RELIABILITY OF RANDOM STRUCTURAL SYSTEMS AND FAILURE COST DESIGN (LOAD SPACE, RESISTANCE, NONLINEAR ANALYSIS, MULTIOBJECTIVE)

Point estimates of the reliability of redundant structural systems are computed in load space. The limit state curve is determined using a nonlinear structural analysis program. Evaluation of system reliability is by simulation of load vectors and identification of safety (load vector within the safe region) or failure (load vector outside the safe region). A distribution function is associated with the limit state curve in the case strength variables are random.;It is concluded that the dimensions of load space can be kept small even for large structures. Realistic and fictitious structures are examined as load space dimension is reduced from seven to three (in one case to two). It is shown that a three dimensional load space formulation provides a reasonable approximation of system reliability of structural systems.;A systematic design approach based on applying weights to modes of failure is proposed. Representative individual failure modes are identified on the limit state surface and modal failure probabilities are linearized around mean plastic moment capacities. Single and multiobjective formulations are used as design tools. The procedure is applied to a simple frame for which optimum plastic moment capacities (single objective case) and a trade-off curve between future (failure) cost and initial cost (multiobjective cases) are found.