Move based heuristics for the unidirectional loop network layout problem

We consider the loop network layout problem in a manufacturing system where n machines must be placed in n available locations around a loop to minimize the total flow distance. The formulation of the problem results in a quadratic assignment problem which is computationally a very hard problem. We discuss the idea of positional moves and local improvement algorithms based on moves or k-way (particularly 2-way) interchanges. Even though the concept of moves has not found its way into algorithmic design in the existing literature, our computational experimentation with two-move based heuristics indicates uniformly superior performance in comparison to the well known pairwise interchange heuristic.