RELIABILITY PERFORMANCE OF LIGHT-FRAME WOOD ROOF SYSTEMS
Primus Vincent Mtenga
Under the supervision of Professor Alain H. Peyrot
at the University of Wisconsin-Madison
ABSTRACT
The current procedure for designing Metal Plate Truss (MPT) roof systems is based on the design of individual components that make up the system. The forces in the components are obtained by analysis of an individual truss assumed to carry loads from its tributary area. Moreover, the proportioning or sizing of the components assumes that the acting forces do not exceed code specified deterministic strength values. This design procedure ignores the following facts: 1) The in-situ performance of a component can be influenced by the performance of other components within the system, 2) The tributary area approach ignores the fact that there is considerable load redistribution within the roof system, 3) Neither the component strength nor the loading on the system is deterministic.
The main objective of this study, therefore, was to analytically investigate the performance of metal plate truss roof systems with plywood sheathing. Then use the results of such investigation to determine what modification factors are to be applied in the design of individual components to reflect the system effect.
A roof system model and computer program (NARSYS) to compute the ultimate strength of metal-plated trusses and roof systems were developed. Parametric studies, conducted using the developed model, revealed considerable strength recovery when a weakened truss is used in a roof system. The degree of strength recovery was found to be dependent on the location of a weak component within the weakened truss.
Ultimate strengths of individual trusses and roof systems were simulated with NARSYS. In the simulations,the variability of the ultimate system strength was found to be less than the variability of the ultimate strength of trusses that make up the systems, which in turn were found to be less than the variability of the strength of the components that make up the trusses.
Methods to determine factors to be used to include system effect in an RBD (LRFD format) component based design procedure are presented. A system factor of 1.15 is recommended as a start value, pending further studies, to be used in the design of components of trusses of the type and span ranges examined in this study. However a value of 1.20 is recommended as a start value, if early connector failures are prevented, through "over-plating" of some critical joints and proper joint detailing. The system factors were found to be insensitive to changes in sheathing thickness.