Two excellent examples of process capability analysis from Quality Magazine

[Bill Levinson]

Article by Dr. Sophronia Ward Capability Analysis

I posted my own opinions of the two capability studies; see if you agree.

 

[Miner]

I agree with your comments. I would like to see the control charts for both as well as understanding their subgrouping scheme. There may have been a process shift or mixed process streams in the bimodal example.

 

[Katana_Clarity]

I agree with your comments. I would like to see the control charts for both as well as understanding their subgrouping scheme. There may have been a process shift or mixed process streams in the bimodal example.

Can I ask the objective behind requesting the control charts and subgroup scheme? The process is not capable for the Key Blank one and not great either for Solenoid housing. What do you exactly mean by process shift or mixed process streams?

 

[Welshwizard]

Bill,

Thanks for posting this.

"How do you as a quality professional explain that the data provided show all parts meet specification?"

The data provided does not with any degree of confidence show that all parts meet specification because no attempt has been made to ascertain the homogeneity of the data and therefore its behavior. Sure, there are clues to what's going on, in particular the lack of agreement between the indexes C and P affecting both location and spread, this indicates a lack of predictability between the past and the present so how could this analysis be extrapolated into the future. The Process status indicates "out of control" so I assume that it means that an observation/s have breached the control limits thus further bolstering the point about homogeneity.

"What questions should be asked about this analysis to understand how to interpret the information available?"

Plot the data on a process behavior chart, if signals appear investigate and fix the issues to ensure that the data only displays random variation

How do you answer management's q

 

[Welshwizard]

Bill,

Thanks for posting this.

Solenoid Housing

"How do you as a quality professional explain that the data provided show all parts meet specification?"

The data provided does not with any degree of confidence show that all parts meet specification because no attempt has been made to ascertain the homogeneity of the data and therefore its behavior. Sure, there are clues to what's going on, in particular the lack of agreement between the indexes C and P affecting both location and spread, this indicates a lack of predictability between the past and the present so how could this analysis be extrapolated into the future? The Process status indicates "out of control" so I assume that it means that an observation/s have breached the control limits thus further bolstering the point about homogeneity.

"What questions should be asked about this analysis to understand how to interpret the information available?"

Plot the data on a process behavior chart, if signals appear investigate and fix the issues to ensure that the data only displays random variation

"How do you answer management's question to their satisfaction?"

It would appear that judging by the histogram, the process has potentially around 40% less width than the spec so if, and only if the process is brought and kept under control and centered the company could enjoy less defects against spec on an ongoing basis.

My answer for the Key blanks is the same.

 

[Miner]

Can I ask the objective behind requesting the control charts and subgroup scheme? The process is not capable for the Key Blank one and not great either for Solenoid housing. What do you exactly mean by process shift or mixed process streams?

The control charts would show whether the process is stable. If it is not stable, it may help answer some of the questions that were raised. For example, one of the histograms appeared to be bimodal. This could be caused by a shift in the process average, which would show up in the control chart. It could also be caused by a mixture from two different process streams with different averages. Examples of process streams are two different production lines, machines, molds, or cavities of molds. Output from different process streams should be kept separate and analyzed separately.

Correcting these issues would improve the capability.