WIP Level Research Articles

Controlling shop floor operations in a multi-family, multi-cell manufacturing environment through constant work-in-process

This paper discusses pertinent issues in applying CONstant Work-In-Process (CONWIP) principles to control shop floor operations in a manufacturing environment characterized by several product families processed along different routes in several production cells. The approach we take is to simultaneously answer two major questions: (1) what is the best WIP level? and (2) how to arrange the backlog list for a given system? The problem is posed as a mathematical programming model and solved via a simulated annealing heuristic. We design an experiment that captures essential elements of the systems under investigation. We then execute an extensive simulation to evaluate the effectiveness of various control schemes in a multi-cell, multi-family production environment. Specifically, we compare two variants of CONWIP control, one where containers are restricted to stay within given cells all the time and the other where containers are allowed to move through the entire system. We demonstrate the superiority of the latter in all the simulated scenarios.

Analyzing closed Kanban-controlled assembly systems by iterative aggregation–disaggregation

In recent years, many companies have begun using WIP-limiting strategies such as CONWIP, Kanban or drum-buffer-rope to control parts flows in complex production systems. This paper analyzes assembly systems with a tree structure, random processing times and a constant WIP control system. We present heuristic version of the exact aggregation–disaggregation theory for finite Markov chains to evaluate the performance of these closed Kanban-controlled assembly systems. Because the approximation is theory based, it provides a framework for further model development, with some possible extensions described in the paper. The approximation has the novel feature of doing simultaneous multiple partitions of the state space, in such a way that the associated aggregate transition rates are mutually consistent. The methodology is a novel approach towards extending aggregation ideas to fork-and-join queueing networks, and it provides several useful operational and analytical insights. Numerical comparisons with simulation show that the proposed approximation computes accurate estimates of the plant throughput. It provides a fast way to assess the performance and economic impact of changes in the total WIP level (or the number of Kanbans), or in the part routes, on the throughput rate of the assembly system. Scope and purpose We develop a new methodology for rapid performance analysis of assembly systems with a tree structure, random processing times, and a constant work in process Kanban control system. Extensive evaluations show that this methodology computes accurate estimates of the plant throughput. It provides a fast way for managers to assess the performance and economic impact of changes in the total WIP level (or the number of Kanbans) and in the part routes, on the overall throughput rate of the assembly system.

The construction of production performance prediction system for semiconductor manufacturing with artificial neural networks

The major performance measurements for wafer fabrication system comprise WIP level, throughput and cycle time. These measurements are influenced by various factors, including machine breakdown, operator absence, poor dispatching rules, emergency order and material shortage. Generally, production managers use the WIP level profile of each stage to identify an abnormal situation, and then make corrective actions. However, such a measurement is reactive, not proactive. Proactive actions must effectively predict the future performance, analyze the abnormal situation, and then generate corrective actions to prevent performance from degrading. This work systematically constructs artificial neural network models to predict production performances for a semiconductor manufacturing factory. An application for a local DRAM wafer fabrication has demonstrated the accuracy of neural network models in predicting production performances.

Setting WIP levels with statistical throughput control (STC) in CONWIP production lines

We develop a simple adaptive production control method for setting WIP levels to meet target production rates in a pull production operating under the CONWIP (Constant work in process) protocol. This method, termed Statistical Throughput Control (STC), uses real-time data to automatically adjust WIP levels (via kanban cards) in the face of noisy estimates of throughput. Because STC does not rely on a steady-state model, it is well-suited to systems subject to environmental changes such as those induced by continuous improvement efforts. Using simulation, we demonstrate the effectivness of STC under a variety of conditions, including single and multiple products, simple flow lines, routeings with shared resources and assembly systems.

Dispatching in a workstation belonging to a re-entrant production line under sequence-dependent set-up times

This paper deals with dispatching of jobs to a workstation, which is part of a re-entrant production line and which is operated under random arrival and sequence-dependent setup times. The problem is raised in a semiconductor plant in Israel. The suggested approach can be easily applied and integrated into the production line. Under the proposed approach, four heuristics were examined by an extensive simulation study. Results show good performance in terms of means of throughput, total set-up time and WIP level.

A dispatching scheme involving move control and weighted due date for wafer foundries

Increasing automation of wafer foundries in Taiwan has heightened the demand to automate dispatching decisions in recent years. To attain a better global performance, dispatching decisions should be made based on critical shop floor information. In this study, we present a dispatching scheme which combines move control and weighted due date concepts, genetic algorithm (GA) and simulation. The move control and weighted due date concept is applied to identify important dispatching decision factors respecting performance indices of due-date satisfaction and throughput. The GA is employed to derive proper weights for those decision factors. A simulation procedure capable of simultaneously considering product mix, reentrant process, and the influence of random events on the shop floor is developed to evaluate the system performance based on the derived weights of decision factors. A case study involving a simplified and revised data set from a Taiwanese wafer foundry is presented. Results show that the dispatching scheme proposed in this paper is effective and outperforms six conventional dispatching rules (FIFO, SPT, SRPT, EDD, SLACK, and CR) on performance measures of due date satisfaction, throughput, WIP level, critical machine utilization, and mean cycle time under both CONWIP and uniform release policies.

Lower In-Process Inventories and Better On-Time Performance at Tanner Companies, Inc.

In the early 1990s, Tanner Companies had a 74 percent on-time delivery rate and high levels of work in process. To increase on-time deliveries and raise customer service levels, the company focused on production planning and scheduling for the first time. In developing a short-term planning system, we began by designing a garment information system which provided the basis for an analytical heuristic-based scheduling model. During the development process, we actively involved users and managers at all levels to ensure employee support and smooth installation. As a result, the scheduling model was a big success. In one year, it decreased WIP levels by $200,000 and increased on-time deliveries to over 90 percent. The garment information and scheduling system has been well received and has become an integral part of daily operations at Tanner Companies.

On the availability of “relevant costs” information in operations management

The operations management (OM) function is, amongst others, held responsible for in-time delivery of customer orders against minimal costs. From a management accounting point of view, all decisions which are made to fulfil this function can be evaluated against relevant costs. This would imply that for each decision, management must take into consideration all possible variable costs and opportunity costs. At the operational level, variable costs consist of capacity costs such as work in overtime and inventory carrying costs. Opportunity costs (may) arise if promised delivery dates cannot be met. In practice, this OM-function is often (partially) performed by materials management, production management and/or physical distribution management. Moreover, in an engineer-to-order production situation, engineering management and related areas, (e.g. process planning) are also involved. In each stage of production, physical or non-physical, decisions concerning (internal) delivery dates, utilization of capacity and inventory may occur. Research was carried out to find out whether management is actually provided with relevant costs information to evaluate these (operational) decisions. Therefore, case study research in eight Dutch industrial firms was carried out. It concerned small to medium sized firms (80–250 employees) firms, all producing (rather complex) machines for industrial markets. Production control situations varied between assemble-to-order and engineer-to-order. The case studies were carried out in a highly structured manner. It occurred that all kinds of relevant costs information was missing or not known to the management in charge. Missing information concerned internal and external delivery performance, actual inventory and WIP levels, and actual utilization of capacity. Furthermore, it arose that all kinds of standards were unreliable. From a management accounting perspective, the results of the case study research indicate that relevant costs decision making is al least not very common in practice. If relevant costs are partially missing, managers will decide using other (irrelevant?) information. We feel that this might perhaps be an explanation for the growing discontent that the management accounting is confronted with.

Throughput of a constant work in process manufacturing line subject to failures

We consider a production line consisting of several machines in tandem operating under a constant work in process (CONWIP) control strategy. We assume that processing times are deterministic but machines are subject to exponential failures and repairs. We model this system as a closed queueing network and develop an approximate regenerative model (ARM) for estimating throughput and average cycle time as a function of WIP level. We compare ARM with mean value analysis (MVA) and develop readily computable tests of the suitability of the two approaches to a given production system. Through comparison with simulations, we show that ARM gives better predictions than MVA in a range of realistic situations.

Cyclic sequencing problems in the two-machine permutation flow shop: Complexity, worst-case, and average-case analysis

In this article, we formalize the cyclic sequencing problem in the two-machine flow shop. When jobs are processed in a repetitive cycle, the size of a scheduling problem is significantly reduced, and the resulting schedule is easy to implement because of its simplicity. Two types of cyclic sequencing problems are considered: the no-wait problem and the minimum-wait problem. The no-wait problem maximizes the throughput rate subject to the condition that there is no buffer space between the two machines. The minimum-wait problem minimizes the average WIP level subject to the conditions that the maximum throughput rate is maintained and that the FIFO dispatching rule is used in the intermediate buffer space. The no-wait problem is a well-known special case of the traveling salesman problem (TSP) and is polynomially solvable. The minimum-wait problem is shown to be NP-hard; therefore, we develop a heuristic procedure along with the analysis of its worst-case and average-case performance. Here, the average-case analysis is based on the expected length of the Hamiltonian tour for this special case of the TSP. The average-case analysis indicates that when the number of jobs in a cycle is small, the derived cyclic schedule yields a low WIP level.

A rule-based simulator for modeling Just-in-Time manufacturing systems (JITSAI)

The design details of a new rule-oriented simulator for modeling the production related issues in a Just-in-time (JIT) system are discussed in this paper. The user interacts with the simulator and provides input data related to a JIT system and the simulation experiment. The simulation model is generated automatically by the computer using the rules selected from a domain-specific rulebase. The machine utilization, system throughput, number and type of kanbans, and average WIP levels are some of the results generated by the simula tor. The design was originally implemented using LISP on an IBM 4381 computer and subsequently extended to Macintosh and Sun workstations.

Analysis of storage policies for miniload AS/RS

Miniload automated storage and retrieval systems can be configured with multiple pickup and delivery positions to provide both storage and material handling functions for assembly of small items. The performance of zoned storage and closest to next location storage for such a system is investigated using a simulation model. Closest to next location storage is found to be sensitive to the level of work in process in the system. At high WIP levels, this rule does not match the performance of a simple zone rule.