Date of Award
Doctor of Philosophy
James E. Bernard
In the design of mechanical systems, the designer is often faced with the problem of satisfying a number of competing requirements, including light weight and limits on displacements, stresses, and natural frequencies. This usually leads to an optimization problem with weight as the objective and the performance requirements as constraints. It may be desirable, in addition to minimizing weight, to minimize or maximize other performance measures rather than enforcing performance limits;This thesis recast the optimization problem as the minimization of multiple objectives, including performance indices and weight, subject to restrictions placed on the size of the design variables. The thesis also developed an interactive optimization procedure to enable design engineers to bring their skill into play during the optimization process;The procedure transforms the multi-objective problem into a single objective problem by taking the weighted sum of the performance indices and the structure's weight. The weighting factors reflect the relative importance to the designer of the various conflicting objectives. The choice of the weighting factors best suited for the problem is not generally obvious and may require several adjustments before leading to an acceptable design. Thus the procedure is set up so that the designer can choose the weighting factors interactively;Optimization software was developed to provide designers with a decision making tool which is easy to use and provides useful information from which designer can confidently proceed. Since the method uses the NASTRAN for the response and sensitivity calculations, it is applicable to almost any structure which can be modelled using finite elements;Two examples illustrate the power of the technique. The first considers the redesign of an automotive engine block where the challenge was to find a low-weight design which has no natural frequencies in undesirable frequency bands. This problem illustrated the trade-offs between the weight and the natural frequencies, and it demonstrated the interactive process wherein the designer found a combination of the twelve design variables that met performance requirements with a very low weight structure. The second example concerned the redesign of the mounting structure of a heavy-duty truck's exhaust pipe. The structural optimization was stated as finding a low weight structure which satisfied limits on maximum displacements and stresses, and shifted the natural frequencies out of an undesirable band. Again, the design engineer interactively found a very nice solution.
Digital Repository @ Iowa State University, http://lib.dr.iastate.edu/
Somayajula, Gopichand, "Interactive optimization of mechanical systems with multiple performance requirements " (1989). Retrospective Theses and Dissertations. 9242.