A heuristic approach to solving facility layout problems under stochastic demand

Facilities planning is a broad area of study encompassing nearly all engineering disciplines. Designing a facility that processes items whose demands vary is an even more difficult problem. Currently industry experts do not have sufficient tools to plan facilities under these stochastic demand conditions. Most existing computer layout algorithms use only deterministic demand quantities and give no concern to variance.; This thesis addresses the issue of facilities planning under stochastic demand conditions. The problem is addressed by implementing quality control techniques long used by the manufacturing sector to improve productivity. By employing a quadratic loss function, which considers both mean and variance of distance traveled, the heuristic attempts to minimize loss associated with distance traveled through a modification of the well-known CRAFT algorithm.; Three algorithms were examined by this thesis to evaluate different adaptations of the CRAFT heuristic. Each adaptation attempts to minimize a different quantity. The original CRAFT minimizes the total volume-distance product, and the others minimize either expected loss or variance.; The modified CRAFT heuristics were evaluated using several well-known facilities layout test problems. The modified CRAFT algorithms improved the robustness of a layout compared to the original CRAFT algorithm. The layouts produced by the new algorithms often lower variance and minimize the effect of demand shifts on the total material handling cost of the layout.