Structural Optimization in a Distributed Computing Environment.

摘要

This report presents the formulation and testing of a Feasible Sequential Quadratic Programming (FSQP-DIS) optimization algorithm customized to a Distributed Numerical Computing environment (DNC). DNC utilizes networking technology and an ensemble of loosely coupled processors to compute structural analyses concurrently. Each iterate of the FSQP-DIS is partitioned for concurrent computations in the direction calculation, and the steplength calculation. The prototype environment is tested on three applications a mathematical programming problem, the design of a two-story planar steel frame, and finally, the optimal design of a two-story three-dimensional steel frame.