Thomas L. Smith, AIA, RRC TLSmith Consulting Inc. Rockton, IL ABSTRACT A terne metal roof was specified and installed on a new church during the winter of 1996 – 1997. The building is located in the eastern part of the U.S., in an area subject to snowfall. Shortly after construction, leakage periodically occurred at clerestory windows and some roof areas. The archi¬ tect, general contractor (GC), and roofing contractor attempted to address the problems, but their attempts were unsuccessful. In 2001, the author was retained by the building owner to investigate the leakage problems. Because the leakage problems could not be effectively repaired, in the latter part of 2001 and early 2002, all of the roof coverings were removed and replaced. The clerestory windows were also removed, new sill flashings installed, and the windows reinstalled. As part of the investigation, the National Research Council of Canada (NRCC) performed exten¬ sive analysis pertaining to corrosion of the metal panels. This paper describes how the architect erred during design and construction, the GC and roofing contractor erred during construction, and how these parties erred in their initial responses to the leakage problems. It also describes the reroofing work. It details many potential pitfalls with terne metal. The paper concludes with a discussion on the successful mediation among the architect, gen¬ eral contractor, roofing contractor, and owner, which avoided litigation. Thomas L. Smith, AIA, RRC Tom Smith is president of TLSmith Consulting Inc. He specializes in architectural technology and research, with an emphasis on roof systems. Smith is a licensed architect and a registered roof consultant. His interest in roofing began in 1974. From 1988 to 1998, he was the research director for the National Roofing Contractors Association (NRCA). Prior to that time, he was in private practice in California and Alaska. He has designed roofs from the arctic to the tropics. Mr. Smith’s wind expertise is based on a unique combination of extensive experience with wind-damage investigation and analysis, involvement in academic and applied research, involvement in standards development and preparation of wind design guidelines, experience in designing roof assemblies in high-wind regions, formal education, and teaching. Smith has been a member of the committee that is responsible for ASCE 7, Minimum Design Loads for Buildings and Other Structures, since 1990. Smith— 87
INTRODUCTION The church is a very unique and aesthetically pleas¬ ing building (Figure 1). Five shed roofs cover the sanctu¬ ary (roof areas 1 -5 on Figure 2) . These roofs have differ¬ ent slopes, from 2-3/4: 12 to 5:12. The eave-to-ridge dis¬ tance also varies, with the greatest distance being 112′. These roofs drained into a stainless steel, built-in gutter. Clerestory windows occur at the walls between the dif¬ ferent roof areas (Figure 3). Five shed roofs cover other areas of the church (roof areas 6 – 10). These roofs also have different slopes, from 1:12 to 3-3/4:12. The entry canopy is also a shed roof that slopes towards the building at a 3/4:12 slope (roof area 12). Roof area 1-10 and 12 had standing seam terne panels with l”-high ribs. Roof 11 has a slope of 1/2: 12; it had a soldered flat locked terne roof. A TPO membrane occurred at the valley between the toe of roof areas 6-10 and the wall below the eave of the sanctuary sheds (Figure 4). The water from roof areas 6-11 drained to two stainless steel internal gutters. The water from roof area 12 drained into a stainless steel internal gutter at the wall juncture. Figure 2 – Roof plan. Figure 1 – General view of the church. The roof assembly over the sanctuary (1-5) was com¬ posed of the following components: • Terne metal panel, field painted on the topside and a mill-applied shop coat on the underside. (Terne is a sheet steel that is factory-coated on each side with an alloy of 80% lead and 20% tin.) • Rosin paper. Figure 3 – View of some of the clerestory windows. Although the roof overhangs provide some protection from rainfall, they do not protect the windows from wind-driven rain. Smith-89 Figure 4 – View of the TPO membrane in the valley area. Water falling from the TPO membrane eroded the paint on the terne below. • #30 asphalt-saturated felt. • OSB • Rigid insulation • Tongue-and-groove wood decking The roof assembly over areas 6-12 was similar to the above, except for the deck, which was metal. The TPO was fully adhered to OSB. INITIAL LEAKAGE AND RESPONSE Initial leakage occurred in the sanctuary. Water dripped from the vicinity of several clerestory windows. The architect made a site visit, but did not perform any destructive observations. His report incorrectly attributed the leakage to the soffit installation. However, no correc¬ tive action related to the soffit was taken. The clerestory windows were composed of a highquality, commercial-grade storefront glazing. The system was designed with an internal gutter designed to intercept water that penetrated past the glazing gasket. The inter¬ cepted water would drain from an opening between the sill flashing and the sill face cap. Not recognizing the purpose of the gap, the GC decided to apply sealant between the sill flashing and the sill face cap, thereby blocking drainage of water that penetrated the gaskets (Figure 5). Application of the sealant exacerbated the window leakage problem. The window leakage was due to lack of upturned ends on the sill flashing, as discussed later. As the window leakage continued, leaks began to develop in other areas of the building. At the time of the Figure 5 -In a misguided attempt to solve the window leakage problem, the general contractor applied sealant (noted by arrow) at the drainage slot between the sill flashing and sill face cap. author’s investigation in February 2001 , in addition to the clerestory leakage in the sanctuary, leakage had occurred at the north end of the TPO roof, the southwest portion of roof 6, roof 11, and roof 12 (at the wall juncture). Because the architect, GC, roofing contractor, and glazing contractor could not determine the leakage caus¬ es, the author was retained by the building owner to investigate the leakage, determine the causes, and recom¬ mend corrective action. PAINT PROBLEMS The manufacturer’s specifications require the panels to be field painted “immediately after application or as soon as proper painting conditions prevail.” Painting soon after application is important to avoid corrosion, for the thin lead/tin coating offers very limited corrosion resistance. The panels were installed during the winter and temperature conditions were much too cold for paint¬ ing. Rather than tent over the building, painting was delayed until May. The paint experienced a widespread peeling failure. The author was not involved in that prob¬ lem, but was advised that after arbitrating the dispute, the roof was repainted in the fall of 2000. During the author’s leakage investigation in February 2001, the new paint had also peeled in several areas. AUTHOR INVESTIGATIONS During the author’s initial investigation, a technical representative of the window manufacturer was on-site, as well as representatives of the GC, roofing contractor, and glazing contractor. The author directed the glazing contractor to remove some of the glazing caps. After Smith— 90 Figure 6 – View of the down-slope end of the sill flash¬ ing after removal of the face caps and glazing. The only provision preventing water running down the sill from entering the building was a fillet-profile sealant joint between the sill flashing and EIFS. The double arrows show a piece of the sealant joint. The sealant was not well-adhered to the EIFS or sill flashing. Water was able to flow past the sealant, where it had an unob¬ structed path into the building. removing the caps, it was apparent that due to lack of upturned ends on the sill flashing, water was able to flow past the window frame (Figure 6). At two locations over the sanctuary, very small cor¬ rosion-induced holes through the metal panels were found. The author initially attributed these holes to lack of paint, but as discussed below, they developed from underneath the panels. The holes were temporarily patched. Figure 7 -A hole through the terne at roof 6. This hole was caused by mechanical damage. At the north end of the TPO roof and roof 11, the author believed the leakage problems were related to the built-in gutters or the drain lines. Instructions were pre¬ pared for the owner to plug the outlet tubes and fill the gutter with water. When both gutters were water tested, it was discovered that when the plugs were removed, leak¬ age occurred due to a piping error. The drain line from the gutter emptied into an open funnel at the end of a major drain line. Other drain lines from the sanctuary roofs also emptied into the funnel. Thus, when there was a deluge of water, water overflowed the top of the fun¬ nel. At the roof 6 leakage area, two small holes through the metal panels were found (Figure 7). These were caused by mechanical damage. A tear was found through the nearby TPO. At another TPO area, the membrane Figure 8 – View of corrosion-induced holes through the terne on roof 3. Holes occurred on both sides of the rib. These holes occurred a few inches upslope of an end¬ lap. Note the buckled rib. See Figure 9. Figure 9 – View of the underside of the panel shown in Figure 8. The circle indicates the upper end of the down-slope panel. A panel clip with two nails occurs in the lower right portion of the photo. Smith— 91 was tom by a nail head that extend¬ ed above the OSB. The leakage at the canopy roof was due to inade¬ quate flashing between the roof and wall. During the summer of 2001 , additional holes were discovered through the metal panels. The author performed a follow-up investigation in August 2001 . During this investi¬ gation, several metal panels and ridge cap flashings were removed. Severe corrosion was observed on the underside of the panels (Figures 8 and 9). Wind-driven water and water from power-washing prior to painting had driven past the ridge cap flashing and flowed underneath the panels. Water was also driven underneath the endlaps (Figure 10). Water had also leaked through some of the ribs on the low-sloped roofs. Figure 10 – The left panel is the down-slope panel. The arrows indicate sol¬ der that was used to attach the lock-strip. Both sides of the lock-strip were very corroded. The right panel is the underside of the up-slope panel. Enough water had reached the underside of the panels and enough time had elapsed by August 2001 that the corrosion was penetrating the pan¬ els in numerous areas. In addition to panel penetration, the corrosion had significantly impaired several of the concealed clips, hence the roof was vulnerable to wind blow-off. Because the corrosion attack was coming from the underside of the panels and it was in an advanced state, the panels could not be effectively repaired. To avoid damage to the OSB and wood decking, the author recom¬ mended that the entire roof area be removed and replaced. Reroofing work commenced that fall. The owner retained the services of another GC and roofing contractor to perform the reroofing work, which is dis¬ cussed below. LABORATORY INVESTIGATIONS The lab determined that the corrosion was caused by the presence of water and oxygen on the underside of the panels. Analysis of the corrosion products did not indi¬ cate the presence of common atmospheric pollutants such as chloride or sulfur. The solder that was used was the correct type. It did not contribute to the corrosion. There was no indication that the flux that was used in the sol¬ dering process contributed to the corrosion. Rosin paper Leachate from wet rosin paper had a pH between 6 and 7, which is similar to water in the natural environ¬ ment. The leachate also increased the conductivity of the water, thereby releasing electrolytes into the water. This increased conductivity helps to facilitate electron transfer and accelerate corrosion. When submersed in water, the rosin paper absorbed a substantial amount of water. Several samples taken by the author were sent to the National Research Council of Canada for various evalua¬ tions to further understand the corrosion problem. The thickness of the steel substrate and metallic top and bot¬ tom coatings was relatively uniform and without pin¬ holes. The tested specimens complied or substantially complied with the manufacturer’s specification. No sig¬ nificant manufacturing deficiencies with the terne were found. Although the rosin paper did not leach very corrosive chemicals, the paper had the ability to hold a substantial amount of water, which if present, would increase the relative humidity in the enclosed space between the paper and metal panels. High humidity on the underside of the panel and the presence of electrolytes from the rosin paper create good conditions for corrosion to occur. In addition, when wet rosin paper is in contact with the metal, it provides a moderately corrosive water film (with a potential pH of 4 to 5) next to the panel. A water film generally contains a high concentration of dissolved Smith— 92 oxygen, which also increases the corrosion rate. Therefore, when wet, the presence of the rosin paper could accelerate the corrosion process. Felt The panel manufacturer’s recommendations adamant¬ ly and repeatedly recommend against the use of asphaltsaturated felt below terne (only rosin paper is recom¬ mended between the metal and the wood substrate). However, felt was installed below the rosin. In a few small areas, the felt was in direct contact with the terne, but in those areas, only minor, superficial “white rust” was found. (White rust was caused by corrosion of the metallic coating. Red rust was caused by corrosion of the steel substrate.) Leachate from the felt had nearly the same pH as the wet leachate from the rosin paper. However, the felt leachate did not increase the conductivity nearly as much as the rosin paper leachate. The felt also absorbed less water when submerged. The felt unlikely promoted cor¬ rosion. ARCHITECT’S ERRORS The architect’s errors were numerous and significant. They began during design and continued through his response to the leakage problem. The following is a syn¬ opsis of the major issues: This was a complicated building. To avoid leakage problems, it was incumbent on the architect to give spe¬ cial attention to: 1) selection of building envelope sys¬ tems, 2) the design of the details, 3) submittal review, and 4) field observation. The architect failed in all four areas. Roof system selection Selection of metal panels for roofs 1 – 5 was appro¬ priate. However, a copper or PVDF painted aluminum¬ zinc alloy (“Galvalume”) standing seam system would have been a better choice. These alternative systems would have been much more resistant to underside corro¬ sion, and they would have been less maintenance inten¬ sive than terne. According to the manufacturer, terne requires repainting “at least once every eight years.” The owner was not made aware of the future repainting costs until after the roof was completed. The cost of repainting this roof was not trivial; the owner was astounded at the future financial demands imposed by the need for fre¬ quent painting. Selection of metal panels for roofs 6-12 was inap¬ propriate. Most of these roofs are low-slope, which pre¬ sent design and installation challenges. These roofs are essentially not visible from the ground, hence there was no aesthetic justification for selecting metal. A membrane roof would have been more reliable and economical. Selection of the single-ply membrane was inappro¬ priate because of susceptibility to puncture and tearing. Metal panels were installed after installation of the TPO; hence, during that time, the risk of damage was great. Also, during painting operations, substantial foot traffic presented damage opportunity. Metal panel specification This section was very deficient; it was only two pages, whereas the much smaller and less complicated single-ply membrane section was five pages. Flat-locked terne was specified, which complied with the manufac¬ turer’s recommendations. However, the SMACNA Architectural Sheet Metal Manual recommends stainless steel or copper for flat-locked seamed roofs. Stainless steel would have been more resistant to corrosion and puncture. The spec did not specify special provisions for those panels on slopes less than 3:12. Drawings The roof plan indicated that the panels at roofs 1 – 5 were to have two endlaps (i.e., three panels from eave to ridge). On three of these roofs, the panel lengths were 30.6′, 33′, and 37’. The manufacturer recommended the use of expansion clips when the panels exceeded 30’. Expansion clips were not specified, nor were they installed. The endlap detail did not specify the manufacturer’s recommendation to field-paint the underside of the end¬ lap. Considering the limited corrosion resistance of the metallic coating, field painting the unexposed portion of the endlap was important and should have been included with the detail. Field painting of the concealed area was not done, and severe corrosion resulted. There were no provisions for ice dam protection at the eaves of roofs 1 – 5. The rake and ridge details were only shown schematically – the design intent was not clear. The transition between the TPO and metal panels was not detailed. Neither the details nor window specifi¬ cation called for upturned legs on the sill flashings. Submittal review The architect’s review was superficial and very inad¬ equate. Expansion clips were not shown for the panels Smith— 93 over 30′ in length, painting of the underside of the end¬ laps and the flat locked panels was not noted, the TPO/metal transition was shown (but it did not allow for escape of water that reached the underlayment), no spe¬ cial provisions were shown for the low-sloped roofs (such as sealant tape in the ribs, as recommended by the manufacturer), the ridge rake flashings were poorly detailed, and several special details were not included. Pre-roofing conference A pre-roofing conference was specified, but it was not held. Considering the complexities of this roofing project, it is bewildering that a pre-roofing conference was not held. The architect should have insisted on the conference. Periodic field observations The architect performed periodic field observations. Either the person conducting the observations was unfa¬ miliar with the manufacturer’s installation recommenda¬ tions or was not diligent in evaluating the work with respect to it. In either case, the roof observations were inadequate. CONTRACTOR ERRORS Although the architect’s errors were the primary cause of the leakage and necessity to reroof, the contrac¬ tor also made errors that contributed to the problems. Submittal preparation As previously discussed, the submittal was very defi¬ cient. While many of the items that should have been shown on the submittals were not addressed in the con¬ tract specifications or drawings, they were addressed in the panel manufacturer’s specifications, which the roof¬ ing contractor should have been aware of. Workmanship Workmanship on the exposed portions of the roof generally exhibited good workmanship (with the excep¬ tion of the field-applied paint, which was applied by a painting contractor). Workmanship errors included lack of paint on the underside of the endlaps and the flat locked panels, poor fabrication of the ridge cap detail, three nail punctures (two through the metal) – either due to nail back-out or they were not flush with the OSB at time of installation, puncture of a metal panel by a small rock underneath the panel , and use of fixed clips on pan¬ els in excess of 30′. As previously discussed, the contractor should not have sealed the window drainage slots. REROOFING The author recommended the following: Roofs 1 -6: Remove the existing metal panels and underlayment, replace damaged OSB, install an APP modified bitumen base sheet and red rosin paper and new standing seam metal panels. A modified bitumen base sheet was specified because it is more robust than a #30 felt. APP was specified because of its great resistance to high-temperature flow. The base sheet laps were sealed for a distance of 9′ above the gutter for ice dam protec¬ tion. Two options were given for the metal: PVDF paint¬ ed aluminum-zinc alloy or copper. At the time the reroof¬ ing was about to commence, the price of copper was rel¬ atively low, so the owner elected to have copper installed. All other roof areas: Remove the existing metal pan¬ els and underlayment, replace damaged OSB, install a mechanically-attached SBS modified bitumen base sheet and white, granule-surfaced SBS cap sheet set in cold adhesive. Walkway pads were specified in the vicinity of the gutter drip line to avoid damage from falling ice. Pads were also specified at the vertical drops from one roof area to another to avoid water erosion. The clerestory windows were removed, new sill flashings with upturned ends installed, and the windows were reinstalled. A local architect, relatively knowledgeable about roofing, was retained by the owner to work with the roofing contractor to develop numerous custom details. The reroofing work was negotiated with a local general contractor and local roofing contractor familiar with standing seam metal and modified bitumen membrane roof application. Leakage problems have not been reported since the reroofing and window work was completed in early 2002. POTENTIAL TERNE PITFALLS There are potential pitfalls with terne and terne II that designers and contractors should be aware of: 1. Owners should be made aware of the manufactur¬ er’s recommendation to repaint at least every eight years and the cost implications thereof. 2. The manufacturer recommends against inclusion of felt underlayment. However, if the manufactur¬ er’s recommendations are followed and only rosin paper is specified, an adequate underlayment will not be achieved. Rosin paper is effective in avoid- Smith— 94 ing bonding between the metal panels and felts, but it is easy to tear during application and it has limited water resistance. An appropriate underlayment system is a layer of asphalt-saturated base sheet (or a modified bitumen base sheet) and a layer of rosin paper. The problem for the specifier is that if the manu¬ facturer’s recommendations are followed, a poor underlayment system occurs. If a good underlayment system is specified, it violates the manufac¬ turer’s recommendations, which is not desirable. Based on the results of the author’s field observa¬ tions and the laboratory testing, the manufacturer should change its underlayment specification. 3. Ventilation: The manufacturer’s specification states that “an air space must be provided under roof deck to facilitate ventilation and eliminate conden¬ sation.” While this recommendation is easy to accommodate when an attic space occurs below the roof deck, it is difficult and expensive to accommodate when there is no attic space. Had a ventilation cavity been provided on the church roof, the panel corrosion problem would not have been affected. Furthermore, a ventilation cavity below the deck will not necessarily prevent con¬ densation on the underside of the panels. As with the underlayment issue discussed above, the manu¬ facturer should revise its recommendations regard¬ ing ventilation so that they are technically valid. 4. Low slopes: For slopes below 3:12, the manufac¬ turer should be contacted for special requirements (in 2002, the special requirements were not pub¬ lished in the manufacturer’s literature). 5. Field painting: Field painting very soon after panel application is important. This should be specified and rigidly enforced. If panel application will occur when it is colder than + 50 degrees F, the roof should be tented during painting, panel appli¬ cation delayed, or a panel that does not require field painting should be specified. Specify field painting of the end laps and rigidly enforce this requirement. 6. Underside corrosion: The manufacturer’s warranty is voided if the terne is “subjected to underside moisture, either from roof leakage or condensation resulting from inadequate ventilation.” Specifiers should be aware that the underside of terne has a much shorter time-to-failure due to underside cor¬ rosion than other common metal roofing panels such as aluminum-zinc alloy and copper. MEDIATION After the church was reroofed, the owner, architect, general contractor, and roofing contractor entered into a one-day mediation. The parties were able to reach a settlement and therefore avoid the cost, time, and uncertainty associated with litigation. Before the mediation, a comprehensive report was prepared by the author. That document, as well as other documents, were provided to the other par¬ ties prior to the mediation. The mediation began with brief statements by various parties. The par¬ ties then adjourned to separate rooms and the mediator met privately with each party and carried settlement offers back and forth between the par¬ ties. The progress was slow and at times it looked doubtful that a settlement would be reached. But late in the day, an acceptable settlement was reached. A long, difficult project for all involved was finally concluded. Smith— 95
