ABSTRACT Roof systems are one of the most commonly damaged elements of the building enve¬ lope during natural disasters such as tornados and hurricanes. Determining if dam¬ age has occurred to a roof and the extent of the damage related to a storm event can be a difficult and controversial issue among building owners, professionals, and insurance companies. This paper will discuss how to utilize wind uplift field-testing procedures as a tool to assess and determine if an adhered roof system has failed from a storm event. The presentation will also review some of the tasks and proce¬ dures to follow when performing a detailed damage assessment. These tasks include collecting weather data, performing code research, documenting visual observations, and performing additional testing such as nondestructive testing (electrical capaci¬ tance meter, infrared imaging, etc.) and destructive test openings. Information gained from these tasks will assist in determining the existing conditions and the extent of damage from the storm event. SPEAKERS Christopher W. Giffin is a licensed architect specializing in the diagnosis and repair of building envelope problems. He has been involved with many roofing- and water¬ proofing-related projects having to do with both historic and contemporary struc¬ tures. Notable projects include the Candler Building, the Grove Park Inn Resort & Spa, Chicago public schools, and U.S. Cellular Field. Giffin has performed numerous building envelope condition assessments and investigations, including storm damage assessments following hurricane and tornado events. He has also managed the design and construction period services for the installation of several new or reno¬ vated roofing and waterproofing systems. James M. Brown is a licensed architect who has been involved with many special¬ ized roofing- and waterproofing-related projects. He has experience with many differ¬ ent types of materials, including masonry, exterior insulation and finish systems (EIFS), and stucco. He has conducted building envelope condition assessments and storm damage assessments on several projects. He has also performed several dif¬ ferent types of field tests for quality control purposes and for diagnosis purposes in investigations, including water testing, adhesion testing, and wind uplift testing. CONTACT INFO: cgiffin@wje.com andjbrown@wje.com or 770-923-9822 Giffin & Brown – 56 Proceedings of the RCI 24th International Convention Ev luating St rm D mage Fl t-Ro f Assemblies Test No. 13 Roof Area No. 1 Test No. 10 Roof Area No. 2 Test No. 9 Test No. 8 Test No. 7 Test No. 6 Test No. 5 «• Roof Area No. 4 * Test No. 4 H Test No. 3 Test No. 1 Test No. 2 Roof Area No. 6 * Test No. 12 mt Test No. 11 Figure 1 – Overall plan of roof area. 8,9 a. $ Roof Area No. 7 at •- ’ Roof Area No. 3 iii! * fl. . 8 Roof Area No. 5 INTRODUCTION Roof systems are one of the most common¬ ly damaged elements of the building envelope during natural disasters such as tornados and hurricanes. Determining if damage has occurred and the extent of the damage that may be related to the storm event can be diffi¬ cult and controversial among building owners, professionals, and insurance companies. This paper will discuss how to utilize wind uplift field-testing procedures as a tool to assess and determine if an adhered roof system has failed from the storm event. The paper will also review some of the tasks and procedures to fol¬ low when performing a detailed damage assessment. These tasks include collecting weather data, performing code research, docu¬ menting visual observations, and performing additional testing such as nondestructive test¬ ing (electrical capacitance meter, infrared imaging, etc.) and destructive test openings. Information gained from these tasks will assist in determining the existing conditions and the extent of damage from the storm event. After a major storm event, the condition of a roof system can be generally summarized as follows: • The roof or building is at total loss. The roof is missing, or the building is dam¬ aged beyond repair. • Some percentage of the roof is missing or a partial loss. Obvious visual dam age because of the storm has occurred. • There is no readily apparent storm damage to the roof assembly. For the first two conditions, evaluating storm damage to the roof system or determin¬ ing what components have been affected due to the storm is generally evident. However, when it appears that there has been no discernable damage, and a claim has been made that the roof system has been compromised, determining if storm damage occurred and to what extent can often be contentious. While in the United States, hurricanes and other severe storm events can occur anywhere along the Gulf or Atlantic coast, this paper will present experi¬ ences and observations made during numerous roof Proceedings of the RCI 24th International Convention Giffin & Brown- 57 Figure 2 – Wind speeds as Hurricane Charlie crossed the Florida peninsula (image, cour¬ tesy FEMA). assessments following events that occurred in Florida in 2004 and 2005. The variety of construction types, evolving and changing building codes over the past decade, and the number of named storms during this period, result¬ ed in several challenges in the assessment of storm damage to flat-roof assemblies. HISTORY OF BUILDING One part of assessing the roof system is acquiring any back¬ ground information and historical records of the building and roof. This is important, as often the assessment involves evaluating or testing something that is old. Since the building codes have evolved to include more stringent wind speeds, newer roofs should be able to better withstand wind events and should perform better than a 25-year-old roof. Historical information may help to deter¬ mine the existing condition of the roof prior to the storm event. Depending on the information available, it may help determine whether the roof can be repaired or if replacement is necessary. In addition, the information will aid in the visual observation portion of the assessment. If available, key information would include building orientation, the age of the building and the roof, the number of roofs, roof geometry, roof height, roof area, type of roof assembly, and the history of mainte¬ nance and re¬ pairs. Considera¬ tion should be given to any unique site condi¬ tions or building geometries that would create lo¬ calized high-pres¬ sure zones, which may require clos¬ er evaluation. Most of this information can be determined from the architec¬ tural drawings, previous assess¬ ment reports, contractor invoic¬ es from repairs, and from inter¬ viewing building owners or facility engineers. Some of this information can also be obtained or confirmed from the visual observations made during the assessment. Developing a roof plan that identifies the various roof areas, types of equipment, and other related components will be useful during the survey and testing portions of the assessment (Figure 1). WEATHER DATA Acquiring weather data on the storm event can be helpful in determining storm-related dam¬ age. The purpose of gathering weather data is to understand the storm and its effects on the build¬ ing. The data may not be available immediately after the storm. Over time, as more data are compiled and made available, the information can assist in analyzing and confirming field observations. The storm information gener¬ ally provided consists of the hurricane category, wind speed, wind gust, location and path of the storm, location and path of tornados, hail, amount of rainfall, flooding, storm surge, and images of the storm and damage (Figure 2). The information is pro¬ vided in various formats from charts, maps, illus¬ trations, photographs, and Table 1. Common Form of Damage Based on Hurricane Category_ Category Wind Speed Common Forms of Damage Storm Surge 1 74 to 95 mph Minimal damage, primarily to trees, foliage, 4 to 5 ft above normal shrubbery and unanchored mobile homes 2 96 to 110 mph Moderate damage such as trees blown down, major 6 to 8 fl above normal damage to exposed mobile homes, and some damage to the building envelope such as roots, doors, and windows ” 11 1 to” 130 mph *’ Extensive damage such as large*trees blown down?* ?*9 to 12 ft above normal ‘ * destroyed mobile homes, and some structural •? damaoe to roofs and small buildings 4 131 to 155 mph Extreme damage such as large trees blown down, 13 to 18 ft above normal complete destruction of mobile homes, and extensive damage to roofs, doors, windows, and complete failure of roofs on small residences Greater”than ‘8^’ Catastrophic damagc’such as complete failure of Greater than 19 ft abdve^l gr – ? 155 mph roofs on residences and industrial buildings. normal I extensive damage to doors and window’s, and some i 7c complete building failure | Giffin & Brown – 58 Proceedings of the RCI 24th International Convention existing flat-roof membrane is still intact and there is no obvious vis¬ ible damage, determining if the roof is damaged is not as clear. In order to determine if dam¬ age has occurred, the definition of “damage” must be considered. A proposed definition of damage for flat-roof assemblies might include the lack of functional integrity, lack of water tightness, or the reduction of the expected service life of the roofing material. Dam¬ age can also be classified into two categories: deterioration and damage from natural weathering, and storm damage. There are dis- V ■ J ■ ■ ■ ■11111 ■ ■ I ■ I ■ ■ ■ Figures 3 and 4 – Examples of natural weathering observed on a flat-roof system. ■—■ images. Storm-weather data can be collected from many sources and agencies. However, the most widely used agencies are the National Oceanic and Atmos¬ pheric Administration (NOAA), the National Weather Service (NWS), and the National Hurricane Center (NHC). Hurricanes are rated from 1 to 5 on the Saffir-Simpson Hurri¬ cane Scale {Table 1). The ratings are based on the hurricane’s susli l«i ■II tained wind speed. The sustained wind speed is the speed of the hurricane sustained over the water for one minute. The rating scale also relates to the type of potential property damage created by the storm. Category 1 and 2 hurricanes are dangerous, war¬ rant preventive measures, and cause moderate damage. How¬ ever, hurricanes reaching Cate¬ gory 3 and above are considered major hurricanes and have a greater potential for loss of life and severe property damage. DEFINITION OF DAMAGE When performing a storm damage assessment on an exist¬ ing roof membrane, defining what is damaged can often be challeng¬ ing. If the roof membrane, insula¬ tion, and structural deck are missing or sitting on the adjacent property, it is easy to determine that the storm event produced the damage. The failure mode that initiated the damage can be var¬ ied; nonetheless, the storm played a key role in producing the dam¬ age. On the other hand, when the tinct differences between these two forms of damage, and they need to be considered and docu¬ mented in the assessment. Natural Weathering Natural weathering of gran¬ ule-surfaced modified-bitumen and built-up roof membranes includes uniform loss of granules, exposed reinforcing fabric, cracked and brittle membranes, blisters, ridges, and splits {Figures 3 and 4). Natural weathering can also be from entrapped water Proceedings of the RCI 24th International Comention Giffin & Brown – 59 within the roof assembly. Generally, water entrapped within a roof system is the result of re¬ peated water infiltra¬ tion into the roof sys¬ tem that occurs over a long period of time. While storm damage from punctures can allow water into the roof system, the en¬ trapped water is typi¬ cally isolated to the point of the damage, unlike widespread areas of entrapped water from natural weathering. Over time, entrapped wa¬ ter within the roof system will decay the underlying materials and weaken the co¬ hesive strength of the material, or loosen the bond or adhesion between the various materials. Granule- and gravel-surfacing loss from natural weath¬ ering is generally uniform throughout the roof area. Storm damage to granule¬ surfaced modified bi¬ tumen- and gravel¬ surfaced, built-up roof membranes typi¬ cally results in local¬ ized areas where the granules or gravel are missing, expos- Figure 5 – Example of storm damage to granule-surfaced modified-bitumen membrane from glass debris. Figure 6 – Example of wind uplift damage to a portion of a smooth-surfaced modified-bitumen roof membrane. from the evapora¬ tion of liquid water and expan¬ sion of water vapor in the blis¬ ter. As the blisters grow, they impart more stress on the roof membrane and can result in splitting of seams or rupturing of felts, allowing more water over time to infiltrate into the roofing system. Natural wea¬ thering of thermo¬ plastic roof mem¬ branes can in¬ clude plasticizer loss, membrane em¬ brittlement, loss of reflectivity, and dirt accumulation. For thermoset mem¬ branes such as ballasted EPDM membranes, there is a tendency to shrink and pull away from the perimeter flash¬ ings. Storm Damage Storm damage to granule-sur¬ faced modified bi¬ tumen and builtup roof mem¬ branes includes punctures and scrapes from foring the underlying bitumen, typically at the corners of the building. Wrinkles and ridges of builtup roof membranes are a form of natural weathering where, over time, moisture absorption by the roofing felts and cyclic fatigue produce the observed wrinkles and ridges. Curled or improperly attached insulation boards can also, over time, telegraph through the roofing membrane as ridges or wrinkles. Blisters are the result of a void that is created between the roof¬ ing plies, or between the roof membrane and the underlying in¬ sulation, and are formed when the roof membrane is installed. Over time, blisters grow in size eign object impact, scoured and missing areas of granules or gravel surfacing, uplifted and detached roof mem¬ brane, broken and damaged roof insulation, and missing areas of the roof assembly (Figure 5 and 6). Storm damage to fully adhered thermoplastic and thermoset membranes can include punc¬ tures, cuts and tears, uplifted and Giffin & Brown – 60 Proceedings of the RCI 24th Internationa! Convention Coping Mansard HMR’- BL BL ® BL Test No. 5 Opening est No. 3 Debris HMR Mastic repairs Inspection Opening Debris in this area Puncture in membrane Damaged gutter Inspection Opening Damaged gutter 85’ damaged siding Possible new flashings at pitch pans Steel frames for elevated HVAC units Evidence of possible ponding water granule loss 75’ damaged siding Possible ponding water granular loss (severe) with exposed reinforcement fibers Roof area and gutter damaged removed and temporarily patched – no cap sheet exposed reinforcement YA 20’ removed damaged coping Legend: Damaged or Removed Coping » m Damaged Gutter Damaged or Removed Siding — — – – Unsealed Penetrations I I Roofing Debris “ ~ Z Z Expansion Joint KI Inspection Opening ■wmwmm Fear in Membrane Test Location (2) Flashing seams 1 __ _ I Mastic Repairs High Moisture Readings (HMR) 1_ 1 Ponding Water Blisters f BL) I . 1 Patched Area Missing Cover to Exchange Fan (MC) r. v.1 Weathered Membrane Figure 7 – Roof survey plan identifying locations of damage. detached roof mem¬ brane, broken and damaged roof insu¬ lation, and missing areas of the roof assembly. Storm damage to the roof mem¬ brane is generally accompanied by damage to other items on the build¬ ing or roof area. This might include damaged and blown-off sheetmet¬ al copings, gutters, or fascias; dented or damaged rooftop mechanical units; or damaged or missing compo¬ nents of the exteri¬ or wall. Other indi¬ cators of stormrelated damage and its intensity in the area can include fallen trees or light poles, broken win¬ dows and doors, or damaged signs and awnings. VISUAL OBSERVATIONS After obtaining and reviewing the historical data, a visual survey of the building and the roof should be per¬ formed. The pur¬ pose of the survey is to identify, lo¬ cate, and docu¬ ment any damage to the building and roof. These observations are critical in determining if the dam¬ age is a result of the storm event, natural weathering, or previous damage. If the damage is stormrelated, the observations are important in determining if defec¬ tive design or installation were a contributing factor to the loss. The visual survey is conducted in a manner similar to a normal roof maintenance inspection. A roof plan should be used to illustrate the location of all the pertinent observations and dam¬ age. Use the most recent version of the roof plan. If a roof plan is not available, one should be drawn up while on the roof. The roof plan should be to scale and should illustrate the locations of different types of roof edges, roof equipment, penetrations, and accessories (Figure 7}. Photo¬ graphs of the observed conditions should be taken. The survey should also include an inspection of the underside of the roof deck, exteri- Proceedings of the RCI 2 4th International Convention Giffin & Brown – 61 or walls, and areas adjacent to the building prior to inspecting the roof. The underside of the roof may reveal signs of water intru¬ sion, rust, dry rot, poor attach¬ ment, roof uplift, or other prob¬ lems that may be the result of previous damage or the storm event. Special attention should be given to roof penetrations and along the perimeter of the exterior walls. If the visual damage to the roof membrane extends to the edge of the roof, thoroughly docu¬ ment the roof-edge detail. Deter¬ mine the materials used along with the fastener types and their relative location and spacing. The observations should be noted and illustrated on the plan so they can be translated to the roof surface. The exterior walls may reveal signs of water staining, cracks, settlement, plumbness, move¬ ment, debris impact, and damage to drainage accessories such as downspouts, gutters, and scupper heads. When inspecting the exte¬ rior walls, observe and document the adjacent areas for storm surge and amount and type of debris. This is important to help under¬ stand the effects of the storm event. The roof membrane and adja¬ cent rooftop features or elements should be inspected for both nat¬ ural weathering damage and storm damage. All deficiencies and defects should be noted on the roof plan. Note the general appearance and condition of the roof, and document the locations and frequency of the deficiencies and defects. Natural weather damage items may include the following: • Blisters. • Membrane slippage. • Fishmouths. • Alligatoring of the flood coat. • Splits. Figure 8 – View of uplift test in. progress. • Ridges. • Granule and gravel surfacing loss. • Ponding water. Storm damage items may include the following: • Debris impact, resulting in punctures and scrapes in the membrane, which can allow water to infiltrate into the roof assembly. • Hail impact damage, result¬ ing in localized granule loss, which can lead to accelerated deterioration and aging of the roof membrane. • Membrane bruising. • Possible exposure of the roof¬ ing felts. • Adhesion loss of the mem¬ brane to the substrate. • Wind scouring, resulting in areas of missing granules or gravel surfacing, which can lead to accelerated deteriora¬ tion and aging of the roof membrane and absorption of water at areas of exposed membrane. • Areas of uplifted and detached roof membrane or substrate materials. When performing the roof sur¬ vey, the following are a few addi¬ tional items to be aware of and to document as part of the storm damage assessment: • Inspect the perimeter flash¬ ings for normal deterioration, granule loss, punctures, tears, open lap seams, wrin¬ kles and ridges, and flashing attachment along the top edge, if any. • Inspect embedded edge metal and gravel stops, as they can tear the membrane due to the differential thermal move¬ ment of the roof membrane and the embedded metal. • Inspect the counterflashings for attachment, rusting, dents, bent sections, punc¬ tures, and open seams that may prevent the counter¬ flashing from protecting the base flashings. • Inspect the copings and cap flashings, as they protect the roofing and wall systems. Check for attachment, dents, rusting, punctures, and open seams. If water bypasses the coping and cap flashings it has a greater chance of infil- Giffin & Brown – 62 Proceedings of tire RCI 24th International Convention trating the roof and wall sys¬ tem. • Inspect penthouse and clere¬ story walls for deterioration, defects, and damage, as they can contribute to water infil¬ tration and damage to the roof assembly. • Inspect the flashings at all roof penetrations. Observe and note the conditions of the lap seams, membrane, seals or sealants, lead flashings, draw bands, and metal rain hoods. Note if the pourable sealer in the pitch pans is weathered, underpoured, or not adhered to the penetra¬ tion substrate. • Inspect the condition and attachment of any expansion joints. Ensure the expansion joint is free from defects and performs in a watertight manner. • Survey the roof equipment. Note the condition of and attachment of the roof equip¬ ment, if any of the equipment is damaged or missing, and if the equipment rests directly on the roof. FIELD UPLIFT TESTING Three standardized roof uplift tests exist. They are as follows: • ASTM E907, Standard Test Method for Field Testing Uplift Resistance of Adhered Membrane Roofing Assem¬ blies. • FM Global Property Loss Prevention Data Sheet, Field Uplift Test 1-52. • Florida Building Code, Test Protocol HVHZ Testing Ap¬ plication Standard (TAS) 124. Each of these tests generally outlines similar procedures to determine the uplift resistance of an adhered roof membrane with either a negative pressure bell chamber or a bonded pull test. When performing uplift tests in a storm damage assessment, the bell chamber test is typically more practical and efficient to perform. These test methods are intended to be used as a measure of the uplift resistance of the roofing system. The tests apply to roof systems with or without rigid board insulation or base plies, which are either adhered or mechanically fastened, and fully adhered membranes. The uplift test is performed by creating a controlled negative pressure on top of the roof surface by means of a fitted plastic cham¬ ber with a pressure-measuring device and vacuum equipment (Figure 8). A 5-ft x 5-ft square plastic chamber is placed over a deflection bar with a dial indicator attached. The perimeter of the chamber is then temporarily sealed to the roof surface. The dial indicator is positioned so that the tip of the dial indicator is in con¬ tact with the roof membrane near the center of the test area. A pres¬ sure-measuring device (manome¬ ter) and the vacuum equipment are attached to the holes provided in the chamber. The vacuum equipment is activated and adjusted to regulate the negative pressure in the chamber to speci¬ fied levels. According to the test procedures, a negative pressure of 15 lbs per sq ft (psf) is created in Table 2. Uplift Test Results at Various Pressures _ | Test 1 No. 15 p<-f 1 plil’l Gauge Gauge at 0 at 60 Seconds seconds (in.) (in.) 22.5 psf l.plili Gauge Gauge at 0 at 60 Seconds seconds (in.) (in.) . 30 pM I pliti Gauge Gauge at 0 at 60 Seconds seconds (in.) (in.) . 45 psi ( pill I Gauge Gauge at 0 at 60 Seconds seconds (in.) (in.) •?»MSS|^’»’“5WB‘W«’-’S»SHSSaagaSSS®S|SS!!S!S^^ * Total ’ Deflection at 45 psi (in.) Cunuiiunj 1 2 3 5 7 3OZ 9 ii IK 13 rrc 15 17 0.0000 0.0569 ‘ 6.0000 753)73980* 0.0000 0.3100 lO.Ob06;X.O.542Ol 0.0000 0.1365 70.00011 0 03887 ‘ 0.0000 O.i32O ’0:0027^07668r 0.0000 0.2561 ‘0.0038 W4021Z 0.0096 0.0162 0.6i20 1.2768 “6.0020X0:2344* 0.0000 0.4472 “0.0030W6.6335* 0.0050 0.4784 0.1156 0.1208 7025390^03630?’ ”0.0407 0.1052 T0r5776^07857rr 0.1836 0.1933 ZO‘.04013E70?0490” 0.2430 0.3123 0.6684^0.9839 „ 0.3740 0.4064 •3)74595 0.671 U 0.0206 0.0236 7X6089 X0.0093* 1.2768 1.4450 70’.2590ir 0.26577 0.5075 0.5961 ”6.0450^r0.047r 0.6190 0.6648 0.1900 0.2222 “0?673b~Wr 0.7000* 0.1370 0.1471 ZWoKbUiC 0.2309 0.2449 “0.0501 W.0629Z 0.3196 0.3564 -1-0724^1. 75 10, 0.5098 0.7889 “0:7920 XO.8044* 0.0276 6.0286 *0.0123 T’oTobs* 1.4450 1.5938 20.2981W3017Z 0.6184 0.6445 *0:0585^0.06167? 0.8252 0.8494 0.3570 0.3985 76:9687^ i:Q640T 0.2140 0.2864 □?3660j|[L64452 0.3042 0.3918 Zo:O668?3g7O.’ 11567 “0.3759 0.4722 *1.2390WL3706* 0.9648 1 1248 *0.9960 .y 1.0354 7 0.0355 0.0381 ZoTmT&oifC 1.5938 1.5940 70’j9467X’07624C 0.7900 1.0119 Z0″.1007^0.104C 1.0750 1.2-450 0.3985 7’1 .oi’4ii 0.2864 £1.6445 0.3918 T0.U56 0.4722 Z13679_ : 1.1248 0.028′ To^h 1.5820 Movement occurred instantaneously to 1.25” and gauges peaked at 1.5940 in. 1.0119 X0.1011 ■ .”-I ‘K-.. 1.2400 Sudden jump during 45 psf – fastener pop Proceedings of the RCI 24th International Convention Giffin & Brown – 63 Table 3. Weather Data During Uplift Test Test Wind Speed Air Temperature Roof Surface Relative Barometric Heat Index Number (mph) (°F) Temperature Humidity (%) Pressure (inHg) (°F) L_F 2 2.0 88.6 100.5 58.0 30.07 ‘ • IB? . …’97.031^^^65.7 ‘ T .. = 30.06^^^:96 1 4 4.0 86.0 95.2 57.0 30.05 ■■r5j»!»5:5>ii»W^^^ 96.orwss»3.(‘ ’ ■ ‘ ’0.04^^BT?L9^;r 6 4.5 88.0 98.2 55.9 30.05 94.7 M.7;^^«5.9^|||!||»88.9 100.7^^^<50.2 ’0’.0611Sa&7.3 • 8 2.5 88.9 102.5 52.0 30.05 97.6 10 1.8 88.5 98.5 55.0 30.03 95.7 12 2.1 82.4 94.6 66.0 30.05 99.7 ^W^68.5 14 10.0 85.0 100.3 68.5 30.05 — Fl< ~ Fs-oT^MBBiE 90.0 – 105.0 ?<> o ; ■ to u? : ‘ :. • 16 5.0 90.0 104.4 56.0 30.05 the chamber and held for one minute. The negative pressure in the chamber is then raised in increments of 7.5 psf and held for one minute at each increment. The maximum negative pressure we created in the chamber was typically 45 psf. The deflection of the roof membrane is measured at the beginning and end of each increment (Table 2). In addition, the air temperature, roof tempera¬ ture, relative humidity, baromet¬ ric pressure, and wind speed were recorded for each test location (Table 3). The test methods state how many uplift tests should be per¬ formed given the size of the roof area. Typically, a minimum of four tests should be performed, with one additional test for every 10,000 sq ft. The selection of the test area should be made careful¬ ly. Locations where tests should be performed are adjacent to visi¬ bly damaged areas, corner condi¬ tions, perimeter or edge condi¬ tions, and interior field condi¬ tions. According to ASTM 907 and TAS 124, failure of the roof mem¬ brane occurs when the roof is uplifted 1 inch or a sudden bal¬ looning occurs. FM 1-52 classifies failure when a quarter inch of roof deflection is achieved. Depending on the deck type, insulation, or membrane system, this amount of deflection may be too limiting. Therefore, during our assess¬ ment, we generally use the 1-in failure classification. In addition, the standards state that the uplift tests are to be performed when the roof surface temperature is between 40°F and 100°F. One of the problems that can be encountered when performing the uplift test is obtaining false results. One of the issues that can be frequently encountered is that the placement of the chamber over insulation joints or between fasteners can potentially skew the results. If the chamber is placed over steel joists or at a beam loca¬ tion, a stiffer roof assembly will be tested compared to placement of the chamber between a series of joists. When performing these types of tests for quality assur¬ ance purposes, factors such as these can be important to know if a roof system passes or fails. However, when performing an uplift test during a storm damage assessment to determine if dam¬ age may exist, the low pressures typically needed to determine if a roof is damaged or not are gener¬ ally not affected by some of these other conditions. These applied pressures are often well below the design uplift pressures for the building, since only the weight of the roof materi¬ als needs to be overcome by the negative uplift force. Therefore, for the purposes of assessing whether uplift damage has occurred to a roof-membrane assembly, the ini¬ tial negative load of only 15 psf will likely be an indicator whether the roof system is adhered. If the roof surface is not adhered, 15 psf of uplift pressure will normally exceed 1 inch. If a roof membrane resists a negative pressure for some period then fails at some higher negative pressures, the roof membrane was initially adhered, not damaged from the storm, and failed due to the nega¬ tive pressures applied during the test. Uplift Test Pressures ASTM E907 states the nega¬ tive pressure in the test chamber shall be increased “until the agreed-upon pressure is reached.” During the course of damage assessments, determining what the agreed-upon pressure is can be difficult. For the purposes of assessing if damage has occurred, a maximum negative pressure of 45 psf is often sufficient. The selection of this test pressure was based upon the rationale that if a Giffin & Brown – 64 Proceedings of the RCI 24 th International Convention fully adhered roof membrane was uplifted and damaged during a storm event, the roof would no longer be adhered to the sub¬ strate, and little negative pressure would be needed to overcome the weight of the roof materials and lift them above the 1-in failure distance. Some testing agencies, own¬ ers, or contractors may elect to use uplift pressures calculated using the latest building-code requirements. Several problems become apparent when using these uplift pressures to measure hurricane damage and define what constitutes failure of the test. The age of the roof on the building needs to be considered. If a roof system is 20 years old, the design pressures at the time the roof was installed were likely less than those used by today’s build¬ ing code. Over the past 10 to 15 years, the wind velocities and gust coefficients have increased in each building-code revision to better reflect the forces that actu¬ ally occur during a hurricane. The weather data obtained from the storm event could also be used to calculate what the likely uplift forces might have been at the time of the storm. This could then be compared to the original design pressures, as well as the uplift testing results. In south Florida, for example, if the roof assembly was installed in 1996, the design wind speed was 110 mph. However, if the same roof were installed using the current building code, the design wind speed would be 140 mph. This results in a significantly higher uplift pressure applied to the building. Using the current building-code design pressures to determine if damage has occurred or if an existing building can meet these requirements is an inappro¬ priate use of the code. One cannot expect a roof that was installed several years ago, using lower design pressures, to be able to Figure 9 — Diagram Type 1 where the roof is well adhered dur¬ ing the uplift test. resist the higher, modern, build¬ ing-code design pressures. In our opinion, this does not meet the definition of storm damage. Definition of Failure The uplift test procedure out¬ lined in ASTM E907 can be used on an existing roof assembly to determine if storm damage occurred. Similar tests are also outlined in FM Global 1-52 and TAS 124. When utilizing these tests on existing roof assemblies, a thorough understanding of the roof composition and existing conditions and the negative pres¬ sures that will be applied need to be fully evaluated. ASTM E907 states in paragraph 9.1, “Most roof systems subjected to a nega¬ tive pressure will exhibit an up¬ ward deflection that will increase as the negative pressure increas¬ es. Poorly adhered systems will exhibit relatively large increases in upward deflections with rela¬ tively small increases in applied pressure. For roof systems that are well adhered, the increase in deflection will be gradual and at a relatively constant rate up to a point at or near failure. When fail¬ ure occurs due to lack of adhesive or cohesive resistance of the roof system, there will be a sudden increase in the upward deflec¬ tion.” In addition, according to the ASTM E907 test method, failure during the test also occurs when the deflection of the roof mem¬ brane exceeds 1 inch, even if no sudden increase occurs. FM Global 1-52 limits the maximum deflection to 1/4 inch. However, for light-gauge metal deck and bar -joist roof systems, the maxi¬ mum limit for deflection of 1 inch is more common and would pro¬ vide a more reasonable measure of storm damage during these types of assessments. During the test, the deflection is measured at each negative pressure increment. If these two variables are plotted on a chart, a stress/ strain diagram can be drawn illustrating the relation¬ ship of deflection to applied load on the roof assembly. Based upon our uplift testing experiences and ASTM definitions, four general types of stress/ strain diagrams can be developed. The first type of diagram illus¬ trates a roof that performs well during the uplift test. This is indi- Proceeiiings of the RCI 24th International Convention Giffin & Brown – 65 Figure 10 – Diagram Type 2 where the roof progressively fails during the uplift test. cated by a shallow sloping line that gradually increases in deflec¬ tion as the negative pressure is applied (Figure 9). The deflection of the roof also does not exceed 1 inch during the pressure incre¬ ments. This test demonstrates that the roof assembly was well adhered and attached to the structure prior to the start of the test and remained attached upon completion of the test. As a result, the tested roof membrane was not uplifted by a storm event. The second type of diagram is an increasingly steeper curve, or an exponential type of curve where the line starts out on a shallow slope, then increases dur¬ ing each pressure increment (Figure 10). This would indicate that the roof assembly resisted the initial negative loads applied, then progressively delaminated cohesively or adhesively as the pressures increased. The deflec¬ tion recorded at the end of the test may or may not have exceeded 1 inch. With this diagram, there is no clear spike or sudden increase in the deflection as the pressure increases. This test demonstrates that the roof assembly was well adhered and attached to the structure prior to the start of the test and failed during the test, if the deflection exceeded 1 inch. As a result, the test also indicates the tested roof membrane was not uplifted by a storm event. The third type of diagram would be one that initially starts out with a shallow line similar to the first graph, and then jumps steeply upward within one pres¬ sure increment (Figure 11). This type of diagram would indicate the roof assembly was well adhered during the initial load¬ ings, then failed suddenly – either cohesively or adhesively – during the test and was no longer attached. Often, the deflection recorded with this type of diagram would exceed 1 inch. This test demonstrates that the tested roof assembly was well adhered and attached to the structure prior to the start of the test, and failed during the test. As a result, the test also indicates that the roof area was not uplifted by a storm event. The fourth type of diagram is one that jumps steeply upward within the first pressure incre¬ ment (Figure 12). This type of dia¬ gram would indicate the roof assembly was not attached, as it could not resist any load. Often, the deflection exceeds 1 inch within the initial 15-psf negative loading. This type of diagram might indicate that the roof was uplifted and damaged during a storm event. It is important to note that further investigation Figure 11 – Diagram Type 3 where the roof suddenly fails at a high load. Giffin & Brown – 66 Proceedings of the RCI 24th International Convention Uplift Pressure (psf) Figure 12 – Diagram Type 4 where the roof membrane is not adhered. and analysis are required to determine the failure plane within the roof assembly and the condi¬ tion of the installed materials. This is often done with inspection openings at the test location to identify why and how the roof failed the test. The delamination and failure of the roof could be the result of installation problems, wet materials, natural weather¬ ing, or from storm damage. The fourth type of diagram is the only one of the four mentioned where actual storm damage might have been detected. Inspection openings are required to verify the test results and to confirm if the lack of uplift resistance was from storm damage. The other three stress/ strain diagrams indicate that the roof assembly was not damaged as a result of a storm. When a roof membrane is subject¬ ed to the negative pressures exert¬ ed on it by the effects of a storm, the roof materials will either resist the pressures or fail and become detached from the substrate. When roof assemblies are uplifted to the point of failure during a storm event, the effects are imme¬ diate and irreversible. The roof system is then no longer attached, and thus cannot withstand any future applied load, either from wind or during subsequent roof¬ uplift testing. Therefore, if a roof assembly has been uplifted and damaged during a storm event, it will not resist much negative applied load, and large initial deflections will occur when tested. As stated in ASTM E907, poorly adhered roof systems or roof systems that have been dam¬ aged or uplifted by a storm event exhibit relatively large increases in upward deflection with relative¬ ly small increases in applied pres¬ sures. These applied pressures are often well below the design¬ uplift pressures for the building, since only the weight of the roof materials needs to be overcome by the negative uplift force. There¬ fore, for the purposes of assessing whether uplift damage has occurred to a roof-membrane assembly, the initial negative load of only 15 psf will likely be an indicator of whether the roof sys¬ tem is adhered. If a roof mem¬ brane resists a negative pressure for some period, then fails at some higher negative pressures, the roof membrane was initially adhered and not damaged from the storm, and failed due to the negative pressures applied during the test. This would correspond to diagrams one, two, and three. In addition, when evaluating an older roof system, the uplift tests might also indicate the existing roofs do not meet the current building code uplift requirements, but this is not damage from a storm event. NONDESTRUCTION EVALUATION METHODS The three main nondestruc¬ tive tests for evaluating the pres¬ ence of moisture within a roof assembly are infrared thermogra¬ phy, electrical capacitance, and nuclear detection. Depending on the type of roof assembly being evaluated, each of these tests has advantages and disadvantages that need to be considered in the evaluation process. Performing these types of tests can be of assistance in determining the extent of entrapped water within a roof assembly. By themselves, these tests may not be able to identify if the entrapped water is a direct result of the storm, only that water is entrapped in the assembly. If a corner or section of the roof has been damaged, the non¬ destructive tests could be per¬ formed on the intact adjacent roof areas to determine if water exists in the roof that likely occurred as a result of the storm. Conversely, if no apparent roof damage can be identified, yet there are a number of natural weathering-related con¬ ditions present, along with areas of entrapped water, it is possible the entrapped water is a result of damage due to natural weathering and has been entrapped in the roof for a significant amount of time. It is also possible that the effects of long-term entrapped moisture can influence the bond of the roof membrane to the Proceedings of the RCI 24th International Convention Giffin & Brown – 67 underlying insulation, which can be misinterpreted as storm dam¬ age. The combination of the visu¬ al observations, uplift testing, and inspection openings can then be used to help determine if these areas have been damaged by the storm. INSPECTION OPENINGS When performing an evalua¬ tion of storm damage on a flat¬ roof assembly, inspection open¬ ings can provide critical clues into whether the roof system has incurred roof damage. Upon com¬ pletion of the uplift test where the roof assembly has exceeded the 1- in failure deflection criteria, an in¬ spection opening should be made. It should also be noted at what pressure the 1-in failure occurred and what type of roof uplift curve was generated, as this can provide clues to the mode of failure. The inspection openings should be approximately the same size as the uplift chamber or 5 ft x 5 ft. An inspection opening of this size will generally allow for exami¬ nation and determination of many conditions, including • The existing roof materials and their condition. • The amount and spacing of insulation fasteners. • Types of fastener plates. • The location of board joints. • The amount of adhesive used and its coverage area. • The presence of moisture in the roof system. The mode of failure should be determined in the inspection opening, and this information can be used to determine if the results of the uplift testing curve are in agreement with the inspection openings. This would confirm if the uplift test damaged the roof assembly, or if the roof was dam¬ aged as a result of the storm event. This information can also be used to compare the estimated uplift forces that may have occurred during the event. In addition to the inspection open¬ ings at the uplift test location, in¬ spection openings taken in an ad¬ jacent or nearby roof area would further support the results of the uplift test and observations of the inspection opening at the cham¬ ber. BUILDING CODE SUMMARY Researching and interpreting the relevant building codes is essential in determining the pos¬ sible repair methods for storm¬ damaged roof areas. In this instance, the applicable building codes are the current building codes at the time of the event and not the building codes used or defined during the design and construction of the building. In some cases, more than one code may be applicable to the building. Be sure to check for both state and local codes and to follow the most stringent applicable code, ensuring all of the amendments and supplements have been col¬ lected. Building codes have an evolving language, and the amendments and supplements may contain changes to the origi¬ nally issued code that can affect the repair method on the building. A thorough review of the code is also needed to determine the classification of, or the level of work to be performed. For exam¬ ple, if using the Florida Building Code, the classification of the work and the level of alteration or combinations of levels must be selected prior to determining the subsequent applicable provisions. Work on existing buildings, in¬ cluding roofing-related work, can be classified as either Repairs, Alteration – Level 1; Alteration – Level 2; or Alteration – Level 3. Each classification has distinct requirements and parameters. CONCLUSIONS When performing a storm¬ damage assessment on an exist¬ ing flat-roof assembly, the follow¬ ing are some of the tasks that could be performed so that a full and accurate assessment can be made: • Determine the history and background of the building and roof construction. • Obtain the weather data sur¬ rounding the storm event. • Determine what damage has resulted from natural weath¬ ering and what damage has occurred from the storm event. • Conduct a visual survey of the roof and surrounding building elements. • Conduct a wind uplift testing of the roof system to deter¬ mine the uplift resistance of the system. • Conduct nondestructive moisture surveys to docu¬ ment entrapped water in the system. • Make roof inspection open¬ ings to determine the mode of failure and type of damage in the roof system • Determine the applicable code requirements for repair or replacement of a damaged roof section. Each of these tasks in the storm-damage assessment can provide vital information in help¬ ing to determine the extent of storm damage. Depending on the roof assembly and potential issues being evaluated, a thor¬ ough storm-damage assessment will often require performing many, if not all, of these tasks. The observations and findings, after each task, need to be com¬ pared against the observations and findings from the other tasks Giffin & Brown – 68 Proceedings of the RCI 24th International Convention to ensure they complement and support the overall assessment of the roof condition. Simply per¬ forming a visual inspection, mak¬ ing inspection openings, or con¬ ducting a series of uplift tests alone may not yield enough infor¬ mation to make a full and accu¬ rate assessment. Wind uplift testing may be necessary to determine if the storm event damaged the attach¬ ment of the roof membrane or insulation to the substrate, which would otherwise be missed if a visual survey of the roof were per¬ formed. Uplift testing can be effec¬ tively used as an evaluation tool but will need to be used in con¬ junction with inspection open¬ ings. Many underlying conditions may affect the results of uplift testing, such as the placement of the chamber over insulation joints, fasteners, steel joists, or beams. These factors can influ¬ ence the stiffness of the roof assembly. Therefore, it is impor¬ tant to perform an inspection opening after an uplift test to doc¬ ument these underlying condi¬ tions. The inspection openings are also useful for determining the existing roof materials and their condition and the mode of failure from the uplift test, if any. Establishing the definition of failure and performing uplift test¬ ing to agreed-upon test pressures is essential in storm-damage assessments. Performing the uplift test pressures on the roof membrane up to the design pres¬ sures or even to failure is often not necessary to determine if the roof membrane has been uplifted or damaged by the storm. The uplift testing can be performed in incremental pressures to deter¬ mine if the roof membrane is adhered to the substrate. Typ¬ ically, incremental pressures of 15 psf, 22.5 psf, 30 psf, and 45 psf will provide enough informa¬ tion on the current performance of the roof membrane (without ultimately damaging the roof at the test area) to determine if storm damage has occurred. Entering the uplift test results into a chart and graphically illus¬ trating the stress/strain diagram will also provide a good indication if the roof is damaged. High deflection of the roof membrane at low pressures will indicate the roof has been previously uplifted and damaged from the storm. Conversely, low deflection of the roof membrane at low pressures and incrementally increasing pressures will indicate the roof is adhered and was not damaged by uplift from the storm. The current standardized uplift tests limit the testing to when the roof-surface tempera¬ ture is between 40°F and 100°F. Depending on the time of the year when the assessments are being performed, this limitation can be difficult to work around. During our own assessments in Florida during August, it was not uncom¬ mon for the roof surface tempera¬ ture to be over 100°F by 10 a.m. The use of portable canopies and tents can be strategically utilized to shade the test area and keep the surface of the roof cool during the test. However, the industry should perhaps consider studying this temperature limitation, as storm events and high winds can occur abruptly when the surface temperatures of the roof are out¬ side of this temperature range. The roof system should still be expected to perform at the same level, no matter if the temperature is 35°F or 135°F. Once the storm-damage assessment is made and the quantity of storm damage is determined, an estimate of the costs associated with repairing the damage is often required, par¬ ticularly when the extent of the damage is in dispute. REFERENCES ASTM Standard E 907, 1996 (2004), Standard Test Method for Field Testing Uplift Resist¬ ance of Adhered Membrane Roofing Systems, American Standards of Testing and Measurement International. West Conshohocken, PA. Property Loss Prevention Data Sheets. No. 1-52. Factory Mutual Insurance Company, 2000. www.fmglobal.com .aspx?id=040 10200. International Code Council. 2008. Florida Building Code, Existing Building. Country Club Hills, IL. Florida Test Protocols for High- Velocity Hurricane Zones. In¬ ternational Code Council, 2008. Country Club Hills, IL. The National Roofing Contrac¬ tors Association. The NRCA Roofing and Waterproofing Manual. 5th ed. Rosemont, IL: NRCA, 2001 Hurricanes… Unleashing Na¬ ture’s Fury. U.S. Department of Commerce, National Oce¬ anic and Atmospheric Admin¬ istration, National Weather Service. Washington, DC, January 2007. U.S. Department of Commerce, National Oceanic and Atmos¬ pheric Administration, Na¬ tional Hurricane Center. www.nhc.noaa.gov. Hurricane Charley in Florida, FEMA 488. U.S. Department of Homeland Security, Fed¬ eral Emergency Management Agency. Washington, DC, April 2005. Proceedings of the RCI 2 4th International Convention Giffin & Brown – 69
