How do you set and adjust control limits for SPC charts?

Limites de controle são a voz do processo, limites de especificação são a voz do cliente. O processo deve ser capaz de atender ao cliente.

SPC is a process "Blind Man's Blunt Tool". Manufacturing and process technology has changed drastically to define and control the process precisely. Use Industrial Statistics to plan and control the production and build quality into the production process. No need of SPC, Cp, Cpk, Six Sigma etc. You know, our antiquated quality policies and systems are killing the manufacturers by imposing unwanted practices that hinders the operational freedom of a manufacturer. Time to change and go for a game changer - Industrial statistics - only for manufacturing industries. There are many areas SPC will work successfully.

This is very common in manufacturing. Especially if the customer can’t define the specifications in a clear quantitative fashion, or a too broad of specification. In this cases, we would define our “production” limits, considering maintenance, consumption and scrap rate. In services, customers rarely state their quantitative specifications, then the quality procedure will “arbitrarily” define the limits and put it to test, until it’s validated by customer feedback. After that, the Continuous Improvement processes will take care of refining until a polished and reasonable specification. Maybe, it enlights the way to an industry based specification.

Further, to make sure the control limits are updated to accommodate expected changes in the system due to change in process or product aspects, use of dynamic change point detection methods is highly recommended. Ensuring SPC based process stability monitoring to be effective on an expected dynamic process is key to its mainstream adaptability.

Pretty good article overall, with the exception of the final section. NEVER EVER compare control limits and specification limits! This is because of a few reasons, the most important is as follows: If you are using an x-bar chart, you are plotting AVERAGES (x-bars) not individual readings. In the example above you take a sample of size 4. These x-bars will fall in a much narrower band (see central limit theorem; about 2x in this example) than the individual readings! So although your x-bar might be within the specification limits, that doesn’t mean the individual (parts)are. Ex: lower spec is 10. Sample of 4: 8,9, 11, 13. Xbar=10.25, yet 1/2 parts are out of spec. Always use the individual readings for your capability analysis.

Understanding process variability and stability is one of the first fundamental things you learn when generating your first control chart. Do not be surprised or disappointed if your process is a unstable and out of control. It is probably not a coincidence that you are looking at this process. For these processes, focus on understanding the outliers. Eliminating or reducing their impact will reduce the process variability. Significant gains in variability reduction can have big benefits to the bottom line on your journey to In Control and Capable.