Reactions to out of control X bar and R charts?

[Josh Smith]

I am curious as to how other organizations have written their procedures for reacting to out of control points up or below the control limits, 7 consecutive points running above or below the centerline, and 7 points trending up or down?

Our reaction plan on the control plan states that when the process has violated these rules. Supervisor will be notified by quality auditor, suspect product will be isolated and verified by other means (i.e 100% inspect), the process will be adjusted to bring back into a state of control, and process changes will be noted on the process change log sheet.

Does this sound reasonable? Or is it painting us into a corner for QS-9000?

Looking for input?

 

[auditor16949]

The procedure is basically OK. The QS-9000 auditor may ask if you have identified and eliminated the special cause of variation acting on the process.

 

[Indywine]

Keep in mind that 7 points in a row usually reflects a change in the process. However, it could be associated with a learning curve after training. I wouldn't suggest blindly changing the process without trying to understand why there is a new trend.

Jim

 

[lyman]

Our procedures which were written by engineering and approved through manufacturing centered the responsibility for initial monitoring of SPC and any related trends with the floor operators. We tied our SPC as much as possible with our computerized manufacturing system so that if the chart showed an OOC point the equipment automatically was no longer available to production until such time that someone from Engineering reviewed the situation and released it. The procedures for trends directed operators to OCAPs which either dictated that a change to equipment settings be made (if appropriate)with the requisite documentation, or that the process be shut down and Engineering and Manufacturing Management be notified. The biggest challenge we faced with this allocation of responsibility was making sure that there was enough information in our computerized record so that Engineering could evaluate the situation to begin effective corrective or preventive action. The second biggest challenge - getting Engineering to document what they did!

 

[Brian Dowsett]

I think you also could make some reference to process capability.
For instance if you have a very good Cpk value, seven points in one direction definitely merits some attention but may not merit 100% inspection.
In a previous job, our reaction plan was in the form of a flow chart, with some decision boxes based on proven process capability.

regards

Brian

 

[ABS]

"We tied our SPC as much as possible with our computerized manufacturing system so that if the chart showed an OOC point the equipment automatically was no longer available to production until such time that someone from Engineering reviewed the situation and released it."

I am looking for ways to connect SPC to the manufacturing system. Looking after your reply, I was wondering if you could elaborate more on how to accomplish this system.


Abhinav.

 

[Caster]

Hmmm.....quality auditor, supervisor, engineering

Everyone involved but the operator? SPC was supposed to be a tool for the machine operator to control her process. Are they out of the loop?

Can't they just make adjustments and carry on? Assuming a decent capability, a trend may not be bad product but just a signal to perhaps change worn tooling.

The more the operator knows about SPC, the better.

At least thats the theory. We don't do any better here, but I hope some people get to run SPC right.

 

[Bev D]

agree with Caster on the capability part. you state that "suspect product will be isolated and verified by other means (i.e 100% inspect)". I hope that you mean only product suspected of being out of tolerance?
As Caster points out a capable process can go out of control and all of the parts can be perfectly in tolerance nd acceptable.

The other comments regarding operator involvement are also pertinant. Keeping in mind that are processes that the operator cannot troubleshoot. I've worked with both and sometimes you just dont' want the operator touching anything...

 

[Steve Prevette]

"We tied our SPC as much as possible with our computerized manufacturing system so that if the chart showed an OOC point the equipment automatically was no longer available to production until such time that someone from Engineering reviewed the situation and released it."

I am looking for ways to connect SPC to the manufacturing system. Looking after your reply, I was wondering if you could elaborate more on how to accomplish this system.


Abhinav.

You are welcome to make use of our trending procedures on the internet at http://www.hanford.gov/rl/?page=1144&parent=169. Just recently (December 2007) the Institute of Nuclear Power Operations published a 20 page extract of our procedures in one of their "best practices".

 

[bobdoering]

Another question is whether an X-bar-R chart is even appropriate for the process. It seems to be rubber stamped on every process. I have seen automated SPC systems choke precision machining operations - because they were using the wrong statistics - and have unplugged them and replaced them with paper charts the correct statistics. Made everyone happy. Don't fall into the X-bar - R trap. It is not a universal tool.