According to the Institute for Business and Home Safety (IBHS), roof cover damage continues to be the largest, most frequent source of non-surge related failures related to hurricanes1. In response, roof covering manufacturers have provided a multitude of products that have demonstrated through independent testing to be able to withstand design event forces. However, it is apparent in the wake of recent windstorm events that much is lost in translation between the idealized laboratory setting and an actual constructed roof. This article will provide a brief review of the current codes, design, and selection process. It will also present a number of common construction-related defects that result in the loss of millions of dollars annually.
The selection of the roof covering in a hurricane-prone region would likely start with the locally adopted building code. For most of the U.S., the building code is likely a recent version of the International Building Code2 with local- and regionally specific, amendments. Some jurisdictions—Florida, for example—have specific codes that apply. The purpose of the code is to set a minimum standard of construction and design for the purpose of life safety. It typically does not purport to address other aspects of the functionality of the roof that might include resistance to water infiltration, service life expectations, hail resistance, and other functions that the facility owner may also require. Another distinction to make is that codes, especially those related to commercial buildings, have become less prescriptive and more performance-based in the last few decades. The code dictates the design criteria but not the number of fasteners per square foot.
Once the basic life-safety design criteria is established, local municipalities and insurance requirements take the design process further, by requiring the selection of assembly that has shown through independent laboratory testing to have a particular resistance to applied wind loads. It is the manufacturer of the specific assembly that undertakes to have it rated by third-party testing. The tested and rated assembly is specific with respect to the materials, attachment, and configuration. Any variation in, say, the type of fastener used, would constitute a different assembly that would require its own rating. Therefore, upon selection of an assembly that is meant to meet building code, local municipality, and, in some cases, insurability requirements, it becomes imperative to closely follow the manufacturer’s recommendations for the installation of the assembly.
Before examining some common construction-related defects that affect the wind-related performance of a roof, here is a quick recap of the way it is supposed to work. For the building-code-derived design wind speeds are used to calculate wind-related pressures imposed on the roof. An assembly is either selected based on its laboratory-obtained rating or designed to be adjoining attached to the structure to resist these pressures. The actual attachment of the roof is done in one of two basic ways, the assembly is either mechanically attached to the structure with fasteners or adhered with adhesive. The base sheet of an assembly may also be mechanically attached and the cap sheet adhered to the base sheet. In either case, the assembly must be attached to the structure or component with a resistance that is greater than the design pressures. Like a chain, the assembly is only as strong as its weakest link. The following examples are used to illustrate some common disconnects between the ideal laboratory conditions used to determine ratings that are then select an assembly to resist the design pressures.
Modes of Attachment
Fasteners are used to mechanically attach components to the structure. These come in many shapes, sizes, and types depending on the application. The fastener will have an anticipated resistance to failure (as determined by the manufacturer of the fastener). Assuming the fastener is properly installed, the resistance of the individual unit divided by the spacing gives an allowable resistance to the design pressure. The number of fasteners, per unit area, then, is critical to the wind related performance.
In the case of a low-slope modified bitumen or single ply membrane over a metal deck, the base sheet and/or the membrane would likely be mechanically fastened to the deck. These fasteners should penetrate the deck and should be observable if access can be obtained to the underside of the deck by lifting ceiling tiles or by accessing a mechanical room. Consider Figures 1 and 2; both are taken in hurricane-prone regions but at different locations. Without performing arduous calculations, suffice it to say the required number of fasteners in a hurricane prone region for a mechanically fastened assembly would probably be counted by the dozens per a 10-foot by 10-foot area (as seen in Figure 1). All things being equal, one can easily discern in Figure 2 that this roof may be lacking the appropriate number of fasteners.
Another common mechanically fastened roof system is the asphalt shingle. Asphalt shingles typically are specified to have either a ‘four-nail’ pattern or a ‘six-nail’ pattern, the latter being meant for regions susceptible to high-velocity winds. The nails are supposed to be installed just below the adhesion strip for three tab and laminate shingles according to ARMA3. The location is critical because with the shingles properly exposed (overlapped), the placement of the nail in this location would penetrate the top shingle and one underneath. Putting the nail too high (Figure 3) on the shingle misses the lower one, in effect, cutting the number of fasteners per shingle in half. Thus, the ‘six-nail’ pattern actually produces 12 nails per shingle holding the shingle to the roof.
Fully adhered systems are required to meet the same resistance standards as the mechanically fastened systems, given that the location and application are the same. Also, like the mechanically attached systems, the adhered applications are only as strong as their weakest component.
With a ‘hot-moped’ application the temperature of the asphalt when applying the insulation or overlying plys of felt becomes critical to its adhesion. Asphalt has an Equiviscous Temperature (EVT) range that is type specific and represents the point at which the asphalt has the ideal viscosity of workability and adhesion between the layers4. Application of the asphalt below the EVT can prevent the asphalt from penetrating the felt and creating the appropriate bond. Working with asphalt above its EVT can breakdown and cause degradation of the asphalt. Figure 4 depicts an example of the blow-off of a membrane where asphalt was applied below its EVP. This is evident by the smooth surface of the asphalt which is indicative of lack of transfer between the asphalt and the adjoining layer. Also noted in Figure 4, proper inter-ply adhesion would be expected to cause some delamination of the insulation.
With fully adhered single ply membranes the attachment of the membrane to the roof is adhesive. The condition of the substrate plays an important role with respect to how well the membrane is attached. The substrate should be clean and free of dust, and debris, or materials not compatible with the adhesive. The ‘Achilles Heel’ of a fully adhered single-ply membrane, however, is the edge attachment. Once the edge is compromised by wind, the system has little resistance to the wind that is now under the membrane and the peeling action of the combined pressure.
Figure 5 shows a roof located in the Gulf of Mexico that failed during a storm that produced wind velocities well below the applicable code required velocities. Inspection of the perimeter condition (Figure 6) revealed that an inappropriate termination bar was used to fasten the edge of the membrane to the parapet. In this case, the flat bar had little clamping ability to adequately hold the edge of the membrane. In effect, the edge of the membrane was only attached at the fasteners and not continuously as is the intent of the termination bar.
Limits of Technology
The requirements of a roof system are many, and are dependent on region and function. Locally adopted building codes and other criteria narrow the search for an appropriate system. Manufacturers then go to great time and expense to have their products rated as appropriate for use in a particular application. This long, involved process requires input from designers, code officials, contractors, and manufacturers. This process is all for naught, however, if the intent of the design is not fulfilled in the field during installation of the roof. How does one ensure that the roof is properly installed in the first place? The answer is through proper procurement and quality control, the discussion of which is entirely outside the scope of this article. The technology does exist to attach a roof to a building to withstand a design event but when determining causation after a loss, it is important to be mindful that the technology does not always make it to the roof.
1 Hurricane Ike: Nature’s Force vs. Structural Strength, Institute for Home and Business Safety, Sept. 2009
2 International Building Code, International Code Counsel
3 Asphalt Roof Manufacturer’s Association, ARMA Residential Roofing Manual, 2010
4 The Roofing and Waterproofing Manual, 5th Edition, National Roofing Contractors Association, 2003