Optimizing Machine Allocation in Semiconductor Manufacturing Capacity Planning using Bio-Inspired Approaches

Having an accurate capacity planning is always an ultimate goal for semiconductor manufacturing. However, as capacity planning is highly affected by the demand forecast uncertainty, there is a necessity to make the gap closer to ensure profitability. The paper defines the problems faced by a semiconductor company in handling capacity planning and balancing the capital investment cost against the risk of losing revenue. Machine allocation in capacity planning is a process to determine mixture of a machine types that satisfy all precedence and resource constraints and minimize the total machines allocation. We adopt constraint-based genetic algorithm (GA) to solve this optimization problem with the focus on mapping the right chromosome representation to the domain problem. The method is chosen to allow the running of GA in original form as well as to ensure the computation is straight forward and simple. Datasets from one of the semiconductor manufacturing in Penang have been used for experiments. The results showed that by using good chromosome representation the machine allocation optimization will be obtained faster.