Reducing variability in equipment availability at Intel using systems optimization

Abstract

Equipment management is an important driver behind operational efficiency, since capital equipment makes up about 40% of the average semiconductor manufacturer's total assets. The main goal of this project is to reduce variability in tool availability by planning for usage-driven preventive maintenance. A method and associated tools are proposed and investigated in the context of the Thin Films area in Intel's Hudson facility. The solution we propose incorporates the following characteristics: Drives towards a balanced preventive maintenance (PM) schedule such that PMs are evenly distributed in time; Enables fast recovery to a normal PM schedule after unexpected events occur on the factory floor, e.g. equipment breakdown, by re-distributing loads on each tool; Facilitates performance tracking and accountability; and Ensures consistency in the decision-making process. We will describe the conceptual method and the implementation process, from prototype deployment to the development of a production application. Alternative solutions using case-based reasoning and rule-based systems will also be discussed. We will conclude by discussing the role of automated decision systems in manufacturing and outline key issues to be considered in choosing an optimal design.