Robust design optimization of structures through consideration of variation

Optimal structural design is considered under the presence of variation in loading, geometry and material properties. A Monte Carlo simulation is embedded in a genetic optimization algorithm to produce an output distribution for the objective function and constraint functions at each design evaluation. A hybrid genetic/non-linear-programming algorithm is used with a multi-objective formulation to locate a design that is optimal under the primary design criteria, but is simultaneously insensitive to variation. Cross-sectional, geometric and topological design changes are considered. Specific examples presented include a truss structure and an automotive inner body panel. The goal is to produce optimal designs that map better to the designerʹs intent.