Analysis of Control Charts

 

Analysis of Control Charts- A control chart is a significant tool applied to monitor and improve quality. It is one of the effective seven basic statistical tools often applied in Six Sigma along with scatter diagram, flowchart, pareto chart, cause-and-effect diagram, check sheet and histogram. All the tools are used to evaluate the business or manufacturing process in an effective way.

Each project of Six Sigma executed by small or large venture follows a distinguished sequence of processes, and all the processes can be evaluated statistically with the control charts.

Control charts are easy-to-use and make it simple to observe both special and common cause variation in a process. The Six Sigma control chart is an effective tool of statistical process and also popularly known as Shewhart chart or process-behaviour chart. Control chart indicates whether the process being monitored is currently under control or not. All control charts have following basic components:

• Central Line: A central line is drawn at the process mean. Control chart omits specification targets.
• Upper Warning Limit: An upper warning limit is drawn two standard deviations above the centre line.
• Upper Control Limit: An upper control limit is drawn three standard deviations above the centre line.
• Lower Warning Limit: A lower warning limit is drawn two standard deviations below the centre line.
• Lower Control Limit: A lower control limit is drawn three standard deviations below the centre line.

The objective of control charts is to allow simple detection of events that are indicative of actual process change. If all process values are plotted and no particular propensity is marked within the upper and lower control limits, the process is considered to be stable and is referred to as “In Control”. If the values of process are plotted outer surface of the control limits the process is referred to as “Out Of Control”. The practice of creating a control chart is as follows:

• Select the process you would want to chart.
• Choose your process sampling plan.
• Collate data from your process.
• Compute the control chart precise statistics.
• Decide your control limits.
• Construct your control chart.

If the entire process of a venture is evaluated with control chart to figure out whether the manufacture or venture is in control, all points would be plotted within the control limits. Any observations deviating from the systemic patterns in a process suggest that its demand or introduction of a new source of variation. Increased variation implies increased quality costs, a control chart signalling the presence of a special-cause requires immediate investigation and if possible adopt strategies to tackle them. The control limits highlights the process behaviour and have no fundamental relationship to any distinguished targets.

Control charts help discern process variation due to assignable causes from those due to unassignable causes. On a control chart, both these types of process variations are depicted. Assignable causes are significant factors of a process and are not always normal. Such variety of causes can be avoided and should be scrutinized. Unassignable causes or chance causes are factors that occur by chance and not always present. These types of causes are normal and expected within a process as they are inevitable.

Control chart omits specification target because of the tendency involved within the process. It is usually done to make a process simple. Typically, control charts use a distribution chart or histogram as well as X bar (mean), and R bar (range), charts. Meanwhile, if a special cause does occur, it may not be of sufficient degree for the chart to produce an instant alarm condition. If an unusual cause crop up, one can depict that cause by assessing the change in the mean and/or variance of the course in question.

Control chart is not free from criticism. Scholars have criticized the control chart on the condition that it contrasts the likelihood principle. However, supporters of control charts argue in general, it is impossible to specify a likelihood function for a process not in statistical control, especially where knowledge about the cause system of the process is weak.

Go back to Tutorial                                                                                Go to Home Page