کنترل WIPمبتنی بر خط تولید مجازی برای سیستم های تولید نیمه هادی

The explosive growth of data acquisition and increasing number of machining processes makes extremely difficult to deliver precise information to the proper users at the right time by relying on a centralized work-in-process (WIP) control system in manufacturing. This article presents a potentially practical solution to a manageable and well-distributed WIP control system by addressing issues such as real time performance, scalability, and reconfigurability. By taking advantage of the advances in distributed computing technologies multiple WIP control instances performing different subsets of control and management responsibilities can be spawned from a server repository, running on geographically dispersed networked computers. These spawned WIP control instances are coordinated and synchronized using the concept of virtual production lines (VPLs). WIP control algorithms for implementing real time “pull” operations, leveraging WIP levels, and resolving resource sharing are proposed. Certain validation has been conducted in an industrial testbed to confirm the applicability of the proposed VPL-based WIP control solution.

As semiconductor business environments are challenged by the rapid changes in business needs and customer demands, customer satisfaction becomes one of the most important factors for manufacturers to stay ahead of competition. One of well-recognized strategies for providing customers with satisfactory services is to continuously improve manufacturing and service responsiveness. A proven successful means in improving the responsiveness of manufacturing systems is to deploy an efficient and effective shop floor work-in-process (WIP) control and management system, which controls the material flow, monitors the WIP level for each machine/stocker, tracks the statuses of materials on the shop floor in a timely way, and also real time responds the requests from the shop floor. It is the availability of pertinent and real time production data that timely well-informed decisions could be made and satisfactory customer services provided. A typical advanced semiconductor manufacturing control and management system is of a hybrid nature (Fig. 1). Between cells,1 a heterarchical architecture nature takes place; while within a cell it locally deploys a hierarchical architecture. A manufacturing execution system (MES) is a shop floor information system that provides floor production control, WIP control and management, and floor accounting functions, and acts as the interface between enterprise planning (i.e., office-level information systems) and shop floor execution (e.g., cell controllers). A cell functions as an autonomous manufacturing entity. Each cell consisting of a group of equipment in the same area may accomplish one or several successive manufacturing processes. Between cells, a material control system performs the shop level material flow control and tracking. The material control system ensures that all the necessary materials will be delivered to cells in an optimal manner. Within a cell, a cell controller supervises a group of equipment. It first refines all the task assignments from the MES. It then executes the refined tasks through coordinating the activities of all the equipment group controllers (EGCs) within the cell. An EGC is considered as an equipment software driver, which provides the interface between a cell controller and physical machines. EGCs command all the equipment to accomplish the detailed manufacturing operations as assigned. An internal material transport system receives material delivery commands from the WIP control and optimally delivers the materials to the right equipment/storage at the right time. Currently most semiconductor manufacturing fabrication facilities use a cell/process layout configuration (Matsuyama and Niou, 1993). An MES2 is configured using one or a few predefined “super production lines” based on such a facility resource model, equipment proximate information, and administrative organization structures (White et al., 2000; SiView, IBM SiView standard). Materials are usually released in batches. For each batch, when one process is done, operators check the MES and get a list of available equipment for next process. If there are many choices, simply it is at operators’ or production managers’ discretion to pick equipment to perform the process. For instance, SiView is a leading centralized MES product deployed by many semiconductor manufacturers (White et al., 2000; SiView, IBM SiView standard). It utilizes client/server architecture. On the client side, SiView provides a thin GUI interface; the major functions provided for users are (1) getting information on lot, consumables, setup, tooling, and process specifications before a process gets started and (2) recording data on the process, consumed consumables, and quality assurance as the process gets completed (White et al., 2000). On the server side, SiView includes best manufacturing practices and application scenarios ranging from MES specification manager and material manager to web reporting. With the quick advance of semiconductor technologies, semiconductor manufacturing processes have become more and more complicated. The resultant manufacturing systems become more competent but also complicated, giving rise to more intensive and complex equipment setup and recalibration in production transitions (Blose and Pillai, 2001). Consequently, a currently and widely adopted centralized WIP management system that runs as a mammoth and monolithic application is losing its designated performance due to the explosive growing need of data acquisition and analyses. In addition, inflexibility and lack of good reconfigurability and scalability are also the drawbacks of a centralized WIP control system (Sturm et al., 1999; Qiu et al., 2002). With the help of advanced network computing technologies, production controls could be managed in a more flexible and agile manner without making a physical change on the shop floor. In other words, as the business environment changes, different management and operational methods can be applied to the same physical configuration of manufacturing systems. Group technology has been adopted by many other manufacturing industries to reduce machine setup and production cycle time (Dowlatshahi and Nagaraj, 1998; Heragu, 1994). It has been widely used to design manufacturing cells and facility layouts and optimize production schedules, but seldom considered in the implementation of WIP control in semiconductor manufacturing. This article proposes an approach to distributed WIP controls by incorporating the group technology philosophy, which makes the implementation scaleable in the sense that is adaptable to current business natures: (1) a continuous growth of manufactures’ productivity in terms of manufacturing capabilities, equipment complexities, and the size of manufacturing system; and (2) a dynamic changing market in light of a variety of product mixes. The remainder of this article is organized as follows: Section 2 briefly discusses a typical back-end semiconductor manufacturing system and how production is operating under the control of existing WIP control and management applications. Section 3 reviews the concept of virtual production lines (VPLs) and discusses how it could improve the responsiveness and flexibility of a manufacturing system; Section 4 proposes the synchronization mechanism and “pull” algorithm for VPLs based WIP control system; Section 5 presents an implementation of the proposed distributed WIP control solution and validation results from simulation runs; finally, Section 6 provides conclusions for this article.

This article proposed a methodology for the design of a distributed WIP control and management application for semiconductor manufacturing. The concept of VPL is applied, which makes a process layout manufacturing system capable of running like a product-family type layout facility. Thus the controlled system could take advantage of both types of layouts, while overcome the drawbacks of a process layout, such as high-level WIP, long production cycle time, and low machine utilization. We discuss the algorithms in supporting the proposed solution, for instance, VPL synchronization mechanisms and WIP control algorithms. The validation of the applicability of the proposed solution is given in light of the improvement of machine utilizations on the average by 7.5% in a test case. From the control and management perspective, through leveraging WIP level and applying “pull” material flow control, the investigation provides a potential solution to the resolution of more complex and dynamic manufacturing system aimed at meeting the future need: reconfigurability, scalability, and system responsiveness. Note that it is still in a preliminary stage. Many relevant issues should be further investigated, for example, more effective algorithms for resolving machine conflicts and setting up WIP cap (i.e. Cijk), a test in a large-scale production environment, and system reliability issues due to the increased complexity of computing technology required in the proposed solution.