The sensitivity of rule combinations for scheduling in a hybrid job shop.

Introduction

Domestic manufacturing firms, large and small, have been under constant pressures of global competition, the competitive pressures include the need to remain at the cutting edge of flexible, automated and computer controlled equipment. Due to many reasons, either financial, technological, or operational, the development of any manufacturing system is incremental. The periods for implementation of these systems may vary from months to years. During these long periods of incremental implementation, combinations of traditional equipment with advanced numerically controlled flexible machines are often used. This gives rise to the hybrid manufacturing environment. In a hybrid manufacturing environment, strategic issues such as product design, tactical issues involving shop floor layout, and operational considerations of scheduling and sequencing must be dealt with. This article will focus on the operational scheduling aspect of the hybrid environment of a job shop.

Much of the research literature tends to look at scheduling problems in purely dedicated job shop or flow shop environments. However, the job shops of today are a combination of dedicated and flexible machines and some sequence of machines resembling a flow shop. To illustrate the issues of production scheduling in a hybrid manufacturing environment, a hypothetical small manufacturing firm is used. The production environment is detailed in the section on research methodology, however, the shop is a combination of conventional machine tools alongside flexible manufacturing technology in a job/flow shop environment.

The performance of a job shop is usually measured in terms of the level of work-in-process (WIP) inventory, its on-time performance and the throughput. In an attempt to improve performance, there is a move to switch from a process to a product oriented layout. However, this involves purchase of additional as well as new automation technologies that can be costly. Some process oriented job shops have been converted, at a very high cost, to product oriented layouts, with the improvements being nominal. The return on investment is too slow and the process of conversion is too lengthy to be cost effective. This article will show that the performance of a process oriented hybrid job shop depends on the turnover in the shop, which means that performance is a function of the scheduling rule, which itself is a function of the variation in job types. A simulation model of a job/flow shop is developed to study the performance of despatching rule combinations under varying arrival rate/processing time scenarios. It is hypothesized that rule combination performances will vary with arrival rate/processing time scenarios in a hybrid shop. It is further hypothesized that for the success of JIT objectives, e.g., minimizing WIP, minimizing early or late job completions, maximizing throughput, one needs to control arrivals and processing times.

Very little work has been done on scheduling in a hybrid job-shop environment. A brief but exhaustive literature review is presented in the next section. Section four presents the job shop environment and the model to be simulated. Subsequent sections present the scheduling rules evaluated, and the performance criteria they are evaluated on. The experimental design for the study is presented in section seven followed by verification and validation of the simulation model. In the last two sections the study results are presented.

Literature review

Numerous literature surveys and expositions have been written concerning scheduling and sequencing models in various production environments[141]. Discrete simulation has been used extensively for testing dynamic despatching rules. It has been shown by researchers that a combination of simple despatching rules in many cases works better than individual despatching rules[12,13].

In the scheduling literature, the conventional processing environment has been decomposed into four major categories:

(1) single stage-single machine; (2) single stage-parallel machine; (3) flow shop; (4) job shop.

The single stage-single machine problem is the simplest and most mathematically tractable of the scheduling environments. In this case all jobs require only one operation which has to be performed on a single machine. These types of problem although unrealistic, help to provide insights into more complex production systems. Good heuristic techniques could be derived from optimum seeking procedures of the single-machine case.

The single stage-parallel machine environment requires the scheduling of jobs that require a single operation on a set of machines capable of performing it. This problem can be further classified into classes of identical, uniform or unrelated machines. The processing times are the same for all identical machines, proportional times for uniform machines and independent processing times for unrelated machines. Study of such systems can benefit the development of models for an automated manufacturing environment where multiple routings are possible.

In the flow shop environment, jobs are modelled to go through multiple stages of processing, where each stage is served by a single machine. It is assumed that all jobs go through the same sequence of manufacture, …