Elsevier

Computers in Industry

Volume 43, Issue 3, December 2000, Pages 237-248
Computers in Industry

Logistics models in flexible manufacturing

https://doi.org/10.1016/S0166-3615(00)00069-5Get rights and content

Abstract

An integrated modeling approach that considers the overall production schedule is needed in order to effectively manage different material flows in a flexible manufacturing system (FMS), where large amounts of data intervene in the dynamic control and decision making process. This study focuses on the development of an integrated FMS control model that includes essential features, such as routing of simultaneously processed work orders and batch dispatching, as well as dynamic vehicle path determination and conflict-free routing. A logistics-oriented modeling methodology for FMS distributed control design is proposed that provides the capability for rapid development and evaluation of the control policy.

Introduction

A high degree of flexibility and quick response times have become essential features of modern manufacturing systems. Flexible manufacturing systems (FMS) allow for more efficient use of resources in terms of increased machine utilization, reduced work-in-progress inventory, increased productivity, a reduced number of machine tools, lower labor costs, shorter lead times and less floor space [1], [3]. This is obtained, however, at the expense of more complex control of these systems. Some of the most recurring and important control tasks are those related to scheduling and dispatching, because it is commonplace in an FMS environment — where many different types of work orders, in varying batch or lot sizes, are produced simultaneously — to find jobs competing simultaneously for the same resource [14], be it an intersection in an automated guided vehicle system (AGVS), machine tool time or access to an automated materials handling system.

In order to maximize total productivity (considered as a mixture of various measures, such as net profit, time delay, and inventory level at local input buffer) of the pre-fixed work orders and their lot sizes, the following issues have to be taken into account.

  • Selection of the best configuration at the design stage.

  • Selection of an optimal route and workload balancing at the production planning stage.

  • Identification of the most efficient scheduling and/or dispatching strategies at the operation stage.

  • Handling the information interchange between the components of the distributed manufacturing environment, and the material and information flows.

Considering the complex nature of the above problems and their domain-dependent solutions, it is virtually impossible to satisfy the requirements, such as manufacturing cycle time, delivery time and batch size limitations, of several simultaneously processed work orders and, at the same time, to decide on the optimal routings, workload balancing, scheduling rules and control strategy. Consequently, a research approach that has been recently followed is to develop a framework for considering all this issues in a unified manner. Examples of such a unified approach include scheduling for a materials handling system [13], as well as work-order routing and workload balancing [6]. This approach offers a broader perspective for evaluating system performance and for describing possible schemes for managing the workflows within the system [10].

In order to characterize production processes from a logistic point of view, close attention has to be paid to the problem of synchronized manufacturing [12]. The concept of synchronized manufacturing is built around the postulate that the shortest throughput time gives the maximum profit rather than following a more traditional view where resources must be kept busy all the time. Its main idea derives from the optimized production technology (OPT) method that predicts where bottlenecks will appear and uses them to control output [11], [15]. In other words, instead of balancing capacity, which is inefficient, the FMS should be a balancing flow. Flow balancing, in turn, guarantees that all resources are produced at the same relative rate while no resource is overloaded, resulting in minimum inventory.

In this study we propose some generalizations of previous contributions to the concept of synchronized manufacturing [4], [5], [8], [16]. We emphasize, in particular, the role of a logistics approach to process control (i.e. the role auxiliary processes such as transportation, materials handling and diagnostics repair processes play in flexible manufacturing) from the point of view of distributed bottleneck (i.e. critical resource) control. The main objective of the study is the development of conditions sufficient for a cyclic steady-state operation of a system composed of a set of sequential cyclic processes, and not only of the regular mesh-like structures [9]. The discussion focuses on the conditions guaranteeing the cyclic steady-state behavior of an FMS, which provide a formal basis for the distributed control strategy. The problem of distributed system control is seen as a problem of defining a set of rules (laws) that locally constrain the way the distributed processes (e.g. workflows, AGVS, repair processes) interact with each other so as to guarantee the desired performance of a whole system. The resulting system may be considered as a self-synchronized system, by which we mean a system capable of returning to a unique steady state from any state it was forced into as a result of an accidental disturbance. The self-synchronization paradigm allows us to construct an FMS that has the required functional properties, and opens up the possibility of building distributed systems that can be dynamically rescheduled.

Section snippets

Logistics of materials flow

Let us consider an FMS composed of workstations linked by an AGVS as shown in Fig. 1. Four kinds of products are processed along with the following production routes.P1=B1,M1,B3,M2,B4,P2=B6,M3,M1,B5,M4,P3=B2,M4,B3,P4=B2,M5,B3

Each machine and buffer is served by a dedicated set of AGVS. Each particular workflow follows a unique path in the AGVS network. The routes RPi, i=1,4 of AGVS dedicated to particular workflows are as follows.RP1=A,B,RP2=D,C,B,RP3=E,F,B,D,RP4=E,G,B,D

Since the workflows

Problem statement

Consider a workflow model of an FMS system, as shown in Fig. 4, composed of a set of sequential repetitive processes competing for access to shared resources. Because alternative process executions lead to different values of performance measure, the problem of selecting an appropriate realization is crucial. The problem can then be formulated in terms of finding and assignment of a relevant dispatching rule (resource access protocol) to a set of shared resources that will guarantee the

Bottleneck control framework

A number of analytical and simulation models dealing with problems of prediction and verification of system performance (such as throughput rates and product cycle times) have been developed to date [2]. Many of these models, while effective in modeling various interrelated subsystems, fail when dealing with performance evaluation of a whole system that depends on the effectiveness of the component elements as well as on the synchronization of their interactions.

Conclusions

The common features of the class of system consisting of a set of sequential and repetitive processes specified by fixed routing are complexity and the lack of adequate performance evaluation tools. The methodology proposed in this study integrates workpiece flow structure design, buffer capacity assignment, and allocation of the dispatching rules that control the workflows. Conditions for designing the procedures that permit achievement of the required performance of the system while

Acknowledgements

The authors wish to acknowledge the support of Natural Sciences and Engineering Research Council of Canada, grant no. OPT 9337.

Professor Z.A. Banaszak received the MS and PhD degrees in control systems from Wroclaw University of Technology in 1973 and 1977, respectively. He received the DSc degree in automation and robotics from Paton’s Institute of Welding, Ukrainian Academy of Sciences, in 1989. Presently he is the Head of the Department of Computer Science and Management. He has international experience from the Tokyo Institute of Technology (1981–1983, 1989), Technical University of Kiev (1987–1988), Edinburgh

References (16)

  • M. Shah et al.

    A production order-driven AGV control with object oriented implementation

    CIM Systems

    (1997)
  • J. Agnetis, A. Alfieri, P. Brandimarte, P. Prinsecchi, Joint job/tool scheduling in a flexible manufacturing cell with...
  • F. Baccelli, G. Cohen, G.J. Olsder, J.-P. Quadrat, Synchronization and Linearity (An Algebra for Discrete Event...
  • Z. Banaszak, Modelling of manufacturing systems, in: M. Zaremba, B. Prasad (Eds.), Modern Manufacturing: Information...
  • Z. Banaszak et al.

    On self-synchronization of cascade-like coupled cyclic processes

    Applied Mathematics and Computer Science

    (1993)
  • M. Fanti et al.

    System approach to design generic software for real-time control of flexible manufacturing systems

    IEEE Transactions on Systems, Man, Cybernetics Part A

    (1996)
  • H.-S. Ham

    Job route selection model for workload balancing between workstations in flexible flow line

    Production Planning and Control

    (1996)
  • http://www.iiz.pz.zgora.pl/wppt/iiz/crp/crp.html, The WWW presentation of the Department of Computer Science and...
There are more references available in the full text version of this article.

Cited by (23)

  • A novel 0-1 linear integer programming model for dynamic machine-tool selection and operation allocation in a flexible manufacturing system

    2012, Journal of Manufacturing Systems
    Citation Excerpt :

    In fact, their model selects machines, assigns operations of part types to the selected machines, and assigns material handling equipment to transport the part types from one machine to another machine. Banaszak et al. [13] extended a logistics models in an FMS that focuses on developing of an integrated FMS control model including essential features, such as routing of simultaneously processed work orders and batch dispatching. Persi et al. [14] represented a hierarchical approach that at the upper level batches are determined and at the lower level batches are sequenced, linked and scheduled.

  • Distributed control for systems of concurrent production flow

    2003, IFAC Proceedings Volumes (IFAC-PapersOnline)
View all citing articles on Scopus

Professor Z.A. Banaszak received the MS and PhD degrees in control systems from Wroclaw University of Technology in 1973 and 1977, respectively. He received the DSc degree in automation and robotics from Paton’s Institute of Welding, Ukrainian Academy of Sciences, in 1989. Presently he is the Head of the Department of Computer Science and Management. He has international experience from the Tokyo Institute of Technology (1981–1983, 1989), Technical University of Kiev (1987–1988), Edinburgh University School of Management (1992), Kuwait University (1992–1994), Carnegie Mellon University (1994), Hull University, Quebec (1995, 1998), Manchester Metropolitan University (1998). His current research interests include theory and applications of Petri nets, discrete event dynamic systems with applications to modeling, and performance evaluation of manufacturing and computer communication systems. He has authored or coauthored over 200 books and technical papers, and he is lecturing production systems design, logistics, and computer integrated manufacturing at the Technical University of Zielona Gora and the Technical University of Wroclaw.

M.B. Zaremba received the MS and PhD degrees in control systems from Warsaw University of Technology in 1972 and 1978, respectively. He is currently a Professor of Computer Engineering at the University of Quebec. He has had industrial experience in distributed control and vision-based robot control. His primary areas of research interest are hybrid systems integrating computational intelligence paradigms, real-time intelligent systems, high-performance distributed computing, and learning control. He has authored or coauthored over 120 books and technical papers. He is currently Associate Editor of Control Engineering Practice and member of the Advisory Board of Concurrent Engineering: Research and Applications. Dr. Zaremba has served in different capacities in several international conferences, symposia, and workshops, chairing CARs&FOF’94, and presiding the International Program Committees for IMS’97 and INCOM’98. He has been involved in the activities of International Federation of Automatic Control (IFAC) as Chairman of the Technical Committee on Advanced Manufacturing Technology (1993–1999) and recently as a member of the Technical Board. Dr. Zaremba is a Senior Member of IEEE, Fellow of ISPE, and a Registered Professional Engineer in the Province of Ontario, Canada.

View full text