Process capability
From Wikipedia, the free encyclopedia
A process a unique combination of tools, materials, methods, and people engaged is producing a measurable output; for example a manufacturing line for machine parts. All processes have inherent statistical variability which can be evaluated by statistical methods. The Process Capability is a measureable property of a process. It is usually reported in terms of the inherent variability of a process and its output.
Two parts of process capability are: 1) Measure the variability of a process, and 2) Compare that variability with a proposed specification or product tolerance.
Contents |
[edit] Measure the Process
The output of a process usually has at least one measureable characteristic. These can be analyized statistically. When a normal distribution is used, the process is described by the process mean (average) and the standard deviation.
A process needs to be established with appropriate process controls in place. A control chart analysis needs to determine that the process in "in statistical control". The process capability involves only common cause variation and not special cause variation.
A large batch of data needs to be obtained from the measured output of the process. This should include a variety of production conditions, materials, and people in the process. With a manufactured product, it is common to include at least three different production runs, including start-ups.
Using proper statistical methodology, the process mean (average) and standard deviation are calculated. The process capability is often described as six standard deviations. With a normal distribution, the "tails" can extend well beyond plus and minus three standard deviations, but this interval should contain about 99.73% of production output.
[edit] Compare Process to a Standard
The output of a process needs to meet a customer requirement, specification, or product Tolerance (engineering). It is useful to compare the variability of the process to the intended requirement to determine its suitability.
Four indices are produced by a Capability Study, the Cp index, Pp index, Cpk index, and Ppk index. The most optimistic of these is Cp, which disregards centering, and is insensitive to “shifts and drifts” (special cause) in the data. It indicates what the capability of the process would be if no such problems existed. The most realistic estimate is Ppk, which indicates how the process really is. Because of their different properties, experienced users can compare these indices, and know what type remedial action a process needs: removal of special cause, centering, or reduction of natural variation.
In the common case, the same formula that produces Cp also produces Pp. The difference is in how standard deviation is estimated. The same is true for Ppk and the Cpk formula.
Generally, Ppk numbers below 1.0 indicate a process in need of work, while numbers as high as 1.5 indicate an excellent process.
If a process is stable and predictable, and has consistently provided high Ppk numbers for a long period, the process is a candidate for removing final inspection, which is likely to produce more defects than it finds. The process should then have the input variables controlled, and undergo periodic audits.
Perhaps the most common Capability Study error is requesting suppliers to provide just Cpk, along with shipped goods. Cpk can easily be manipulated by selectively changing the order of the data. To get spectacularly inflated values, simply sort the data. If you are to rely on a single index, then Ppk is the appropriate choice. Authoritative texts allow the substitution of Cpk for Ppk if the process is stable and predictable. This is allowed because in that case, they are equal.
[edit] See also
- Statistical process control
- Six Sigma
- Control chart
- Normal distribution
- Corrective and Preventative Action (CAPA)
- Tolerance (engineering)
[edit] External links
- NIST/SEMATEK discussion of capability
- The Six Sigma Zone
- Cpk and Ppm - www.SiliconFarEast.com
- Process Capability Tool - download "Lean6" to analyze capability and track variability
[edit] References
- Pyzdek, T, "Quality Engineering Handbook", 2003, ISBN 0824746147
- Bothe, D. R., "Measuring Process Capability", 2001, ISBN 0070066523
- Godfrey, A. B., "Juran's Quality Handbook", 1999, ISBN 007034003
