Methodologies for Tolerance Intervals
Birna P. Kristinsdottir, Zelda B. Zabinsky, Tibor Csendes, Mark E. Tuttle · 2013
Algorithms for finding large feasible n-dimensional intervals for constrained nonlinear optimization are presented. The n-dimensional interval is iteratively enlarged about a seed point while a checking routine maintains feasibility. Two checking routines are discussed: an interval subdivision method and a global optimization method. Both checking routines can be used in the overall methodology to generate a feasible suboptimal interval. Such an interval is useful when examining manufacturing tolerances in design optimization. Numerical results are presented for a practical application in the optimal design of a flat composite plate and a composite stiffened panel structure.