AbstractThe purpose of this study is to examine the relative efficacy of several promising heuristic approaches to a classic problem of component placement. Four "construction" and nine "improvement" algorithms were chosen for investigation and compared experimentally on a CDC 6400 computer. The improvement methods were selected to test some basic strategies of pairwise-interchanging of components and the construction procedures were chosen primarily to evaluate the effects of the quality of starting solution on the improvement methods. The algorithms were tested on 75 problems generated from the literature and compared with respect to the produced solution quality and CPU run-time requirements. A construction approach due to Graves and Whinston produced the best results, both when used to generate starting solutions for the improvement methods and when evaluated on its own merit against the improvement methods using other starts. Construction approaches have previously been regarded in the the past as relatively inferior techniques.