Calculation Mean for Control Chart -X Double Bar

In control chart X Double bar is calculated by taking the average of no of subgroup mean (X Bar).This is the normal practice we are following for calculating Upper and lower control limits.

But one of our colleague saying that we have to take the Drawing specification nominal value.

I think it is not correct.Pl.give your views

What is your process and measured characteristic?

Process:Aluminium pin Forming process
Measured characteristicin height

The specification nominal is NOT included in the calculations for statistical* process control charts. If you are using an Xbar, S or Xbar, R chart the center line is the grand average. If your subgroup size is always the same then you can simply average the subgroup averages. If your subgroup size varies then you have to calculate the average using all of the individual points and not just the subgroup averages.



*There are some control charts - not statistical - that utilize specifications and I'm sure Bob will point you to one for tool wear...

I agree with you, the SPC chart is an indication of process stability rather than compliance with drawing specification.

To measure accuracy against specification you would need to calculate the Process Capability ratios Cp & Cpk.

Bev D said:

The specification nominal is NOT included in the calculations for statistical* process control charts. If you are using an Xbar, S or Xbar, R chart the center line is the grand average. If your subgroup size is always the same then you can simply average the subgroup averages. If your subgroup size varies then you have to calculate the average using all of the individual points and not just the subgroup averages.



*There are some control charts - not statistical - that utilize specifications and I'm sure Bob will point you to one for tool wear...

Click to expand...

Thanks for your reply

Markaich said:

I agree with you, the SPC chart is an indication of process stability rather than compliance with drawing specification.

To measure accuracy against specification you would need to calculate the Process Capability ratios Cp & Cpk.

Click to expand...

Thanks for your reply

Bev D said:

*There are some control charts - not statistical - that utilize specifications and I'm sure Bob will point you to one for tool wear...

Click to expand...

Hey, now - it is statistical - just using different statistics. It is for non-random continuous variation - and the other, traditional SPC charts are for true random variation (or their underlying statistics are not applicable based on their assumptions). But one would need to review the total variance equation to see what factors of variance contribute to the variation on the chart. If the process was controlled enough to be just tool wear, then yes the tolerance is more important than the mean - in fact the mean is meaningless in control of that process.


So either answer could be right...it depends.

sorry I should have clarified: statistical limits/statistical charts in the context of SPC. The OP was asking a specific question about a specific type of chart, the Xbar chart. For this chart, the center line is teh grand average of the subgroups not the nominal of the tolerance.

statistical analyses will almost always point you to which chart is best for your situation. Those charts may have limits that are based on the statistical calculation of the process variation; some may utiliize only specification limits or other limits that do not statistically derive from the process variation itself.

Bev D said:

sorry I should have clarified: statistical limits/statistical charts in the context of SPC.

Click to expand...

I prefer the term "Traditional SPC".