Adding/Removing Machines Based Dynamic Layout Optimization

Dynamic Layout Problem (DLP) is very important for manufacturing plants under dynamic business environment. Generally, it can be divided into two situations: (1) certain demand based; (2) uncertain demand based. In this thesis, a new kind of dynamic layout problem is studied, which is based on un-equal area and continual representation of layout and allows machines added into the plant or removed from it.As the basis of the thesis, problem is defined first, and then evaluation system is set up. Costs here are divided into four parts: material handling cost, relocation cost, area-utilization cost and shut-down cost. A free space searching algorithm is proposed, based on it, heuristic rules of adding and removing machines are discussed.Under certain demand situation, disregard of random factors in system, two mathematical models-dynamic programming based and shortest path based-are set up, then, two modified heuristic algorithms-Dynamic Programming based Simulated Annealing (DPSA) and Shortest Path based Simulated Annealing (SPSA) are proposed to solve these two models. In order to improve poor comparability between cases, an evaluation index framework is set up. A number of numerial tests show that SPSA can find better solutions, however DPSA is faster.Under uncertain demand situation, according to that production is equal to demand, with constraint of system shut-down, arriving speed formula of new batches is deduced. Then mathematical model under uncertain demand is set up. Consequently, based on the study of two kinds of random events and three dispatching rules in system, a simulation system is set up, which allows to add and remove machines in the production. By designing interface module, simulation module and optimization algorithm module are joined together to be a simulation based optimization system. Finally, an example is studied to verify this simulation-based system and simuation system.