John Wells, RRO Wells Klein Consulting Victoria, BC, Canada ABSTRACT This paper looks at the investigative process of the failure of a single-ply roof on a new high-end, high-rise condominium building. The failure resulted in major damage claims for the insurance company and a replacement roof on a building less than six months old. The initial failure appeared to be wind-related, but a more complete investigation proved that components of the roof had actually failed prior to the wind events that drew attention to and emphasized the failure. The investigation pointed out fundamental problems with the roof design and the building envelope design, all of which were exacerbated by poor workmanship by the roofing contractor and per¬ haps pressures from the developer to “value engineer” the building process. SPEAKER John Wells is the senior partner and president of Wells Klein Consulting Group Inc., a British Columbia, Canada-based company that specializes in providing solutions to roofing, waterproofing, and related building envelope problems for commercial and institutional building owners throughout British Columbia and Western Canada. Prior to starting his consulting practice in early 1992, John gained international recognition for his work as technical manager of the Roofing Contractors Association of B.C., a position he held until the fall of 1991. John has authored technical manu¬ als, published technical columns and articles in trade magazines, and has been a guest lecturer and speaker at universities, technical societies, and national conven¬ tions. Wells was the founding president of the British Columbia Building Envelope Council and was an elected director for ten years. John joined RCI in 1997 and earned the RRO certification in 1998. Contact Information: Phone – 250-658-3143; E-mail -jwells@wellsconsult.com Wells – 184 Proceedings of the RCI 23rd International Convention
INTRODUCTION In this paper, we will review the investigative process of the premature failure of a new single¬ ply roof on a high-end, high-rise condominium building in Victoria, BC, Canada. Once again we learned that “things ain’t always what they seem to be!” The roof failure resulted in interior water damage claims well in excess of $150,000 to an insur¬ ance company and the necessity of a replacement roof (projected cost in excess of $100,000) on a high-end condominium building that was less than six months old. We will lead the reader through the investigative process¬ es that led to our conclusions on this failure as we review informa¬ tion and gather clues leading to a perhaps startling conclusion. It was not the conclusion that many wanted to hear. Wells Klein Consulting Group’s main work is in the investigation, remediation, and replacement of roofs, due diligence inspections, roof condition assessments, roof¬ top quality control, maintenance planning, and inspections, etc. Forensic investigation projects don’t come along very often, but are a favorite of our business. Our practice is mostly com¬ mercial, institutional, and indus¬ trial buildings with only a smat¬ tering of residential condominium work and virtually no single-fami¬ ly construction. We operate pri¬ marily in southwestern British Columbia in the Vancouver area, and on Vancouver Island. WKGCI and its predecessor company, J.W. Wells Consulting Inc., has been in practice for 16 years. Figure 1 – The roof had failed after only six months in service. British Columbia is a very so¬ phisticated roofing market, given the influence of the Roofing Con¬ tractors Association of BC. The climate varies from coastal to se¬ vere. Lots of rain and wind are common in winter months on the South Coast area, where we oper¬ ate. Unlike other parts of Canada or even the interior of BC, winters in southwestern coastal BC are pretty much snow-free. For those unfamiliar with our geography, think Seattle, because we are only 80 miles (130 km) northwest of the Emerald City. For the purposes of this paper, we are focusing on Victoria, BC, the capital city of British Colum¬ bia. Victoria is located on the southern tip of Vancouver Island, the largest of British Columbia’s 6,500 islands. It is a community often chosen for its temperate cli¬ mate, natural beauty, recreational sites, and superior economic opportunities. According to the 2006 census, about 380,000 peo¬ ple make their home in the greater Victoria area. Located in a sub-Mediterran¬ ean zone, Victorians enjoy some of the most moderate weather in all of Canada. Victoria boasts an average of 2,183 hours of sun¬ shine yearly, and an eight-month, frost-free season. Average annual rainfall is only 26.2 in (65.6 cm), compared to some 48 in (121.9 cm) for Vancouver and 37 in (94 cm) for Seattle. Victoria, however, can be a windy place. The real stormy sea¬ son of late October through January usually produces a num- Proceedimjs of the RCI 23rd International Convention Wells – 185 ber of wind events with storms often exceeding 50 mph (80 kph). Wind gusts of up to 65 mph (100 kph) are not unusual, and occa¬ sional gusts to 75 mph (120 kph) or more are not uncommon. Storm wind direction is southeast to west. THE FACTS MA’AM, JUST THE FACTS The subject building is a brand new, 11-story, high-end residential condominium, of rein¬ forced concrete structure. The penthouse portion has a curved, cast-in-place, concrete roof deck shaped similarly to an airfoil or aircraft wing. We are advised this was an architectural feature rather than an attempt to provide a spoiler effect to keep the build¬ ing firmly on the ground! There are two main roof areas. Only the failed roof area on the airfoil is addressed in any detail in this paper. The roof plan of the building is shown in Figure 2. The drain loca¬ tions are highlighted. A section view of the upper roof portion fol¬ lows in Figure 3. THE PRELIMINARY REVIEW AND CLUES Figure 2 – Drains shown in white circles or arrows. We received the first call on Figure 3 – Section of the upper roof. the project on January 9, 2006, from an insurance company adjuster in Victoria. We were told briefly that a roof had failed (“blown off’ was the term) due to wind and were we interested and qualified to perform an investiga¬ tion and provide a report. The adjuster said he had been told the roof membrane was single-ply TPO (thermoplastic polyolefin) and asked if we knew anything about that material. We advised in the affirmative and our bona fides were supplied and a written instruction to proceed was quick¬ ly issued. We were advised in the initial phone discussions that the roof had partially “blown off “ the building a few days previously, that water ingress and damage had occurred, and that a contrac¬ tor had supplied and installed a number of concrete pavers as bal¬ last to hold the loose roof mem¬ brane in place. Following are some excerpted sections from our preliminary report (italicized within quotation marks). Our first impression in driving up to the property was that it was a very exposed location to all pre¬ vailing storm winds. It is close to the Victoria “Inner Harbor” area, a prime location. We observed a very unique feature on the roof. The top of the penthouse looks like an airplane wing, an airfoil, in fact. This feature, with its atten¬ dant possibilities, was not lost on the writer, a former pilot. “We attended this property on January 10, 2006, in order to perform a prelimi¬ nary assessment on the condition of the roofs on this building. You advised that subsequent to recent windstorms, portions of Wells – 186 Proceedings of the RCI 23rd International Convention the roof, specifically the curved or “airfoil” section, had suffered damage. ” “Weather conditions during our preliminary review were showers, breezy, with a temperature of 9°C (48°F). Steady rain settled in as we left the site at about 3:00 p.m. “Prior to accessing the roof, we viewed a partial set of reduced architectural draw¬ ings. Project specifications and other information such as roofing warranties were not available. The property manager also advised that the roof problems were first noticed or reported on or about Jan. 1, 2006. Also prior to accessing the roof, we were shown some of the suites where water damage had taken place. The points of water entry into the building were drain holes in the concrete (the roof deck) slab above the suites. We were advised that the “drains had been pulled out of the deck pene¬ trations and allowed water in¬ gress.” We observed interior dam¬ age was severe and that drying and repairs were still underway. The interior damage was extensive, indicating a significant volume of water had entered the building. All the flooring, hard¬ wood, and carpet for three floors was being removed for replace¬ ment, and the lower portions of walls (both wood and drywall) were wet. The suites are valued from $700,000 to over a million dollars. We were able to access loca¬ tions underneath the deck and took photographs of the partially repaired drains. It had been re¬ ported that the drains had been pulled out of the deck, tearing the membrane. Figure 4 shows one of the drains on the east side of the deck. Figure 4 – One of the drains on the east side of the deck. One of the things we observed was the difference between the size of the drain line and the size of the chase. We also observed that other chases or holes in the concrete for penetrations were all significantly larger than the pene¬ tration would require. Our preliminary look at the drawings showed the lower or main roof was a 2-ply, SBS, modified- bitumen assembly over poly¬ isocyanurate insulation. The air¬ foil roof was originally specified as a TPO membrane, fully adhered to a treated gypsum-board overlay, adhered with a well known poly¬ urethane adhesive to 4 in (200 mm) board-stock polyisocyanu¬ rate insulation adhered to a con¬ crete deck with the same well known polyurethane adhesive. We were advised that the TPO assembly had been installed by a roofer subcontracted to the prime roofing contractor. We then accessed the airfoil roof and were quite frankly (par¬ don the expression) “blown away” by what we saw. Photos 5 and 6 may help to explain why. We counted well over 150 pavers, at 160 kilos (100 lbs) a piece, that had been brought up to hold the roof membrane in place! Back, briefly, to our report: “The complete assembly is no longer “fully adhered” but is now completely loose and only held on the building by concrete pav¬ ers (installed as emer¬ gency ballast) and a perimeter fixation bar from the original installation. We were advised (by the property manager) that the pavers move during wind events and this may repre¬ sent an extreme safety hazard if a paver fell from the roof level. “Subsequent to our site visit, another windstorm occurred on January 12, and we were advised again by Mr. Ennis that the pavers were moving and becoming displaced by membrane movement in the wind.” Proceedings oj the RCI 23rd International Convention Wells – 187 Figures 5 and 6 – Over 150 pavers of 100 pounds each held the roof membrane in place. In fact, we observed that es¬ sentially none of the membrane was adhered. “We observe that the insu¬ lation has moved around and p’iled up’ under the membrane. “Water entry into the suites took place at the locations of cast drain penetrations through the concrete deck. The cast drains were likely ‘pulled out’ of these locations by the membrane billowing and luffing in severe wind conditions. “We advise it seems unlikely that the drains were ever mechanically fastened to the deck, but more likely, the cast iron drain flanges were just placed on top of the insul¬ ation without adequate mechanical fixation. No clamping rings were evi¬ dent. Downpipes were at¬ tached with MJ-type cou¬ plings, which apparently failed when the drains were pulled and moved upward and water entry took place. ” Our Preliminary Opinions “The roof membrane, how¬ ever, [was] intact at the time of our preliminary review (due to the tensile strength of the membrane) and water entry was through the concrete deck subsequent to the failure of the membrane at the drain connection. “The roof membrane is in danger of blowing off the building and is held in place by the concrete pavers. These pavers may represent an extreme health and safety hazard, should one or more fall from the roof in high wind conditions. The pavers also represent a danger to the membrane proper from abrasion or physical pene¬ tration. We observed some broken pavers with sharp, pointed corners on the roof. “If the membrane is pene¬ trated, further severe water ingress may be expected. “The exact failure mode cannot be established without further investiga¬ tion and calculation. The membrane must be cut open and the membrane, insulation, and roof deck examined for clues. “We suspect there were deficiencies, both in the application of the insula¬ tion to the deck, and of the membrane to the insula¬ tion. Viewing the compo¬ nents after opening the system would confirm (or deny) our suspicions. ” In fact, the only things that had held this roof on the building (prior to the paver installation) were the perimeter fixation bars. The whole membrane would flap or flutter in the breezes, and that was the force that pulled the improperly attached roof drains in place. Wind Conditions As previously mentioned, windy conditions are not uncom¬ mon in Victoria, especially during winter. I note, however, that this winter had not been particularly windy up to the point of our pre¬ liminary investigation. Winds of 40 to 50 kmh (25 to 31 mph) are commonplace year round. My idea of a “wind event” is only when the wind exceeds at least 70 kmh Wells – 188 Proceedings of the R CI 23rd International Convention (44 mph) . A wind chart of Canada is pretty self explanatory, with Victoria highlighted with a red star in the 100 mph (160 kph) coastal zone. “A review of recent wind conditions in Victoria Har¬ bor since December 1, 2005, is attached. I have only included the days when gusts exceeded 35 kmh. We note that wind conditions at the building site are likely exacerbated by shape, height, and exposure. “It would appear from the property manager’s re¬ marks that the major prob¬ lems took place in a wind event on December 29. Harbor gusts that day were recorded as high as 50 kmh (30 mph). Another wind event with one gust at 48 kmh was recorded on January 1, 2006. We noted a more prolonged wind event on January 7, 2006, and another, more severe event on January 12. Records are attached”. 1 None of these events is out of the ordinary, and would normally be no cause for problem or alarm. “Our initial or preliminary opinion of the reasons for the failure of this roof sys¬ tem includes poor or im¬ proper workmanship of the membrane assembly (primarily the insulation and membrane adhe¬ sive), improper installa¬ tion of the roof drains, coupled with the possibil¬ ity of under or inadequate roof-system design for the conditions. “Further and much exten¬ sive investigation is re¬ quired to confirm our opinions. ” And there it was for the first “go round” – some suspicions and opinions based on initial observa¬ tions and experience. The Destructive Testing On January 21, we were invit¬ ed back to observe cut testing of the roof. We also viewed copies of more drawings and the original specifications. It became quickly apparent that there had been some “value engineering” in that the original specifications, poor as they may have been, had been changed or not followed. Our inspectors’ observations recorded this day include: “Airfoil roof area is/ was .060 TPO fully bonded with bonding adhesive to 3.2- in x 4-ft x 4-ft board stock polyisocyanurate in¬ sulation adhered to con¬ crete deck with foam poly¬ urethane adhesive in rib¬ bons at +/- 9 in oc. “At each end of the roof, the backslope from the 5 ft perimeter band to the drain line was installed using plywood and fram¬ ing instead of the concrete shown on the drawings. There are 3 +/-25-mm (1- in) diameter holes to drain from under the wood backslope to the exterior. Water was draining from those at the bottom of the roof. “Most of the south half of the roof and some of the north half has had all of the insulation moved around and piled up under the membrane. ” The Key Clues “A 2 ft x 4 ft section near the ridge on the north¬ west quadrant was opened. “About 1/3 of the insula¬ tion facer was attached to the membrane. Over the rest of the area, the facer sheared parallel to the membrane, resulting in facer being attached to Figure 7 – About 15% of the insulation facer was attached to the membrane, complete with a thin layer of insulation. The remainder of the membrane had the top half of the insulation facer attached. The attachments referred to are from the original report, not this presentation paper. Proceedings of the RCI 23rd International Convention Wells – 189 Figures 8 through 11 – A total of 480 pavers kept the membrane on the roof; some had been flipped off the roof by flapping membrane, falling over 100 feet to the ground. both the membrane and the insulation. The portion of the facer still attached to the insulation ap¬ peared to be well ad¬ hered. The facer attached to the membrane had been set in a full bed of bonding adhesive. “The insulation facer on top of the insulation was wet. “The joint between the insulation panels was irr¬ egular and about 3/8-in wide maximum. “When the insulation was removed, this cut test was over the edge of a faced gypsum board panel. Or¬ iginal spec called for the faced gypsum, board to be installed on top of the in¬ sulation. Faced gypsum board that was installed was on the concrete deck, contrary to the specifica¬ tions. The requirement for faced gypsum board was deleted on the first day of installation (according to the roofer) after a few rows of faced gypsum board had been installed. “The insulation was ad¬ hered to the faced gyp¬ sum board and the con¬ crete with beads of ure¬ thane adhesive at about 8- in centers. In some areas of the glue bead, it is obvi¬ ous that the insulation was well adhered, and irv others, that there was no contact between the glue bead and the insulation. “At the interface between the faced gypsum board and the uneven concrete deck, there is a gap of up to about 5/8-in under the insulation. “The insulation facer on the bottom of the insula¬ tion and the concrete deck was dry.” Three other locations cut open exhibited similar conditions. Wells – 190 Proceedings of the RCI 23rd International Convention “A third 2-ft x 4-ft opening was cut in the membrane where the insulation had been shuffled, leaving the concrete deck exposed under the membrane. “About 15% of the insula¬ tionfacer was attached to the membrane, complete with a thin layer of insu¬ lation. The remainder of the membrane had the ‘top half of the insulation facer attached. ” (See Figure 7.) “The adhesive ribbons se¬ curing the insulation to the rough concrete deck were spaced at intervals varying from 8 into 14 in. Some of the ribbons had parts of the insulation facer attached; other sec¬ tions had been in contact with the insulation; and still others had never been in contact with the insulation. ” The concrete deck was diy and had a very rough and uneven surface. “Another (fourth) opening was made at the south edge of the insulated roof area. The 2 2×6 wood blocking was in place (although the roofer had said that some of this block had come away from the deck). The 2-3/8- in barbed metal plates securing the perimeter RPF strip were still installed in the wood blocking and had the rem¬ nants of TPO membrane under them. The mem¬ brane had torn out from under the metal plates. “There was bonding ad¬ hesive on the concrete upstand (parapet) to the perimeter band, but no membrane attached.” Figure 12 – The concrete deck, had humps and hollows of more than an inch. We had not been retained to consult; only to investigate and report. It was subsequently decid¬ ed (by others) to try to do a tem¬ porary repair on the roof as the weather was unlikely to cooperate for months. The “roof’ was now not leaking as long as it stayed on the building. Additional pavers (now 480!) were brought up and then an elaborate system of steel cables anchored at the perimeter was installed to keep the pavers on the roof. Some had been flipped off the roof by the flapping membrane, falling over 100 feet below. Some balcony railings were damaged, but, thankfully, no injuries had occurred to humans. See Figures 8 through 11. Most of the clues are in. But you may have noticed one partic¬ ular thing that has not been iden¬ tified or mentioned. This paper may have been available, and you may have read it prior to my pre¬ sentation! Here’s another clue. Figure 12 is a picture of the concrete deck to which the assembly was installed. This photo was taken after the failed assembly had been removed for replacement roofing. The concrete deck is far from level. In fact, a straight edge would reveal humps and hollows of more than 25 mm (1 in). Some of the adhesive strips are undis¬ turbed, due to the rigid insulation boards spanning hollows and never coming in contact with the adhesive. There is still, however, one very important clue that we haven’t mentioned yet! A review and comment on observations “Observation #10 – About 1/3 of the insulation facer was attached to the mem¬ brane. Over the rest of the area, the facer sheared or failed in cohesion parallel to the membrane and insulation. The portion of the facer still adhered to the insulation appeared to be well adhered. The facer attached to the membrane had been set in a full bed of bonding adhesive. “The insulation facer on top of the insulation was Proceedings of the RCI 23rd International Convention Wells – 191 wet. The joint between the insulation panels was irregular and about 3/8 in maximum. ” Photo 13 was actually taken six months after the failure. This material is being removed for the installation of a replacement roof. Note the insulation facer is fairly well adhered to the back of the TPO membrane. The facer is wet or even saturated with water. At this time, the odor of the facer was quite significant, as well! The wet or damp insulation facer is of considerable concern and would be a contributing fac¬ tor in the failure due to the decreased intra-laminar strength of the facer. The facer is a glassreinforced saturated paper and the strength would be significant¬ ly reduced by moisture. We con¬ sidered that the source of mois¬ ture might have been water ingress when the drains failed. Upon further investigation and consideration of the configura¬ tion, this was discounted. The likely source of moisture would be from condensation of moist air from within the building itself, with the airflow uninhibited by the lack of an air barrier or air seal at the deck-level penetrations and exacerbated by the stack effect of the 11 -story building. Warm, moist air could flow up into the roofing assembly at the drain locations as well as other unsealed penetrations. We ob¬ served significant air gaps around drains and other visible roof pen¬ etrations in our site visit of Jan¬ uary 10. We did not observe any requirement or mention of an air seal or vapor retarder at the deck level in the project specifications, section 07530. In my opinion, the lack of an air seal at the deck would be considered a fundamen¬ tal roof design issue as well as an installation issue that contributed significantly to the failure. Observation #10 continued: “When the insulation was removed, this cut test was over the edge of a faced gypsum board panel. The specifications called for the faced gypsum board to be installed over the insula¬ tion. The faced gypsum board, which was in¬ stalled, was on the con¬ crete deck under the insu¬ lation, contrary to the spec¬ ifications. “This was deleted on the first day of installation (ac¬ cording to the roofer), after a few rows of the Dens Deck had been installed.” OPINION The specification actually calls for the faced gypsum board to be installed over the insulation, not under. In my opinion, this (over) was the proper installation, as the faced gypsum board then provides protection to the insulation layer beneath the membrane from foot traffic and increases the mechan¬ ical resistance of the assembly. The polyisocyanurate insulation is quite friable under point loads and the friability contributes a void under the membrane at that location. I note the faced gypsum board has little resistance to moisture vapor drive and would allow the diffusion of air upward to the insulation and membrane. We were not advised why the roofer began putting the faced gypsum board under the insula¬ tion and we were not privy to any written change orders allowing the elimination of the faced gyp¬ sum board. In Canada, this is often referred to as “value engi¬ neering” or expediency or some other reference to cheapening building assemblies. This, in my opinion, is an installation issue that contributed to the failure. Please note we are not criticizing the faced gypsum board itself. It did not contribute to the failure. Further Observations The insulation was adhered to the concrete deck with ribbons of urethane adhesive at about 8-in centers. In some areas of the application, it is obvious that the insulation was adhered; and in many others, that there was no contact between the adhesive rib¬ bon and the insulation. At the interface between the insulation (or faced gypsum board in some locations) and the con¬ crete deck, there is a gap of about 5/8 inch under the insulation – all due to spanning between high points. The insulation facer on the bottom of the insulation and the concrete deck were dry. The adhesive ribbons securing the insulation to the rough con¬ crete deck were spaced at inter¬ vals varying from 8 in to 14 in. Some of the glue beads had parts of the insulation facer attached; other sections had been in contact with the insulation, and still oth¬ ers had never been in contact with the insulation. Opinion The observation that “there was no contact between the adhe¬ sive ribbon and the insulation” indicates that portions (we sus¬ pected significant portions, much later confirmed) of the roof assem¬ bly were never or were poorly adhered at best. Irregularities in the deck level, in conjunction with the rigid insulation boards (which in themselves are not always flat and tend to “cup”), coupled with the failure of the contractor to “walk in or weigh down the insu¬ lation boards into the adhesive (until the adhesive is cured),” would all contribute to this situa¬ tion. This, in our opinion, was an installation issue, which also con¬ tributed significantly to the failure of the assembly. This would at Wells – 192 Proceedings of the RCI 23rd International Convention least partially explain why the insulation was able to move around and “pile up” under the membrane. There was an adhe¬ sion failure, both at the mem¬ brane level and the deck level, so the insulation could move under when the membrane flapped or fluttered in the wind. Lastly, in our preliminary report, we advised, “It seems unlikely that the drains were ever fastened to the deck, but, more like the drain flanges, were likely placed on top of the insulation without adequate mechanical fix¬ ation. Downpipes were attached with MJ-type couplings, which apparently failed when the drains were pulled and moved upward by wind; and, subsequently, water entry took place.” This suspicion was verbally confirmed in our site visit of January 21. Although not a pri¬ mary failure mechanism, in our opinion, this would have provided an additional, larger channel for more air to get into the assembly, and was also the primary point for water ingress into the building. What Happened Here? The “top” or “airfoil” roof is a single-ply TPO membrane. The system is designed as an insulat¬ ed assembly and was designed to be installed “fully adhered.” We observed a complete system fail¬ ure on this roof, although the membrane integrity proper was still essentially intact. The complete assembly was no longer “fully adhered,” but was now completely loose and only held on the building by concrete pavers (installed as emergency ballast) and a perimeter fixation bar from the original installation. Subsequently, a system of steel cables was installed to hold the pavers and membrane on the roof. We observed that the insula¬ tion had moved around and “piled up” under the membrane. Water entry into the suites took place at the locations of cast drain penetrations through the concrete deck. The cast drains were likely “pulled out” of these locations by the membrane bil¬ lowing and “luffing” in severe wind conditions. OUR CONCLUSIONS We confirmed a total system failure of the fully adhered TPO roof section. The roof would have to be removed and replaced at some time in the very near future. The assembly was not, in our opinion, “repairable,” The roof membrane integrity, however, was still intact at the time of our preliminary and sub¬ sequent reviews (due to the tensile strength of the membrane), and water entry was through the con¬ crete deck subsequent to the fail¬ ure at the drain connection. The roof membrane was in danger of blowing right off the building and was held in place by the concrete pavers and a system of steel restraining cables. Subsequent to our first visit, an elaborate system of steel cables was installed over the membrane and pavers as a tem¬ porary measure to hold the roof membrane in place. We under¬ stand the assembly was to have been eventually removed and replaced in the summer of 2006. We advised our concern for the adequacy of the design. The roof is an airfoil shape, 12 stories high, in an exposed building loca¬ tion, and is subject to significant wind loading. The specifications only contained one minor refer¬ ence to wind design. We believe no wind engineering was ever done in the design stages. As mentioned earlier, there seem to have been no requirements in the specifica¬ tion for an air/vapor seal at the deck level. We advised that the membrane manufacturers’ speci¬ fications often call for certain components like an air barrier or a very specific roof assembly, depending on the use and design of the building; but none was used. We were not able to view any project correspondence that approved any of the apparent changes made during construc¬ tion. We further advise that in our opinion, wind conditions actually had little to do with the failure of this roofing assembly. All the fail¬ ure mechanisms were in place prior to any significant wind events, and, in fact, the assembly had failed prior to any significant wind event. The wet insulation facer prevented any resistance to wind uplift; and, given the eventu¬ al wind and the shape of the building, flutter and failure were inevitable. The wind events that “proved” the assembly failure were not that significant. The wind events only indicated that the failure had taken place and additional or more severe wind could then fur¬ ther damage the assembly and building. SUMMARY Our opinions of the reasons for the failure of this roof system remained unchanged from our preliminary review. The failure mechanisms included poor or improper workmanship of the membrane assembly (primarily the insulation and membrane adhesive), improper installation of the roof drains, coupled with fun¬ damental problems with the roof and the building envelope design. THE AFTERMATH Subsequent to our report, there was, predictably, a lot of fin¬ ger pointing. The manufacturer declined warranty coverage due to winds beyond the usual warranty small print. We questioned the wind information supplied by the Proceedings of the RCI 23rd International Convention Wells – 193 manufacturer, but we were never advised of the results. The manu¬ facturer offered to supply new materials to the owner at a reduced price, but assumed no liability whatsoever. We used to call this a “policy allowance,” in my old days with a major manu¬ facturer. A “quit claim” always accompanies this kind of offer. Reminds me of my favorite warranty expressions: “the LARGE PRINT GIVETH and the small print taketh away.” War¬ ranties aren’t printed on water¬ proof paper. The developer and roofing contractors are, to my knowledge, in litigation. The insurance com¬ pany is attempting to subrogate. Some months after our initial involvement, the insurance com¬ panies’ lawyers asked that we again visit the site to report on the new roof system that was being installed. We observed further information about the initial fail¬ ure as the old roof was removed. In its place, the roofer was installing a new vapor retarder (2- ply #15 felt in hot asphalt), new 3- in isocyanurate mopped in hot asphalt, a wood fiberboard over¬ lay, mopped and a 2-ply SBS membrane with a mopped base sheet. We reported our observations, and that was our last involvement in the project. We understand that the roofer was denied a progress payment on finishing the base sheet and that subsequently, the roof sat with just the base sheet installed for nearly a year. I was recently (October 2007) advised by anoth¬ er consultant that the cap sheet had finally been installed a year later. We need not say that this kind of “phased” construction is not usually considered acceptable roofing practice. Ah well, eveiy street we drive down is a street of opportunity! Wells – 194 Proceedings of the RCI 23rd International Convention
