When can you stop root cause analysis - Is the answer never?

Are there any guildelines on when root cause analysis can be stopped...or is the answer never?

Say you have a product where you weren't doing complete testing. Because of this some small level of rejects reached the customer.

You now add a new test that prevents the rejects from reaching the customer. But, the 8D process requires that you find the root cause of that failure mode.

After going so far in root cause analysis, you find the problem is a random defect, and the defect level is in the low ppm range. But, you haven't been able to find the exact location of the defect or what process step is causing the defect. Knowing that the defect is random, it is in low ppm level, you have general programs to reduce defects, you can test out the defect, and that the defect doesn't come close to being your number one yield loss, can you close root cause analysis?

Are there any publications that provide guidelines on this?

Root cause analysis does not always result in the definitive root cause. The process you used seems to me to be technically correct. I think that it can be closed sighting random variation.

I would apply the rule in 4.14 "...shall be to a degree appropriate to the magnitude of problems and commensurate with the risks encountered". If you have reduced the problem and eliminated the risk, work on something more significant.

If you are QS compliant, apply the rule in 4.13.2.1 for nonconforming product "The supplier shall quantify and analyze nonconforming product and establish a prioritized reduction plan..."

Eliminating a problem is great, when it is "normal". It sounds like it is no longer "normal".

[Kevin Mader]

What is a reasonable risk?
What level of resources can be committed to this problem?
What is required of your Customer?

In my experience, CAs are issued too liberally. I believe this is what creates many closure problems. People get them and don’t know what to practically do about the problem. All systems and processes produce some level of nonconformance. Sometimes it matters. Sometimes it does not. This is where managed risk needs to enter the equation and the reasonable expectations of your customer. This should suggest the level or degree of resources that should be spent on rectifying the situation.

Using the 5 WHYs or 8D are good tools. I personally try to trace the root cause back to a system/process parameter or a Quality Program oversight. I believe in the 94/6 principle, that most problems are caused by the system/not the people.

Regards,

Kevin

I agree with Kevin. CA's should be reserved for systemic problems, for example a recurring nonconformance of a particular type. If you follow this rule, then the root cause analysis is more likely to get to a system (rather than people) problem. When you overuse the CA system, it has a tendency to become a nuisence and often ineffective. Sound root cause analysis is usually the first thing that suffers.

As to when to stop root cause analysis, I generally suggest you keep going until you get to a system issue. If the root cause analysis was insufficient, the audit of the action taken should show you that, and a new CA can be initiated.

I, would agree, once you reduce the PPM to a very low level ie;0-15, and it is a random occurence, you should close the 8D. Having said that, if you have the budget, or your customer will not accept the risk of that level of quality, you should try to investigate some more.

regards,

Gerald

To add to Kevin and Dan,

In these days of lean manufacturing we have to allocate our resources wisely and try to make sure there will be a tangible ($$$) benefit to the expected outcome. It comes down to defining what type of situations really need to go through a formal and costly corrective action process.

Also as said before, it is the systemic issues that need to be addressed through the formal corrective action process. Hopefully, if their is a robust system in place that constantly uses that wonderful tool of ongoing FMEA review, issues for corrective action can be determined.

All that said, there is the dreaded customer complaint that requires a formal response. They want an answer and they want it now.

We are now in crunch mode even if it is an isolated incident. Don't skimp on the basics, go through the process of containment, brainstorming, fishboning etc... and all important, the 5 why's. Now is the time to physically visit the customer and show that they are important to us.

I have found that having a scheduled visit to each customer that includes interraction with line personnel, the lower the complaint rate. I believe that this is because most complaints start at the production floor and work their way upwards. Keep the line people happy, show you care about them and your fallout rate can be decreased.

(Some aspect of human nature that other members of the forum can expand upon)

ASD...

[This message has been edited by Al Dyer (edited 28 March 2001).]