Semiconductor Capacity Planning Problems Under Uncertainty Empirical Research

Semiconductor manufacturing is one of the most complicated manufacturing industries. The capacity planning is nowadays popular and hot issue for semiconductor manufacturer, because of many working procedures to finish one semiconductor product, high requirement of the equipment, the special character of the re-entry procedure, expensive equipments and the long lead time for buying the equipment.At present, the department of industry engineering is in charge of long, medium and short capacity planning in manufacturing factory. The key factors in the long capacity planning process in semiconductor manufacturing industry are the equipment demand and production space in factory. Equipment demand impacts on the factory space directly by the space equipment needed. The main variables influencing the factory space are the forecasting demand from market, the character in product working process and the machines' running statement which are uncertain with time moving on forward in the long capacity planning. So those uncertainty factors result in the uncertainty of factory space in planning process.This paper tries to identify and quantify the uncertainty in the capacity planning process in form of risk to help semiconductor manufacturer make decision. First, this paper does sensitivity analysis on equipments to find the key factors which impact on factory space in all machines. Then based on the principle of support vector machine and according to the optimal hyperplane set up, an identify function built up is used to identify or forecast if the risk exists about space enough or not enough at the future time point, meanwhile, The very popular tool to calculate the invest in financial field VaR(Value at Risk) is introduced to quantify the risk. According to VaR value, we can get one upper limit about the space demand at the future time point, under the given risk bearing and given lead time, which can help enterprise to make correct decision.