RROS reliability redundancy optimization solver: A Microsoft Excel add-in

Abstract

This thesis proposes an implementation of an improved algorithm for solving nonlinear separable integer programming problems, as they arise in the field of system reliability. The reliability redundancy optimization solver (RROS) improves on a hybrid technique of dynamic programming with depth first search with tighter variable bounds. The implementation takes the form of an Excel add-in, which will be familiar to users with little formal OR/MS training. The user interface design is discussed. Example applications with benchmarks show that the RROS offers accurate solutions in a comparatively efficient way. This thesis further introduces a surrogate method to solve nonlinear problems with non-separable objectives, where a separable constraint is used as a surrogate objective. This is shown to yield solutions to a number of difficult cases.