Use SPC to Make Better Parts

Statistical process control (SPC) is a tool for monitoring and controlling manufacturing processes. Dr. W. Edwards Deming claimed that the majority of variation in a process is due to operator over-adjustment. SPC gives operators a tool to determine when a statistically significant change has taken place in the process or when a seemingly significant change is due to chance causes.

SPC involves:

• Determining the critical process parameters that need to be monitored.
• Setting up an initial control chart and confirming that the process is in-control.
• Collecting and plotting future data on the chart and interpreting the chart to determine if the process has gone out of control.

There are a number of reasons why companies use SPC. Often an internal champion initiates the use of control charts and other SPC techniques to reduce variation and to improve manufacturing processes. Sometimes companies implement SPC to satisfy customer requirements or to meet certification requirements. SPC itself will not make improvements. However, SPC will give the operator a tool to identify when a special cause of variation has entered the process so that the special cause can be eliminated. With this tool, constant adjustment of the process is eliminated. In addition, SPC can be helpful in identifying opportunities for improvement that can lead to reduced variation and processes that are better aimed at their target.

MeasurLink® is a data collection and real-time SPC suite from Mitutoyo America Corp. (Aurora, IL). The software is designed to detect and display patterns and provide additional statistical information. Many patterns can be detected on SPC charts, including:

• Cycles
• Trends
• Mixtures
• Grouping of measurements
• Gradual change in level
• Sudden shift in level
• Instability (abnormally large fluctuations)
• Stratification (abnormally small fluctuations)
• Interactions (two or more variables acting together)
• Systematic variation

Make good parts

To ensure manufacturers make good parts, SPC starts at the point of manufacture. When manufacturing a part and checking features critical to quality, it is important to collect the data immediately and as physically close as possible to the point of manufacture. When recording and charting data, the operator is going to see that things such as assignable cause and corrective action. The only time you have good understanding of these is at the point of manufacture. A week later, no one remembers.

SPC systems provide records of when things are going right—the data that’s in control and in tolerance—and when things go wrong–the assignable cause and corrective action. That’s important in using data for statistical projections. Statistical predictions can only be made on stable processes.

Statistics is all about numbers, however, sometimes looking at the numbers is not as easy to interpret as a graphical representation of those same numbers. As the saying goes, a picture is worth a thousand words, or in this case, a picture is worth thousands and thousands of numbers. SPC systems allow you to do both, allowing a quick check of a process, as well as the ability to drill down as necessary for more information when necessary.

For example, you want to do a comparison between the tolerance limits and the control limits. If you have control limits that are well within the tolerance limits, great. You want to look at statistics to validate what you see in the graph, but the graphs tend to give you big-picture information on the overall process.

All the while, manufacturers are trying to maintain a process, make sure it is stable over time and continues to be so in the future.