4
Wind Damage to Buildings

This chapter concentrates on areas that sustained major wind damage. These consisted of the cities of Gulfport and Biloxi in Harrison County, Mississippi; the cities of Ocean Springs, Pascagoula, Moss Point, and unincorporated areas, including the community of Gautier, in Jackson County, Mississippi; and Dauphin Island, Alabama. Because of differences in the nature of the communities and the development of building regulations, the damage in Mississippi and Alabama are considered separately.

WIND DAMAGE IN MISSISSIPPI

Building Regulations

Prior to Hurricane Camille in 1969, only the cities of Gulfport and Biloxi in Harrison County and Pascagoula in Jackson County in this coastal region had adopted and enforced building codes. Jackson County had adopted a building code but apparently had no means of enforcing it. The codes in use had been steadily tightened in the light of hurricane experience and by 1969, Gulfport, Biloxi, and Pascagoula had adopted the Standard Building Code. Following the devastation of Hurricane Camille, the Governor of Mississippi's Emergency Council was established because it was determined by the state that "local governments would not or could not institute, maintain, and enforce regulations necessary to protect life and property against disasters through adequate construction requirements and land use regulations" (Leyden, 1985).

A subcommittee of the council, the Gulf Regional Planning Commission, was given the responsibility of drafting a hurricane building code. This document required suitable elevation of structures above the flood level and



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 4 Wind Damage to Buildings This chapter concentrates on areas that sustained major wind damage. These consisted of the cities of Gulfport and Biloxi in Harrison County, Mississippi; the cities of Ocean Springs, Pascagoula, Moss Point, and unincorporated areas, including the community of Gautier, in Jackson County, Mississippi; and Dauphin Island, Alabama. Because of differences in the nature of the communities and the development of building regulations, the damage in Mississippi and Alabama are considered separately. WIND DAMAGE IN MISSISSIPPI Building Regulations Prior to Hurricane Camille in 1969, only the cities of Gulfport and Biloxi in Harrison County and Pascagoula in Jackson County in this coastal region had adopted and enforced building codes. Jackson County had adopted a building code but apparently had no means of enforcing it. The codes in use had been steadily tightened in the light of hurricane experience and by 1969, Gulfport, Biloxi, and Pascagoula had adopted the Standard Building Code. Following the devastation of Hurricane Camille, the Governor of Mississippi's Emergency Council was established because it was determined by the state that "local governments would not or could not institute, maintain, and enforce regulations necessary to protect life and property against disasters through adequate construction requirements and land use regulations" (Leyden, 1985). A subcommittee of the council, the Gulf Regional Planning Commission, was given the responsibility of drafting a hurricane building code. This document required suitable elevation of structures above the flood level and

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 the use of the 1969 edition of the Standard Building Code (Southern Building Code Congress International [SBCCI], 1969) amended for areas within 1,000 ft of the shore to include the wind load provisions of the then-current South Florida Building Code (Dade County Board of County Commissioners, 1959). Initially, it was intended that enforcement of the code should be on a regional basis, but after a short time this became politically unpopular and control passed to local governments. At the time of Hurricane Elena, all jurisdictions had adopted the Standard Building Code and were enforcing it with full-time staff. Considering that building codes had been in use in the area from at least 1969 and a serious attempt had been made to enforce the regulations, the extensive damage observed suggests one or both of the following causes: Wind pressures and forces are in excess of those specified by the building code. The acceptance of structural systems is incapable of resisting the design wind loads. These possible causes are now examined in detail. Design Wind Speeds and Pressures The wind force exerted on a component of a building depends upon the local wind velocity and the location of the tributary area transferring the load to the component. In general, where F = force in lb, p = pressure in lb/ft2, Cp = pressure coefficient, q = velocity pressure = 0.00256 V2(lb/ft2), A = tributary area in ft2, and V = local wind speed in mph. By convention, Cp, is positive when the resulting force is directed toward a surface and negative when directed away from the surface. When a component is subject to pressure from both sides, as in a roof or wall element, the force on the element will be the result of the difference between the pressures on the two surfaces. The earliest building codes simply specified a pressure to be used at a particular location and height, which then had to be multiplied by a shape factor, essentially a pressure coefficient. The Standard Building Code used this type of procedure until the 1974 revisions to the 1973 edition of the code. A slight improvement on this is to use a basic wind speed derived from meteorological data to determine a velocity pressure that varies in an appropriate manner with height. This is then multiplied by a shape factor based on measurements of pressure coefficients made in wind tunnels. Since these are normally mean-pressure coefficients, the gusting effect of the wind has

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 to be taken into account by using peak wind speeds, typically averaged over about 3 seconds. Unfortunately, the National Weather Service (NWS) keeps its records in the form of fastest-mile wind speeds, a value based on the time taken for a "mile" of wind to pass an anemometer. Thus, the averaging time depends on the wind speed. In typical hurricane conditions (74 to 100 mph), the averaging time would be between 49 and 36 seconds. These wind speeds are about 20 to 30 percent lower than the 3-second gust speed. Basic design wind-speed maps are produced in the form of fastest-mile wind-speed maps for different probabilities of occurrence. Their use in various building codes and standards has produced considerable confusion over the years. The Standard Building Code is a typical example. When the 1974 revisions to the 1973 code were introduced (SBCCI, 1974) to make use of this type of wind-speed map, a probability of exceedance in a given year of 0.01 was chosen, commonly known as the 100-year fastest-mile wind speed. The map prescribes approximately 110 mph for the Mississippi coast. However, the code used shape factors that were, in fact, mean-pressure coefficients. Thus, the buildings were, in essence, being designed not for fastest-mile wind speeds of 110 mph, but for gust wind speeds of 110 mph. This gust wind speed is associated with a fastest-mile wind speed of approximately 90 mph, which has a recurrence interval of less than 30 years. These problems were compounded by the fact that the Standard Building Code provided velocity pressures for design wind speeds taken from the wind-speed map. These pressures are based on wind speeds at the midheight of each zone. While this procedure is acceptable for high-rise buildings, it is unsuitable for low-rise buildings in a range where wind speeds change rapidly with height. For buildings approaching 30 ft, the velocity pressures would be underestimated by approximately 25 percent. Two factors did help to ameliorate these problems locally. First, the South Florida Building Code was used for areas within 1,000 ft of the shore. Although suffering the same problems regarding fastest-mile and gust wind speeds as the Standard Building Code, it did use a design wind speed of 120 mph and provided velocity pressures at 10-ft instead of 30-ft intervals. Second, all buildings were assumed to be in open country, generally overestimating wind speeds near the ground and taking no account of shelter provided by trees or adjacent buildings. It is unfortunate that this incorrect use of wind speed was adopted by the Standard Building Code, because a much better standard, the American National Standards Institute A58.1 (American National Standards Institute [ANSI], 1972), was available for adoption. The ANSI standard treated the fastest-mile wind speeds correctly and, perhaps even more importantly, considered the very high negative pressures that can occur locally near the edge of a roof. The simpler Standard Building Code could have been justified had it led to conservative design pressures, but it is shown below that in many cases it seriously underestimated them.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 In the late 1970s, an extensive study of wind pressures on low-rise buildings was conducted in a modern boundary-layer wind tunnel at the University of Western Ontario, Canada (Davenport et al., 1977, 1978). The results of these tests formed the basis of revisions to the Standard Building Code. First introduced as an alternative procedure in 1982 (SBCCI, 1982), the provisions became mandatory for low-rise buildings in 1986, finally providing for a more rational treatment of wind pressures on low-rise buildings. Unfortunately, most of the buildings affected by Hurricane Elena were built to the older versions of the Standard Building Code. Notable exceptions to this were several preengineered metal buildings. The upgrading of the Standard Building Code took place as a result of pressure from the metal building industry, and many manufacturers took advantage of the new regulations when they became an alternative procedure in 1982. The shortcomings in assessing design wind pressures are best illustrated by reference to the design pressures that would have been used in the design of buildings that sustained the most serious damage in Elena—namely, flat-roofed buildings 10 to 20 ft high. Table 4-1 shows the design pressures prescribed under normal conditions by the three versions of the Standard Building Code and the South Florida Building Code for a building located on the Mississippi coast. Table 4-2 shows the pressures that might have been used by a prudent designer who anticipated the possibility that an accidental opening might occur in a wind-ward wall as a result, for example, of window damage. It should be pointed out, however, that this eventuality is rarely considered in the design of such buildings. Note that the decking load from Table 4-2, SBC3, is nearly twice the SBC2 value in Table 4-1, the standard to which most of the structures were constructed. The effect of this underdesign is illustrated more clearly in Table 4-3, taken from Sparks (1987a). Sparks used the best available wind-tunnel pressure coefficients for the roof of a single-story, 80 ft by 80 ft building in a suburban area to estimate the mean recurrence intervals for wind pressure in excess of those prescribed by the Standard Building Code between 1974 and 1986 (SBC2). In this analysis, wind-speed recurrence intervals were taken from Batts et al. (1980), and the wind was assumed to have an equal probability of occurrence from any direction. The corner and edge conditions, appropriate for a roof covering, considered only the external pressure condition. The three alternative quarter-roof conditions, suitable for the design of a roof framing system, included appropriate internal pressure coefficients. Clearly, buildings of this type were being designed not for events with a 100-year recurrence interval, as the code implied, but for events with a much higher frequency of occurrence. Table 4-4 shows the factors of safety required to prevent failure in hurricanes with mean recurrence intervals between 50 and 500 years. The fast-

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 TABLE 4-1 Normal Design Pressures (lb/ft2) Codea Wallb Roof framec Deckingd Membranee SBC1 ± 25 -31 -31 -31 SBC2 ± 26 -24 -24 -24 SBC3 ± 25 -30 -38 -61 SFBC ± 30 -27 -27 -27 a SBC1—Standard Building Code prior to 1974; SBC2—Standard Building Code 1974 to 1986; SBC3—Standard Building Code 1986 to present (alternate from 1982); and SFBC—South Florida Building Code. b Tributary area = 100 ft2. c Tributary area > 500 ft2. d Tributary area = 50 ft2. e Tributary area = 10 ft2. TABLE 4-2 Open-Side Design Pressures (lb/ft2) Codea Wallb Roof framec Deckingd Membranee SBC1 ± 25 -31 -31 -31 SBC2 ± 26, -36 -36 -36 -36 SBC3 ± 34 -38 -46 -70 SFBC ± 27, -41 -41 -41 -41 NOTE: See notes in Table 4-1. TABLE 4-3 Mean Recurrence Intervals for Wind Loads in Excess of the Standard Building Code (1974 to 1986, i.e. SBC2) Location Recurrence Interval (years) Corner 12 Edge 35 Quartera 38 Quarterb 28 Quarterc 15 a Minor opening in any wall. b Major opening in only one wall. c Major opening in any wall. SOURCE: Sparks (1987a).

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 TABLE 4-4 Factors of Safety Required to Prevent Failure   Recurrence Interval by Years Location 50 100 200 500     Factor of Safety Corner 1.57 1.77 2.17 2.52 Edge 1.37 1.56 1.90 2.22 Quartera 1.13 1.25 1.55 1.81 Quarterb 1.21 1.37 1.67 1.95 Quarterc 1.47 1.66 2.02 2.37 a Minor opening in any wall. b Major opening in one wall. c Major opening in any wall. SOURCE: Peter Sparks, 1987a. est-mile wind speeds in Hurricane Elena were, at worst, in the mid-90-mph range, with a recurrence interval of less than 50 years. In addition, many buildings undoubtedly received some shelter from the full force of the wind. With this in mind, the extensive damage observed suggests that not only was the building code inadequate, but typical construction practices provided rather low factors of safety that were insufficient to compensate for the inadequacies of the code, particularly with regard to resistance to high local pressures. Wind Resistance of Structural Systems Load Combinations It has been traditional in structural design to increase the allowable stresses in materials by 33 percent for load combinations that include wind effects. Presumably, the logic behind this is that it is unlikely that the design wind load, live load, and dead load will occur at the same time and/or that the peak wind load will be of short duration. But the procedure is also applied to conditions when the critical loading is a combination of only dead and wind load, as in the case of roofs subjected to uplift. In these circumstances, the design wind load will occur in the presence of the full dead load. In situations where light roofing systems are subjected to high uplift forces, certain steel members might be sized to carry the design wind loads at stresses very close to the yield strength of the material, providing little or no safety margin to accommodate incorrect design pressures or unforeseen occurrences such as window damage.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Wind-Load-Resisting Systems In a high-rise building the wind-load-resisting system provided by moment-resisting frames, braced frames, or shear walls is clearly identifiable, and failure of part of a roof or wall will have only a marginal effect on the load-resisting system. In a low-rise building, especially a single-story one, the roof and walls resist the wind loads in a complex manner. For example, a roof deck may resist wind forces applied directly to it, may act as a diaphragm to distribute wind forces applied to the walls, and may supply lateral support to those walls. Unfortunately, these structures are often designed as a set of isolated components without regard for their combined role in the complete system or an appreciation for the major structural consequences of an apparently minor component failure, such as window breakage or door loss. The Standard Building Code also contains some statements that in the light of recent research (Leland, 1988) appear to be incorrect regarding critical structural details—in particular, the anchoring of roofs to masonry walls. It is unlikely that in hurricane conditions the recommended anchorage using an embedded rod in masonry without a bond beam would be sufficient for any light roof system, relying as it does on the weight of a few masonry units and their bond strength. A second form of anchorage mentioned in the code that uses a bond beam would prove satisfactory only if the bond beam were properly tied to the foundation by vertical wall reinforcement. The Steel Joist Institute recommends that joists be anchored to walls by welding them to plates attached to either one 3/4-inch-diameter steel rod embedded 1 ft into the wall or, in the case of walls with parapets less than 2 ft high, two 3/4-inch-diameter anchor bolts. No mention is made of how or to what these anchors should be bonded. In wood construction, the Standard Building Code suggests the use of three 8d nails driven at an angle through a roof truss or rafter into the top plate of the wall, although hurricane requirements contained in the code's Appendix D state that approved hurricane anchors must be used for wooden truss rafters. Fortunately, most coastal jurisdictions have recognized the inadequacy of the toenail connection and require the use of proper anchors, at least for residential construction. Workmanship and Materials Few of the professionally designed buildings damaged in Elena were prestigious structures in which careful control of the construction would have been carried out by the designer. Additionally, some of these buildings were exempt from inspection by local building inspectors. Roofing systems, the primary areas of damage, are notorious for their variability in quality. Good inspection is required for the actual structure to

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 TABLE 4-5 Maximum Pressures of Underwriters Laboratory 580 Specification (lb/ft 2) Class Top Pressure Bottom Pressure Total Uplift 30 -24.23 +20.77 45.00 60 -40.38 +30.62 75.00 90 -56.54 +48.46 105.00 meet the design specification. A classification system for roofing systems, including the waterproof membrane, decking, insulation, and supporting system, is available based on the successful resistance to a series of tests conducted in accordance with Underwriters Laboratories (UL) 580 specification (Underwriters Laboratory, 1980). These tests apply a sequence of negative pressures to the upper surface of the roof and positive pressures to the underside of the roof. Table 4-5 indicates the maximum pressures applied during these tests. Based on the requirement for resistance to high local suctions specified by the Standard Building Code for the top surface of the roof, the most stringent classification, 90, would be required for flat roofs in this area. Even still, the test might not exceed the expected suction on the corner of a roofing membrane. In view of the pressures specified in the earlier editions of the Standard Building Code, it is likely that designers would have accepted a Class 30 roofing system or, perhaps conservatively, Class 60. In exposed locations during Hurricane Elena, the suction on the roofing membranes near the corners probably exceeded the tested capacity of Class 30 and 60 roofs. They may have even exceeded the capacity of a Class 90 roof. Not surprisingly, many buildings that performed satisfactorily in other respects suffered severe damage to the waterproof membrane. On the other hand, even after windward openings were established, the total uplift probably did not exceed the tested uplift capacity of even a Class 30 roof. However, the UL test does not include the roof-to-wall connections, and it was at these locations that major failures of roof structures appeared to have been initiated. "Nonengineered" Structures Despite the existence of a building code, residential and small commercial structures are often built without any formal consideration of the wind resisting system. Compliance with the building code is essentially determined by rules set forth by the local building officials. Following Hurricane

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Camille, hurricane straps were required on all small residential structures, but suitable, roof anchors did not appear to have been specified for small commercial structures with masonry walls. These buildings, when located in an exposed area, were often severely damaged in Hurricane Elena. Detailed Damage Descriptions of Classes of Structures Traditionally, damage surveys have classified buildings into fully engineered, marginally engineered, nonengineered, and preengineered structures. While preengineered structures can be readily identified as a class, the distinction between fully, marginally, and nonengineered structures has become indistinct in many hurricane-prone areas. At one end of the spectrum, even single-family dwellings are often built to prescriptive building code requirements that indirectly reflect the effect of wind loads on the structure. At the other end, major structures that would be expected to be fully engineered are often designed as a set of components with little regard for the way in which these components might combine to form a complete structural system. For this reason, in describing damage, the majority of buildings are classified by function, with the exception of metal building systems (preengineered structures) and structures that were obviously fully engineered. Schools The satisfactory performance of schools in a hurricane is important, since they are often used as evacuation shelters. However, they are, in fact, more vulnerable than most buildings to damage. Although they have the advantage that they must be designed by a registered architect or engineer in accordance with the Standard Building Code, their plans are not checked, nor is construction inspected by the local building official. This is the responsibility of the state s Department of Education. Construction is often on a tight budget, and the buildings are usually located in open areas where they receive the full force of the wind. In addition, when a jurisdiction adopts a particular form of construction, often involving lightweight flat roofs and masonry walls, the design is often repeated, sometimes using the same building orientation. As a class of structures, these buildings performed very poorly during Elena. The Gulf Coast Sun on September 4, 1985, reported the following descriptions furnished by Mississippi school officials: Gulfport—Expected to get only the high school operational by the end of the week. Biloxi—Roof and interior damage to all schools. Harrison County—More than 100 classrooms with ceiling damage.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Ocean Springs—Damage estimated at $3 million. Moss Point—Extensive damage. Pascagoula—Every school sustained some damage. Jackson County—Most schools damaged. Some damage was also reported in Hancock County, Pass Christian, and Long Beach. Damage ranged from roof leakage due to the sucking off of the waterproof membrane to serious roof damage and the collapse of walls. Leakage Almost all flat roofs and some pitched roofs on schools were leaking after the storm. This was probably the result of a combination of factors, including the use of an inappropriate building code, poor design specification, poor installation, and deterioration; it was probably not the result of excessive wind speeds. Individual leakage apparently was minor, but the widespread occurrence of this type of failure caused considerable damage and disruption. Figure 4-1 shows a typical example. Failure of Roof Decking As indicated earlier, while the loads on these roofs may have exceeded the design level as a result of a window failure, a properly designed and installed system should have accommodated this overload. Figure 4-1 Loss of roof covering—West Elementary School.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Failure probably indicates noncompliance with the code in use at the time of construction. Failure of Roof-Supporting System This failure generally resulted from the pulling out of the joist anchor from a masonry wall, often following the failure of a nearby window. This probably indicates failure to comply with the wind-loading provisions of the code while incorporating a reasonable factor of safety. Unfortunately, in some instances, compliance with the anchorage requirements of the code still resulted in failure. In other cases, the anchorage met neither the wind-lead nor anchorage requirements of the code. Wall Failures The specification of the wind loads on the walls has generally been quite adequate, except perhaps when a high internal pressure develops. However, two, major failures took place when walls were blown inward where, internal pressure was not a factor. In one case, a tornado may have caused locally high wind pressures; in the other, poor construction and loss of roof support were probably the cause of failure. Descriptions of some of the more serious school failures are found in Appendix A. Commercial Structures Masonry-walled commercial structures with light metal roofs tend to suffer the same problems as similarly constructed schools. The problems, however, are accentuated. Roof spans are usually longer and walls higher. In many cases, such structures are more exposed because of extensive parking areas and are usually more vulnerable to high internal pressure because of large glazed areas. During Elena, most of the flat roofs of commercial structures leaked, and loose gravel from the roofs sometimes caused serious damage nearby. Figure 4-2 shows a new shopping mall in Gautier. Note the removal of the roofing membrane in the areas of high suction. A shopping center, Jackson Square, opposite this mall suffered much more severe damage (Figure 4-3). In some parts the roof membrane was removed. In others, usually where the glazing system had failed, the decking was also removed. In one location where the wall was not protected by the canopy, the bar joists failed, and the wall, lacking top support, collapsed. This wall was reinforced with steel columns whose hold-down bolts failed in tension (see Figure 4-4). This serious damage was probably due to some very minor details. The high internal pressure probably resulted not from glass breakage but from failure of the wall-to-window-frame connection (Figure 4-5). The bar joists

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Figure 4-41 Damaged building with low-pitched roof (left) and undamaged A-frame house (right), Dauphin Island. ally the result of failing to follow the nailing schedule of the Standard Building Code. Note in Figure 4-42 the advantage to be gained by using a hipped roof, in which the areas of high suction are avoided. Shingles were quite frequently stripped from edges of roofs, again as a result of locally high suctions. Special care is needed in these regions, but an architect or engineer using the Standard Building Code prior to 1986 to specify the uplift requirements would probably have underestimated the need by at least a factor of two. Glass breakage was common in unprotected windows. In some cases, patio doors were stripped from their mountings, but wood-framed walls generally did not collapse unless the top support provided by the roof was lost. Paneling was sometimes sucked from the corners of buildings. Major damage and collapse appear to have been restricted to those buildings without hurricane anchors and those using light-duty anchors in which windows or doors had broken or porches had been poorly secured (Figure 4-43). Some recent wind tunnel experiments have shown, in retrospect, that this pattern of damage was to be expected. Based on tests of models of buildings with 24-ft span roofs and 4:12, 6:12, 8:12, and 12:12 gable and hipped roofs, predictions were made (Table 4-6) about the probability of serious damage based on the form of roof-to-wall connection, degree of

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Figure 4-42 Roof damage showing the influence of roof shape, Dauphin Island. Figure 4-43 Repeated damage patterns in similarly constructed buildings, Dauphin Island.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 TABLE 4-6 Risk of Major Structural Damage in a Design Hurricane   Damage Risk by Structure Type and Class   Type 1b Type 2c Conditionsa Class Ad Class Be Class Cf Class Ad Class Be Class Cf Sheltered, secured Low Low Low Low Low Low Sheltered, unsecured Low Low Low Medium Low Low Open, secured Low Low Low Medium Low Low Open, unsecured High Low Low High Medium Low Severe, secured Low Low Low Medium Low Low Severe, unsecured High Low Low High Low/ High medium a Sheltered—wooded areas, densely packed subdivisions, and centers of towns; open—flat, open country with few obstructions; severe—flat areas adjacent to the sea; secured-windows protected against damage, porches and carports secured against uplift forces; and unsecured—all other buildings with porches and carports or with windows exceeding 5 percent of the wall area. b Type 1—hip roofs with slopes greater than 25 degrees. c Type 2—all other roofs. d Class A—ordinary toenailed connections. e Class B—light-duty hurricane anchors. f Class C—heavy-duty hurricane anchors. SOURCE: Peter Sparks et al., 1988. protection of windows, or size of roof overhang and wind exposure (Hessig, 1986; Sparks et al., 1988). Using the classification in Table 4-6, nearly all older buildings on Dauphin Island would have been class A (using toenailed connections). Mobile County's building regulations do not specify the type of hurricane strap or anchor to be used, but virtually all of the buildings that sustained serious damage used the type of anchor described in class B construction (light-duty hurricane anchors), and this was apparently considered acceptable by the local building inspector. It is not known on what basis the type of acceptable hurricane anchor was determined. However, a major tightening of regulations came after Hurricane Camille, at which time the South Florida Building Code (SFBC)

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 was widely considered to be superior to the earlier editions of the Standard Building Code (SBC1). It can be seen from Table 4-7 that for a typical single-story house 24 ft by 48 ft, elevated 8 ft above the ground and adjacent to the ocean on Dauphin Island, SBC1 was in fact more appropriate. Furthermore, the amended Standard Building Code (SBC2) dangerously underestimated the forces on steeper-pitched roofs because it ignored the suction generated when the wind is parallel to the ridge of the roof. Fortunately, most buildings on Dauphin Island were not professionally designed. Had they been designed in accordance with SBC2, houses with steeper-pitched roofs would have stood little chance of survival. The experimental values given at the bottom of Table 4-7 represent the expected upper bound of the forces. They use the measured exterior pressure coefficients from Hessig (1986) and combine them with an internal pressure coefficient for a major windward opening from ANSI (1982). Values for a coastal location as described in ANSI (1982) using both external and internal pressure coefficients from normal and major openings are also given in Table 4-7 (ANSI—fully enclosed and ANSI—partially enclosed). Note that these are considerably larger than those recommended by any of the earlier editions of the Standard Building Code or the SFBC. Using anchors with 500-lb capacity, as suggested by ANSI for normally enclosed buildings, would probably provide sufficient reserve capacity to accommodate the extra force created by the development of a major windward opening. Anchors generally have capacities in the range of 250 to 350 lb or 500 to TABLE 4-7 Maximum Roof Anchor Forces (lb) Roof Pitch 4:12 6:12 8:12 12:12 SBC1 430 419 407 376 SBC2 244 176 80 Compression SFBC 251 217 211 192 SBC3 (fully enclosed) 398 387 343 291 SBC3 (partially enclosed) 560 552 509 457 ANSI (fully enclosed) 499 474 464 442 ANSI (partially enclosed) 819 766 895 736 Experimental (hip-partially enclosed) 510 560 587 565 Experimental (gable-partially enclosed) 640 730 673 641

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 600 lb. Clearly, SBC2 or SFBC would permit those with the lower capacity. It is unlikely that such anchors would survive the development of a windward opening. SBC1 and SBC3 would permit only the higher-capacity anchors similar to those required by ANSI. SBC3 also gives the designer the option of considering the building to be partially enclosed. These values are also given in Table 4-7. Note that these are still lower than the experimental values, but this code uses only 80 percent of the measured values determined by Davenport et al. (1977, 1978) and assumes only flat, open terrain (exposure C). It must be emphasized that Table 4-7 refers to a rather modest span of 24 ft, with joist spacing of only 16 inches. As land values have risen, beach houses have become larger, and prefabricated trusses spaced at 2 ft on center have become more common. For such a spacing and a 40-ft span, the uplift forces per connection might be as high as 1.5 kips, or more than four times the allowable capacity of a light-duty hurricane anchor. Clearly, something more than an instruction to ''hurricane strap each rafter to plate'' (Figure 4-37) is required. The correct design of the roof-to-wall connectors is vital to the integrity of the complete structure. If the roof separates from the walls, complete structural collapse is likely to follow rapidly. In most cases, buildings survived this storm because they had roof-to-wall connections capable of resisting the uplift forces imposed on them. These forces, however, vary considerably depending on the shape of the roof, the weight of the roof, the degree to which the windward wall is open, the size of roof overhangs, and the local wind speed and direction. Ironically, those structures that failed could probably have been saved had they used anchors of slightly higher capacity or doubled up on the ones used. Compared with the losses sustained, the initial cost would have been negligible. During construction, anchors can be purchased and installed for less than $1.50 each, and the difference between a light-duty anchor and a heavy-duty anchor is a few cents. In most houses there are fewer than 100 connections to be made. Other Structures A few other types of buildings did sustain some damage. The Isle Dauphin Country Club lies to the south of the forest cover, among the dunes in an exposed location. The roofing membrane had been removed from most of the buildings. Also, a large portion of a wood roof deck was removed from the most windward building (Figure 4-44) in an area with a large overhang. Figure 4-45 shows a detail of the roof-to-wall connection, clearly unsuitable for the high uplift forces. To the east of the country club lie two similarly exposed sets of build-

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Figure 4-44 Loss of roof, Isle Dauphin Country Club, Dauphin Island. Figure 4-45 Detail of toenailed roof-to-wall connection associated with damaged roof in Figure 4-44.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Figure 4-46 Dauphin Beach Surf Club (left) and Sand Castle Condominiums (right), Dauphin Island. ings, the Dolphin Surf Beach Club and the Sand Castle Condominiums (Figure 4-46). The former was built about 1970. The buildings are not elevated above the 100-year flood level. Although the high-water mark did not reach the buildings in this storm, they were damaged by water in Hurricane Frederic. Metal straps could be observed tying the roofs to the walls, but the masonry appeared to have been inadequately tied to the stud walls. Similar observations could be made for the upper siding. In one location, the loss of the windward wall precipitated the loss of the the roof. The Sand Castle Condominiums had just been completed. Figure 4-46 shows the extensive roof damage. The roof trusses had apparently been well secured to the framing system, but the roof sheathing had not been properly nailed to the trusses. Several reinforced masonry buildings, presumably built in accordance with Corps of Engineers specifications, are located on the easterly tip of the island adjacent to Fort Gaines (Figure 4-47). These buildings survived Hurricanes Camille and Frederic intact, apart from loss of roof coverings. Similar behavior was observed following Elena. The built-up roof was stripped from a high percentage of the buildings by the storm, resulting in water damage to almost all units. Two preengineered metal buildings were sited near the bridge connecting

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 the island to the mainland. A metal-clad, timber-framed boat storage facility had been completly destroyed in Frederic and was replaced with the preengineered building shown in Figure 4-48. Only superficial damage to cladding was observed, in contrast with the severe damage to the adjacent sign. Of interest is the fact the building had been designed in accordance with the 1974 revisions to the 1973 Standard Building Code (SBC2), but for a wind speed of 120 mph for the cladding and 140 mph for the primary framing; the pressure coefficients selected were based on one wall being open. As seen in Figure 4-48, the walls are free of cladding near the ground. This would have served to prevent any significant increase in internal pressure during the passage of Elena. Other preengineered building was located a short distance from the boat storage facility and experienced minor loss of trim (Figure 4-49). Considering the damage sustained by many schools on the mainland, it is interesting to note that the small wooden school on Dauphin Island survived with very little damage, as it had in Hurricanes Frederic and Camille (Figure 4-50). Yet, it probably experienced the highest wind speed of any school affected by Elena and probably had the smallest amount of professional input in its design. It did, however, have a fairly heavy, steep-pitched hip roof that generated little if any uplift. This is in stark contrast to the Figure 4-47 Fort Gaines area showing damage to flat-roofed buildings, Dauphin Island.

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Figure 4-48 This preengineered metal boat storage facility on Dauphin Island suffered only minor damage in Hurricane Elena. (Note damage to sign in foreground.) Figure 4-49 Undamaged modern metal building, Dauphin Island. (Note properly designed canopy.)

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Figure 4-50 Dauphin Island School. lightweight fiat roofs of the damaged schools that would have been subjected to significant uplift forces. Comparison with Hurricane Frederic During the survey, several residents of Dauphin Island reported that many buildings severely damaged by Hurricane Elena had survived Hurricane Frederic with little or no damage. This was confirmed by reference to aerial photographs taken after Hurricane Frederic (U.S. Army Corps of Engineers, 1981). An analysis of the wind data from the Dauphin Island Bridge by Reinhold and Mehta (1981) determined that the fastest-mile wind speed during Frederic, reduced to started height and exposure, was 106 mph. The fact that very few houses were damaged on the island was considered a triumph of good building regulations. It therefore appears strange that such extensive damage took place during Elena when the wind speed, reduced to the same averaging time, height, and exposure at Fort Gaines within a few miles of the bridge, was only 96 mph. As indicated earlier, most of the damage in Elena was on the exposed western end of the island. The eye of Hurricane Frederic passed directly over this area (Figure 1-2), and so it is probable that these buildings did not experience the high winds measured at the bridge. Most of the parts of the island that did experience winds similar to those on the bridge were heavily

OCR for page 41
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 forested and provided protection to the buildings. Some damage did take place where buildings were exposed. In contrast, the eye of Hurricane Elena passed offshore, but approximately parallel to the island (Figure 1-2). The exposed western end of the island experienced high winds approaching over the sea for several hours. Thus, although Mobile County had been diligent in requiring hurricane anchors on houses, the difference in the patterns of damage during Frederic and Elena was probably attributable not only to structural factors but also to unusual meteorological and topographical circumstances.