ABSTRACT The 2006 International Building Code disallows the use of loose-laid aggregate on roofs in hurricane-prone regions for fear of damage caused by blow-off. This restriction has had a significant impact on the roofing industry, as loose-laid aggregate has historically been used as an economical material to increase ultraviolet, fire, hail, and traffic resistance to roof membranes. In 2009, under the auspices of the Asphalt Roofing Manufacturers’ Association (ARMA) et al., Jay H. Crandell, PE, developed a design methodology, based on the 1970s work of Kind and Wardlaw, to avoid aggregate blowoff from roofs of buildings of all heights. As a follow-up to the development of this methodology, this researcher has inspected 20 spray-applied polyurethane foam roofs in the Stuart, Florida, area that were covered with loose aggregate and subjected to the combined 2004 wind events of hurricanes Francis and Jeanne. Each roof was examined for the variables required in Crandell’s Method (i.e., building height, parapet height, aggregate size, wind speed, surface roughness) and the likelihood of aggregate blow-off back-calculated to test the methodology; this likelihood was then compared to actual performance. Background Aggregate as a roof surfacing material has a long track record of providing an economical protective material for resistance to ultraviolet radiation, hail, and surface traffic. Additionally, aggregate can provide aesthetic qualities and a predator-free nesting place for certain bird species. However, a well intended but overly restrictive building code change in the 2006 International Building Code (IBC) severely restricted the use of loose aggregate as a roof surfacing material in hurricane zones. Specifically, it stated the following: 1504.8 Aggregate. Aggregate used as surfacing for roof coverings and aggregate, gravel, or stone used as ballast shall not be used on the roof of a building located in a hurricaneprone region as defined in Section 1609.2, or on any other building with a mean roof height exceeding that permitted by Table 1504.8 based on the exposure category and basic wind speed at the site. Section 1609.2 de fines hurricaneprone regions as: Hurricane-Prone Regions. Areas vulnerable to hurricanes defined as: 1. The U.S. At lan tic Ocean and Gulf of Mexico coasts where the basic wind speed is great er than 90 mph (40 m/s) and 2. Hawaii, Puerto Rico, Guam, Virgin Islands, and American Samoa. [Cited from the 2009 IBC.] Thus, as illustrated in Figure 1, the IBC eliminated a historically economic and reliable roof surfacing material from a substantial geographical segment of the roofing industry. Beginning in late 2008, ARMA sponsored Jay Crandell of ARES Consulting to revisit the wind tunnel work of Canadian researchers Kind and Figure 1 – Hurricane-prone regions (Atlantic and Gulf coasts as defined in IBC Section 1609.2) 18 • I N T E R FA C E J U LY 2011 This paper was originally presented at the RICOWI Fall Symposium, November 11, 2010, in Rock Hill, SC. It is published for the first time here, courtesy of RICOWI. Wardlaw (1976 and 1984). Crandell completed his work in May 2009 and presented his findings at several symposia (Crandell, 2009; Crandell and Fischer, 2010). Crandell reviewed the Kind-Wardlaw work and developed a modified design method incorporating the effects of parapet height and gravel size. Note that the Kind- Wardlaw work and design methodology were oriented primarily toward avoiding scour on ballasted single-ply roofs, whereas the Crandell work focuses on avoiding blowoff. (Crandell’s Modified K-W Design Method is reproduced in Appendix A.) The purpose of this paper is to report the findings of a field study of gravelcovered, spray polyurethane foam (SPF) roofs that experienced hurricane-force winds during Hurricanes Francis and Jeanne in 2004. Introduction Twenty aggregate-covered, SPF roofs at 19 locations (two roofs at one location were combined in the data set) were inspected in the Stuart, Florida, area to determine the degree of aggregate blow-off and scouring following the 2004 hurricanes. The physical characteristics of each roof were used to determine the adjusted critical wind speeds (Vcr’) in accordance with Crandell’s Method; these were then compared to the estimated actual wind speeds that occurred at the roof (Vroof). According to Crandell, roofs where the estimated actual wind speed exceeded the adjusted critical wind speed (i.e., where Vcr’ > Vroof) would be at risk for aggregate blow-off. Determination of Estimated Wind Speeds Obtaining an accurate analysis of hurricane wind speeds is not a simple task. Hurricane Jeanne was chosen as the more critical storm event (Jeanne was a Category 3 storm, whereas Francis was a Category 2 storm). For this study, wind speeds were estimated using a combination of available Hurricane Jeanne surface-wind speed maps and adjusting those speeds to basic wind speeds of three-second gusts at 33 ft (10 m) above ground, Exposure C. The net result is the wind contour map/study area map in Figure 2. Roof Inspection Procedures For each roof/building location, the following data were gathered: • Building name, address, coordinates • Building height • Roof dimensions • Parapet description and dimensions • Exposure category • Aggregate sampled • Wind effects on aggregate • Inspection photos • Google Earth overviews Wind speeds were estimated based on the Figure 2 map for each building location. The aggregate samples were sieved (to determine average diameter) and measured for specific gravity. The wind effects on the aggregate were based on observations during the inspections and discussions with building owners and the roofing contractor. Effects were classified as No Scour, Minor Scour, Scour, and Gravel Loss. Data Analysis Crandell’s Method requires the following data inputs: Vmap: The gust speed from the ASCE 7 wind maps. For this study, the estimated wind speeds derived from Figure 2 were used for Vmap values. I: The building importance factor for all of the buildings in this study was 1.0, as all of the buildings were classified as Occupancy Category II. h: The measured building height was used. hg: The gradient height for site wind exposure was set at Exposure B, 1270 ft; Exposure C, 900 ft. All but one building were Exposure B (Roof No. 13 was Exp. C). Roof No. 14 could be either B or C, depending on wind direction; Exp. B was used for the calculations. α: The power law terrain roughness parameter was set at Exposure B = 6.2; Exposure C = 9.5. Figure 2 – Study area with Hurricane Jeanne wind contours. Yellow squares represent inspected roofs. Wind speeds are three-second gusts, 33 ft above ground, Exposure C. (Map from Topo North America 9.0, 2010, © DeLorme, Yarmouth, ME. Used with permission.) J U LY 2011 I N T E R FA C E • 1 9 H: Setting the parapet height for this study presented a dilemma: Most of the study building had partial parapets or mixed parapet heights. This is discussed further in the Discussion section of this report. Parapet heights ranged from 0 to 78 inches. d: Aggregate diameter was determined from sieve analysis of the samples obtained from the study roofs. Average aggregate diameter varied from 0.245 to 1.5 inches. Using Crandell’s Method, Vroof was determined according to Step 2, using the estimated wind speed from the Figure 2 map in place of Vmap. Next, a critical blow-off wind speed was determined (Crandell’s Step 3) based on parapet height. Then, using the measured average aggregate diameter, the adjusted critical speed was determined (Vcr’). As a final step (corresponding to Crandell’s Step 5), the quantity (1.1 x Vcr’) – Vroof was calculated (labeled “X-value” for lack of a better designation); the lower the result (the more negative), the more likely aggregate blow-off would occur. DISCUSSION Parapet Height Parapets, partial parapets, or porous parapets dramatically affect the wind velocities and pressure differentials across roof surfaces (Canon et al., 2002; Phillips, 2003). Many of the study roofs had discontinuous parapets or variable parapet heights. Crandell’s Method offers no suggestions regarding mixed parapet heights. Figure 3 – Discontinuous parapet on Roof No. 2. 800.877.6125 www.hydrotechusa.com The key to our Garden Roof® is our Monolithic Membrane 6125®, a seamless rubberized asphalt membrane with a 45+ year track record for critical water-proofing and roofing applications world-wide. From concept to completion 20 • I N T E R FA C E J U LY 2011 For this study, an “effective” parapet height was used in the calculations. An effective parapet height was defined as the greatest parapet height at the apparent origin of the aggregate scour. This provided the best correlation between observations and Crandell’s Method. For example, Roof No. 2 (Figure 3) drained at the back of the building. The parapet height varied from 10 inches in front to 36 inches (due to roof slope) at the rear where the parapet stopped at the draining edge. Aggregate scour originated at the junction of the draining edge and the 36-in parapet; an effective parapet height of 36 inches was used in the calculations for this roof. General Findings X-values (i.e., 1.1 x Vcr’ – Vroof) of the various roofs varied from a low of -52 (aggregate blow-off) to a high of +88 (no scour). Table 1 lists the details of each roof in the study. Of the twenty roofs in – spected, the overall observations are listed in Table 2. All structures in this study were less than 30 ft in height, and all the roofs were SPF with looselaid aggregate coverings (there was no embedment of the aggregate). All structures were Occupancy Category II: neither an essential facility nor a substantial hazard (IBC, 2009: Table 1604.5). No Scour Some of the roofs exhibiting no scour were completely expected. • Roof No. 8 (X-value: +39) had a 42- in parapet wrapping the entire roof structure, providing ample protection to the aggregate. • Roof No. 19 (X-value: +88) had been a ballasted single-ply; when this had been converted to an aggregatecovered SPF roof, the ballast stone was reused. This roof had the greatest aggregate diameter of the study roofs (1.5 inch average). Addi tion – ally, while this roof had a draining edge, the 38-in-high parapet Roof # ID City Avg. Gravel Building Parapet Ht. Exposure Est. Wind Vroof Vcr Vcr’ 1.1 x Vcr’ – Experience Size (in) Height (ft) Eff. (in) Category Speed Vroof (X-value) 1 Treasure Coast Stuart 0.511 20 35 B 96 68.9 121 96 37 Scour Commerce Center 2 Treasure Coast Stuart 0.464 20 36 B 96 68.9 122 95 35 Scour Commerce Center 3 Whiting Construction Palm City 0.508 15 0 B 98 67.2 60 48 -14 Minor Scour 4 B & A Industrial Park Stuart 0.469 21.5 0 B 100 72.6 60 47 -21 Scour Building A 5 B & A Industrial Park Stuart 0.350 16 0 B 100 69.2 60 42 -23 Scour Building H 6 Deggeller Building Stuart 0.496 25 0 B 100 74.4 60 48 -22 Scour 7 Arlington Electric Stuart 0.268 16 0 B 100 69.2 60 39 -27 No Scour 8 Seacoast National Bank Stuart 0.482 30 42 B 99 75.9 133 104 39 No Scour 9 Stuart Brake and Auto Stuart 0.466 15 0 B 101 69.2 60 47 -18 No Scour 10 Don Ramon’s Stuart 0.345 10 0 B 101 64.8 60 42 -19 Minor Scour 11 Wayne’s Auto Repair Stuart 0.442 15 0 B 100 68.5 60 46 -18 Scour 12 Mayfair Plaza Stuart 0.620 25 38 B 101 75.2 126 107 43 No Scour 13 Granada Plaza Stuart 0.245 14 0 C 103 93.2 60 38 -52 Gravel Loss 14 Island Village Jensen Bch 0.462 15 2 B 106 72.6 63 49 -19 Scour 15 Rent-a-Center (et al.) Fort Pierce 0.520 14 0 B 110 74.5 60 48 -21 No Scour Strip Mall 16 Virginia Avenue Plaza Fort Pierce 0.605 15 0 B 110 75.4 60 51 -20 Minor Scour 17 Fort Pierce Business Fort Pierce 0.448 15 15 B 107 73.3 86 66 -1 Scour Park 18 Town Center Port St. Lucie 0.458 15 18 B 104 71.3 91 70 6 No Scour 19 Blockbuster Video Port St. Lucie 1.500 14 38 B 104 70.5 126 144 88 No Scour Table 1 – Roof inspection observations and adjusted critical wind speed (Vcr’). Observation Number X-Value Range of Roofs [1.1 (Vcr’)] – Vroof No Scour 7 -27 to +88 Minor Scour 3 -20 to -14 Scour 9 -22 to +37 Blow-off 1 -52 Table 2 – Comparisons of roof observations. J U LY 2011 I N T E R FA C E • 2 1 wrapped around and protected the corner (see Figure 4). A few of the no-scour roofs were surprises. Roof No. 7 (X-value: -27) would have been expected to exhibit some degree of scour: There were no parapets, and the average gravel diameter was 0.268 in. Why this roof, with such a low X-value, would have performed as well as it did is not understood. The contractor posited that this particular roof has a gravel stop that tends to slow the drainage; during the hurricane(s), the standing water at the roof perimeters may have absorbed the kinetic energy of the wind, protecting the submerged aggregate. Scour and Minor Scour Roofs experiencing some degree of scour ranged in X-values from -23 to +37. Scour patterns were not unexpected. Exposed corners exhibited the classic “heart-shaped” or “V” pattern; corners with partial or discontinuous parapets exhibited a “teardrop” pattern. (See Photos 5 to 7.) Aggregate Blow-off Only one roof in the study was observed or reported to have had a loss of aggregate. This particular building (Roof No. 13) had the smallest aggregate diameter of the study (0.245 in) and was classified as Exposure C (Figure 8). The roof has a high parapet on the west exposure (66 inches), but the eastern side (exposure was an open field) had no parapet. The aggregate loss apparently caused no damage, as it was never determined where it landed (likely in the adjoining field). The X-value for this roof was the most negative of the study at -52. Figure 6 – Typical scour pattern (Roof No. 4). Figure 7 – Typical scour pattern (Roof No. 2). Photo courtesy Google Earth. Figure 5 – Typical minor scour (Roof No. 16). 22 • I N T E R FA C E J U LY 2011 Figure 4 – Roof No. 19 with 1.5-in aggregate and a wrap-around parapet. Aggregate Characteristics Samples of aggregate were removed from each of the study roofs. The samples were sieved to determine the size distribution from which average diameters and the size classifications were determined. Size classifications are shown in Table 3. Average diameter varied from 0.245 to 1.50 in (Photo 9). The largest size was aggregate recycled from a ballasted single-ply roof. The smallest sizes (7, 78, and 8) were locally described as “pea gravel.” Fines were found to be minimal and well within ASTM D448 specification. Aggregate was composed of various mineral species, but the specific gravity was remarkably uniform. Specific gravity averaged 2.59, with a range from 2.40 to 2.72. Variation in specific gravity was insufficient to draw any conclusions as to its effect on scour or blow-off. CONCLUSIONS Crandell’s Method The major intent of this study was to determine the validity of Crandell’s Modified Kind-Wardlaw Design Method for Buildings Figure 8 – Roof No. 13 exhibited aggregate loss. The roof had no parapet on the side facing Exposure C. Aggregate loss was from the northern half of this roof. Photo courtesy Google Earth. ASTM D448 Number Size of Roofs 3 1 5 2 67 11 68 2 7 1 78 1 8 2 Table 3 – Measured aggregate sizes. Enhanced Performance Roof Hatches for Green Building Applications For more information call 800-366-6530 LEED is a registered trademark of the or log on to www.bilco.com U.S. Green Building Council Available in a number of standard sizes Energy effi ciency exceeds standard hatches by more than 48% Meets LEED® standards for recycled content High solar refl ective index (SRI) Ozone-friendly polyisocyanurate insulation EPDM gasketing with enhanced weather-resistance Our Most Energy-Effi cient Model Yet! J U LY 2011 I N T E R FA C E • 2 3 Figure 9 – Aggregate varied in average diameter between less than one-half inch and 1½ inches. Most roofs had Size 67 (nominally one-half-inch diameter). of All Heights. All of the inspected structures in this study were less than 30 ft high; conclusions, therefore, are confined to this height limitation. An X-value calculation was determined to compare the adjusted critical wind speed (Vcr’) to the actual estimated wind speed (Vroof). Per Crandell’s Method, a positive Xvalue would be “safe” from the standpoint of aggregate blow-off. Indeed, this was consistent with the observations. In fact, Crandell’s Method appears to be quite conservative since 12 of the 20 roofs observed had negative X-values but no observed or reported aggregate blow-off. The single roof that did experience blow-off had an X-value of -52. While this might suggest that Crandell’s Method has a “safety factor” of about 50 mph wind speed, this is only one sample, and there were multiple uncertainties in this analysis. Effects of Parapets and Edge Details The effect of parapets on the generation of corner wind vortices is quite profound. It appeared from this study that discontinuous parapets that abruptly stopped at a draining edge had significant effects on vortices. Several roofs exhibited scour patterns that suggested that corner vortices, which might otherwise lead to aggregate blow-off, instead were disrupted with aggregate scouring in toward the field of the roof and away from the edge. Roofs with mixed parapet heights or with discontinuous parapets presented a challenge to model under Crandell’s Method. After trying a number of variations on what parapet height to use, this study found best correlation with Crandell’s Method using an “effective” parapet height as defined as the greatest parapet height at the apparent origin of scour. There may be other subtle yet significant factors affecting the likelihood of aggregate blow-off (and scour), including edge details (Lin, et al., 2008). Roof No. 7 should have experienced aggregate scour to some degree. The theory expressed by the roofing contractor that standing water stabilized the aggregate is speculative though plausible. It’s apparent that the microenvironment of the roof corner needs further study. ACKNOWLEDGEMENTS The author greatly appreciates the support and assistance of NCFI Poly urethanes, Mount Airy, NC; and Whiting Construction, Stuart, FL. Cool Tools INDEPENDENT STANDARD FOR AERIAL ROOF MEASUREMENT REPORTING 866-659-8439 | www.eagleview.com EagleView ® continues to change the industry with Mobile Apps! Stay productive on the go with EagleView Apps for iPhone® and ™. Get roof measurement reports, place and receive orders, EagleView It! © 2011 EagleView Technologies, Inc. check order status and Androidme M cont ! view past orders all from your m easurement rec iPh obile tinues mobile phone. ceive oneSTANDAR 866 659 8 RD MEASU REMENT 24 • I N T E R FA C E J U LY 2011 REFERENCES 2009 International Building Code. Inter – na tional Code Council: Country Club Hills, IL. ASTM D1863, “Standard Specification for Mineral Aggregate Used on Built- Up Roofs,” ASTM International, West Conshohocken, PA. ASTM D448, “Standard Classification for Sizes of Aggregate for Road and Bridge Construction,” ASTM Inter – national, West Conshohocken, PA. Richard P. Canon, Blake S. Joplin, and S. Thomas Watson, “SMARF Build – ing Cape Canaveral Air Force Sta – tion, Florida: Seven Years Later,” Interface, Vol. XX, No. 9, September 2002, pp. 24-36. Jay Crandell and Michael Fischer, “Winds of Change: Resolving Roof Aggregate Blow-Off,” paper presented at the RCI International Con – vention and Trade Show, Orlando, March 25-30, 2010. Jay H. Crandell, “Design of Aggregate- Surfaced Roofs to Avoid Aggregate Blow-off: Development of a Modified Kind-Wardlaw Design Method for Buildings of All Heights,” paper presented at the RICOWI Fall meeting, Portland, Oregon, October 1, 2009. R.J. Kind and R.L. Wardlaw, “Design of Rooftops Against Gravel Blow-off,” National Research Council Canada, Ottawa, 1976. R.J. Kind, M.G. Savage, and R.L. Wardlaw, “Further Model Studies of the Wind Resistance of Two Loose- Laid Roof-Insulation Systems (High- Rise Buildings),” National Research Council Canada, Ottawa, 1984. J.X. Lin, P.R. Montpellier, C.W. Tillman, and W.I. Riker, “Aerodynamic De – vices for Mitigation of Wind Damage Risk,” paper presented at Advances in Wind and Structures (AWAS) conference, Jeju, Korea, May 29-31, 2008. Mary Katherine Phillips, “Evaluation of Mitigation Measures for Reducing High Suction Pressures on Roof Corners,” masters thesis, Clemson University, Clemson, SC, 2003. RCI 2012 DOCUMENT COMPETITION NEW RCI DOLLARS INCENTIVE! Interface LARGE PROJECT | SMALL PROJECT | REPORT nd 3rd - The entry deadline is October 31, 2011. SEE WHAT IT TAKES TO GET INVOLVED! C J U LY 2011 I N T E R FA C E • 2 5 Based on the evaluation in the previous section, the modified K-W design method is presented as follows for aggregate-surfaced BUR and SPF: STEP 1: Determine mapped basic (design) wind speed (mph, gust) for standard conditions (33-ft elevation and flat, open terrain – Exposure C) using the ASCE 7-05 wind map (ASCE 2005). STEP 2: Adjust mapped wind speed (Step 1) to a design wind speed at roof height using the following equation and terrain roughness parameters: Vroof = {[h/hg](1/α)/[33/900](1/9.5)} x Vmap • I x Kd = 1.4 x [h/hg](1/α) x Vmap x I x Kd where, Vroof = gust wind speed at roof height (mph). Vmap = gust wind speed from ASCE 7 wind map (mph). I = building importance factor (use 0.75 for Category I buildings; 1.0 for Category II; and 1.1 for Categories III and IV).* h = building roof height (ft). hg = gradient height for site wind exposure (Exp. B use 1270 ft; Exp. C use 900 ft; Exp. D use 700 ft).** α = power law terrain roughness parameter (Exp B – 6.2; Exp. C – 9.5; Exp. D – 11.5).** Kd = 0.9 = wind speed directionality factor consistent with ASCE 7 wind load directionality factor.*** *Importance factors per ASCE 7-05 have been modified to apply to wind speed in lieu of wind load. **Parameters for Exposure B are based on “typical” values per ASCE 7 commentary. ***For a discussion of the inclusion of the directionality factor, see Kind and Wardlaw, 1976, Appendix A.2.6, page 28. STEP 3: Determine critical (blow-off) wind speed for roof system design as follows: Vcr = 20.8 (H) + 60 where, Vcr = critical wind speed (mph, gust). H = parapet height above roof surface (feet). STEP 4: Adjust critical wind speed (Step 3) for aggregate size when different than 1-in nominal diameter as follows: Vcr´ = Vcr x (d)1/3 where, Vcr´ = aggregate size-adjusted critical wind speed (mph, gust) d = aggregate nominal diameter (inches) (Nominal aggregate diameter is based on mean aggregate size – see examples below.) *ASTM D1863 #7 aggregate has a mean size similar to #67 aggregate, but with a maximum aggregate size of ½ in instead of 3/8 in. **ASTM D448 aggregate is not typically specified for BUR and SPF roof systems. STEP 5: Verify that Vroof ≤ (1.1 x Vcr´). If the above design check is not satisfied, increase aggregate size or parapet height and reevaluate starting at Step 3. (Note: The 1.1 factor is a calibration factor derived and justified in Crandell, 2009.) Roger V. Morrison, RRC, PE, is the president of Deer Ridge Consulting, Inc., Ararat, VA, specializing in issues related to construction uses of spray-applied polyurethane foam. Prior to that, he served as senior staff engineer with NCFI Polyurethanes for 21 years. Roger currently chairs the Technical Committee and the Roofing Committee of the Spray Polyurethane Foam Alliance. Roger may be contacted at roger@deer-ridge-consulting.com. Roger V. Morrison, RRC, PE 26 • I N T E R FA C E J U LY 2011 APPENDIX A MODIFIED K-W DESIGN METHOD (Reproduced with permission from Jay Crandell, ARES Consulting, West River, MD.) ASTM Nominal D1863 Size Diameter (in) #7 3/8* #67 3/8 #6 ½ ASTM D448 Size** #4 1 #24 1½ #2 2
