Selection of a pull production control policy under different demand situations for a manufacturing system by AHP-algorithm

https://doi.org/10.1016/j.cor.2008.03.006Get rights and content

Abstract

A multistage serial production system is considered in the present paper. A generalized model has been developed with the use of probabilistic demand situations for the end product. The demand situations considered are binomial, exponential, lognormal and Poisson. These demand patterns are used as input parameter for various production control policies. The output values for performance parameters are obtained by simulation. The production policies analyzed are Kanban, CONWIP and Hybrid as alternatives for controlling the engineering manufacture. In the presence of several performance measures, it often becomes difficult for the management to select the most appropriate policy. Analytic hierarchy process (AHP) has been implemented for the purpose of selection. Computational results have been reported along with the sensitivity analysis after designing and conducting various experiments.

Introduction

In a highly competitive environment, manufacturing organizations need a suitable production policy for the control of the production line so that the resources including machines as well as the buffer can be better utilized in conducting the manufacturing activities. The best possible performance in terms of better utilization of resources can be achieved by reducing non-productive time and increasing the customer service level, maintaining throughput and reducing the work-in-process (WIP) inventory from the implemented policy. It is very difficult to achieve the entire objectives by a manufacturing unit. Therefore, one has to decide on the trade-off among the performance measurements. This is again a difficult task in terms of both understanding the nature of the problem to find the most effective production control policy and its logical selection methodology that develops from this understanding.

With the emergence of just-in-time (JIT) manufacturing, production control systems that react rapidly to actual occurrences of demand are gaining importance. Several pull-type control policies e.g. Kanban, CONWIP (Constant WIP) and Hybrid have been proposed by the experts, but it is very difficult to quantify how good these policies are as well as understand the performance parameters that make them desirable.

Earlier, some of the researches demonstrated only the comparative performance results of a manufacturing system in different production control policies’ environments, but no research paper is found which shows the trade-off among the parameters and selection of the best policy in different stochastic demand scenarios. Therefore, in this paper, a new simulation model integrated with AHP has been proposed in order to present experts with a set of choices so that they can decide on the trade-off. Simulation programs are used to predict the performance of the considered pull policies with different demand patterns as input characteristics. The application of the AHP-algorithm calculates the performance weight of the alternatives to show the effectiveness of the production system. Thus, the combined model of AHP and simulation provides a complete solution to the considered selection problem.

In other words, outputs of the proposed simulation model are incorporated with AHP, and AHP is used to rank the alternatives. For this purpose, we consider a single-item, multi-stage serial and unreliable manufacturing system with internal buffer storage between N serial workstations which is needed to support the smooth operation of manufacturing.

Chan [1] demonstrates that there are several conflicting performance criteria which are used to determine the optimal production control policy. Among these are: service level, throughput, WIP, lost demand, machine utilization, etc. It is desirable to have high level of throughput and customer service by the maximum utilization of the machine but at the same time, WIP and production costs should be low with the minimum utilization of the buffer to reduce the waiting cost. These goals are in conflict and affect each other because throughput and service level or fill rate can be maximized only after maximization of utilization of the machine by adding more buffer inventories at high waiting cost [2].

In the case of single product manufacturing system, all the machines need to be set up once; hence, the set-up time is insignificant as compared to the total processing time. Thus, the batch size and the amount of set-up time will not seriously affect the quantity of product to be processed. Therefore, we try to observe the performance with respect to significant parameters.

The objective of the present paper is to develop a new generic model with different stochastic demand situations. Several performance measures have been incorporated in the study. In the referred relevant literature, there is a separate comparison of the policies. Selection of the best production control policy is difficult in the presence of several performance parameters/criteria. Another contribution of the present paper is the application of the analytic hierarchy process (AHP) in the discussed environment.

The complete paper has been divided into a number of sections: Section 2 describes various pull production control policies i.e. Kanban, CONWIP and Hybrid. Section 3 presents literature review related to control policy, AHP and its applications in the field of manufacturing. Section 4 presents the considered manufacturing system with assumptions and performance parameters. The proposed simulation model with problem statement and generation of data has been described in Section 5. Selection by AHP methodology and Results analysis have been presented in Section 6. For the effects on the relevant output parameters, sensitivity analysis is carried out and described in Section 7. Section 8 concludes the paper.

Section snippets

Production control policies (PCP)

On the basis of circulating kanban, pattern of information flow, control characteristics and the resulting behavior of inventory, the pull control policies are studied exclusively for the production line of a manufacturing industry. These are as follows:

Literature survey

There are a number of existing studies about comparing of Kanban & CONWIP, CONWIP & hybrid, kanban, CONWIP and hybrid systems. Several authors have shown comparisons of the policies through both simulation and analytical models considering various performance parameters. In a flow line model based on an actual system in a Toyota assembly factory, Bonvik et al. [9] showed the comparison in some specific situations. While comparing the production policies, the hybrid control policy demonstrated

The manufacturing line model

As shown in Fig. 1, the proposed manufacturing system is governed by probabilistic demand. Various demand patterns like binominal, exponential, lognormal and Poisson have been considered. Typically, binomial distribution is used where a single trial is repeated over and over. Exponential distribution is used to represent the service times of a specific operation. The lognormal distribution can be sampled very efficiently and can represent low variability distributions without numerical

Experimental design

In this section, we present an explicit stochastic production control policy selection problem related to the proposed manufacturing system as shown in Fig. 1 (Section 4) with four workstations. The proposed manufacturing system follows various demand patterns like binominal, exponential, lognormal, and Poisson with block seed=3, standard deviation=1.0 and mean=1.1 giving n (number of trials)=12, p (probability of the event occurring)=0.0917 where p(x)=[n!/{x!(n-x)!}]px(1-p)n-x for binomial

Selection by AHP methodology

It is very difficult to predict which PCP for the proposed system could be implemented; parameters are in conflict with each other. Here, eight significant parameters have been identified for the optimal selection process: SL(t), WIP(t), TH(t), LD(t), TC(t), MU(t), BU(t) and BC(t). An attempt is made in this section to identify and quantify the relationship among the parameters for the optimal choice of control policy by initializing the AHP. The basic steps for using AHP are same in all the

Sensitivity analysis

Many parameters in the system dynamics model represent quantities that are very difficult to measure to a great deal of accuracy in the real world. After building a system dynamics model, the modeler is usually at least somewhat uncertain about the parameter values he chooses. Therefore, in this section, with the help of sensitivity analysis, we provide an evaluation of the confidence in the model, possibly assessing the uncertainties (or what if) associated with the modeling process and the

Conclusion

After identification of a need for the appropriate generic simulation model, the probabilistic demand situations, i.e. binomial, exponential, lognormal and Poisson have been incorporated in the context of a multi-stage unreliable serial production system with manufacturing blocking concept. This paper has investigated the behaviors of kanban, CONWIP and hybrid systems and compared the performance of a manufacturing system with respect to different performance criteria, such as average

References (41)

  • T. Ertay et al.

    Integrating data envelopment analysis and analytic hierarchy for the facility layout design in manufacturing systems

    Information Sciences

    (2006)
  • L. Rabelo et al.

    Value chain analysis using hybrid simulation and AHP

    International Journal of Production Economics

    (2007)
  • E.W.T. Ngai et al.

    Evaluation of knowledge management tools using AHP

    Expert Systems with Applications

    (2005)
  • S. Tayur

    Structural results and a heuristic for Kanban control serial lines

    Management Science

    (1993)
  • W.J. Hopp et al.

    Factory physics

    (1996)
  • M.L. Spearman et al.

    CONWIP: a pull alternative to Kanban

    International Journal of Production Research

    (1990)
  • S.G. Koh et al.

    Comparison of DBR with CONWIP in an unbalanced production line with three stations

    International Journal of Production Research

    (2004)
  • Dallery Y, Liberopoulos G. A new Kanban-type pull control mechanism for multi-stage manufacturing systems. In:...
  • A.M. Bonvik et al.

    A comparison of production-line control mechanisms

    International Journal of Production Research

    (1997)
  • M.L. Spearman et al.

    Push and pull production systems: issues and comparisons

    Operation Research

    (1992)
  • Cited by (44)

    • The state-of-the-art integrations and applications of the analytic hierarchy process

      2018, European Journal of Operational Research
      Citation Excerpt :

      To enhance the fitness of AHP, the scholars applied the simulation-based approach to build a group consensus instead of forming point estimates that were aggregated from individual preference judgments. Sharma and Agrawal (2009) developed the integrated AHP-simulation approach to evaluate the performances of production control policies. Simulation programs were used to predict the performance of policies with different demand patterns.

    • Development and evaluation of a Basestock-CONWIP pull production control strategy in balanced assembly systems

      2018, Simulation Modelling Practice and Theory
      Citation Excerpt :

      The advantages of these pull PCS lie in their ability to eliminate waste, reduce WIP and improve customer service levels [13]. According to previous studies on PCS that were performed in single product systems [7,8,14,24]; it is shown that pull control strategies (KCS, BSCS, CONWIP and HK-CONWIP), achieved high service levels and low WIP levels which led to a reduction in production cost, inventory, lead time and an increase productivity and fill rate. In multi-product systems, pull PCS can be operated by two different policies called Kanban Distribution Policies (KDP); Shared Kanban Distribution Policy (S-KDP) or Dedicated Kanban Distribution Policy (D-KDP) (Bayant et al. [2], Faccio et al. [6] and Lolli et al. [16]).

    • Evaluation of multi-product lean manufacturing systems with setup and erratic demand

      2015, Computers and Industrial Engineering
      Citation Excerpt :

      This issue of product volume variations often results in flow line congestion, long lead times and low throughput rate in MPLME, such that the performance goals and the principle of lean manufacturing is undermined. A majority of the studies on pull production control strategies considered single-product production lines (e.g. Ang, 2015; Deleersnyder, Hodgson, King, O’Grady, & Savva, 1992; Geraghty & Heavey, 2005; Kleijnen & Gaury, 2003; Koh & Bulfin, 2004; Krejewski, King, Ritzman, & Wong, 1987; Lee, 1989; Sharma & Agrawal, 2009; Spearman, Woodruff, & Hopp, 1990; Weitzman & Rabinowitz, 2003; Özbayrak, Papadopoulou, & Samaras, 2006). These studies assumed that research results on single-product lean manufacturing environments are scalable to multi-product lean manufacturing environments.

    • Environmental performance evaluation of implementing EMS (ISO 14001) in the coating industry: Case study of a Shanghai coating firm

      2014, Journal of Cleaner Production
      Citation Excerpt :

      Developed by Saaty (1977), AHP is a decision-making tool for analyzing complex problems with multiple criteria. The literature on AHP in various applications is very rich (Chan and Chan, 2010; Chang et al., 2007; Dagdeviren, 2008; Mendoza, 1997; Sarminento and Thomas, 2010; Sharma and Agrawal, 2009; Wind, 1987). In this paper, the fuzzy comprehensive evaluation combined with the AHP and membership degree analysis was employed as the study methodology.

    • Multi-Criteria decision-making methods in manufacturing environments: Models and applications

      2023, Multi-Criteria Decision-Making Methods in Manufacturing Environments: Models and Applications
    View all citing articles on Scopus
    View full text