Virtual Design and Production Using Distributed Simulation–based Optimization

Virtual engineering utilizes various computer–based simulation techniques for the analysis of complex systems such as mechanical structures, environmental systems, production processes, and even whole production plants. One main goal of virtual engineering is to find the optimum for mechanical structures, facilities, or production processes in terms of cost, time, energy consumption etc. Hence, simulation–based optimization is an integral part of virtual engineering. The paper deals with the distributed solution of simulation-based nonlinear constrained optimization problems in engineering on a network of workstations. This type of nonlinear optimization and control problems in engineering can be generally characterized as non–convex and non-smooth. Additionally, the involved simulation introduces certain numerical ”noise“ to the solution process. These characteristics lead to very time consuming solution processes, because of long–running computations for simulation and the difficulties in finding a global optimum. The resulting requirements for the corresponding software architecture and software integration of the simulation and optimization components are discussed. Implementation of this software architecture is presented using the OpTiX system for optimization and various finite-element packages for simulation. The application of the distributed solution concept to industrial problems from various engineering domains shows the feasibility of the proposed system.