Practice Periodical on Structural Design and Construction

A comparison of the economies of the structural design of wood members subjected to axial loads using the allowable stress design (ASD) and reliability-based [load and resistance factor design (LRFD) format] codes is presented. Analyses considered the combinations of dead load only and dead plus live loads due to floor occupancy, snow, wind, and seismic activity. Ratios of live-to-dead loads ranged from zero to 10. The relative economies of the ASD and LRFD codes show the combined influences of the time-related factors assigned to the various load types, load and resistance factor values, and live-to-dead load ratios. The behavioral equations in both codes are now virtually the same. Since the allowable material stresses of the LRFD code are equal to the ASD value multiplied by a constant, the resulting economies are not species- or grade-dependent. For tension and zero-length compression loading, the LRFD format provided the more optimal design for all loading combinations except snow and seismic loadings when the live loads exceed three and four times the dead load, respectively. For columns with the potential for buckling, the LRFD format yielded larger buckling adjustment factors for wind and seismic loadings than did the ASD code. For all other loading arrangements, the ASD format produced larger factors. For floor loading, a 20% average over the LRFD code was observed for very slender columns.

As in the companion paper for axial loading, a comparison of the economies of the structural design of wood members subjected to flexural loading using the allowable stress design (ASD) and reliability-based [load and resistance factor design (LRFD) format] codes is presented. Analyses considered the combinations of dead load only, as well as dead plus live loads resulting from floor occupancy, snow, wind, and seismic activity. Ratios of live-to-dead loads ranged from zero to 10. Other factors that were found to have some influence on the efficiencies of beams susceptible to lateral torsional buckling were the ratio of modulus of elasticity to bending strength, unsupported beam length divided by beam depth, beam aspect ratio, and coefficient of variation of the beam modulus of elasticity. For laterally stable bending and flexural shear, the LRFD format provided the more efficient design for all loading combinations except snow and seismic loadings with live load-to-dead load ratios greater than three and four, respectively. For nominal 2 × 4 members, there was generally little difference between codes when the bending members were laterally unbraced. For the deeper sections, the ASD code proved to be more optimal. This was especially true when the variation in modulus of elasticity was low (i.e., for machine-stress-rated lumber).

A study was conducted to evaluate the two design codes currently in use for the design of mechanical connections subjected to both lateral and withdrawal loading in wood structures. The allowable stress design (ASD) code and the new reliability-based code using the load and resistance factor design (LRFD) format were evaluated to see which offers the greatest efficiency (i.e., economy) with respect to the design of mechanical fasteners. When the allowable dead load plus live load (due to floor occupancy, snow, wind, and seismic activity) combinations were calculated in each format, comparisons revealed that the LRFD code proved to be more economical over a wider range of the two relevant input variables [dead-to-live load ratio (N) and effect of time under load]. When snow or floor live load governed the design, the LRFD format was always optimal. The amount of increase in load-carrying capacity exceeded 40% for floor live load with very small values of N. In the more commonly seen range of load ratios, the magnitude of increased capacity associated with the LRFD format was approximately 10–15%. The LRFD was also more efficient for wind and seismic loading for N < 3 and 4, respectively. For greater N values, the ASD was more advantageous. However, even when N approached 10 (well above the most common applications), the advantage derived from using the ASD format never exceeded 5%.

It is the goal of a computer model to accurately represent a structure mathematically. However, limitations exist that inhibit the ability to achieve an “exact” mathematical model. One such limitation is the number of nodes/members that the computer program can utilize. To help alleviate this problem for structures with trusses, a method has been developed to replace trusses with beam elements thereby reducing the size of the computer model required for analysis. An example structure is presented.

Using the computational neural network model developed recently by the writers, a comprehensive parametric study is performed for global optimization of cold-formed steel I-shape beams based on the American Iron and Steel Institute specification. Design curves are presented for global optimum values of the thickness, the web depth-to-thickness ratio, and the flange width-to-thickness ratio.