Two new optimization methods, the Percent Selection method and the Percent Elimination method are presented. The Percent Selection method models discrete optimization problems as continuous problems. A gradient based procedure is used to determine the optimum given two or more distinct options. This is done with averaged properties. It is shown that the process is much more efficient than discrete searches. This method is applied to determining the optimum end fixity of a member, and the optimum flange alignment. The Percent Elimination method is an adaption of the Percent Selection method for eliminating members from a design. Existing optimization procedures can become trapped at local optimum solutions. The Percent Elimination method overcomes local optima.
Link to paper