Which is the Weakest Link – How Can it be Stronger? Lesson Learned From 20 years of Hurricane Investigations Bas. A. Baskaran, PhD, PEng, and Helen Yew National Research Council 1200 Montreal Rd., Ottawa, ON, Canada K1A 0R6 613-990-3616 • bas.baskaran@nrc.ca David L. Roodvoets DLR Consultants 6710 Lakefront Dr., Montague, MI 49437 231-893-1291 • 1dlrconsul@charter.net IIBEC 2020 Virtual International Conve ntion & Trade Show | June 12-14, 2020 Bask aran and Roodvoets | 267 268 | Bask aran and Roodvoets IIBEC 2020 Virtual International Conve ntion & Trade Show | June 12-14, 2020 ABSTRACT As insurance claims increase, wind-induced failure is a major concern for building enclosure designers. To understand the weakest links on the roof assembly, the Roofing Industry Committee on Weather Issues (RICOWI) launched a Wind Investigation Program (WIP) in 1996. WIP missions are as follows: 1. To investigate the field performance of roofing assemblies after major windstorm events 2. To factually describe roof assembly performance and modes of damage 3. To formally report the results for substantiated wind speeds Keys to the RICOWI investigations are teams that are balanced, unbiased, and trained in wind damage assessment. Teams typically are made up of a manufacturer, a roof consultant, and a university or insurance organization representative. Several WIP investigations have been completed over the past 20 years. This presentation focuses on the performance of low-sloped roofs. Among others, three weak links were found critical in the failure of the roofing systems, namely: • Securement of roof edges • Role of rooftop equipment • Integration of roof/wall interface Each of the above weak links is scientifically analyzed, followed by field observation. Correlations are developed for roof wind design. In addition, wind design data from the North American codes of practice are also calculated and compared to show the impact of science and field observation on durable roof design. With these illustrations, this paper offers recommendations to advance the roof system design for hurricane-prone regions. Bas Baskaran is a group leader researching the wind effects on building enclosures through experiments and computer modeling. As an adjunct professor at the University of Ottawa, he also supervises grad students. As a professional engineer, he is a member of RICOWI, IIBEC, ASCE, SPRI, ICBEST, and CIB technical committees. He is a research advisor to various task groups of the National Building Code of Canada (NBCC) and a member of ASCE’s Wind Load Committee. He has authored over 300 research articles and received over 25 awards, including the Canadian Roofing Contractors Association’s Frank Lander Award and ASTM’s Carl Cash Award. Dr. Baskaran was recognized by Her Majesty Queen Elizabeth II with a Diamond Jubilee medal for his contribution to fellow Canadians. David L. Roodvoets is an independent consultant, a past chairman of RICOWI, technical director of SPRI, and he is on the board of directors of the Cool Roof Rating Council. He was previously employed as an associate development scientist for the Dow Chemical Company and technical director for the T. Clear Corporation. Roodvoets has worked with major research institutions and conducted extensive wind tunnel testing of roofing systems. Recently, he has worked on developing fire and wind standards for vegetative roofs, ventilation requirements for attics and cathedral ceilings, and hurricane investigations by RICOWI. He has presented at IIBEC symposia and published in industry magazines. Nonpresenting Author: Helen Yew SPEAKERS INTRODUCTION Natural wind hazards, such as typhoons and hurricanes, have been catastrophic in the past decade, causing significant loss of life and property damage. Figure 1, with data taken from the National Oceanic and Atmospheric Administration (NOAA, 2017), shows the trend of the damage amounts of the most costly hurricanes in the United States. Hurricanes Hugo and Andrew created awareness of roof failures. There were concerns in the building enclosure design community that the root causes of failures and the types of products that failed were distorted (Cook and Soltani, 1993; Smith et al., 1992). Following Hurricane Hugo, two workshops were held to identify and discuss roof wind uplift issues and solutions (ORNL, 1989). These workshops resulted in the establishment of two groups. The first is the Roofing Industry Committee on Wind Issues (RICOWI). The second is the Special Interest Group for Dynamic Evaluation of Roofing Systems (SIGDERS), a North American roofing consortium. RICOWI’S WIND INVESTIGATION PROGRAM RICOWI is made up of 14 sponsors representing the major roofing associations and 83 affiliates representing general interested parties (visit www.ricowi.com for current membership details). Because there was a great deal of controversy about roof performance after Hurricane Andrew, the leadership of RICOWI decided that they needed a greater understanding of windrelated roof system performance. A major training program was carried out at Oak Ridge National Laboratory in 1996, where experts explained the fundamentals of wind loads on roofing systems and trained over 100 interested investigators. RICOWI set up the Wind Investigation Program (WIP) with plans, funding, and development of a database (Baskaran et al., 1997). WIP has the following objectives: 1. Investigate the field performance of roofing assemblies after major windstorm events 2. Factually describe roof assembly performance and modes of damage 3. Formally report the results for substantiated wind speeds OBSERVATIONS FROM THE RICOWI WIP DATABASE There were no major hurricanes making landfall in a largely populated area until 2004. Starting with the inspections after Hurricanes Charley and Ivan in 2004, RICOWI has sent out balanced teams to inspect roofs after a total of six major hurricanes: Charley, Ivan, Katrina, Ike, Irma, and Michael. Teams investigated both low-sloped and steep-sloped roofs. As the only large-scale roof investigation program of its kind, the total cash and inkind budget was approximately $2 million for the RICOWI investigations. RICOWI teams collected building-specific information from numerous locations, including type of structure (use or occupancy), wall construction, roof type, roof slope, building dimensions, roof deck, insulation, construction, and method of roof attachment. All inspections were documented using standard forms and photographed. At the end of each investigation day, reports were completed and provided to an administrator. A feedback session also occurred so that teams could follow up on interesting leads. Teams typically worked from the highest to lowest wind-damaged areas so there was no attempt to get randomized data. All of these data have been compiled into reports (RICOWI 2006, 2007, 2009, 2017, and 2019), which are available for free download at the RICOWI website. By going through the WIP reports, prevalent trends in roof failure can be identi- IIBEC 2020 Virtual International Conve ntion & Trade Show | June 12-14, 2020 Bask aran and Roodvoets | 269 Which Is the Weakest Link – How Can It Be Stronger? Lessons Learned From 20 Years of Hurricane Investigations Figure 1 – Historical data for the estimated loss due to hurricanes (source: NOAA). fied. Damage investigation showed that nearly 50% of the failures on low-slope roofs were initiated in a corner of the roof. Damage, however, may occur from other causes, with edge metal damage in 58% of the roofs inspected. Some roofs were damaged from more than one cause. Wind-borne debris—usually from failed air-handling units—was responsible for 27% of the damage recorded in the database. Fasteners often failed because they were weakened by corrosion, but in a few cases, fasteners backed out due to membrane flutter. Over the course of these hurricane aftermath investigations, the teams have inspected more than 225 individual roofs with a total low-slope roof area of over 8 million sq. ft., as well as steep-slope roofs and metal roofs. From these observations it can be concluded that even with high hurricane wind speeds, the damage consisted mostly of partial failures of the building enclosure components rather than complete system failures. This is an encouraging observation for the engineering community, since it demonstrates that prevention of these failures is possible by applying sound engineering principles and design methodologies while controlling the quality of the workmanship in the field. PRESENT CONTRIBUTION This paper focuses on the investigations of low-sloped roofs. The objective of the paper is not to present several photographs and information related to specific roof configurations. Rather, efforts were made to scrutinize all these photographs and field observations towards developing a relationship with the existing science in the wind and roofing fields. A segment of the findings was already presented in previous peer-reviewed publications (Baskaran et al., 2007; and Baskaran et al., 2010), and this paper amalgamates the current state of the art with updated code provisions. 270 | BaskSKaran and Roodvoets IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 Figure 2 – Wind aerodynamics on roof edges. Figure 3 – Simplification of forces acting on roof edge metal. ROOF EDGE WIND AERODYNAMICS AND FAILURE INVESTIGATIONS Wind flow around buildings creates pressure and suction on the building envelope (Figure 2). As the building obstructs the flow movement, pressures are induced on the windward side, and as the flow separates from the side walls, suctions are created. For the roof assembly, most suctions are induced due to flow separation from the edges. However, depending on the building aspect ratio (e.g., for a slender building in which the length of the roof is much longer than the width), the flow can reattach the roof along its length and, thus, can induce positive or negative pressure on the roof surface. Moreover, depending on the rooftop obstructions, the wind flow gets modified and develops complexities in the flow aerodynamics. As shown in Figure 3, these complex wind aerodynamics can be simplified into three force segments (F) as follows: 1. F1 is the outward force acting on the edge system due to the positive pressure. Basically, this force can unlatch the coping or edge flashing from the cleat and lead to edge failure. 2. F2 is the uplift force acting on the whole edge system due to the strong flow separation. This can cause complete failure of the edge system by initiating the failure on the weakest link. 3. F3 is the force caused on the edge system due to the membrane behavior. This can be the dominant failure factor in the case of mechanically attached systems where the membrane billows (flutters) between the mechanical attachments. This can also affect loose-laid systems where the membrane can be exposed to wind due to displacement of ballast. Figure 3 shows that these three forces are simply acting as perpendicular components to the surface. However, these three forces are not the only vectors; all of the force vectors can vary in time, as well as along the surface of the edge system. Figure 4 shows the typical causes of roof edge metal failures, starting from the nailer attachment—the nailer is not attached to the deck (Figure 4a)—to inadequate cleat engagements (Figure 4d). In all these cases, the roof membrane is not damaged, and the weakest link is the edge system. This demonstrates that these types of failures are not dependent on the membrane type. IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 BaskSKaran and Roodvoets | 271 Figure 4 – Typical causes of roof edge metal failures. STATE OF THE ART OF THE ROOF EDGE METAL DESIGN Perimeters and corners of low-sloped roofs have been recognized as the most vulnerable areas of the roof to wind damage (Smith, 1990). Failure continues to occur at these locations of buildings because both design and installation practices can be inadequate. Review of the RICOWI investigation database shows that one in four roof failures was due to the failure of the edge. This is likely because prior to 2004 there were no code requirements for roof edge attachment. Neither the NBCC nor the widely used American Wind Standard ASCE 7 specifies wind load requirements for the roof edges. Recently, this has been corrected in the NBCC and ASCE 7. Figure 5 compares the new ASCE 7 provisions with ANSI/SPRI/FM 4435 ES-1, Test Standard for Edge Systems Used with Low Slope Roofing Systems (ANSI/SPRI, 2017). The test standard was introduced as a code requirement in 2003 to the International Building Code (IBC) for resistance evaluation. Presentations (LeClare, 2010) have shown that with a minimal increase in the installation cost, major improvements in the resistance of the roof edges can be achieved. As shown in Figure 5, the horizontal and vertical suctions used by ES-1 are lower than the current ASCE 7 provisions. ES-1 also omits positive pressure requirements for the edge metal evaluations. As shown in Figures 3 and 4, this is critical in order to test the adequacy of the cleat engagement with the coping. Note that ASCE 7-16 provides provisions for only horizontal loads (P2 and P4), whereas provisions for the vertical load (P8) will be incorporated into ASCE 7-2022. WHAT IS MISSING IN THE CURRENT EDGE METAL TEST METHOD? The negative forces at the perimeter must be resisted by adequate mechanical attachment and/or bonding of the roofing membrane to the substrate. Many designs allow pressurization of the underside of the roofing system, which significantly adds to the loads that must be resisted. The characteristics of the load to be resisted are dynamic, and most tests used to evaluate roofing systems are static or quasi-static. Current test methods, such as ANSI/SPRI/FM 4435 ES -1, only focus on the vertical force and/or outward force in evaluating the mechanical attachment. None of the existing test methods simulate the membrane peel forces. In current testing, the first mechanical failure (screw withdrawal) or separation of the membrane stops the test. In nature, roofs survive with small amounts of initial failure if the peel forces are resisted. If the peel forces are resisted, catastrophic damage 272 | BaskSKaran and Roodvoets IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 The negative forces at the perimeter must be resisted by adequate mechanical attachment and/or bonding of the roofing membrane to the substrate. Figure 5 – Comparison of the ASCE 7 vs. ES-1 load provisions. is less likely. All the existing wind uplift standards (CSA A123-21-19, ETAG 2016, FM 4474, NBI 160, and UL 580) are applicable only for the roof assembly; thus, it is clear that there are major limitations in the current testing and design requirements: 1. There are no requirements for peel loads at the roof edge. 2. There is no consensus-based test method that can simulate simultaneously all the wind-induced peel and uplift forces on roof edges. 3. Current edge metal testing is not dynamic; considering the flexible nature of many coping and edge flashing designs, the influence of dynamic loading is not reflected in current testing. Implementation of the ES-1 specifications improved the edge metal resistance; however, there has been an increase in nail failure, which is the weakest link (Smith, 2015). RESILIENCE DESIGN WITH HIGH PARAPET Figure 6 shows the beneficial effect of parapets on the wind uplift resistance of a roof assembly. The investigated building was a six-story condominium building with stucco wall construction with a poured-concrete roof deck. A granule-surfaced BUR system was over-mopped in place with perlite insulation. The roof was approximately 37,000 ft.2 (3400 m2). The parapets were designed well, and the field measurements indicated they were about 5 ft. (1.5 m) high. The parapet corners were also well designed and capped with a single piece of metal coping. Wind-related damage was limited to approximately 5% of the wall/parapet cap and the underside of soffit. No damages were observed to the roof. As shown in the photographs in Figure 6, the building is relatively new and exposed to the waterfront. This exposure condition is considered to be the most severe exposure IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 BaskSKaran and Roodvoets | 273 Figure 6 – Beneficial effect of parapet on the wind uplift resistance of a roof system. condition. Due to high parapet configuration, the wind suctions were reduced at the perimeters and, therefore, no damage was caused to the roof assembly; this will be further discussed in Figure 7. Also, there were many individual HVAC units that were mounted on stands, and all of these were in excellent condition. No movement was apparent for any of the HVAC units. However, the surrounding buildings in the same neighborhood—which had no parapets—experienced severe damage. One 274 | BaskSKaran and Roodvoets IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 Figure 7 – Benefit of high parapet in reducing the wind uplift load. Figure 8 – ASCE 7 and NBCC reduction allowance for high parapet. such building with aggregate blow-off from the roof is also shown in Figure 6. The effect of parapets on pressure coefficients differs from one region of a flat roof to another—similar to the different parapet configurations on the roof edges (Kind et al., 1987). Figure 7a tabulates the data from a wind tunnel, demonstrating: High parapets (over 3 ft.) can mitigate the high corner suctions. Reduction depends on the type of parapet configuration. For roofs with solid perimeter parapets, it can be as high as 50%. For variations in parapet configuration, further reductions were measured. Overall, the parapets are beneficial in reducing the suctions, irrespective of their configurations (Baskaran et al., 1988). Figure 7b shows a roof design using a guardrail and flat panels to reduce the peak wind uplift in the corner of the roof on the 400-ft.-tall Cape Canaveral Vehicle Assembly Building. The design led to a reduction from 400 lbs. per sq. ft. to 40 lbs. per sq. ft. This allowed the construction of a safe and stable roofing system. Wind tunnel studies showed that the system with multiple panels works by inducing a multiple vortex pattern, creating interference and negating some of the load. However, the authorities having jurisdiction (AHJ) approval is needed for load reductions that differ from the ASCE 7 provisions. Acknowledging that corners of roofs have the greatest uplift and, for many reasons, are the most vulnerable, it is prudent to install parapets. The IBC requires 30-in. parapets in Section 705.11 (there are many exceptions in the code). If there is regular IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 BaskSKaran and Roodvoets | 275 Figure 10 – Peel-force propagation on the failure of roof assembly. Figure 9 – Failure of roof edge due to air movements at the roof/wall interface. roof traffic, the Occupational Safety and Health Administration (OSHA) requires guardrails of 42 in. ±6 in. Factory Mutual Global Property Loss Prevention Data Sheet 1-28 and other FM documents reduce the uplift requirement when 36-in. parapets are installed. Parapets and designed guardrails reduce corner uplift loading. Parapets contribute to roof fire safety (ICC-IBC Section 901.7, 2018). They add pedestrian safety to the roof. Both ASCE 7 and the NBCC (Figure 8) allow roof wind lift load reductions for roofs with high parapets. RESILIENCE DESIGN WITH PEEL STOP Figure 9a demonstrates how the positive internal pressure can lead to roof failure. Figure 9b shows a loose-laid EPDM roof assembly that had a parapet of 4.5 ft. (1.4 m) high with a well-designed and installed edge system. However, at the wall/roof junction, there was a gap of more than 0.5 in. (127 mm) along the length of the roof (220 ft. [67 m]). During the field investigation, it was noticed that strong air currents were emerging along that gap. This internal positive pressure lifted the loose-laid membrane along the edge and displaced the aggregate ballast. The exposed membrane experienced tension when it was subjected to wind uplift. Since the membrane tear resistance was lower than the strength of the edge metal, the membrane ruptured, and led to the roof failure. Figure 9c shows failure of a roof as a typical installation issue at the roof/wall junction. Poor bonding of the membrane at the parapet vertical surface caused delamination during the wind event, which induced fatigue resulting in the seam failure, causing water ingress to the roof assembly. In both case studies, the roof edge was properly designed and installed. In examining several such failures, it has been observed that the intensity of the internal pressure build-up is time-dependent. As shown in Figure 10, the extension of the roof failure is directly related to the pressure equalization time and intensity. This can clearly be seen from the figure: The membrane peel was stopped after a certain distance from the roof edge. What is also interesting is the role of the roof divider acting as a peel-stopper. This is evident from the fact that the extent of the peel is much longer along the length in comparison with the width of the roof. This is mainly due to the divider, which was running across the roof width, preventing additional peeling. Two approaches to mitigate peeling are shown in Figure 11. Controlling the air movements at the roof/wall interfaces (Figure 11a), a “peel-stop” bar should be placed over the roof membrane near the edge of the flashing/coping to provide secondary protection against the membrane lifting and peeling, as shown in Figure 11b (ASCE, 2019; and FEMA P-424, 2010). The bar has to be anchored to the parapet or deck. The spacing is recommended between 4 to 12 in. on center. A few inches of space should be left between each bar, and the bar should be stripped over with a stripping ply. LESSONS LEARNED FROM FIELD INVESTIGATION BY THE PRESENT STUDY 1. Design load should be calculated based on the ASCE 7/NBCC for parapet component and edge metal design. 2. Edge system resistance should be evaluated based on ANSI/SPRI/FM 4435 ES-1 to demonstrate: system resistance > load requirement. 3. Roof edges should be designed and installed as a system rather than as separate components. 4. Proper design and installation of high parapets, air movement control at roof/wall interface, and peel stop can increase resilience of the roof systems. ROOFTOP UNIT (RTU) FAILURE AND ITS IMPACT ON THE ROOFING SYSTEM In addition to the waterproofing function, rooftops are used as platforms for air handling, HVAC, and other equipment. With increased emphasis on energy efficiency, roofs produce solar energy and moderate the greenhouse gas levels with vegetation. Our investigations exclude the study of vegetated or solar roofs. Several roofs that were properly designed and constructed performed well. However, failure of the RTU is one of the major causes of a roof to lose watertightness of the building. Investigations grouped these failures into four categories: 1. Blow off of the complete RTU due to wind uplift 2. Blow off of only RTU coverings/parts due to wind forces 3. Sliding of the RTU when wind forces exceed the self-weight 4. Combinations of the above three. Complete Blow-off of RTU Figure 12 shows the complete failure of a mechanical unit and its impact on the 276 | BaskSKaran and Roodvoets IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 Figure 11 – Air movement control and peel-stop design to improve resilience of roofs (Figure 11b from FEMA P-424). IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 BaskSKaran and Roodvoets | 277 Figure 12 – Complete blow off of an RTU and its impact on a roofing system. roof assembly. The retail building, with aggregate-surfaced BUR, is at elevation of 35 ft. (10.7m) above grade. There was a penthouse on the roof, and it was 15 ft. (4.6 m) above the main roof level. The penthouse had a mechanical unit, which was held in place with metal L angles and attached using nails to the nailers. Failure of the nail attachment caused the blow-off of the mechanical unit, which caused major punctures on the penthouse roof. Subsequent wind flow dynamics caused the mechanical unit to come apart, which affected the main roof by puncturing the membrane at several locations. The debris path is evident from Figure 12, as well as the propagation of membrane puncture failures until the debris blew off the roof. RTU Cover Blow off This two-level roof had loose-laid EPDM on the lower level and mechanically attached PVC roof on the upper level (Figure 13). Both roof systems performed well during the wind event. The bases of the mechanical units and encircled paver were still in place; however, the sheet metal unit enclosures (cabinets) were missing (Figure 13). The plywood used to temporarily repair the units is visible in the figure. The metal debris from the units caused several punctures to the single-ply membrane where some of them were temporarily patched and some of them were unnoticed. Similar to the previous case study, the trail of punctures was observed across the roof surface, ending at the roof perimeter. These are classic examples of the RTU covers/claddings becoming the weakest link for wind forces, and failure of that resistance link in a roof assembly leads to roofing failure. Shearing of RTU Figure 14 shows failure of several mechanical units. The mode of failure is missing attachment between the unit and supporting frame. Wind forces exceeded the self-weight of the unit. Toppling of these units caused membrane localized punctures, leading to loss of water-tightness of the roof. Combined Failures Similar to roof edge failures, rooftop equipment failures are not dependent on the membrane attachment method nor the 278 | BaskSKaran and Roodvoets IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 Similar to roof edge failures, rooftop equipment failures are not dependent on the membrane attachment method nor the type of membrane used. Figure 13 – RTU covering blow off and its impact on a roofing system. type of membrane used. Authors also reported success stories of the surviving rooftop equipment in the previous publication (Baskaran et al., 2007). Figure 15 gives an example for the RTU uplift and shear loads for various wind speeds based on the ASCE 7 provisions. Both of these loads are higher compared to the roof uplift load. The structural attachment of the RTU to the roof deck and the coverings need to be designed for both shear and uplift loads in order to minimize failures, as demonstrated in Figure 14. LESSONS LEARNED FROM FIELD INVESTIGATION BY THE PRESENT STUDY 1. Roof design should include integrating RTU with the structural element of the building, and the design loads should be calculated based on the ASCE 7/NBCC. 2. Equal design care should be taken for the RTU covering to resist the calculated wind loads. 3. RTU should be evaluated for wind uplift resistance. CONCLUDING REMARKS This paper presented an over- all view of RICOWI’s WIP and its findings. As one of its kind, this program investigated and documented roofing system performance. Reports containing fac- tual data are available for free download from the RICOWI website. This paper amalgamated the findings from the last two de-cades into those highlighted in the “lessons learned” sections of this paper that contributed significantly to the failure of the roofing systems. ACKNOWLEDGEMENT SIGDERS was formed from a group of partners who were interested in roofing design. These partners included: Atlas Roofing Corporation, Canadian Roofing Contractors’ Association, Dupont Performance Building Solutions, Duro-Last® Roofing Inc., EXP Inc., Firestone Building Products Company, IKO Industries Ltd., International Institute of Building Enclosure Consultants, Johns Manville Inc., OMG Roofing Products, Roofing Contractors Association of British Columbia, Rockwool Group, Sika Sarnafil Inc., Soprema Canada Inc., Carlisle SynTec, Tremco Inc., and Trufast Corporation. IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 BaskSKaran and Roodvoets | 279 Figure 14 – Mixed failure mode of several RTUs and loss of watertightness. REFERENCES ANSI/SPRI/FM 4435 ES-01. “Test Standard for Edge System Used with Low Slope Roofing Systems.” American National Standards Institute and Single Ply Roofing Institute, 200 Reservoir St., Suite 309A, Needham, MA 02194. 2017. ASCE 7-2016, Minimum Design Loads and Associated Criteria for Buildings and Other Structures. 2016. American Society of Civil Engineers. https://www.asce.org/structural-engineering/asce-7-and-sei-standards/ ASCE “Pre-standard for Performance-Based Wind Design.” American Society of Civil Engineers. 2019. A. Baskaran and T. Stathopoulos. 1988. “Roof Corner Wind Loads and Parapet Configurations.” Journal of Wind Engineering and Industrial Aerodynamics. V 29, pp. 79-88. A. Baskaran, A. Desjarlais, D. Roodvoets, and P. Wood-Shields. “Strategic Plan for the Wind Event Investigation Program.” 8th U.S. Wind Engineering Conference, Baltimore, MD. pp. 1-5. June 1997. A. Baskaran, S. Molleti, and D. Roodvoets. “Understanding Low-sloped Roofs Under Hurricane Charley from Field to Practice.” Journal of ASTM International. 4 (10), Nov/Dec., pp. 1-13. November 01, 2007. A. Baskaran and D. Roodvoets. “Lesson Learned from Hurricane Wind Investigations of Low-Sloped Roof Assemblies—A Researcher Perspective.” NRCA. 2010. R.A. Cook and M. Soltani. “Hurricanes of 1992—Lessons Learned.” Proceedings of the Symposium by American Society of Civil Engineers. Dec 1-3, 1993. CSA A123.21-19, Standard Test Method for the Dynamic Wind Uplift Resistance of Membrane-Roofing Systems. Canadian Standards Association. 2019. ETAG 2016, Guideline for European Technical Approval of Mechanically Fastened Flexible Roof Waterproofing Membranes.” EOTA, Kunstlaan 40, Avenue des Arts, B-1040, Brussels, Belgium. Factory Mutual Research, Approval Standard: Class I Roof Covers (4474). 270 Central Avenue, P.O. Box 7500, Johnston, RI 02919-4923. FEMA P-424, Design Guide for Improving School Safety in Earthquakes, Floods, and High Winds. Federal Emergency Management Agency (FEMA). Risk Management Series. December 2010. International Building Code. International Code Council. Falls Church, VA. International Code Council. 2000 and 2018. R.J. Kind, M.G. Savage, and R.L. Wardlaw. “Further Wind Tunnel Tests of Loose-Laid Roofing Systems.” National Research Council of Canada, Report LTR-LA-294. April 1987. R. LeClare. “Metal Edge Systems for Low-Slope Roofs.” Presentation at the RICOWI Fall Seminar. November 2010. Available at www.ricowi.com. “The Deadliest, Costliest, and Most Intense United Sates Tropical Cyclones (and Other Frequently Requested Hurricane Facts.” National Oceanic and Atmospheric Administration (NOAA). Technical memorandum, NWS TPC-5. 2017. NBI 160, Roof Coverings–Dynamic Wind Load resistance.” Norwegian Building Research Institute. Finland. 1990. NRC 2015, National Building Code of Canada, Part 4. National Research Council of Canada. Ottawa, ON, Canada. 2015. ORNL. Proceedings of the Roof Wind Uplift Testing Workshop. Oak Ridge, TN. November 8-9, 1989. Roofing Industry Committee on Weather Issues. “Charley – Ivan – Katrina – Ike – Irma and Michael, Wind Investigation Reports.” www.ricowi.com. T.L. Smith. “Hurricane Hugo’s Effects on Edge Flashings.” International Journal of Roofing Technology. pp. 65-70. 1990. T.L. Smith. “Facing the Wind.” Professional Roofing. March 2015. T.L. Smith, R.J. Kind, and J.R. McDonald. “Hurricane Hugo: Evaluation of Wind Performance and Wind Design Guidelines for Aggregate Ballasted Single-Ply Membrane Roof Systems.” Proceedings of the VIII International Roofing and Waterproofing Congress. p. 598. 1992. UL 580, Standard for Tests for Uplift Resistance of Roof Assemblies. 5th Edition. Underwriters Laboratories Inc. 2600 N.W. Lake Rd., Camas, WA. 2006. 280 | BaskSKaran and Roodvoets IIBEC 2020 Virtual International ConveVEntion & Trade Show | June 12-14, 2020 Figure 15 – ASCE provisions for the attachment of the RTU and its coverings.
