Managing Bottlenecks

Introduction

Assembly lines can be classified into single-model, batch-model, and mixed-model lines. In a batch-model assembly system, a few product models are produced in batches, with one product at a time on the same line, and a certain amount of changeover time is allotted to ready the line for the production of another model. In recent decades, the need for more versatile and flexible production has forced assembly line production systems to change from fixed assembly lines to mixed-model assembly lines, where the output products are variations of the same base product and only differ in specific customisable attributes. Considering mixed-model manual automobile assembly systems, setup times between models can be reduced enough to be ignored, so that intermixed model sequences can be assembled on the same line.

A procedure is needed to determine a particular configuration for the products to be produced on the line that will not only minimise the balance delay or number of workstations, but also satisfy other conflicting criteria such as production rate, variety, minimum distance moved, division of labour, and quality. The key input to a line balancing problem is accurate standard time derived from a time study or from any other work measurement technique.

For a traditional direct time study, a time-study analyst divides a job that usually takes a long time into basic tasks that are easier to measure and analyse; and then the analyst observes a qualified person using the best method to perform the job. The time standard is the time required by a skilled operator, working at a normal pace, to perform a specific task using a prescribed method, allowing time for personal needs, fatigue, and delays. It can be used for a work assignment, for evaluating the numbers of workers, the type and capacity of machines, the overall productivity, the total cost for product manufacturing, and so on.

Bottlenecks for one product are not automatically bottlenecks for other products in the same manufacturing line. This is due to variations in the processing time for the different products on different machines. An important trend in the queuing theory literature that can be used in bottleneck analysis is the development of laws that connect the system content and customer delay. The most well-known result is Little's law, which is valid for any arrival process, service process, or scheduling discipline, but only deals with the first moments of system content and delay. According to Masood, increased throughput and higher machine utilisation in an automotive plant can be achieved by managing bottlenecks through line balancing. Das discusses the conceptual overview of a simulation methodology to evaluate assembly line balancing with variable operation times based on the next-event analysis. Pourbabai proposes a methodology for the design of a flexible assembly line system while controlling the bottleneck problem. Plenert demonstrates how geometric programming can be used to solve an industrial bottleneck with an unlimited number of products and multiple constraints. Wang proved that a data-driven approach can enable the modelling and simulation of a complex assembly plant in a real-time fashion, and thus effectively improve the responsiveness and flexibility of the production line.

The aim of this paper is to use a discrete-event simulation package to illustrate that bottlenecks can be managed by ensuring high workstation utilisation, reducing queue lengths before stations, and reducing station downtime. The paper begins with background information on the case in point. The second section of the paper comprises the work approach from time studies to quantitative analysis of mixed-model assembly lines. A simulation model is then developed using Showflow simulation software. The model is verified and validated using information derived from actual historical data. A detailed bottleneck analysis is then conducted, after which decisions are made for optimal bottleneck management. The paper also demonstrates that simulation can be a tool for determining the makespan for specified vehicle demand so that due dates for orders from customers are met.