Friday 30 November 2018

Control Chart in Production Process - Statistical Process Control Charts

Control Chart series

As a person engaged in the business of goods or services, of course you understand that any goods or services produced, the results will not be identical to 100 percent. Then, is the existence of this variation something normal and tolerable, or vice versa, to be eliminated?

Walter Andrew Shewhart, a physics and statistical engineer from America who is also known as the father of SPC alias Statistical Process Control, said that the variation is normal and reasonable. But he divides the type of variation as something that can be controlled (controlled variation) or something uncontrolled (uncontrolled variations). The second difference is that we should pay attention.

Controlled variation is a variation due to common causes, which occurs naturally as predictable and stable. While uncontrolled variation is a variation caused by special causes, i.e. variations that occur when an abnormal event enters a process and produces unexpected and unpredictable changes.

Furthermore, the second variation does not sound so good. This is what prompted Shewhart to create a tool known as the Shewhart Chart or Control Chart. This "tool" was developed with colleagues and was first introduced in 1924. A tool aimed at making the process of providing goods and services more predictable and more consistent.

In other words, the Control Chart will tell you if there are things that deviate in the process, so you can immediately take action and restore the process on the right track. Since this tool is able to provide great benefits to the company, it is still often used in Lean Manufacturing practice to determine whether a manufacturing process and the quality of the products it produces are stable based on the graph they produce.

Shape and Control Type Chart

Control Chart will provide guidance in making improvements to the process, and should be able to help the process owner out of bad habits, the habit of taking action based on the latest data only.

If you look at the latest data, the info you get will not be complete enough to make a good decision. The Control Chart will tell you, when to conduct a follow-up investigation of an aberration, and when you just leave it alone. In short, the Control Chart will help you become more productive, not wasting time and energy for unnecessary things.

In Control Chart, we will perform the process performance planning in graphical form. The chart will have three lines: Upper Control Limit (UCL). If your data is still within the 3rd limit of this line, then it can be assumed that there is no significant problem in your current process.

In Control Chart, there are many types and forms that you can find. Some of them are:

  • Individual Control charts

  • I & R MR Control Charts

  • X-Bar & S Charts

  • X-Bar & R Charts

  • P-Charts

  • U-Charts

Fortunately, all of the Control Charts have the same three-line format: midline, UCL, and LCL. The purpose of making all the control charts is to make plotting data so you can immediately see the case - special causes that occur in the process, if any. The goal is of course to make decisions about the actions that need to be done.

All types of charts provide data on process performance over time. The difference lies in the calculation of limits for different data types.

Interpret Control Chart

As a process owner, you should be able to interpret the Control Chart appropriately, as the chart will show what action is necessary or not necessary.

Variations will always be in progress. Your task is to separate the variations into two categories, namely common cause and special cause. Then use the mean line, the control limit, and the data plot to form your interpretation.

Often, the process owner made the mistake of ordering his team to make a lot of Control Chart. When their boards are filled with many good charts, no one is responsible for interpreting and taking action on the chart. In fact, acting on a single chart will be much better than a board filled with many nonfunctional charts.

Special Cause on Control Chart

Control Chart Normal Distribution


Interpreting the Control Chart is important because the actions you take in the future will be based on this interpretation. In Control Chart we will encounter two variations of the problem, the first special cause, this means you simply use a special action, while the common cause requires action using the Lean Six Sigma methodology, in the form of improvement.

Any event that has a repeating pattern, not just a random variation, can be classified as a special cause. The statisticians have created several signs to detect this special cause.

The following four are the most used:

Outliers -

Outliers are data points located at UCL and under LCL. Since control limits are calculated based on probability theory, an outlier is rarely found in a process that has only common cause variations.

Shift -

The likelihood that a stable process will produce nine consecutive points on the same side is tantamount to throwing coins and getting 'head' as many as nine times in a row. It is possible, but very rare. The existence of nine consecutive points on the same side of the center line indicates a shift on the mean. This is a strong indicator that signifies the process has changed and requires investigation.

Trend -

a trend is defined as six points that appear respectively, each higher than the previous point. Trends indicate a special cause with a gradual effect. Look for process changes that may start at the beginning of a trend.

Cycle -

Repeating patterns calls cycles are mark with 14 consecutive points that alternate up and down. This pattern signifies cyclical changes that are repetitive in the process and certainly requires investigation. Possible cases include, over-adjustment, shift-to-shift variations, and machine-to-machine variations.

If no special cause is find out, we can conclude the process is still under control. Means that the process is still stable and unchanged, and only the variation of common causes affecting his behavior. Well, now, what about your production process?

No comments:

Post a Comment