A COMPLEX PHENOMENON
DYNAMIC WIND-UPLIFT PERFORMANCE OF COMPOSITE METAL ROOF ASSEMBLIES
WIND LOADING on roofs remains a complex phenomenon and one of the most common factors in roof failures. This mainly is because wind-flow conditions around buildings and wind-induced effects on such buildings have dynamic characteristics that are time and space related. From a space perspective, the wind's characteristics are affected by aspects such as building geometry, site topography and architectural features. From a time perspective, the wind dynamic characteristics are affected by wind speed, turbulence intensity and wind direction. It also must be noted roofs are subjected to wind-induced negative pressures on the exterior and positive pressures acting on the underside. This is shown in Figure 1.
Within a roof assembly, each component of that assembly offers a certain resistance to the wind-uplift forces. Figure 2 provides an illustration of the force-resistance link concept. For a roof to adequately resist the wind-uplift forces, all resistance links should remain connected. Failure is considered to have occurred when the wind-uplift force is greater than the resistance of any one or more of these links. To identify the weakest link, designers need wind-uplift test protocols that mimic the wind-load characteristics, as well as the types of roof failure modes resulting from such winds.
Static vs. Dynamic and Panel vs. Assembly
On the North American scene, two main test procedures mostly have been in use for determining the wind-uplift resistance of metal roofs:
1Conventional ASTM E1592-01 "Structural Performance of Sheet Metal Roof and Siding Systems by Uniform Static Air Pressure Difference" from ASTM International, West Conshohocken, Pa.
2Recent CSA A 123.21 "Standard Test Method for the Dynamic Wind Uplift Resistance of Mechanically Attached Membrane-Roofing Systems" from the Canadian Standards Association, Mississauga, Ontario. CSA A 123.21 represents the only North American test procedure for assessing the wind-load uplift resistance under dynamic wind-load conditions. The development of this test procedure was the result of the work undertaken by the Special Interest Group for Dynamic Evaluation of Roofing Systems, a consortium of manufacturers, building owners, trade associations and researchers (www.sigders.ca).
*The project's hypothesis is to develop the load-sharing relationship among the roofing components by quantifying the pressure-equalization process in the assembly.
ASTM E1592 static testing is inadequate from the standpoint of true representation of wind action, particularly in a storm. A proper simulation of the wind, which is a main characteristic of CSA A 123.21, is necessary for developing an appropriate presentation of its dynamic features through rationalized dynamic load cycles that were developed through wind-tunnel testing of full-scale roof assemblies. This resulted in qualifying and quantifying the effects induced by wind.
One of the major limitations of ASTM E1592 lies in the fact that it is limited only to the assessment of the strength of metal panel and its fastening techniques while one of the major characteristics of CSA A 123.21 is that the testing is not limited to the covering material. Rather, it considers the effects on the whole roof assembly, along with its components and the interaction among them (structural elements, retarders/barriers, deck fastening, insulation fasteners and roof covering).
A joint research project is ongoing between the National Research Council of Canada, Ottawa, and Metal Building Manufactures Association, Cleveland, as well as American Iron and Steel Institute, Washington, D.C.; Copper Development Association, New York; and Metal Construction Association, Glenview, Ill., with the main objective of evaluating the performance of composite roof assemblies with metal covering. The project's hypothesis is to develop the load-sharing relationship among the roofing components by quantifying the pressure-equalization process in the assembly. PEP depends on the air permeability of the various components, as well as field construction details. Therefore, assembly (rather than material) evaluations are accomplished.
*To identify the weakest link, designers need wind-uplift test protocols that mimic the wind-load characteristics.
This project investigated 12 assemblies in three groups:
> Group 1: Composite assemblies with plywood deck
> Group 2: Composite assemblies with steel deck
> Group 3: Composite assemblies with air retarders
Experimental Setup and Test Protocol
Investigations were carried out at NRC's Dynamic Roofing Facility. Figure 3 shows a typical composite assembly layout and installation at the facility. The facility consists of a bottom frame of adjustable height upon which roof specimens and a movable top chamber are installed. The bottom frame and top chamber are about 240 inches (6100 mm) long, 87 inches (2200 mm) wide and 32 inches (800 mm) high. The top chamber is equipped with six windows for viewing and a gust simulator, which consist of a flap valve connected to a stepping motor through a timing-belt arrangement. All assemblies were subjected to CSA A 123.21-04 dynamic test protocol.
Performance of Composite Assemblies Under Dynamic Loading
When the assembly is subjected to suction at the top, the panel deforms and transfers the forces through the fastener to the structural deck. Figure 4 shows typical wind-uplift performance of composite assemblies. It has two data sets, one representing a composite assembly without an air retarder and the other with a retarder. For each set, about five wind gusts are presented at a pressure level of 4.3 kPa (90 psf) and each have a duration of about eight seconds. Within the eight-second period, the panel experiences the maximum pressure of 4.3 kPa (90 psf) for two seconds and there is no loading for a period of two seconds.
Such time varying gust simulation creates fatigue on the assembly. What is interesting is the ratio of sharing of the applied suction between panel and insulation. In the case of assembly with no retarder, the panels resist the applied wind uplift whereas the pressure difference across the insulation is only instantaneous. While in assembly with a retarder, the insulation boards share about 90 percent of the applied suction because of the airflow resistance. This mechanism of pressure sharing between the components is indicative of the composite behavior of the assembly, and its effect will be reflected in deflection and clip load of the metal panel. This insulation contribution partly reduces the stress on the panel clip fasteners and facilitates to increase the wind-uplift rating of assemblies. To reinforce this observation, measured fastener forces were analyzed.
During the wind test, panel clip fasteners were instrumented with load-measuring sensors. Figure 5 shows the measured loads at various pressure levels. Data from different assemblies are grouped into two segments; one segment to represent the panel assembly and a second to represent the composite assembly. It is clear the clip loads are reduced in the case of composite assemblies. This is because of the fact that in composite assemblies, as already discussed, the load sharing mechanism of the insulation contributes to lower clip load of the metal panel. The percentage of load shared by the insulation varies depending on the insulation thickness and type of air retarder above the deck.
Failure Modes of Composite Assemblies Under Dynamic Loading
Failure modes of the assemblies also were investigated after the wind test. Typically, three failure modes were observed. Assemblies from Group 1 with wood deck configurations failed because the net load on the deck exceeded the strength of the deck-to-joist connection. In other words, the fastener attachment of the deck pulled out from the joist. In the case of Group 2, panel seams opened or separated during wind testing. The weakest link of this group is the seam-locking mechanism of the metal panels. For the Group 3 configurations with air retarders, mostly clip attachment fasteners pulled out and caused the cracking of the insulation. This preliminary failure triggered the panel seam opening as a secondary failure. In addition, deck pullout from the structural joist also was observed in one assembly.
Wind-uplift performances of composite assemblies are different from the performance of just panel behavior under static loading. In other words, it is clear the CSA A 123.21 dynamic test offers advantages compared with the conventional ASTM E1592.
Observation of this ongoing investigation can be summarized as follows:
> In noncomposite assemblies, only the metal panel resists the simulated wind uplift. For the same pressure level, panel deflection of a noncomposite assembly is twice compared to a composite assembly. This excessive panel deformation causes the panel bucking at the middle span between panel clips leading to a failure mode of panel cracking around the clips.
> Wind-uplift resistance of composite assemblies are higher than the panel wind-uplift rating. Inclusion of an air retarder in a composite assembly significantly improves the wind-uplift rating.
> Load sharing by other components (deck, insulation, etc.) is evident in assemblies with an air retarder.
Dr. Bas Baskaran is a group leader for the Roofing Sub-Program at the Ottawa, Ontario-based National Research Council of Canada, Institute for Research in Construction. At NRC, he is researching the wind effects on building envelopes through experiments and computer modeling. Baskaran also acts as adjunct professor at the University of Ottawa. He is the vice chairperson for the Roofing Committee on Wind Issues and a member of American Society of Civil Engineers, SPRI, and the International Council for Research and Innovation in Building and Construction technical committees.