A FEW WORDS ABOUT SPC

Under the Company-Wide Quality Control (CWQC) philosophy, all parts of an organization cooperate to improve all aspects of the organization's operations. The main characteristic of CWQC is a customer-first orientation. Customer satisfaction is seen as the company's highest priority. It is recognized that customer satisfaction can only be obtained by providing a high-quality product, and consequently, continuous improvement of the quality of the product is seen to be the only way to maintain a high level of customer satisfaction. To achieve this goal, maximum involvement by all the company's employees in the quality improvement process is essential. On-going education and training of all employees supports the drive for quality.

CWQC is a way of life for a company, so it is natural that CWQC be led by top management. Their leadership of the quality improvement process is essential. Setting the performance measures consistent with the objectives of the company is a key task for top management.

As well as recognizing the link between customer satisfaction and product quality, CWQC also recognizes that product quality is the result of process quality. As a result, there is a focus on continuous improvement of the company's processes. This will lead to an improvement in process quality. In turn this will lead to an improvement in product quality, and to an increase in customer satisfaction.

Elimination of waste is a major component of the improvement approach. There is also a strong emphasis on prevention rather than detection, hence an emphasis on quality at the design stage. Another component of the approach is the continual reduction of product and process variation. This is supported by the use of statistical methods such as Statistical Process Control (SPC).

Traditionally, Statistical Process Control has been used during the production phase of the product life cycle to reduce the product and process variation and help correct whatever is wrong. All production processes may fluctuate over time. Provided they are stable, they will stay within certain well-defined limits known as control limits. If a process gets out of control, the fluctuations go beyond the control limits.

SPC techniques are simple statistical techniques to help identify process problems. The following seven statistical problem-solving techniques are grouped under the SPC acronym:

The control chart is a simple two-dimensional easy-to-understand representation showing process performance over time. There are several types of control chart, e.g. X-bar, R, c and p. On a X-bar chart the process operator routinely plots sample values on a chart that shows the expected value and the control limits. The resulting chart shows process performance over time, and whether action has to be taken. It can also be reviewed at a later date to see if more subtle trends exist. (If the process is under control, 99.7 % of samples fall within the control limits).

The cause-and-effect diagram, also called the fishbone chart, or the Ishikawa Diagram, shows the relationship between a problem and its possible causes. For example, for a problem with faulty orders, four 'bones' (i.e. potential causes) on the fishbone chart could be materials, machines, manpower and methods. Potential causes on the manpower `bone' could be poor communication, lack of understanding, new employees, poor liaison with sales, lack of training, and insufficient personnel. If the cause-and-effect diagram is developed systematically and in detail, all potential causes are identified, thus providing a framework for a possible solution to the problem.

The scatter diagram graphically shows the relationship between two variables, for example, wafer thickness and yield.

Pareto analysis is based on the observation of Signor Pareto that most of the money is in the hands of a few people. Similarly it is often found that 20% of the problems cause 80% of the trouble. Pareto analysis aims to identify the vital few problems and set the priorities for further analysis. The Pareto chart is a bar chart displaying the contribution of each cause to a problem.

The histogram is a bar chart which graphically shows the frequency distribution of data.

The process flow chart.

The check sheet.

The need for SPC in production should decrease with the increased use of techniques such as QFD and Taguchi techniques at the design stage. As these aim to eliminate errors and variation at the design stage, they should prevent a lot of problems getting through to production. SPC can play a role in setting tolerances for manufacture. SPC data can be stored in a data base that product developers can access during component design to help them find out what can be achieved.

Statistical analysis methods are just as effective in an office environment as they are on the factory floor. Just as SPC has been used to improve manufacturing processes, it can be applied to virtually any administrative, technical, or business function. It can be used to help reduce errors in engineering change releases, purchase orders, technical manuals, data entry, and inventory records, and to shorten the time to issue a work directive, pay an invoice, or process an order.

The data integrity of the Bill of Materials (BOM) is critical to ensuring that the right part is in the right place at the right time. In one company, a 99% accuracy level was being maintained at the cost of multiple inspection of each BOM, which is just as costly at a desk as it is on the factory floor. By using control charts to measure process performance, and Pareto analysis to identify the most frequently made errors, the BOM team was able to focus attention on the areas most in need of improvement. After making corresponding improvements, the BOM accuracy rose to nearly 100%, and the 300% inspection previously performed was reduced to random sample audits.

The SPC approach is based on the use of objective data, and provides a rational rather than an emotional basis for decision making. The statistical approach to process management in both manufacturing and white collar areas recognizes that most problems are system-related, and are not caused by particular employees. The approach ensures that data is collected and placed in the hands of the people who are in the best position to analyze it, and then to take the appropriate action to reduce costs and prevent non-conformance.

At the heart of SPC is the statistical analysis of engineering and manufacturing information. If the right information is not available, then the analysis, whether it be of shop floor data, or engineering test results, can not take place, errors can not be identified, and consequently errors can not be corrected. An EDM/PDM system can make sure that the right information is available, at the right time, to the right people. In many organizations, SPC is now part of the culture, and comprehensive test bases have been built up. Unfortunately, these are rarely connected to the Engineering Department, so product developers can not learn as much as possible from their mistakes. An EDM/PDM system spanning product development and testing can help overcome this deficiency.

Copyright 1998 by John Stark