Introduction

There are numerous challenges facing the industry today. One of these challenges is a dedication to quality control and the management of quality in production that has come to be epitomized by such Japanese products as automobiles and consumer electronics. In response to this challenge, the philosophy and techniques of quality control and management are becoming widespread in manufacturing. In addition, the rapidly growing circle of TQM for services as well of applications of Total Quality Management (known by the acronym TQM) has expanded all the way from manufacturing industries to service industries such as health care and legal services.

An on-line system in Tata Steel

Statistical techniques can be applied in process control by the simple application of Information Technology to bring Statistical Process Control tools to the shop floor and the process engineer. Control charts typically reside on the shop floor and are easily understood and clearly readable by all. It is essentially a form of traffic signal: green for good, amber for possible trouble and red to stop production to avoid defects. Control by variables takes into account the accuracy and precision of a process distribution. While data analysis and SPC have become a standard being practiced in the industry, the novel methodology of translating data into an on-line-graphical representation of control charts is the focus of this paper.

Quality and Variability

What is quality? It is an abstract concept, yet it often lends itself to quantification! There are many approaches to understanding and dealing with quality. In the industrial context it has a strong link with variability. This is best illustrated by examples. When you see an advertisement for a "high-quality automobile," do you conjure up images of such luxurious options as leather seats and fancy sound systems? Most of us do connect luxury and quality. But expensive leather seats don't mean very much if the engine won't start on a cold morning, and the latest noise-reduction technology is hard to appreciate if the tape deck starts chewing-up your tapes. These examples show us that it's important to separate the notion of luxury from quality. In fact, some of the cheapest items in everyday life can have very high quality. Consider the paper used in a copying machine. For little more than 20 paise a page, you can buy smooth white paper, less than one hundredth of an inch thick and of uniform size. You have come to expect such high quality in copier paper that you don’t examine individual pages before loading it into a copier. You wouldn’t think of measuring the thickness of each page to make sure that it was thin enough not to jam the copier, but thick enough so that you could print on both sides and not have the two images interfere with each other. The copy paper example gives us a clue to a working definition of quality. Things that are of high quality are those that work in the way we expect them to. As quality expert Joseph M. Juran has put it, quality implies fitness for use. In this sense, quality means conformance to requirements. Note that this is not quite the same as conformance to specifications. The idea of “things that work in the way we expect them to” points out that quality is defined by customers as well as by producers. Meeting the needs of customers is central to TQM. Working definitions of quality vary in different contexts, especially when we contrast goods and services. But in keeping with our notion of conformance to requirements, most working definitions of quality will include the concepts of consistency, reliability, and lack of errors and defects. When mass production became common during the nineteenth century, it was soon realised that individual pieces could not be identical - a certain amount of variation is inevitable. But this leads to a problem: with too much variation, parts that are supposed to fit together won't fit! In this sense, you can see why variability is the enemy of quality.

Statistical Process Control

A control room in the Cold Rolling Mill

When the output of some process is found to be unreliable, not always conforming to requirements, the process must be carefully examined. In the 1920s, Walter A. Shewhart, a researcher at Bell Labs, created a system for tracking variation and identifying its causes. Shewhart's system of statistical process control (or SPC) was developed further and championed by his one-time colleague, W. Edwards Deming. For many years, Deming was a prophet without honour in the United States, but when Japan was rebuilding its economy after World War II, Japanese- managers incorporated Deming's ideas into their management philosophy. Many American industries, including automobiles and consumer electronics, encountered severe competitive pressures from the Japanese in the late 1970s and 1980s. As a result, the contributions made to quality control by Deming and others were reconsidered by American managers.

There are two kinds of variations that are observed in the output from most processes, in general, and from our automatic lathe, in particular:

• Random variation (sometimes called common, or inherent, variation)

• Systematic variation (sometimes called assignable, or special cause, variation)

These two kinds of variation call for different managerial responses. Although one of the goals of quality management is constant improvement by the reduction of inherent variation, this cannot ordinarily be accomplished without changing the process. And one should not change the process until it is sure that all assignable variation has been identified and brought under control. Therefore, the approach is: If the process is out-of-control because there is still some special cause variation present, identify and correct the cause of that variation. Then, when the process has been brought in-control, quality can be improved by redesigning the process to reduce its inherent variability.

The essence of statistical process control is to identify a parameter that is easy to measure and whose value is important for the quality of the process output (the shaft diameter in our example), plot it in such a way that we can recognise non-random variations, and decide when to make adjustments to a process. Figure 1 shows the process management paradigm.

X Bar and R Charts

A very common control chart that is used widely in industry is the X Bar and R Chart shown in Figure 2. These charts are plotted after calculating control (limits upper and lower) for the process variable being examined. There are standard steps for calculating the limits and can be referred to in any book of statistical quality control. A set of sample control charts are shown in Figure 3.

Figure 3 also shows how the charts can be interpreted and how processes can be controlled by proactive action taken at appropriate times. This methodology illustrates the prevention model and its inherent advantage over the detection.

HOME