Reliability Assessment of Roof Sheathing Performance for Typical North American Roofs Subjected to Wind Pressure

Open Access
Amini, Maral M
Graduate Program:
Civil Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
December 14, 2010
Committee Members:
  • Dr Scanlon, Dissertation Advisor
  • Andrew Scanlon, Committee Chair
  • Bohumil Kasal, Committee Chair
  • Daniel G Linzell, Committee Member
  • John Fricks, Committee Member
  • Peggy Ann Johnson, Committee Member
  • Roofs subjected to wind pressure
  • Stochastic modeling of low-rise roofs
  • Roof sheathing Failure
Low-rise buildings encompass the majority of residential structures in the United States. Predominantly, this category of structures is constructed with dimension lumber. Investigations after natural disasters report that during high intensity wind pressures, low-rise buildings with wood-frame construction are at immense risk of damage with the first sign of damage initiated at the roof. Developing a comprehensive understanding of low-rise structures' behavior when exposed to wind loads is essential for maintaining public safety and preserving structural integrity during high winds. This study develops a better understanding of wind load distribution on the roofs of classified low-rise structures and presents a methodology for reliability assessment of roof sheathing performance for typical North American roofs in light wood frame structures subjected to wind pressures. The first part of the study focuses on analyzing wind pressure distribution on the roof surface of a typical North American structure. Experimental data is obtained from a full-scale facility located at the University of New Brunswick, Fredericton, Canada. Two dimensional surface fitting is used to fit higher order curves to predict pressure variation across the roof of classified low-rise structures. Later, pressure coefficients are computed from the predicted pressure time histories and are then fitted with probability density functions (PDF). The developed pressure coefficients can be used for any location for which wind speed data is available in order to compute the roof wind loads. The second part of the study consists of developing a methodology for reliability assessment of typical North American roofs and building shapes. This methodology can be used as a tool to predict and evaluate the performance of low-rise structures during high wind events. Stochastic finite element models of selected roof types are constructed for analytical Monte Carlo simulations. Probabilistic analyses are used to assess the effects of uncertainties which can have various possible sources. Employing probabilistic analysis allows us to account for the effects of uncertainties that influence the outcome of an analysis based on a set of input variables. Random input variables for this project include the nail withdrawal capacity, number and location of missing nails on roof panels due to workmanship, wind induced pressure coefficients, and wind speed and direction. Statistical distributions are used to describe and quantify the uncertain input variables. Next the probability of roof failure is computed for a selected construction quality. Roof failure is defined by exceeding a maximum limiting displacement based on a construction quality. It should be noted that the computed probabilities of failure are based on the global geometric parameters selected for this study. The applicability of the developed methodology is evaluated by employing the presented methodology to predict the probability of roof failure for an inventory of low-rise structures subjected to wind speeds reported during Hurricane Andrew. Finally, the predicted probabilities of failure are compared with the probabilities of roof failure reported after Hurricane Andrew. Higher predicted probabilities of failure are obtained with the developed reliability assessment methodology. In the stochastic finite element simulations, all the structures are exposed to the maximum observed peak gust speed, whereas not all the structures in the region experienced the maximum reported wind speed during Hurricane Andrew; therefore, lower probabilities of failure are recorded for Hurricane Andrew. Similarly, the developed reliability assessment methodology is utilized to compare the probability of roof sheathing failure for the computed pressure coefficients for the full-scale experimental structure and the ASCE/SEI7-10 pressure coefficients. ASCE/SEI7-10 pressure coefficients resulted in lower probabilities of failure than the experimental pressure coefficients. The presented reliability assessment methodology can be used as a tool by Federal Emergency Management Agency 'FEMA' or insurance companies to evaluate/predict the performance of low-rise structures during high wind events for a set of predefined random input variables such as fastener capacity, fastener workmanship, and wind pressure coefficients, velocity and direction. FEMA utilizes standardized methodologies for mitigation and cost analysis estimates for losses after disasters such as hurricanes, earthquakes, and floods.