| The design of many practical mechanical systems usually requires dealing with large scale problems incorporating different physical phenomena with diverse behavioral constraints which are generally nonlinear. It is not uncommon in mechanical systems that the system's parameters have to be selected from a discrete set of values imposed by the physical characteristics, standards, availability and production specifications.; This thesis presents a generalized approach to aid the designer in making such problems tractable through appropriate segmentation. Accordingly, the interactions between the different components and parameters become clearer and modeling improvement and design decisions can be made serially with immediate feed back on how they influence the design objective.; Several case illustrations are considered and treated with the proposed unified methodology. In all cases, it was possible to find feasible solutions without significant computing effort. There is no guarantee, however, that the optimum design can be achieved by this approach although in some cases the obtained solution was either optimum or near optimum. The obtained solutions however would provide excellent initial inputs to available large-scale optimization programs which are appropriate to the particular class of problem at hand; The design problems considered for illustrations are a multiple reduction gear system, a four bar mechanism, a hydraulic system, a discrete gear train system and large scale structures; An important feature of the proposed approach is the ability to quickly identify programming errors, constraint inconsistencies and inherent infeasibilities which may exist in large scale systems models... |
