of residential structure of different age and quality.this is the first formal attempt to present a fullscale model that dictates which mitigation measures applied together will result in the highest economic benefit for specific regions in the State of Florida. Different sets of mitigation measures were investigated that combined improved roofing materials, improved roof to wall connections, and different kinds of opening protection. This mitigation measures were applied to typical timber box and masonry residential structures of different age and quality of construction, from weak pre-1970 to stronger post 2002 construction. In each case, a detailed cost analysis of the unmitigated and mitigated building was performed and the relative cost effectiveness of mitigation was assessed. 2 FLORIDA PUBLIC HURRICANE LOSS PROJECTION MODEL The work was possible thanks to the utilization of the Florida Public Hurricane Loss Projection Model (FPHLPM), a hurricane risk analysis model previously developed with the joint collaboration of Florida Tech, and the University of Florida, among others, under the leadership of Florida International University. The model is a risk-assessment system that can analyze portfolio files of any insurance company to compute the expected annual losses, or actual scenario losses, of the insured policies, for single family residential houses. The output from this analysis can then be used for the validation of other models, rate making, and other tasks. In particular, the FPHLPM can be used to analyze the effectiveness of mitigation measures. The model consists of 3 distinct parts that are integrated into a computer platform: the wind field model, the vulnerability model, the actuarial component. For more information on the FPHLPM see Pinelli et al. (2004, 2006). 1505
Table 1. Non-mitigated/basic models weak, medium, and strong. Garage Shea- Roof-wall Roof Model door thing connection shape Shutters Weak 30 psf 6d nails Toe nails gable none Medium 30 psf 8d nails clips gable none Strong 52 psf 8d nails straps gable none hip In particular, the FPHLPM models the losses for the most common types of single family homes in Florida, which are masonry homes with timber truss roofs and timber box like structures, also with timber truss roofs. Each type of structure is itself represented in the FPHLPM with differing strengths as a weak, a medium, and a strong model based on the age. The weak variant corresponds of course to older structures while the strong one corresponds to the newest building built according to the latest building codes. The models that were considered in this study are listed in Table 1. The numbers in the garage door column indicate the wind pressure the door is rated for. 3 MITIGATION MEASURES The authors examined the cost-effectiveness of applying different sets of mitigation measures to residential homes. A mitigation set is a combination of different mitigation elements, applied as a retrofit to an existing home, or as an upgrade to a new home. Retrofitting in this sense refers to the act of removing an older element, and replacing it with a new improved element, or simply adding the upgraded element if there is no need to remove previous material. Individual mitigation measures were selected based on the recommendations of the Florida Commission on Hurricane Loss Projection Methodology (2006) and included: Bracing of gable ends Choice of a hip roof 110 mph rated shingles Roof barrier membrane Roof plywood joint sealing Roof taping Nailing of deck using stronger 8d nails Strap timber roof-to-wall, and timber wall to floor connections Vertical reinforcing of masonry walls Steel shutters for windows, doors, and skylights Use of laminated glass or impact glass windows An extensive market research was performed in order to obtain the individual cost of these mitigation measures. The primary costing resource used to obtain these numbers were industry contacts, where costs were obtained from local contractors and people in the construction industry. Other costing resources employed for this task were the CEIA Cost, (Langedyk et al. 2002), and the National Renovation & Insurance Repair Estimator, Russell (2004). In this case, adjustment factors were applied to the costs where applicable (these factors account for the age of the costing resource and economic issues like inflation). Additional costing literature, like manufacturers web pages, were also employed to determine the costs. For most items, there are three costs: Cost A: the unit cost of the improved or mitigated component, including the labor for its installation. Cost B: the unit cost of the removal of the old (or damaged) component in case of a retrofit if anything needs to be removed (e.g. removal of old non rated shingles if upgrading to rated shingles). Cost C: the unit cost of the non-mitigated or base component; this includes the unit cost of the material, and the labor for its installation. The cost of the mitigation in each case is the difference between (A + B) and C. For example, in the case of shingles: Cost A: $2.37/sq ft for rated shingles Cost B: $0.31/sq ft for removal of old shingles Cost C: $1.96/sq ft for non-rated shingles Therefore, the additional cost due to the mitigation is $0.72/sq ft for a retrofit and $0.41/sq ft for a new construction. The individual mitigation measures were then combined into different sets based on constructability and on incremental effectiveness. For example, improving roof cover and roof nailing might not be as effective if the proper roof to wall connections are not used, and the entire roof can be blown out during a hurricane. Similarly, wall reinforcement is not a feasible option for a retrofit. Different sets of mitigation measures were selected based on the quality and age of the construction (weak, medium, or strong) and the type of construction (concrete masonry walls with timber roof, or timber box structure). For all the structures, all the sets included rated shingles, opening protection (either steel shutters, laminated, or impact resistant windows), and roof water barrier (either roof under-layment barrier, joint sealing, or roof taping). In addition, for weak and medium structures, all the sets included re-nailing the roof deck with 8d nails, and bracing the gable end. Also, in the case of weak structures, the mitigation sets were divided into 2 subsets: one included the retrofit of the roof to wall connection from toe nails to straps, while the other did not. Finally, the authors also 1506
investigated the benefit of using a hip roof as opposed to a gable roof for new strong homes, and the benefit of reinforcing all the walls for new strong masonry homes. 4 COST BENEFIT ANALYSIS The FPHLPM was used to get the expected annual losses (EAL) for each building type, mitigated and unmitigated, for each zip code in Florida. The difference between the unmitigated EAL and the reduced mitigated EAL was defined as the benefit of mitigation. The cost of mitigation was then transformed into an annuity, and the variation of (benefit minus cost) for each zip code was color mapped for the entire State of Florida, for every building type investigated, in a way similar to the methodology presented by (Porter et al. 2006) for seismic mitigation. The expected annual loss is basically the average loss per year over a long period, to be expected at a particular location. The EAL of a particular home is expressed as a percentage of the total home s value. In order to generate EALs, the FPHLPM requires some specific input, which consists of a portfolio containing the distribution and number of modeled homes for each zip code in the entire state of Florida. The authors generated a hypothetical portfolio containing an identical fixed set of modeled homes in each zip code of the state: four gable masonry homes (one weak or pre-1970, one medium or pre-1990, one strong or pre- 2002, one strong or post-2002 with reinforced walls); one strong hip masonry; three gable timber homes (one weak, one medium, one strong); one strong hip timber. In reality, each zip code has a different, arbitrary home distribution. That is, the hypothetical portfolio is not a measure of the actual distribution of homes in the state. But this fake portfolio repeated identically for each zip code allowed the researchers to investigate the variation of the cost effectiveness of the mitigation measures for the different types of building represented in the portfolio, across the entire State. Therefore, obtained EALs could not be mixed or combined in any way. In the end, the result are separate maps of (benefit minus costs) for each set of mitigation, for each structural type. The portfolio contains data regarding the value of each of the model homes at each zip code, their respective hurricane insurance deductibles, and insured limits for building, contents, appurtenant structures and additional living expenses (ALE). For the purpose of this study, since the researchers at this stage were not concerned with insurance issues, all deductibles were set as $0 (no deductibles), and the insured limits for building and contents were set to be the same as the home and content value (which is assumed to be 50% of the value of the building). Since the mitigation measures have no influence on the appurtenant structures, appurtenant structures are not considered. Therefore, the limit for appurtenant is set to $0 in the portfolio file. The limit for ALE is set at 20% of the value of the home. This value was set to be $100,000 for all the homes in the portfolio. Before the mitigation costs could be subtracted from the benefits they had to be transformed into annuities. An annuity converts all costs incurred during a specific range of years into a constant year distribution of equivalent value. In order to do the conversion, the authors had to define planning periods for the mitigation components of each set, and a return rate applicable to the time under consideration. For the purpose of this first study, and for simplicity, the entire set of mitigations was assumed to last for 30 years. At the time of the study, mortgage rates for a home equity line were defined at an average 8%by several banks in the State of Florida (Bank of America, Lloyd s Bank, and Wachovia, to mention some). Engineering News Records, in its monthly electronic publications, established the inflation rate material cost index at 6.6%. The resulting rate of return was set at 1.4%. 5 RESULTS AND DISCUSSION The final result of the mitigation cost effectiveness study is a series of (benefit minus cost) maps. These maps are a tool to facilitate the visualization of the (benefit cost) output, as an alternative to pure numbers. They are based on simple density models that assign different shades to specific ranges of values. These allow the analyst to pinpoint locations of interest and general trends and to quickly draw comparisons between mitigation sets. The results are presented for the case of weak timber structures in Figures 1 to 4. The colored zip codes represent areas where the benefits exceed the cost of mitigation. The darker the color, the more cost effective is the mitigation. White areas are areas where the mitigation is not cost effective (cost exceeds benefits). There seems to be an anomaly in all the maps, in the fact that a few zip codes in the central south region (below Lake Okechobee) seem to exhibit systematically high (Benefit minus Cost) differences, which defies the logic of the decrease of values away from the coast. This is due to the fact that these large zip codes have a small roughness, since they correspond to uninhabited open terrain country side. Therefore they are theoretically subjected to higher wind speeds than some of the coastal urban zip codes with large roughness. Map1 shows the distribution of cost effectiveness for a set of mitigations that includes retrofitting the roof to wall connections from toe nail to strap connections, in addition to renailing the roof sheathing with 8d nails, installing an extra roof underlayment membrane, installing rated shingles, bracing the gable 1507
Aven CH189.tex 17/5/2007 10: 58 Page 1508 Figure 1. (Benefits Costs) distribution. Map1 (including retrofitted connections). Figure 2. (Benefits Costs) distribution. Map2 (no connection retrofit, shutters). ends, and installing steel shutters. Map2 shows the same distribution for the same set of mitigation but excluding the retrofit of the roof to wall connections. The retrofit of the roof to wall connections of an Figure 3. (Benefits Costs) distribution. Map3 (no retrofit connections, laminated glass windows). Figure 4. (Benefits Costs) distribution. Map4 (no connection retrofit, impact resistant glass). existing older home is an expensive proposition, and this is reflected in the fact that it is cost effective only along the south east coast of the State, which has the highest hurricane risk. On the contrary, the simpler 1508
retrofit of the roof nailing, roof cover and use of steel shutters is cost effective for older weak homes in all the zip codes of south and central Florida. Maps 3 and 4 show the cost effectiveness of two alternatives to steel shutters, for opening protection: laminated glass (map 3) and impact resistant glass (map 4). The range of zip codes where these alternatives are cost effective is greatly reduced with respect to steel shutters (map 2). The maps sow that in the northern part of the State, none of these mitigation sets is cost effective. In other words, given the lower hurricane risk in that area, it might be more cost effective to simply use shutters without additional mitigation. On the other hand, in the southeast, the risk is so much higher that almost any combination set of mitigation will be cost effective. Similar maps were also obtained for the case of masonry structures and for medium strength and strong buildings. They are not repeated here in the interest of brevity. 6 CONCLUSIONS The reader is warned that a lot of variables carry substantial uncertainty in this study. In particular, the costing of the different mitigation elements is not an exact science, and the value of the cost annuities depends greatly on the interest rate and inflation rate adopted. However, some general conclusions can still be drawn from the results, with reasonable validity. The resulting maps showed that for most mitigation measures and for most structural types, the mitigation is really cost effective only in the Southeast region of Florida. These preliminary conclusions did not take into account insurance deductible and limits, nor possible tax credits or cash incentives like insurance premium reductions. The study also showed that the results have a high degree of uncertainty attached to the definition of the actual costs, the interest rate, and the inflation rate. These variables could have a significant effect on the cost effectiveness of certain mitigation measures, and need to be further investigated. In general, the study has showed how individual components can affect the overall structural performance of a given mitigation set. For example, upgraded structural roof-to-wall and wall-to-floor connections (straps, clips) are critical for any old construction located in the Southeastern region high wind velocity zone. The use of this structural upgrade should always be accompanied by other critical components upgrades -rated shingles, bracing of gable ends, and 8d deck nailing- plus the use of shutters for opening protection. Older homes located in North, Central & Southwestern Florida may use a cheaper option, since the upgraded connections do not seem to be cost effective in this areas of lower risk. For medium and strong homes the use of roof taping or joint sealing in place of the more expensive roof underlayment option is recommended. For homes built in 2000 and later, the standard mitigation measures already incorporated into the building code plus the use of shutters should be enough to provide adequate structural protection. The work presented in this paper has valuable data for both homeowners and insurance companies. The results and conclusions are a preliminary study on the cost effectiveness of standard mitigation measures. This information is much needed, and more detailed and refined studies are to follow. Insurance companies could use this kind of material to promote mitigation among homeowners in exchange for premium relief. Homeowners could inform themselves of what options would best fit their current structural conditions from the kind of analysis performed in this project. Fabricators and material vendors could rely on this type of work to convince clients to buy their materials for the sake of annual benefit. ACKNOWLEDGEMENTS The financial support for this project of the Florida Sea Grant Consortium through grant R/C-S-45 is gratefully acknowledged. The authors wish also to thanks Dr. Shahid Hamid of the International Hurricane Research Center, at Florida International University for the use of the Florida Public Model. REFERENCES Florida Commission on Hurricane Loss Projection Methodology: Report of Activities as of November 1, 2006. Langedyk, R. & Ticola, V. 2002. CEIA Cost, Construction Estimating Institute, Sarasota, FL. Russell, J. 2004. National Renovation & Insurance Repair Estimator, Craftsman Book Company, Carlsbad, CA. Pinelli, J-P., Simiu, E., Gurley, K., Subramanian, C., Zhang, L., Cope, A. & Filliben, J. 2004. Hurricane Damage Prediction Model for Residential Structures, Journal of Structural Engineering, ASCE, Vol. 130, No. 11, pp. 1685 1691. Pinelli, J-P., Subramanian, C., Artiles, A., Gurley, K. & Hamid, S. 2006. Validation of a probabilistic model for hurricane insurance loss projections in Florida, Proceedings, ESREL 06, Estoril, Portugal, September 18 21. Porter, K., Scawthorn, C. & Beck, J. 2006. Cost-effectiveness of stronger woodframe buildings. Earthquake Spectra, v 22, n 1, pp. 239 266. 1509