Peter Kalinger, BA, MA Canadian Roofing Contractors Association Ottawa, Ontario, Canada ABSTRACT Air barriers are key components in the proper functioning of the building envelope. Building science has proven the construction of an airtight building envelope is a fun¬ damental requirement for acceptable performance of occupied buildings. A recent study by the National Institute of Standards and Technology concluded that heating and cooling costs could be significantly reduced by building an airtight envelope. It has also been proven that the incorporation of an air barrier in a roof assembly can enhance its wind uplift resistance by as much as 50%. Although it is widely accept¬ ed that a correctly built roof air barrier is essential for the proper functioning of build¬ ings, there is much confusion about their necessary performance attributes for lowsloped roofs, what materials can be used, and how they are configured. This paper will explore the various performance requirements of roof air barriers in low-slope roofing assemblies, with particular focus on their correct design and construction. SPEAKER Peter Kalinger has been the technical director of the Canadian Roofing Contractor’s Association since 1993 and has over 35 years of experience in the roofing industry. He has published several papers in various technical and scientific journals and pre¬ sented at national and international symposia on roofing. He is active on numerous Canadian and international roofing-related standards and design committees. He is a member of RCI, Inc. and Construction Specifications Canada. Mr. Kalinger holds a bachelor’s degree and a master’s in public administration from Carleton University. Contact Information: Phone – 613-232-6724; E-mail – p.kalinger@on.aibn.com Kalinger- 134 Proceedings of the RCI 23rd International Convention
Figure 1 – Bituminous membrane air barrier being installed on walls of recreational complex. It has been clearly demon¬ strated that air barriers are key components in the proper func¬ tioning of the building envelope. Building experts with the National Research Council of Canada’s (NRCC) Division of Building Research identified adverse con¬ sequences of air leakage as far back as the early 1950s. In his seminal treatise on design princi¬ ples of exterior walls, presented to the Engineering Institute of Canada in 1953, Dr. Neil Hutcheons proposed that air leakage contributed to concealed moisture accumulation and condensation within walls and the control of air flow was a fundamental require¬ ment (see Figure I). 1 In 1963, in his monograph, titled “Premature Failure of Builtup Roofing,” Frank Joy of the Col¬ lege of Engineering at Pennsyl¬ vania State University, discussed the need to build an air-tight, as well as vapor-tight barrier above humid spaces to prevent moisture accumulating within the roof assembly. 2 Subsequently, build¬ ing scientists have added to our understanding of air, thermal, and moisture transport and rein¬ forced the principle that construc¬ tion of an airtight building enve¬ lope is essential for acceptable building performance. Much of the earlier work on air barriers focused on their role in preventing deterioration of building components from unde¬ sired condensation and moisture accumulation. More recently, the effectiveness of air barriers in reducing heat loss during the heating season and heat gain dur¬ ing the cooling season has gained the attention of building profes¬ sionals throughout North Amer¬ ica. A recent study conducted by the U.S. National Institute of Standards and Technology evalu¬ ated the energy savings obtained through the installation of an effective air barrier in typical com¬ mercial buildings in different cli¬ matic regions in the United States. The study concluded that by building an airtight envelope, energy consumption from both heating and cooling could be reduced significantly. 3 Recognition of the role of air leakage in creating condensation in the building envelope under cold winter conditions was first introduced into the National Building Code of Canada (NBCC) in 1965. Part 4 of the Code called for a continuous vapor and air barrier on the high pressure side of the insulation, thereby encour¬ aging the use of materials and designs that would perform the dual functions of vapor and air Proceedings of the RCI 23rd International Convention Kalinger – 135 leakage control. 4 The 1985 revi¬ sions to the NBCC explicitly rec¬ ognized the distinct functions of controlling vapor diffusion and air leakage. In the United States, Wisconsin has required air barri¬ ers in state-owned projects since 1985 and in 2001 Massachusetts became the first state to require them by code. Today several states have, or intend to require air barriers in commercial build¬ ings as part of their energy or building codes. The 2006 and prior versions of the International Building Code (IBC) do not compel the use of an air barrier in exteri¬ or walls. However, in 2006, the American Society of Heating, Re¬ frigerating and Air-Conditioning Engineers Inc. (ASHRAE) ap¬ proved a revised version of ASHRAE 90.1, “Energy Standards for Buildings Except Low-Rise Residential Buildings” that in¬ cludes requirements for air barri¬ ers. Since ASHRAE 90.1 is adopt¬ ed by reference in the IECC, which in turn is adopted by IBC, it is likely that the next version of the IBC, expected in 2009, will incorporate ASHRAE 90.1, there¬ by making air barriers mandato¬ ry- 5 Material and performance cri¬ teria for air barriers are found in a variety of standards, including ASTM E 2178, “Standard Test Method for Air Permeance of Building Materials; E 1677, Stan¬ dard Specification for Air Barrier (AB) Material or System for Low- Rise Framed Building Walls”; and E 2357, “Standard Test Method for Determining Air Leakage of Air Barrier Assemblies.” Although the requirements for air permeance of air barrier materials may differ from jurisdiction to jurisdiction, it is generally accepted that the per¬ meability of an air barrier materi¬ al should not exceed 0.02 L-s/m2 at a pressure differential of 75 Pa (0.004 cfm/ft2 @ 1.57 psf). As of the date of this writing, there are no prescriptive requirements for maximum allowable air leakage of air barrier assemblies or whole buildings, although recommended air leakage rates are contained in the non-mandatory appendix of the NBCC, based on the interior relative humidity.6 The principal function of an air barrier is to prevent both the infiltration of outdoor air into a building and the exfiltration of indoor air to the outside. 7 The Air Barrier Association of America (ABAA) states that the purpose of installing air barriers is to control the unintended movement of air into and out of a building enclo¬ sure. Air barrier systems are com¬ prised of a number of materials that are assembled together to provide a complete barrier to air leakage through the building enclosure. The National Building Code of Canada (2005) expands their role in the performance of buildings by stating that an air barrier is also required between environmentally dissimilar interi¬ or spaces. In northern climates, the primary function of the air barrier is to prevent condensation that may result from warm, mois¬ ture-laden air entering the colder portions of the exterior building envelope during the heating sea¬ son. Although it is now universally accepted that an airtight envelope is essential for the proper func¬ tioning of occupied buildings, the proper construction of air barriers in low-sloped roofing is still mis¬ understood within the industry. There are several factors that con¬ tribute to this confusion among building practitioners. A review of the energy and building code requirements, as well as the cur¬ rent air barrier standards, reveals that although they do not distin¬ guish between the various ele¬ ments of the building envelope, the primary focus is on wall assemblies. By example, E 2357 contains three references to roofs, two of which address roof-to-wall tie-ins while the other deals with roof and wall interface. In con¬ trast, there are 45 references to walls. Much of the discussion centers on whether roofs are sim¬ ply horizontal walls, or whether they perform differently with respect to air leakage. It is the author’s opinion that roofs are not simply horizontal walls, and it is inappropriate to generalize infor¬ mation and requirements relating to wall performance to low-slope roof assemblies. The question of which compo¬ nents of the roof assembly should perform as the primary plane of airtightness and where it should be located within the assembly continues to be debated. This is of particular concern where the roof is constructed as a compact roof in which the waterproofing mem¬ brane is on top of the insulation and exposed to the weather. The position of the air barrier and the configuration of the assembly are as important as the air properties of the air barrier materials. Article 5.4. 1.1 of the NBCC states that where a building component or assembly separates interior con¬ ditioned space from exterior space, their properties and posi¬ tion shall be such that they mini¬ mize the accumulation of conden¬ sation. To comply with the Wisconsin Enrolled Commercial Building Code, the air barrier must be located at any point on the “interior side of the wall insu¬ lation.”8 Our knowledge of mois¬ ture transport leads us to con¬ clude that the risk of condensa¬ tion resulting from exfiltration can be greatly reduced by posi¬ tioning the air barrier on the warm side of the assembly, or in the case of an insulated compact roof where the membrane is exposed to the elements, below the insulation. Yet millions of square feet of compact low-slope roofing are installed annually where the exposed roof membrane functions satisfactorily as the air barrier, even though located on the cold side of the assembly. Kalingcr – 136 Proceedings of the RCI 23rd International Convention Figure 2 – Modified bitumen membranes perform as air barriers. Ironically, although various types of roofing membranes are widely used as air barriers, their relevant material standards gen¬ erally do not require the testing or reporting of air permeance prop¬ erties. Perhaps this is a result of the fact that many of the air bar¬ rier materials used today in wall constructions are value engi¬ neered roofing products. There is little doubt that almost all roof membranes used in low-slope roofing will exceed the air perme¬ ance requirements of an air barri¬ er material, having the necessary properties of strength, durability, and continuity, and depending on the membrane type – rigidity (see Figure 2). 9 In a low-slope roof, the waterproof membrane, impervi¬ ous to rain and melting ice and snow, will almost certainly per¬ form as an air barrier, provided it is well constructed. Its effective¬ ness, and that of the entire build¬ ing envelope are, of course, dependent on how well the roof membrane is connected to the exterior wall air barrier. In most compact roof systems, the space between the deck and the mem¬ brane does not differ environmen¬ tally from that below the deck. In some systems, however, uncon¬ trolled air leakage into the roof assembly can lead to moisture problems. Even if a roof membrane has the ideal properties of an air bar¬ rier, it does not necessarily mean that it will make an effective air barrier. As Dr. Straube, Professor of Building Science at the University of Waterloo, has point¬ ed out, “the plane of airtightness labeled by the designer… may not, in fact, act as the ABS (air barrier system).” 10 Depending on the con¬ figuration of the roof and the material properties, the mem¬ brane may be effective as an air barrier, preventing air from leav¬ ing or entering the building through the roof, but it may not inhibit the unwanted accumula¬ tion of moisture within the assem¬ bly due to air leakage. Nature abhors a vacuum, and in the case of water vapor in buildings, it doesn’t care for empty spaces much either. Whenever moisture is generated and the interior conditions differ from those outside and a pressure differential exists, some moisture will inevitably find its way into those empty spaces, be they large cavities, as in the case of the roof space in flat roofs insulated below deck, or into the pores of open-cell insulation. Good building practice dictates that we prevent the move¬ ment of moisture-laden air into assemblies by providing air impermeable barriers, or provide venting to the exterior to get rid of the moisture before it can do any harm. Although a compact roof with a continuous vapor barrier forms, more or less, a sealed con¬ tainer, complete airtightness over a large area is practically impossi¬ ble to achieve. In wall construc¬ tion, it is widely accepted that wherever possible, venting should be provided around the outer cladding so that any moisture entering the assembly from the inside can be readily dissipated to the outside. This principle is acknowledged in Section 5.4. 1.1. Required Resistance to Air Leakage of the current edition of the National Building Code of Canada (NBC 2005) that states: “Where a building compo¬ nent or assembly separates interior conditioned space from exterior space, interior space from the ground, or environmentally dissimilar interior spaces, the proper¬ ties and positions of the materials and components Proceedings of the RCI 23d International Convention Kalinger – 1 37 Figure 3 – Vapor retarders are often penetrated by insulation fasteners. moisture from the vented roof space and neither is very efficient in a flat roof. The problem is exac¬ erbated in winter by the poor moisture- holding capacity of cold air. 12 Under ex¬ treme winter con¬ ditions, very sig¬ nificant ventila¬ tion rates are needed to ensure the removal of any moisture that finds its way into a vented roof space. For these rea¬ sons, and due to the economics of construction, most flat roofs in non-residential construction are built as “comor assemblies shall be such that they control air leakage or permit venting to the ex¬ terior so as to… minimize the accumulation of condensa¬ tion in and the penetration of precipitation into the building component or assembly. The reference to venting was added in recognition of the role of venting in achieving the intent of the requirements of 5.4.1. Pitched roofs are relatively easy to ventilate. During winter, any warm, moist air that moves up through the ceiling into a big open attic space will mix with the large available volume of colder air and be carried to the outside through vents before it can do any harm. Stack-induced flow (the chimney effect) increases with slope as well as decreasing tem¬ peratures, thereby promoting ven¬ tilation in properly constructed steep roofs. Flat roofs are much more diffi¬ cult to ventilate adequately. Natural ventilation, which relies on the chimney effect, is difficult to achieve because of the lack of slope over long spans. The height of the roof space also influences the effectiveness of natural venti¬ lation. For any difference in tem¬ perature between the inside of the roof space and the exterior, the chimney effect is directly related to the difference in height between the intake and exhaust openings. On flat roofs, the vent openings are generally at the same level. Even if the vents are raised to different heights by mounting the exhausts on curbs or extending exhaust pipes, by example, this usually amounts to only a few feet difference in eleva¬ tion at most. Diffusion and windinduced ventilation are the only mechanisms left for removing pact” roofs, where the waterproof membrane is placed on rigid insulation over the deck. A separate vapor retarder placed between the insulation and the deck normally provides vapor control by retarding the relatively slow process of vapor diffusion into the roof assembly (see Figure 3). This vapor impermeable layer may be airtight, but in many instances, it is not, with numer¬ ous seams and overlaps and often perforated by insulation fasteners or other penetrations. The outboard roof membrane is often far more air and vapor impermeable than the inboard va¬ por retarder, creating a potential vapor trap and problems, should the uptake of moisture during the heating season through imperfec¬ tions in the vapor barrier exceed the downward drying during the drying season. Over time, this may result in the accumulation of sufficient quantities of moisture Kalinger – 138 Proceedings of the RC1 23rd International Convention Figure 4 – The impermeable membrane is on the warm side in PMR assemblies. to adversely affect the roofs per¬ formance. Building a vapor trap between a vapor impermeable layer on the warm side and an air and vapor impermeable layer on the cold side can be avoided by constructing a protected mem¬ brane roofing (PMR) assembly. The PMR assembly has many advantages with respect to air and moisture control. The roof mem¬ brane serves as both the vapor barrier and the plane of air-tight¬ ness. 13 Being located on the warm side of the assembly, the risk of condensation is virtually eliminat¬ ed (see Figure 4). If any moisture does find its way through the roof membrane, it is transmitted to the outside where it cannot do harm. However, a PMR assembly may not be feasible or economical in all circumstances. The industry has witnessed the growth in populari¬ ty of mechanically fastened and loosely laid flexible membrane systems in the past few decades due to several features of these systems. Mechanically fastened systems are much lighter in terms of bulk and weight of materials required than more traditional fully adhered roofing systems. They are relatively easy to install and can be easily inspected and repaired. In addition, because hot bitumen and adhesives are not required, the hazards and environmental concerns associat¬ ed with their use are avoided. Even where weight is not of pri¬ mary concern, PMR assemblies may not be the most cost-effective design alternative. Loosely laid ballasted systems have proven to be veiy economical and have a proven record of satisfactory per¬ formance. PMR assemblies incor¬ porating loosely laid membranes have not been particularly suc¬ cessful, primarily due to the diffi¬ culty in finding and repairing leaks, should they occur. Although PMR assemblies may appear to be the “ideal” as¬ sembly with respect to the func¬ tioning of an air barrier, countless numbers of compact roofs have been constructed without conden¬ sation problems where the out¬ board roof membrane is the desig¬ nated air barrier. The reason for this is that in compact roofs, where the membrane is fully adhered to rigid insulation, there is little chance of the occurrence of convective currents or appre¬ ciable air leakage paths through the opaque portions of the assem¬ bly (see Figure 5). This is particu¬ larly true where relatively stiff membranes, such as built-up roofing and modified-bituminous membranes, are used. Because the components are built tightly together and in intimate contact, channel airflow is virtually elimi¬ nated. In such assemblies, the roof membrane, even though on the cold side, can perform as an effective air barrier if properly tied to the other parts of the building envelope. Moisture-related prob¬ lems have occurred, primarily manifested by interply blistering or blistering between the mem¬ brane and the substrate or insu¬ lating layer. The source of this moisture can be construction moisture (from moisture bound in Proceedings of the RCI 23 rd International Convention Kalinger – 1 39 the materials, moisture that enters the assembly during construction such as rain, or snow) or moisture from the building interior. Once formed, blister growth can be exacerbated by air leakage into the affected area under pressure from below or by suction, from the top or bot¬ tom, through a small crevice. The addition of a vapor retarder is usually sufficient to control harmful condensa¬ tion from vapor diffusion. Where such systems are built over air-impermeable decks, such as cast-in-place con¬ crete, the deck and/or vapor barrier may act as the air bar¬ rier, providing redundancy. Steel decks account for approximately 70% of all nonresidential low-slope roof Figure 5 – In compact roofs, convective air flow is minimized. decks. 14 Although sheet steel is vapor- and air-imperme¬ able, steel decks contain numerous seams, rendering them air leaky and vapor-per¬ meable. 15 The many seams, overlaps, and flutes of the deck provide unobstructed pathways for air flow (see Figure 6). In addition, in order to satisfy wind uplift resis¬ tance requirements for roof assemblies over steel decks, many roof system design and approval organizations re¬ quire mechanical fastening of some or all of the above-deck components to the steel deck. 16 This results in the pen¬ etration of the vapor/ air retarder by the numerous fas¬ teners, thereby compromising the vapor and air imperme¬ ability of the retarder. The surest way of prevent- Figure 6 – Seams, perforations, and flutes provide unobstructed air flow in metal decks. ing air leakage through mechanical attachment points in these systems is by construct¬ ing a “split” system, where a thin layer of insulation or rigid board is first mechanically fastened to the deck onto which a two-ply asphalt felt vapor barrier is mopped. Insulation is adhered by a mopping layer of asphalt. A split system may also be advantageous in loosely laid systems. The vapor barrier may consist of a polyethyl¬ ene sheet with sealed laps laid loose over a suitable layer of gyp¬ sum board or other suitable board material. The insulation and membrane are laid dry and bal¬ lasted. In both systems, the vapor Kalinger – 140 Proceedings of the RCI 23rd International Convention Figure 7 – Air can move freely under membrane in mechanically attached systems. tem is not an option. Due to the extensibility and flexibility of the membrane, wind and mechanical pressurization can cause them to flutter and billow between the attach¬ ment points, pumping large quantities of air, and with it moisture, from the building interior into the assembly (see Figure 7). The problem is most acute with systems that do not contain vapor retar¬ ders between the insulation and deck, but can also occur if the vapor retarder fails to effectively retard the flow of air into the assembly from below. During cold weather, if the membrane temperature falls below the dewpoint tem¬ perature, the moist air from the building interior will con¬ dense on contact with the cold undersurface of the membrane. retarder can perform as an air barrier. Wind resistance does not rely on mechanical attachment to the supporting deck, thus elimi¬ nating the need to penetrate the air/ vapor barrier. tened flexible membrane systems where wind uplift resistance is obtained by fastening through all the layers of the assembly into the deck at discrete attachment points. With these, the “split” sys- Strictly speaking, the problem is not one of air leakage through the building envelope, but one of air intrusion into the roof assembly. It is a function of the flexibility and extensibility of As reliable as the “split” system is, there is some ques¬ tion as to whether it is neces¬ sary. There are thousands of roofs in Canada where the insulation has been mechani¬ cally fastened through the vapor barrier into the steel deck that are performing sat¬ isfactorily without condensa¬ tion or other moisture-related problems. When properly installed, the vapor barrier is squeezed tightly between the insulation and the deck by fasteners and the vapor and air leakage through the attachment points is limited. 17 In most low-rise and normal¬ occupancy buildings, perfora¬ tion of the vapor and air retarder will not pose a seri¬ ous problem. This is not the case with mechanically fas- Figure 8 – Under wind action, air is drawn into the roof space in mechanically attached systems. Proceedings of the RCI 23rd International Convention Kalinger – 141 the membrane, its rein¬ forcement, its attach¬ ment, and the spacing between the membrane fastener rows (see Fig¬ ure 8). The wider the spacing between fasten¬ er rows and the more extensive the mem¬ brane, the greater the ballooning that will oc¬ cur. As the membrane balloons under the ac¬ tion of wind and/or mechanical pressuriza¬ tion, the volume of the roof space increases with a corresponding drop in pressure, draw¬ ing more moist air from the interior. In loosely laid and ballasted systems, al¬ though membrane deflection may not be a Figure 9 – Air can move freely under membranes in loose-laid and bal¬ lasted systems. concern, there are num¬ erous pathways for air to flow laterally throughout the assembly as the components are not built tightly together (see Figure 9). It is evident that unre¬ stricted flow of air into and through the roof assembly and deflection of the membrane with its corresponding pumping of air, in both mechanically attached and loosely laid flexible mem¬ brane systems, must be prevented to avoid the accumulation of moisture within the assembly. The role of air retarders in enhancing the wind resistance of roofing systems has been well established. 18 Research conducted at the NRC’s Institute for Re¬ search in Construction (IRC) for the Special Interest Group for the Dynamic Evaluation of Roofing Systems (SIGDERS) consortium has demonstrated that the wind uplift resistance of mechanically attached flexible membrane sys¬ tems can be increased by as much as 50% by the inclusion of a vapor retarder, regardless of the vapor retarder type. 19 Part of this in¬ creased uplift resistance is due to the reduced membrane deflection under load when a vapor barrier was incorporated in the assembly. The vapor barrier acts as an air retarder, restricting the airflow into the roof system from below deck. In flexible roof membrane systems, the inclusion of a vapor barrier may serve to limit the deflection of the membrane and the volume of air pumped into the assembly from below to levels where it will do no harm. Max¬ imum allowable air leakage rates through the retarder, and corre¬ sponding maximum allowable increases in volume of the roof space to limit moisture entry and condensation potential have yet to be determined. Up to the present, little research has been undertaken to determine the actual air leakage rates into mechanically attached compact roof assemblies under varying pressure loads, with most of the air barrier research having been conducted on low-rise wall assemblies. Our knowledge of air leakage through these roof sys¬ tems is being expanded by work being carried out by NRC/IRC and its industry partners through the Special Interest Group for Dynamic Evaluation of Roofing Systems (SIGDERS) consortium. Testing is currently underway to quantify the actual air leakage rates of various roof assemblies under varying pressure loads. The objective of this investigation is to determine the actual performance requirements for roof air barriers and to develop the appropriate test methods. In 2004, researchers at NRC conducted a series of small-scale tests on three test roof assemblies to measure the air leakage through them. 20 One assembly incorporated a polyethylene sheet vapor barrier over the steel deck, the second a self-adhering, polymer- modified sheet vapor barrier, and the third no vapor barrier material. In all three, rigid foam insulation was fastened to the deck. The assemblies were devoid of any barrier and insulation Kalingcr – 142 Proceedings of the RCI 23rd International Convention joints, representing the best-case conditions. The airflow through the assembly was determined by creating a negative pressure dif¬ ferential of 240 Pa to 2400 Pa (5 to 50 psf) in increments of 240 Pa (5 psf). The testing revealed that the air leakage rates of both test assemblies that incorporated a vapor barrier fell below the NBCC recommended maximum allow¬ able air leakage rate of 0.15 L/s-m2 at a pressure of 75 Pa for air barrier systems where the interior relative humidity is less than 27%. It is also interesting to note that the small-scale testing indicated that even thin-film ma¬ terial such as polyethylene (even though penetrated by fasteners), provided lower air leakage trans¬ mission rates than those recom¬ mended by the Code. The following year, another series of tests was carried out on a larger test table with dimen¬ sions of 2 m x 6 m x 0.8 m (79 in x 236 in x 32 in). 21 Five assemblies with the following configurations were tested for air leakage: Assembly 1 – steel deck with one layer of 50 mm (2 in) poly¬ isocyanurate insulation, Assembly 2 – steel deck and two layers of 50 mm (2 in) insulation, Assembly 3 – steel deck, asphalt-saturated felt as vapor barrier, one layer of 50- mm (2-in) insulation, Assembly 4 – steel deck, one layer of self-adhesive, modified- asphalt membrane, one layer of 50-mm (2-in) insula¬ tion, and Assembly 5 – steel deck, 6- mil polyethylene, one layer of 50-mm (2-in) insulation. In each assembly, the insula¬ tion was mechanically fastened to the deck. Again, the airflow through the assembly was deter¬ mined by creating a negative pres¬ sure differential between the underside and top of the test assembly. The differential pres¬ sures ranged from 480 Pa to 2870 Pa (10 psf to 60 psf). It should be noted that in both series of tests, the pressure differentials of the test procedure were significantly higher than those contained in ASTM E2357-2005, Standard Test Method for Determining Air Leakage of Air Barrier Assem¬ blies. 22 However, these greater pressure differentials are more representative of actual in-service conditions. Unlike the previous series of tests, all vapor barrier assemblies, with the exception of the polyethylene, had a continu¬ ous overlap in the long direction of the seam. In this series of tests, only the assembly with the poly¬ ethylene vapor barrier had an air leakage rate that fell below the NBCC recommended maximum of 0.15 L/s-m2. This is most likely the result of the polyethylene hav¬ ing been applied in a single and continuous sheet without any overlaps. CONCLUSION It appears, from the limited testing to date, that including a vapor barrier will not only im¬ prove wind uplift resistance, but will also enhance the air barrier properties of the flexible roofing membrane. However, limited test¬ ing at differential pressures simi¬ lar to those that would be encoun¬ tered from wind forces impacting on compact roofing systems indi¬ cated that a single layer of thinsheet vapor barrier material over a steel deck is unlikely to perform as an effective air barrier, particu¬ larly in flexible membrane sys¬ tems. The numerous laps, seams, and penetrations from fasteners resulted in air leakage greater than 0.15 L/s-m2. Most roofing membranes, although located on the cold side of the insulation, will perform as effective air barriers provided that: 1. All penetrations and open¬ ings are sealed and made airtight. 2. Continuity is provided by tying in the roof mem¬ brane to the other (wall) air-barrier elements. 3. Vapor barriers are in stalled in mechanically fastened, flexible-membrane systems to limit vapor diffusion, air intru¬ sion, and membrane deflection. 4. Vapor barriers are in¬ stalled in loose-laid and ballasted systems to limit vapor diffusion and air intrusion into the roof space. When selecting the appropri¬ ate roofing system for a particular project, all of the performance requirements must be considered, including the system’s role as an air barrier in the building enve¬ lope. Some configurations meet this requirement more easily than others. In the protected mem¬ brane roof assembly, the roof membrane acts as the principal plane of airtightness and is locat¬ ed at the optimal location, the warm side of the assembly. It also provides venting to the outside of any water vapor that may pass through it. However, on many projects, a PMR assembly is nei¬ ther feasible nor cost-efficient. Compact roofing systems, whe¬ ther fully adhered, mechanically fastened, or loosely laid, have been widely used with few air leakage-related performance problems. Empirical evidence indicates that in fully adhered compact roof systems, particular¬ ly those with relatively stiff mem¬ branes, there is little airflow through them. In these systems, the roof membrane performs the function of the air barrier effec¬ tively, provided it is made contin¬ uous with the other parts of the building envelope. In mechanical- Proceedings of the RCI 23rd International Convention Kalinger – 143 ly fastened flexible membrane and loosely laid systems, the role of the roof membrane as the prima¬ ry air barrier may be short cir¬ cuited if air is allowed to enter the assembly. There is some evidence that a vapor barrier between the deck and the insulation will provide ef¬ fective air leakage control. How¬ ever, much research is required before the impact of the roof mem¬ brane properties and penetration by fasteners is fully understood. When a compact, mechanically fastened or loosely laid roof sys¬ tem is selected, the designer must carefully consider where the plane of airtightness is to be located. This must be based on those fac¬ tors that will influence the ther¬ mal and pressure gradients encountered in service, as well as the physical properties of the roof membrane and all other assembly components, the anticipated inte¬ rior conditions, the exterior envi¬ ronment, and a numerous other factors that will influence the thermal and pressure gradients that will be encountered in ser¬ vice. FOOTNOTES 1. Hutcheon, N.B. “Funda¬ mental Considerations in the Design of Exterior Walls for Buildings.” Pre¬ sented to Annual Meeting of the Engineering Insti¬ tute of Canada, Halifax, May 1953, NRC 3057. 2. Joy, F.A., Premature Fail¬ ure of Built-up Roofing. Building Research, The Pennsylvania State Uni¬ versity College of Eng¬ ineering, September, 1963. 3. Emmerich, S.J. et al. In¬ vestigation of the Impact of Commercial Building Envel¬ ope Airtightness on HVAC Energy Use. NISTR 7238, National Institute of Standards and Technol¬ ogy, June, 2005. 4. Submission to the Second Commission of Inquiry into the Quality of Condo¬ minium Construction, Ur¬ ban Development Institute Pacific Region. February, 2000. 5. Farahmandpour, K., “Air Barriers, Vapor Retarders, and Weather-Resistive Barriers: Are They All the Same?” Masonry, July, 2002 6. A-5. 4.1. 2.(1) and (2) of Appendix A of the National Building Code of Canada, 2005, recommends a max¬ imum allowable air leak¬ age rate for opaque, insu¬ lated portions of the build¬ ing envelope of 0.15 L/(s-m2) where the relative humidity of the interior at 21 °C is less than 27%. 7. Quiroutte, R.L., “The Air Barrier Defined.” Building Science Insight ’86. NRCC, 1986. 8. “Air Barrier Update,” Tech¬ nology in Brief Interna¬ tional Masonry Institute. Annapolis, January, 2004. 9. Research undertaken by Canada Mortgage and Housing Corporation dem¬ onstrated that the air leak¬ age rate of smooth-sur¬ faced and modified-bitumen membranes was nonmeasurable. See CMHC Technical Series 98-109, Air Permeance of Building Materials. 10. Straube, J.F. “Under¬ standing and Controlling Air Flow in Building Enclo¬ sures,” Affordable Comfort 2001 Conference Pro¬ ceedings, Milwaukee, WI, May, 2001. 11. National Building Code of Canada, 2005. National Research Council of Canada, Ottawa, 2005. 12. At a temperature of 20°C, one cubic meter of air at saturation (RH=100%) can hold 17.236 grams of water. If the temperature falls to -20°C, the same volume of air can only hold 0.565 grams. 13. Johnson, G. “Flat Roof or Wall: is There a Differ¬ ence?” Atlantic Construc¬ tion Journal, January, 2006, Dartmouth, N.S. 14. Booth, R.J. “Field Exper¬ iences Versus Standards and Designs,” Proceedings of the Third International Symposium on Roofing Technology, Gaithers¬ burg, Md, 1991. 15. Desjarlais, A.O. “Self-Dry¬ ing Roofs: What! No Drip¬ ping!” Proceedings of the ASHRA E/DOE/BETEC Thermal Performance of Exterior Envelopes of Buildings VI. 1995. 16. See Loss Prevention Data Sheet 1-29, Roof Deck Securement and Above¬ Deck Roof Components, FM Global, 2006. 17. Tobiasson, W. “Vapor Re¬ tarders For Membrane Roofing Systems,” Pro¬ ceedings of the 9th Con¬ ference on Roofing Tech¬ nology, NRCA, 1989. 18. “Research Needs: Wind Resistance Testing of Roofing Systems,” Pro¬ ceedings of the Roof Wind Uplift Testing Workshop, Oak Ridge National Lab¬ oratory. Oak Ridge, TN, November, 1989. 19. Baskaran, B.A., Ko, S.K.P., Which is the Weak- Kalinger – 144 Proceedings of tbeRCI 23rd International Convention est Link? Wind Perfor¬ mance of Mechanically- Attached Systems. NRCC- 45693, National Research Council, 2006. 20. Baskaran, B.A., Molleti, S„ Booth, R.J., “Under¬ standing Air Barriers in Mechanically Attached Low-Slope Roofing As¬ semblies for Wind Uplift.” NRCC, Ottawa, 2004. 21. Moletti, S., Baskaran, B.A., Air Leakage Quanti¬ fication of Roofing As¬ semblies over Steel Deck. NRCC, Ottawa, 2005. 22. In accordance with ASTM E2357-2005, the air leak¬ age rate of air barrier assemblies is determined by measuring the air leakage rate at pressure differences across speci¬ mens, of 0,5, 1, 1.6, 2.1, 3.1 and 6 psf (25, 50, 75, 150, 250, 300 Pa). Proceedings of the RCI 23rd International Convention Kalinger – 145
