Research And Implementation Of Capacity Planning System For Multiple Semiconductor ATM Factories

The difficulty of capacity planning in semiconductor Assemble and Test Manufacturing (ATM) factory is how to quantify the production capacity using the volume of products. Production capacity planning of semiconductor ATM factory can split production task equally in multiple factories and optimize capacity allocation for each ATM factory,which is in the midterm master production planning period of supply chain. It has to access inventory holding, available machines, available resources and other bottle-neck capacities to answer the question whether enterprise's available production capacity is enough to support the whole market demand. In order to ensure all of ATM factories can support production plan, it will formulate a detailed production capacity and resource requirement plan for each ATM factory when its available production capacity is not enough to support the splitted production task from virtual factory, and it will assess production capacity for virtual factory and each site.Considering different series of products have the same production processes, and the only difference among series is that different operations will use different machines and tools in the same process, this paper will use tool group to define the combination set of machines and tools used by each operation, and calculate out the quantity of products processed by each unit of tool group in a specified time period in the most economical way, which is called LOT in production planning area and can be calculated out by industrial engineers based on actual production processes and operations. The mathematics model proposed by the paper will use LOT as the basic unit of production process, which is very quickly to convert demand of products in market to demand of tool groups inside ATM factory. To further reduce the complexity of capacity planning, the paper will consider available tool groups as the available capacity and calculate out the demand for each resource based on the demand of tool groups, and it will access the volume of resources and tool groups that can be obtained inside ATM factory to calculate out how many products be supported and how many shortage exists, and it will develop a detailed requirement plan for tool groups and critical resources.The paper will study the solution how to build decision support system in production capacity planning area for semiconductor ATM factory based on Capacity Response Analysis Optimize (CRAO) System, which is about to adopt Production Process Model (PPM) to simplify the material allocation in semiconductor ATM processes. CRAO system also uses Multi-Objective Linear Programming as the decision support tool to construct mathematic model for material, resource demand and other business objects. CRAO system can execute production capacity planning automatically with fully considering material, capacity, resource utilization, inventory holding, lead time , yield and other constraints, it's main objective is to maximize the quantity of products can be supported by virtual factory and optimize the allocation of capacity. It will also propose a detailed allocation plan for products demand and capacity and a detailed requirement for additional resources and tool groups. Meanwhile, the paper will introduce business model, mathematic model and software design and implementation of CRAO System respectively. ILOG CPLEX is very suitable for production capacity planning and other optimization application in industrial, so CRAO system chooses ILOG CPLEX engine to solve mathematic model, which is very strong to search the optimal solution for large-scale linear programming and mixed integer linear programming problems in a short time. The prototype system has been on trial in a company, the capacity response analyzing time is reduced from 2 weeks to 2.5 days.