CONTROL CHARTS


CONTROL CHARTS

Control charts separate special from common variation. They are graphical tools that help people to study the type and amount of variation present in a manufacturing system. Summary statistics are plotted on the charts and compared with a pattern that would be expected if only common variation were operating in the system. Control limits are boundaries of expected or common variation (see Figure 6.5).

click to expand
Figure 6.5: A typical process in control.

Figure 6.5 shows graphically a quality characteristic that has been measured or computed from a sample versus the sample number or time. The chart contains a center line that represents the average value of the quality characteristic corresponding to the in-control state (only chance causes.) Two other horizontal lines, called the upper control limit and the lower control limit, are also shown on the chart. These control limits are chosen so that if the process is in control, nearly all of the sample points will fall between them. As long as the points plot within the control limits, the process is assumed to be in control, and no action is necessary. On the other hand, a point that plots outside of the control limits is interpreted as evidence that the process is out of control and that corrective action is required to find and eliminate the assignable causes responsible for this behavior. (There are some exceptions to this, and we are going to discuss them in Chapter 11). It is customary to connect the sample points on the control chart with straight line segments, so that it is easier to visualize how the sequence of points has evolved over time.

The control chart is a device for describing in a precise manner exactly what is meant by statistical control. It is this characteristic that allows us to use this chart in determining both whether the past data came from a process that was in control and whether future subgroups from this process indicate statistical control.

The expected pattern is based on past observations and probability theory. Specifications have nothing to do with assessing the stability or control of a process. When analyzing charts, ask the question, "If nothing changed within the manufacturing system, would it be reasonable to expect this plotted value?" If the answer is "yes," the process is labeled in control. The process or operation is considered stable, repeatable, and predictable from past data. If the answer is "no," something in the system has changed. The operation is labeled out of control and is considered unstable, inconsistent, and unpredictable from previous data. Out-of-control operations must be investigated. The cause of change must be identified and assessed. The change may be a result of an improvement or a decay of the system.

Changes that are better than the established system should be made a part of the regular operating system. When unexpected improvement is made a regular part of the process, it is no longer a rare event (special variation); it is expected (common variation). For example, for many months, the operators who assembled the motor end plates found 5 to 10 misaligned bushings per 1000 end plates. The process was in control because there were always 5 to 10 misaligned bushings. One day, a new person accidentally applied too much air pressure (special variation). However, no misaligned bushings were produced that day. This unexpected improvement caused the process to be out of control. The assembly operators noticed the improvement and asked whether the bushing press could always be set up with extra air pressure. When this change was made a part of the regular setup procedure, there were no misaligned bushings. The extra air pressure was a cause of special variation but became a part of common variation.

Causes of special variation that are worse than the typical system performance must also be identified and assessed. These factors should be eliminated from the system so that unnecessary burdens are removed from the work environment.

The control chart has a unique ability to detect and identify causes. First the pattern is tested for evidence of randomness. Unnatural patterns are then associated with appropriate causes. These causes are extraneous disturbances or rather influences that interfere with or change the ordinary behavior of the process. These causes are called assignable because the reasons for their existence are either identified or assigned. Furthermore, they are always associated with unnatural behavior in the process. (More about this will follow in Chapter 11.)

PROCESS CONTROL AND PROCESS CAPABILITY

Engineering specifications describe acceptable products. Product engineers develop these limits to represent the customers' wants. There are specifications for each characteristic of a product (e.g., hardness, smoothness, size , voltage, and torque).

A process capability study uses engineering specifications to evaluate a manufacturing system. When evaluating a process, ask, "Is the process capable of producing at least 99.73% acceptable products?" This question is very different from that asked during stability assessment. To be in control, the process only has to repeat itself. However, to be capable, the process not only must be repeatable (in control) but also must be able to produce virtually all acceptable products. The level of common variation should be so low that all of the individual parts made by the process meet the customers' wants.

Control charts tell whether a process repeats itself; process capability describes how well a process meets the customers' wants (specifications). Another way of saying this is that the control chart tells the behavior of the process, whereas the capability addresses the needs of the customer. Figure 6.6 illustrates an in-control and capable process.

click to expand
Figure 6.6: A typical in-control and capable process.

A process that is capable is stable, repeatable, and predictable and is completely within the specification limits.

CONTROL CHART PARTS

All control charts, whether variable or attribute, have the following components .

Control limits: Control limits are mathematical limits used to interpret the pattern on a control chart. Control limits are derived from a knowledge of distribution theory and are based on a ±3 sigma. It is important not to confuse control limits with specification limits, or with the so-called "natural limits" of the process, which show the natural spread of individual units. To make sense of the limits, which are calculated numbers in relationship to the data, generally , we use a reasonable scale specific to the process and characteristic. The scale has a purpose. The subgroup range and averages are plotted to give the clearest possible picture. Scales for the graph must provide enough space to plot points (discrimination). You do not want to see the variation from the process in a straight line, nor do you want to see the variation so large that it is plotted off the page.

Therefore, there are some general tips for deciding the scales, recognizing that sometimes it may be necessary to change them. These are the most common rules used for all charts, except of course the standardized ones, which have ±3 sigma.

For the Xbar, the vertical scale should include two times the difference between the highest and lowest subgroup averages (Xbar).

For the R chart, values should extend from a lower value of zero to an upper value of about 1.5 times to 2 times the largest range (R). The scale spacing for the range chart should be twice that of the average chart.

Natural process limits: Three-sigma limits for the individual units produced by a process in control are sometimes called the "natural limits" of the process. These limits have no necessary connection with specification limits or any other arbitrary limits. Natural limits may be either broader or narrower than the specifications set by the manufacturing department. Finally, these limits are the bounds of the process when it operates normally and is under no influence of assignable cause.

Centerline: The centerline on a control chart is a line that passes through the center of a real or assumed set of fluctuating observations from which the centerline was calculated. As a general rule, the centerline is the average of the observations of the specific characteristic, and as such, it may or may not pass through the actual plotted points. This is so because the average is affected by extreme observations, and one has to be aware of such data so that other statistical techniques (for example, median) can be used.

Level: The level on a control chart is a line that passes through the center of the series of points actually plotted. The line may be drawn on the chart or it may be imaginary. The level may or may not be the same as the centerline, because a level is always related to the actual plotted points, whereas the centerline may have been obtained from some other source. It is possible for the same control chart to show more than one level in its patterns.

CONTROL CHART GOALS

  • Identify unusual process performance (special variation). With this information, one can identify the causes of special variation and decide whether to eliminate or incorporate them. Reacting to special variation improves process productivity and product quality. Control charts prevent unnecessary process adjustments. A control chart can identify background noise and abnormal variation, something that no operator can do. If the operators adjust the process based on periodic tests unrelated to a control chart, chances are that they will overreact to the background noise and make unneeded adjustments.

  • Define typical process performance (common variation). If the process is in control, the system is operating as it was designed to do. Systematic changes, not adjustments, are required to improve the process. Control charts are effective in defect prevention. Because, by definition, the control chart helps the process to be in control, by default, one lives up to the philosophy of doing it right the first time. Because sorting is expensive, the effort of identifying a controlled process is beneficial and indeed to one's advantage, as fewer nonconforming products or services are produced.

  • Assess the process's ability to meet the customers' wants (process capability). If the process is capable, 99.73% of all products are within specifications. Control charts provide information about process capability. Information about the value of important process parameters and their stability over time is tracked, which allows an estimate of process capability to be made. The benefit of this characteristic is of great value to the product and process designer.

  • Guide the never-ending improvement efforts. Control charts provide diagnostic information. The pattern of observations contains diagnostic information to the eye of the trained operator or engineer. It also shows the effect of change in the process, as performance changes.

  • The process team uses control chart data to plan improvements to the process. These changes should reduce common variation and improve process capability (see Figure 6.7). A committed and successful control chart program will reduce scrap and rework , which are the primary productivity killers in any operation. As these killers are reduced, production capacity increases .

    click to expand
    Figure 6.7: A typical improvement to the process.




Six Sigma and Beyond. Statistical Process Control (Vol. 4)
Six Sigma and Beyond: Statistical Process Control, Volume IV
ISBN: 1574443135
EAN: 2147483647
Year: 2003
Pages: 181
Authors: D.H. Stamatis

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net