Am I the only one who doesnt understand. You do a Control Chart and find the process to be out of control with special cause variation. You research and identfy the reasons. Then what?…remove them, keep them,..whats the next step in the rules of Six Sigma engagement?. All the books I read, simply say research the reasons for the Special Causes and not what to do next. Also, If the process wasnt out of control to start with, I wouldnt be doing the project anyway…so wheres the beef here?
Situation: there is a problem / a reason for a project.You investigate with a control chart and see:
Process has Special Causes (several types possible). Decide whether they are part of your problem (i.e. 1% SC when 35% rejects => forget them).If so find the Root Cause of them (expertise, tests).Now the ACTION: make somehow (expertise) sure that they cannot happen again (or much reduced) and/or use a Control Loop to negate the effect. So in effect: solve the problem now you know how it is caused.
Process has ‘no’ Special Causes but the UCL/LCL don’t agree with the Customer Wish (specs USL/LSL).Decide whether the Mean or the Sigma is the problem (or both).Now the ACTION: improve it somehow (expertise)
Mean: find a knob (=X) to shift the mean of your process.
Sigma: find the X’s that contribute and how much the impact of each is.
Both of the above mentioned problems: SC is general the best choice for first improvement (get the process stable); then Sigma (get Cp and Pp right) and lastly Mean (get Cpk and Ppk right).
For both FMEA & DoE is the most commen tool-combination used to investigate which X’s and how they influence the Y.
Hope this helps? Why didn’t your teacher cover this in the course?
I am not sure why it was not covered. Still not sure I grasp this. I know it should be simple.
What is a control Loop?
with Control Loop I mean Feedback loop or Feedforward loop.
Suppose you have Y= Function (X1,X2,X3,..) where Y = CTQ of Customer and X1,X2,…. are process parameters. So you know how the CTQ is influenced by the causes (find out by DoE/Expertise).
Feedback: Measure the Y. If you see that it is not on Target: correct with one or more of the X’s to get it back on Target.This is a reactive control loop
Feedforward: X1 cause variation in the Y and X1 can not be controlled by you (e.g. an incoming supplier parameter). Measure the X1 value BEFORE the process and change process settings (X2,…) to compensate such that Y will get on Target when the process makes the product.This is a proactive control loop
Obviously 2. is better than 1. because you prevent rejects (instead of preventing the next one). Als 2. will notwork if you have singular outliers. But 1. is often more difficult to realise.
Hope this helps.
To cut to the chase, fix it! Now you have a new process.
Deka, I am usually very polite on here. But, I must ask, are you doing quality improvement, or what?
Do you think we look at a process just to see what interesting thing will happen next? The whole point of what we do here is to improve processes. Of course you eliminate the special causes. Do it in a way that they can never come back.
“To reduce my message for management to just a few words, I’d say it all had to do with reducing variation.” W. Edwards Deming.
Thanaks James..So you never simply remove the Data points?
Thanks..So..Back to basics…If you find SCV..and fix it..that may very well(most of the time?) be the entire project?..or it may be something the boggs the entire project down..SCV..may be extremely hard to fix…correct?
Hey Michael. I did a control chart on waste and I had an out of control point on the low side indicating a special cause variation. I took your advice and eliminated the special cause to be sure it never comes back. Only kidding…point is that we investigate special cause variation. Sometimes special cause variation is desirable and we wish to incorporate what ever occurred not eliminate it. Key is the investigation and appropriate action.
If it is valid, why would you remove it? If you are calculating control, sometimes people remove out-of-control points.
To your point Darth. If you have a point out on the favorable side, you are right. As a general rule, the only time I recommend changing control limits is when there is an improvement–thus “locking it in.”
The forum ‘General’ is closed to new topics and replies.