CONTROL CHARTS
Introduction:
Control charts may be used partly to control variation and partly in the identification and control of the causes which give rise to these variations.
Causes of quality variations:
Shewhart divided the causes of quality variations into the following groups:
1. Specific causes
2. ‘Random’ causes (system causes).
Random causes: Random’ causes are characterized by the fact that there are many of them and that the effect of each of these causes is relatively small compared to the special causes. Shewhart’s use of the word ‘random’ is somewhat unfortunate—we prefer system causes instead.
Specific causes: As opposed to system causes, there are only a small number of specific causes and the effect of each specific cause may be considerable. This being so, it is possible to discover when such specific causes have been at work which, at the same time, allows us to locate and thus eliminate them. An example of a specific cause is when new employees are allowed to start work without the necessary education and training.
Process control charts:
- A process control chart is a graphic comparison of the results of one or more processes, with estimated control limits plotted onto the chart.
- Normally, process results consist of groups of measurements which are collected regularly and in the same sequence as the production the measurements are taken from.
- The main aim of control charts is to discover the specific causes of variation in the production results.
- When specific causes are affecting the production result because the measurement of this result lies outside the control limits plotted onto the chart. The job of having to find the cause or causes of this brings us back to the data analysis, where the Pareto and cause-and-effect diagrams can be invaluable aids.
BASIC CONSTRUCTION OF THE CONTROL CHART:
- This chart shows the average measurements of a production process plotted in the same sequence as production has taken place, e.g. the five last-produced units of each hour’s production could be measured. The average of these five measurements is then plotted on the control chart.
- The control limits are known as UCL (Upper Control Limits) and LCL (Lower Control Limits), which are international designations.
- Exactly how these limits are calculated depends on the type of control chart used, many different kinds having been developed for use in different situations.
- As a rule, control limits are calculated as an average of the measurements plus/minus three standard deviations, where the standard deviation is a statistical measure of the variation in the measurements.
Reading the control charts:
1. There are two points outside the control limits. This is a sign that the process is out of statistical control.
2. Each of these two points has had a special cause which must now be found. If a point outside the control limits represents an unsatisfactory result, then the cause must be controlled (eliminated).
3. This of course does not apply if it represents a good result.
4. In such a case, employees and management, working together, should try to use the new knowledge thrown up by the analysis to change the system, turning the sporadic, special cause into a permanent system cause, thus permanently altering the system’s results in the direction indicated by the analysis.
Two important principles:
1. If we can accept the variation which results from system causes, then we should not tamper with the system. There is no point in reacting to individual measurements in the control chart. If we do react to individual measurements of a process in statistical control, the variations will increase and the quality will deteriorate.
2. If we are dissatisfied with the results of the process, despite the fact that it is in statistical control, then we must try to identify some of the most important system causes and control them.