Elsevier

Computers in Industry

Volume 64, Issue 6, August 2013, Pages 694-707
Computers in Industry

Multiple-objective scheduling for interbay AMHS by using genetic-programming-based composite dispatching rules generator

https://doi.org/10.1016/j.compind.2013.03.009Get rights and content

Highlights

  • A mathematical model for the scheduling of interbay AMHS is formulated.

  • Several system parameters are simultaneously considered for multiple-objective optimization.

  • The developed approach has better performance in most scenarios compared with other methods.

  • The proposed genetic-programming-based CDR generator significantly improves the computational efficiency.

Abstract

Semiconductor wafer fabrication system (SWFS) is one of the most complicate discrete processing systems in the world. As the wafer size grows from 200 to 300 mm and then to 450 mm in recent years, the interbay automated material handling system (AMHS) has been widely adopted. How to improve the overall efficiency of AMHS has therefore become a crucial and urgent problem to wafer manufacturers. However, the large-scale, dynamic and stochastic production environment significantly substantiates the complexity of the scheduling problem for interbay AMHS. Aiming to meet the demand of multiple-objective optimization, composite dispatching rules (CDR) are applied. The system parameters, including wafer cassettes due date, waiting time, and stocker buffer status are simultaneously considered. In order that the composite dispatching rules can be used in real-life dynamic production, a genetic programming based CDR generator is proposed. Discrete event simulation models are constructed using the eM-Plant software to simulate the 300 mm SWFS. The numerical study indicates that by using the generated composite dispatching rules the transport efficiency is improved, meanwhile, the wafer throughput is increased and the processing cycle time is shortened. The experimental results also demonstrate that the GP-based generating algorithm is effective and efficient for a dynamic environment. Further comparisons with other scheduling methods show that the proposed approach performs better in most scenarios.

Introduction

Due to the great market potential, semiconductor manufacturing has becoming one of the most rapidly developing, capital-intensive and technology-intensive industry in nowadays world [1]. As one of the most complicate discrete processing systems, semiconductor wafer fabrication system (SWFS) has distinct features such as reentrancy, mass and mixed products in process, flexible product routes, coexistence of single-processing and batch-processing equipment, and long production cycle times [2]. These characteristics lead to a high level of wafer cassette transportation within semiconductor fabrication lines [3]. In wafer fabrication, a certain quantity of wafers is grouped in a standard container which is called a wafer cassette. In a typical 300 mm semiconductor fabrication line, there can be as many as 5000 wafer cassettes waiting for about 100 types of tools to finish thousands of manufacturing steps [4]. Each wafer cassette repeatedly goes through deposit film, pattern film, etch film, wafer testing and cleaning process 20–40 times, about 200–600 operations totally [5]. The move between operations requires transportation, thus each wafer cassette travels approximately 8–10 miles to complete all the manufacturing operations. The material handling cost has already stand for 20–50% of the total manufacturing cost, and the material handling in semiconductor fabrication line becomes critically important [6].

With the advent of 300 mm and 450 mm wafer fabrication in recent years, the automated material handling system (AMHS) has been widely adopted and become essential to maintain high tool utilization and guarantee system performances in terms of wafer cassettes’ cycle times and throughput. From the survey of Kaemph [7], the automated transport systems in the wafer fab were evaluated. The simulation results showed that the automated transport of wafers could reduce the transport time and could increase the throughput as well. The main benefits of AMHS in SWFS were also concluded by Davis and Weiss [8].

AMHSs must be embedded with effective and robust material handling policies, so as to maximize the effectiveness of such AMHSs. Typically, there are two types of AMHS in SWFS: one is the interbay system, and the other is the intrabay system (also called tool-to-tool AMHS). The interbay material handling systems transport wafer lots between process bays such as etching bays, diffusion bays, photolithography bays, and so on; the intrabay material handling systems transport wafer lots within one process bay [9]. Compared with intrabay material handling, interbay material handling involves larger number of automated vehicles, more complex transportation tasks, and more stochastic transportation demands with higher risk of transportation deadlocks and vehicle blockages.

Over the last few years, automated vehicle dispatching in material handling systems have received considerable attention [10], [11], [12], [13]. However, most of traditional vehicles dispatching approaches are usually for single-objective optimization, which only consider the wafers’ process time and waiting time, vehicles’ location and travel time, and so on. In order to satisfy the demand of multiple-objective optimization, some other factors which radically influence the overall system performance and customer satisfaction, such as factors regarding wafer cassettes’ priority, have to be considered. On the other hand, the interbay material handling system is a dynamic and stochastic system in real-life semiconductor fabrication lines. Static attribute weights in traditional dispatching rules limit the ability for the material handling systems to manage unexpected environmental changes. Therefore how to deal with the stochastic events such as the unexpected vehicle blockages needs to be answered.

This article investigates the scheduling problem of interbay AMHS in a 300 mm wafer fab. A multiple-objective scheduling model of interbay AMHS is established. Aiming to meet the demands of dynamic adjusting and multiple-objective optimization, a genetic programming (GP) based algorithm is proposed to generate composite dispatching rules (CDR). By using the CDR the waiting cassettes (tasks) are prioritized and assigned to vehicles. Simulation models are built to evaluate the performance of generated composite dispatching rules.

The rest of the article is organized as follows. In Section 2, a brief literature review of scheduling for interbay AMHS is addressed. Section 3 presents the definition and formulation of the scheduling problem of interbay AMHS. In Section 4, a GP-based generator for composite dispatching rules is proposed to realize the multiple-objective and dynamic scheduling for interbay AMHS. In Section 5, simulation experiments and numerical results are reported and discussed. Finally, conclusions and recommendations for future work are outlined in Section 6.

Section snippets

Literature review

It is shown in Fig. 1 which an interbay AMHS typically consists of three subsystems: transportation rails, stockers and automated vehicles. The transportation rails are usually monorail systems, where automated vehicles move wafer lots along the rails between different stockers. Each stocker system acts as an input/output port and provides temporary storage of wafer lots for corresponding intrabay systems. Automated vehicles are used to transport wafer lots between stockers. A detailed

Problem description

In this article, a typical single-loop, spine-type interbay AMHS adopted in a 300 mm foundry fab is discussed. The layout of the material handling system is described in Fig. 2.

During the production of wafer fabrication, interbay AMHS is transporting the wafer cassettes between different bays in accordance with the processing requirements. Once a cassette needs to move to the destination stocker, a vehicle is requested. The cassette faces a decision point to determine which vehicle has the

GP-based CDR generator GP-based composite dispatching rules

In this section, a genetic programming (GP) based generator to develop composite dispatching rules (CDR) is proposed.

Genetic programming is an evolutionary algorithm-based methodology inspired by biological evolution to find computer programs that perform a user-defined task [38]. It is a specialization of genetic algorithms (GA) where each individual is a computer program.

In this article, GP is used to generate the composite dispatching rules to optimize the interbay material handling system.

Simulation model and experimental design

To evaluate the composite dispatching rules generated by proposed GP-based algorithm, discrete event simulation models are constructed using the eM-Plant software. The layout of the 300 mm wafer fab is shown in Fig. 9. There are 14 bays, 127 sets of equipment for 23 types of process (Table 3). Such data are obtained from a local semiconductor manufacturer in Shanghai. The layout parameters of the interbay material handling system are summarized as follows:

  • Interbay rail loop: 370 m;

  • Length of the

Conclusion

The characteristics of SWFS, such as re-entrant production flow, long lead-time, enormous process complexity and WIP, capital-intensive tools and unforeseen interruption make interbay material flow control very challenging. The vehicle dispatching problem of interbay material handling system is complex due to large-scale manufacturing environments, dynamic, stochastic and mass transportation demands, random transportation deadlocks and vehicle blockages.

This article developed a GP-based

Acknowledgments

The authors would like to acknowledge financial supports of the National Science Foundation of China (No. 60934008, No. 51275307) and the National High Technology Research and Development Program of China (No. 2012AA040907).

Wei Qin is currently a postdoctoral fellow at School of Mechanical Engineering, Shanghai Jiao Tong University in China. He received the BS degree from Shanghai Jiao Tong University, China, in 2004 and the MS degree from Tsinghua University, China, in 2006. In 2011, he received his PhD degree from The University of Hong Kong. His current research interests are production planning and control, supply chain management, and intelligent manufacturing.

References (45)

  • S.L. Chung et al.

    An overview of semiconductor fab automation systems

    IEEE Robotics and Automation Magazine

    (2004)
  • J.R. Montoya-Torres

    A literature survey on the design approaches and operational issues of automated wafer-transport systems for wafer fabs

    Production Planning and Control

    (2006)
  • J.A. Tompkins et al.

    Facility Planning

    (1984)
  • U. Kaemph

    Automated wafer transport in the wafer fab

  • F. Davis et al.

    Addressing automated materials handling in an existing wafer fab

    Semiconductor International

    (1995)
  • J.T. Lin et al.

    Simulation analysis of dispatching rules for an automated interbay material handling system in wafer fab

    International Journal of Production Research

    (2001)
  • K.J. Wang et al.

    Agent-based interbay system control for a single loop semiconductor manufacturing fab

    Production Planning and Control

    (2007)
  • U. Kaempf

    Automated wafer transport in the wafer Fab

  • J.J. Jang et al.

    An AGV routing policy reflecting the current and future state of semiconductor and LCD production lines

    International Journal of Production Research

    (2001)
  • T. Le-anh et al.

    On-line dispatching rules for vehicle-based internal transport systems

    International Journal of Production Research

    (2005)
  • G. Cardarelli et al.

    Simulation tool for design and management optimization of automated interbay material handling and storage systems for large wafer fab

    IEEE Transactions on Semiconductor Manufacturing

    (1995)
  • J.T. Lin et al.

    The maximum loading and the optimum number of vehicles in a double-loop of an interbay material system

    Production Planning and Control

    (2004)
  • Cited by (32)

    • Dynamic dispatching for interbay automated material handling with lot targeting using improved parallel multiple-objective genetic algorithm

      2021, Computers and Operations Research
      Citation Excerpt :

      The IPMOGA based solution strategy is compared with the previous scheduling methods in three performance indicators (average delivery time, maximum delivery time, and average processing cycle). In the relevant literature, the previous scheduling strategies such as the earliest due date (EDD) first, the first-encountered-first-serviced (FEFS) rule, and the reassignment based longest wait time (RLWT) first have been widely adopted in vehicle scheduling of interbay AMHS and can be considered as feasible and effective methods (Qin et al., 2013a, 2013b). Besides, a dynamic dispatching method based on modified Hungarian algorithm and fuzzy-logic based control (MHAFLC) developed in Qin et al. (2013a), and the genetic programming based composite dispatching rule GPRCG-R1 presented in Qin et al. (2013b) are also highly competitive.

    • Big data analytics for cycle time related feature selection in the semiconductor wafer fabrication system

      2020, Computers and Industrial Engineering
      Citation Excerpt :

      Wafer fabrication is the most complex phase of semiconductor manufacturing, with such characteristics as the large production scale, the diversity of machine types (Qin et al., 2013a,b), the complex process route (Lu et al., 1994) and highly dynamic environment of the semiconductor wafer fabrication system (SWFS) (Eling et al., 2013; Uzsoy et al., 2018).

    • A multi-attribute, rank-dependent utility model for selecting dispatching rules

      2018, Journal of Manufacturing Systems
      Citation Excerpt :

      In their study, the Pareto-front analysis is applied to assess the trade-offs among makespan, normalized total weighted tardiness, and mean absolute percentage error. Quin et al. [13] developed a multi-objective method for a semiconductor wafer fabrication system that combines GP with a simulation to generate composite dispatching rules. They compared the performance of the proposed method against traditional dispatching rules while taking into consideration nine simulated scenarios.

    View all citing articles on Scopus

    Wei Qin is currently a postdoctoral fellow at School of Mechanical Engineering, Shanghai Jiao Tong University in China. He received the BS degree from Shanghai Jiao Tong University, China, in 2004 and the MS degree from Tsinghua University, China, in 2006. In 2011, he received his PhD degree from The University of Hong Kong. His current research interests are production planning and control, supply chain management, and intelligent manufacturing.

    Jie Zhang is currently a professor at School of Mechanical Engineering, Shanghai Jiao Tong University in China. She received her PhD in mechanical engineering from Nanjing University of Aeronautics & Astronautics, China in 1997, her BS and MS degrees from Jiangsu University of Science & Technology, China in 1984 and 1991, respectively. From 1998 to 1999, she was a postdoctoral fellow in Department of Industrial and Manufacturing Systems Engineering, Huazhong University of Science & Technology, China. From 2000 to 2001, she was a research assistant in Department of Industrial and Manufacturing Systems Engineering, University of Hong Kong, China. In 2002, she joined the School of Mechanical Engineering, Shanghai Jiao Tong University in China as an associate professor. Her current research interests are scheduling and planning in manufacturing systems, performance analyzing of manufacturing systems, and supply chain management.

    Yinbin Sun received the BS degree from Jiangsu University, China, in 2008 and the MS degree in mechanical engineering from Shanghai Jiao Tong University, China, in 2011. His current research interests are automated material handling system scheduling, complex manufacturing system modeling.

    View full text