Control Charts

November 27, 2009

Control charts are tools that help determine if a process being monitored with the help of a given chart is stable or not. Stability is defined by the fact that the data points are within some given limits. The theoretical basis is the fact that if a process for some activity is within control the output will be within specifications. A control chart typically has a plot that indicates the mean or the expected value. Two other plots indicate the lower and the upper specification limits and two other plots indicate lower and upper control limits. These limits are tighter than the specification limits (closer to the mean or expected values).

Control limits usually specify the plus or minus 3 sigma variations from the mean. The monitored values are plotted against these limits. PMBOK fig 8-5 is a good example of control charts. If a data point exceeds one of the control limits it is considered a loss of control situation. Similarly if seven consecutive data points are above or below the mean the process is considered out of control. Statistically one would expect random variations and the consecutive points represent a trend. The criteria to decide when the process is out of control would depend on the exact situation and are determined from statistical principles.

Control Charts

In production processes we monitor how well products meet specifications. Deviations from target specifications and excessive variations from item to item are the problems one has to look for. How this is done is to collect samples of items produced and carry out measurements of the characteristics governed by the specifications. The mean values found for each specification item and the range (or the variations of that dimension) are noted and plotted against desirable values and variations that are not statistically significant. There are several different charts in use but the basic principle is the same. If the measured values go beyond the 3 sigma (or six sigma in some cases) limits or there is excess variability or there is a trend of a drift away from the mean then we have a problem. The fig 8-6 of PMBOK is an example where defects in each sample is counted and plotted against desirable values with control limits.

Control charts can monitor a range of variables. Though, most commonly, they are used to determine if something is being manufactured correctly. Figure 8-6 in PMBOK is an example. These charts could easily be used for trend analysis in monitoring project cost and time variances, frequency of scope changes and other management processes. The figure 8-5 actually monitors the time taken in the project. The last three points monitored are consistently above the mean and seems to indicate a trend that activities are actually taking more time than estimated. The process is still in control, it was completely following the desired values initially. However, the last three points indicate a likely problem. If future readings too indicate the same kind of deviations then obviously the time estimates are off base.

Popularity: 1% [?]


Leave a Reply

Your email address will not be published. Required fields are marked *