Survival Analysis of Elevated Homes on the Bolivar Peninsula after Hurricane Ike Tori Tomiczek 1 ; Andrew Kennedy 1, Spencer Rogers 2

108 Survival Analysis of Elevated Homes on the Bolivar Peninsula after Hurricane Ike Tori Tomiczek 1 ; Andrew Kennedy 1, Spencer Rogers 2 Abstract Hurricane Ike was the third costliest hurricane to hit the United States, causing almost $30 billion in damage after making landfall on the Bolivar Peninsula, Texas, in September, 2008. Following the disaster, a case study of the peninsula was conducted to evaluate hurricane-generated wave and water level effects on wood framed, pile-elevated coastal residences. This study provided information about the structural characteristics most influential to the survival of these homes and may be used to validate future wave force prediction and fragility models for elevated coastal residences. A total of 1922 homes on the Bolivar Peninsula were sampled, and hindcasts of significant wave heights and water levels during the storm were run to relate environmental factors and structural attributes to a home s probability of survival. Freeboard, calculated as the difference between the elevation of the lowest horizontal structural member and the combined wave crest and storm surge elevation, was shown to play a critical role in determining survival: survey data indicated that a positive freeboard resulted in a near 100% survival rate for homes subjected to significant wave heights above 3.5 ft (1.07 m). Survival was also strongly influenced by maximum significant wave height itself; below significant wave heights of approximately 3.5 ft, the freeboard requirement for survival dropped sharply. Uncertainty of numerical hindcasts of significant wave heights was considered when evaluating survival as a function of wave height; however, these estimations were considered more accurate than assuming depth-limited breaking waves at each house location. Homes experienced much larger survival rates in areas subject to small significant wave height conditions than in locations affected by larger waves. This analysis was the first step in accurately determining expected damage resulting from specific wave and water level loadings. A more generalized analysis will mitigate possible effects of construction techniques, survey bias, debris impact, and environment-specific wave height estimations on quantitative survival criteria. Introduction Hurricanes are responsible for seven of the ten costliest disasters in the United States since 1980 (Lackey 2011). Much of this cost is due to structural failures of buildings, bridges, and other coastal structures. Elevated coastal residences often fail as a result of wave forces on the underside of pile-elevated buildings. Accurately predicting these wave forces is critical in order to ensure the safety and stability of coastal 1 Department of Civil Engineering and Geological Sciences, University of Notre Dame, Notre Dame, IN, USA 2 Coastal Construction and Erosion Specialist, Department of Civil Engineering, N.C. State University and Sea Grant Marine Advisory Service, P.O. Box 130, Kure Beach, NC

109 Figure 1: FEMA damage estimator for a single family residence as a function of depth to lowest horizontal structural member (Coastal V-Zone, no obstruction). As seen in Figure 2, the gradual increase in damage with increasing inundation does not reflect the almost complete destruction or survival of homes observed on the Bolivar Peninsula. communities. This case study represents the first step in evaluating the effects of such forces by correlating storm-generated wave heights and water levels to damage suffered by pile-elevated residences on the Bolivar Peninsula, Texas. Wave prediction and damage estimation models currently do not accurately predict success or failure of elevated coastal residences in response to combined storm surge and wave loads. FEMA s current V-Zone depth-damage curve for a single family residence, reproduced in Figure 1, depicts a gradual increase in damage to an elevated residence as the home is impacted by a greater depth of water (FEMA 2011). This model is limited in that it does not consider wave height; a home impacted by a small stillwater elevation and large waves will be subject to larger forces than a home in a location with smaller waves and deeper stillwater elevation. On the Bolivar Peninsula, this model does not adequately predict the actual damage sustained to wooden, pile-elevated coastal residences following Hurricane Ike. Figure 2(a) depicts a home sampled in a March, 2009 reconnaissance survey that survived a water level and wave height combination rising 9 ft above its lowest structural member with almost no structural damage. This survival contradicts Figure 1, which estimates that for a 9 ft water depth above a home s lowest structural member (-9 ft freeboard), such a home should sustain approximately 72% overall damage. While this home will have experienced interior water damage, the structural integrity of the house indicates that the current depthdamage estimation curve significantly overestimated the overall damage. Similarly, the curve predicts that homes with water rising 5 ft above the lowest structural member should experience 60% damage. Figure 2(b) depicts the remains of a home where the water level reached 4.9 feet above its lowest structural member. As seen in the photograph, the home was completely destroyed; only the piles and concrete slab remained. Firsthand accounts (Henley, 2010) and further evidence from a field survey performed by Andrew Kennedy, Spencer Rogers, and a team of researchers after the storm (Kennedy et. al., 2011) confirmed that homes on the Bolivar Peninsula were typically completely destroyed or else exhibited very little structural damage

110 a b Figure 2: Homes sampled on the Bolivar Peninsula, Texas, following Hurricane Ike s inundation of the peninsula in September, 2008. (a). House ID B201, (freeboard -9 ft), was almost completely intact following Hurricane Ike. (b) House ID C 028, which was destroyed by the storm (freeboard=-5ft). Both homes were in the V-Zone. Photographs were taken by AECOM and are owned by Spencer Rogers. following the storm. Hence, the gently increasing relationship between percent damage and depth to lowest structural member observed in Figure 1 must be revised. In order to develop a wave fragility model for pile elevated, wood-framed coastal residences, case studies, based upon known or estimated wave and storm surge elevations, as well as structural characteristics, such as floor elevations, must be performed. These case studies allow hindcasts of wave forces associated with a storm s water levels and wave heights to be related to a home s survival, hence allowing the correlation of estimated wave forces with structural damage. Following Hurricane Ike s devastation of Texas Bolivar Peninsula, survey data collected statistics indicating the damage sustained and the structural elevation of pile-elevated wooden residences on the peninsula. Latitude and longitude coordinates provided means to numerically estimate wave heights and water levels for sampled homes on the peninsula (Kennedy et. al., 2011). Analysis of this data set combined characteristic data of these residences, such as elevations of the ground and lowest structural member at each home, with characteristics of the storm, such as storm surge water levels and significant wave heights. This combination of forcing variables allowed the investigation of the relationship between hurricane wave and water level forces and damage to elevated coastal residences. Results of this case study indicated a strong influence of structural elevation as a survival criterion for an elevated coastal residence. Historically, much structural design for hurricane resistance has focused on the stability of homes against wind loadings. Hence, fragility models have been refined to provide reasonable estimates of expected damage based on a specific combination of wind loads and structural characteristics. Pinelli, et.al. (2004) presented a damage estimation model for multiple damage states of masonry homes due to windstorms in Florida, and Ellingwood et. al. (2004) developed a method for determining the fragility of wood framed construction in response to severe wind and seismic events.

111 In addition to analysis of windstorms, tsunami fragility curves have been improved to accurately estimate the vulnerability of coastal homes to tsunami inundations. Reese et. al. (2011) developed empirical fragility curves for timber residences following a reconnaissance survey taken in Samoa and American Samoa after the South Pacific tsunami in September of 2009. In addition, Suppasri et. al. (2012) used survey data from the March 2011 Great East Japan Tsunami to classify wood framed and mixedtype homes into four damage states and subsequently relate the probability of a certain damage classification to the inundation depth caused by the tsunami. Wind and tsunami risk assessments provide reasonable vulnerability estimates for homes in coastal communities where such events are possible. However, wave fragility models must be improved and coupled with these fragility models to provide homeowners with a complete analysis of the risk associated with all possible factors contributing to structural damage during a hurricane event. Hurricane-generated wave force prediction and resulting damage estimation methods are still relatively new; therefore, in order to verify and improve such fragility models, combined wave and water level forces, as well as their impacts on various engineered and non-engineered structures, must be fully understood. Estimating the forces exerted by a water wave on a structure is complicated by uncertainties in the wave environment and the resulting interaction with a structure s complicated geometry. Morison et. al. (1950) first analyzed water wave forces against structures by combining the drag force of the wave around an object with the inertial force arising from the water displaced by the object. Since this relationship was defined, numerous tests and analyses have been conducted to observe and model wave forces on coastal structures. Wave loads on horizontal decks were modeled by El-Ghamry (1965) and Wang (1970), indicating that uplift pressures can contribute to failure of a horizontal slab. Bea et. al. (1999) further investigated the complex wave-structure interaction for horizontal platforms, proposing a procedure for estimating the performance of platforms subject to impact forces resulting from wave slam as well as the drag, lift, and inertial forces exerted by a wave on a structure. Cuomo et. al. (2007) investigated wave loadings on jetties in two and three dimensions, showing that wave forces, although a function of numerous geometric and environmental factors, were strongly influenced by wave height. Based on their analyses, they developed methods to determine quasi-static and impact forces expected by waves acting on vertical walls (Cuomo et. al. 2009) and exposed jetties (Cuomo et. al. 2007). Yeh et. al. (1999) also studied wave action on vertical walls, specifically considering the response of breakaway walls, which are commonly found below the lowest horizontal member of pile-elevated residences, to loadings caused by breaking waves. Such walls are designed to fail at a predetermined wave or water level force so as to allow floodwater to pass through the structure, therefore minimizing the forces exerted on the elevated home. Laboratory tests and analytic

112 failure modes were compared to study the response of these walls to breaking wave forces. In a case study of multiple bridges that were inundated by Hurricane Katrina s storm surge, Robertson et. al. (2007) analyzed the effects of bridge elevation above surface water level. This elevation may be compared to the freeboard experienced by an elevated coastal residence. They also analyzed other structural design methods effects on increasing a structure s resistance to the combined horizontal and uplift forces caused by inundation and trapped air under the structure. While these studies advance the prediction of wave and flood forces on coastal structures, there currently exists no defensible method to relate such forces to the fragility of wood framed, elevated coastal residences. Case studies, such as the one conducted on the Bolivar Peninsula, must precede force estimation and fragility curve development in order to identify factors most influential to survival of a woodframed, pile elevated home. Knowledge of these factors will provide the basis for a more thorough risk assessment of coastal communities, hence allowing these communities to be designed and constructed with enhanced resistance to significant hurricane events. Study Area and Hurricane Ike As seen in Figure 3, the Bolivar Peninsula is located off the southeastern coast of Texas in the Gulf of Mexico. Bolivar s population as of 2007 was 4148 people (Bolivar Chamber of Commerce, 2011). Hurricane Ike made landfall on the west side of the peninsula at 2:10 a.m. on September 13, 2008 as a Category 2 hurricane, with wind speeds averaging 110 mph. However, Ike was unique in that its large size, combined with the broad continental shelf off the coast of the Bolivar Peninsula, caused a massive storm surge setup to inundate the peninsula 24 hours before the storm made landfall. Large wave heights, together with this rise in water level, completely devastated structures on the peninsula. Residents reported wave action on a b Figure 3: a. GIS Image of Study Area (GIS data obtained from TNRIS (2011)). b. Zoomed-in view of the Bolivar Peninsula. Locations of sampled houses are plotted in the figure (TNRIS 2011).

113 their ceilings; one witness described the destruction caused by the hurricane as resembling a war zone (Henley 2010). ADCIRC and SWAN hindcasts of the storm surge and significant wave heights generated by Hurricane Ike across the peninsula ranged from 3.6-5.2 m (11.8-17.1 ft) and 0.75-2.2 m (2.5-7.2 ft), respectively. The $30 billion in property damage caused by Hurricane Ike on the Bolivar Peninsula and other areas of the southern United States made Ike the third costliest storm in U.S. history (DeBlasio 2008). Data Collection The data set obtained and analyzed by Kennedy et. al. (2011) considered structural and environmental factors, described below, which contributed to structural failure or survival of wood framed, pile-elevated coastal residences. In March, 2009, a team sampled locations on the peninsula to assess the severity of damage and to obtain important structural attributes. Houses were classified as surviving, destroyed, or significantly damaged by wave action. As mentioned above, the vast majority of houses either experienced minimal structural damage or were completely destroyed. Additional survey characteristics measured included longshore and cross shore buffer distance, flood zone designation, and ground level, lowest structural member, bottom of the floor joist, lowest floor, and minimum base flood elevations for each sampled home. Of these elevations, the elevation of the lowest structural member was selected as the representative elevation for freeboard determinations for each home. The March 2009 survey and subsequent data collection trips resulted in a total of 1922 sampled homes on the peninsula. On-site elevations were measured using surveyors rods; however, some structural elevations were impossible to obtain in the field. These elevations were determined using Google Earth s Street View and relating known heights of objects to the elevation of the house. Therefore, a degree of uncertainty is inherent in all subsequent elevation calculations; however, these estimates can be assumed accurate to within one foot of the true elevation; thus, they minimally impacted freeboard calculations and subsequent survival analyses. In addition to the attributes described above, the precise latitude and longitude of each sampled house was determined. These coordinates were inserted into numerical models to generate hindcasts of wave heights and water levels caused by Hurricane Ike. At each house location, the ADCIRC Coastal Circulation and Storm Surge Model was run to determine the storm surge elevation, and the SWAN wave model was used to simulate the maximum significant wave height (H s ). Structural Survival Analysis of Surveyed Residences Relative inundation depths experienced by each of the 1922 sampled houses were determined by combining the lowest horizontal structural member s elevation above grade with the water level generated by the significant wave height and storm surge elevation at each location. Using these variables, the freeboard of each home was calculated and plotted against environmental forcing factors such as water level,

114 significant wave height, and distance from the shoreline to determine freeboard s effect on a home s survival. Freeboard was defined as: FB=LHE-WCE (1) where FB refers to the freeboard, LHE refers to the elevation of the lowest horizontal member, and WCE refers to the wave crest elevation, including the local ground elevation (G). Wave crest elevations were determined using maximum wave height (H w ) and stillwater flood height (h s ) as: WCE= (0.70) (H w ) + (h s ) + (G), with the stillwater flood height defined as the difference between the ADCIRC storm surge hindcast elevation and the ground elevation at each location, and H w taken as the minimum of either depth-limited breaking wave height, 0.78*h s, or the H 1/100 calculated using a Rayleigh Distribution: H w = min These definitions differ from the FEMA Coastal Construction Manual s definitions of freeboard and wave crest elevations (FEMA 2011). FEMA calculates the wave crest elevation assuming that waves are always limited by breaking depth, using the following relationship: WCE=G+1.55(h s ), with all variables defined previously (FEMA 2011). However, smaller waves impacting coastal residences will not have depth-limited heights; therefore, the wave crests may not reach as high as assumed by this definition in areas where the observed or hindcast wave height is less than 0.78*h s. At such locations, wave impact force calculations may overestimate the actual force experienced by a structure. For this reason, using hindcasts of significant wave heights and limiting (but not requiring) these wave heights by maximum breaking wave height provides more accurate estimates of wave loads on pile-elevated coastal residences. However, it is important to note that although numerical methods provide more accurate significant wave height estimates than assuming depth-limited breaking waves, the SWAN model likely overestimated significant wave heights on Bolivar, especially moving inland from the coast, where actual waves dissipated due to vegetation and shielding from other houses and debris. Therefore, the estimated significant wave heights must be taken as preliminary values and later validated to accurately reflect the wave environment on the peninsula. (2) (3) (4) Following the calculation of freeboard at each location on the Bolivar Peninsula, the effect of this variable on the survival or destruction of a home was evaluated by plotting freeboard vs. wave height, distance from the Gulf of Mexico, and water level

115 a b c Figure 4: Freeboard of surviving and destroyed homes plotted against (a) significant wave height, (b) distance from the Gulf of Mexico, and (c) storm surge height. A clear break exists between the freeboard experienced by surviving and destroyed homes. for each sampled home. Results are illustrated in Figure 4; as seen in the figure, a clear break between surviving and destroyed homes exists for small increases in freeboard for a given significant wave height, surge level, or distance inland. With increasing distance from the Gulf of Mexico, decreasing significant wave height, and decreasing storm surge elevation, the freeboard requirement for survival similarly decreases. This relationship was expected, as all three variables are related: wave heights and storm surge levels are expected to decrease as they propagate inland. Above wave heights of around 3.5 feet, the freeboard requirement for survival leveled off; below wave heights of 3.5 feet, the freeboard requirement for survival dropped sharply. This phenomenon suggested that wave impact forces from small waves do

116 not significantly affect a pile-elevated structure, thereby eliminating a required freeboard for survival. For larger waves, on the other hand, a positive freeboard can almost guarantee survival for a wood framed, pile elevated coastal residence. Due to the interdependence of wave height, storm surge, and inland distance from the Gulf of Mexico, analysis was condensed to analyze the combined effect of significant wave height and freeboard on a home s probability of survival. Homes were grouped into bins based on specific freeboard- significant wave height combinations; bins were defined between significant wave heights of 2 and 8 ft using a wave step of 0.5 ft and freeboards between -15 and +10 ft using a freeboard step of 1 ft. The percentage of homes surviving in each bin was then determined by taking the ratio of homes classified as surviving or wave damaged to the total number of homes in each bin. Results are illustrated in Figure 5. As seen in the figure, a distinct curve defines the required freeboard for a desired survival rate; this requirement sharply increases at significant wave heights between 3.5 and 4.5 ft and levels off at larger wave heights. Homes with positive freeboards experienced a near 100% survival rate. Conclusions The data set obtained from the Bolivar Peninsula following Hurricane Ike permitted a case study of the effects of significant wave heights, water levels, and structural characteristics on pile-elevated, wood-framed houses survival of a large scale storm. Hurricanes generate significant forces due to wave heights and storm surge elevations on these pile-elevated residences; homes on Bolivar provide evidence that these additional loadings can lead to structural failure of such residences. Figure 5: Percent surviving based on specific combinations of freeboard and significant wave height. Freeboard requirements for survival sharply increase at wave heights near 3.5 feet before leveling off at near positive freeboards. Small wave heights were shown to have a minimal impact on the survival or destruction of an elevated coastal residence, regardless of the freeboard experienced by that home. Based on SWAN hindcast significant wave heights, analyses indicated that wave heights below around 3.5 ft did not significantly affect survival criteria for elevated coastal residences. However, these wave height limits cannot be reported without first evaluating the accuracy of the numerical models used to calculate such wave

117 heights. Although the SWAN hindcasts for significant wave heights are an improvement over the assumption of a depth limited breaking wave across the peninsula, actual wave heights on the peninsula may be smaller in some areas than predicted by the model as a result of dissipation and location-specific conditions that are not reflected in the numerical model. For larger wave heights, freeboard was observed to play a critical role in determining an elevated home s chance of survival; as indicated by the data surveyed by Kennedy et. al. (2011), homes that were built above the combined water and wave levels exhibited a greater probability of survival than those homes that were impacted by wave crest elevations rising above the lowest structural member. Design values for freeboard or lowest structural elevation needed to ensure a specific survival rate given a storm similar to Hurricane-Ike on any coastline cannot be obtained without a large degree of uncertainty. This uncertainty is due to the environment-specific conditions faced by homes on Bolivar, in addition to the possible bias toward sampling surviving homes in post-ike evaluations. Limitations of numerical models, such as potential overestimations of wave heights, must also be taken into account, as discussed above. Finally, the effects of improved construction techniques and materials over time must be considered, as more modern homes are expected to be less vulnerable than older homes for the same freeboard-significant wave height combination. However, this case study serves as the first step in understanding wave and water level effects on non-engineered, wood framed residences. Fully understanding these effects will make possible the validation of future wave force and fragility models for elevated coastal residences so as to estimate the vulnerability of coastal communities to given wave and storm surge loadings. These models, in turn, will improve the reliability of design standards, thereby enhancing the safety and vitality of coastal communities. Acknowledgements The authors extend their thanks to those who made this research possible: Michael Hartman, the Galveston County Appraisal District, and the Galveston County Engineers office. References Bea, R.G. Xu, T., Stear, J., Ramos, R. (1999). Wave Forces on Decks of Offshore Platforms. Journal of Waterway, Port, Coastal, and Ocean Engineering, ASCE, 136-144. Bolivar Chamber of Commerce (2011). Bolivar Peninsula Census Data. [Data File] Retrieved from www.bolivarchamber.org. Cuomo, G., Tirindelli, M., and Allsop, W. (2007). Wave-in-deck Loads on Exposed Jetties. Coastal Engineering: An International Journal for Coastal, Harbour, and Offshore Engineers, ELSEVIER, 657-679.

118 Cuomo, G., Tirindelli, M., and Allsop, W., (2009). Breaking Wave Loads at Vertical Seawalls and Breakwaters. Coastal Engineering: An International Journal for Coastal, Harbour, and Offshore Engineers, ELSEVIER, 424-439. DeBlasio, Stephen M. (2008). Foreword by Federal Coordinating Officer. Hurricane Ike Impact Report. Ed. Emergency Support Function #14 Longterm Community Recovery FEMA, ii. El-Ghamry, O.A. (1965). Wave Forces on a Dock. Technical Report HEL -9-1, Hydraulic Engineering Laboratory, Institute of Engineering Research, University of California. Ellingwood, B.R., Rosowsky, D.V., Li, Y., Kim, J.H. (2004). Fragility Assessment of Light-Frame Wood Construction Subjected to Wind and Earthquake Hazards. Journal of Structural Engineering, ASCE, 1921-1930. FEMA. (2011). Coastal Construction Manual. (FEMA P-55). Washington, D.C. Henley, B.C. (2010). Winds Over Bolivar. CreateSpace, United States of America. Kennedy, A., Rogers, S., Sallenger, A., Gravois, U., Zachry, B., Dosa, M., Zarama, F. (2011). Building Destruction from Waves and Surge on the Bolivar Peninsula during Hurricane Ike. Journal of Waterway, Port, Coastal, and Ocean Engineering, ASCE, 132-141. Lackey, Mark. National Climactic Data Center (2011). Billion Dollar US/Climate Disasters [Data File]. Retrieved from http://www.ncdc.noaa.gov/img/reports /billion/disasters2010.pdf. Morison, J.R., O Brien, M.P., Johnson, J.W., Schaaf, S.A. (1950). The Forces Exerted by Surface Waves on Piles. Petrol Trans., vol.189, AIME, 149-157. Pinelli, J.P., Simiu, E., Gurley, K., Subramanian, C., Zhang, L., Cope, A., Filliben, J.J., Hamid, S. (2004). Hurricane Damage Prediction Model for Residential Structures. Journal of Structural Engineering, ASCE, 1685-1691. Reese, S., Bradley, B.A., Bind, J., Smart, G., Power, W., Sturman, J. (2011). Empirical Building Fragilities from Observed Damage in the 2009 South Pacific Tsunami. Earth Science Reviews, ELSEVIER, 157-173. Robertson, I.N. Riggs, H.R., and Yim, S.C., Young, Y.L. (2007). Lessons from Hurricane Katrina Storm Surgeon Bridges and Buildings. Journal of Waterway, Port, Coastal, and Ocean Engineering, ASCE, 463-483. Suppasri, A., Mas, E., Koshimura, S., Imai, K., Harada, K., and Imamura, F. (2012). Developing Tsunami Fragility Curves from the Surveyed Data of the 2011 Great East Japan Tsunami in Sendai and Ishinomaki Plains. Coastal Engineering Journal, World Scientific, 1-16. Texas Natural Resources Information System. Original Texas Land Survey. [shapefile]. Austin, Texas: Texas Water Development Board, 2011. Retrieved from http://data.tnris.org/datadownload/download.jsp. Wang H. (1970). Water Wave Pressure on Horizontal Plate. Journal of the Hydraulic Division, ASCE 1997-2016. Yeh, S.C., Rogers, S.M., Tung, C.C., and Kasal, B. (1999). Behavior of Breakaway Walls under Wave Action. Journal of Structural Engineering, ASCE, 1162-1169.