Just another thought here.
What method is recommended for corrective action effectiveness, or root cause verification (just in case they are not the same)?
I have been told in the past, to truly determine whether true root cause was found as well as if the corrective action implemented truly eliminates the cause. Is to implement CA, wait for a determined amount of time (if this is a reoccurrence then you should wait the same amount of time between the occurrences). Then you would take away the CA for the same amount of time with a higher frequency of inspection, if the problem occurs during this time, then your CA has been effective.
This takes a long time, especially if your last occurrence was 2 million parts ago or 2 years. And none of my customers would want me to keep a CAR open for that long. My customers generally give me 15 days to show a completed CA Report, but if I have a problem 1 out of 2 million times over the course of 2 years how do I verify anything within 15 days?
:frust:

Should we be required to verify the effectiveness of a corrective action, as if it was some special activity related to a specific non-conformance ?

Wouldnt our normal continuous monitoring of process performance, complaints, etc tell us if the problem re-occurs ?

Hi Avogel,

I don't really get the problem...but let me describe the situation at our plant...If there has been an non-conformity inspected at the customers site..we fill in a complaint document and take the appropiate CA...

To measure if it's effective we periodically analyse the CA's taken...and look if the non-conformity has reaccured...if it's not then we assume it was effective...an other way (like undoing the CA) is almost impossible because most of the time we can't recreate the same processes in which the non-conformity evolved....

Hope this helps...
Greetings,
Martin

Dear Avogel,
I think that many companies, our customers among them, confuse corrective action, which includes a root cause analysis, with remedial action. Remedial action is usually possible in 15 days, but verification of a corrective action typically takes much longer. As long as the remedial action works and the problem doesn't recur, customers don't often follow up to verify the effectiveness of your corrective action.

I think a good corrective action system includes both short term (remedial) and long term (root cause correction) actions. I think Martin's process of systemically monitoring the effectiveness of corrective actions is a good approach. You might even be selective about which ones you follow closely, based on your knowledge of your company's strengths and weaknesses.

I think Rosie A said it well. I agree.

/Claes

Neither I see a serious problem.

We fill in a CAR form in the cource of corrective action implementation. It is N/C + C/A + C/A verification in one report. First we describe the NC in detail. Then we register, what we have decided to do with the NC (Control of non-conforming product etc..). Then we register what causes for the NC we identified. Then we put down what CA we decided to implement to eliminate the cause. Then we report on the results of the CA.

If the cause (the one we identified no matter right or wrong) is eliminated we close the NC for the time being. If not we investigate why and re-work and repeat the CA. If yes, but the NC repeats, we investigate to re-identify the cause.

The latter should be performed even if the lag between similar NC's is very long. Tehre must be a mechanism for retrieval of old NC's by keword. Otherwise we would not be able to learn by experience (mistakes).

This is how we do it.

regards,

Anton

Many times verification is an educated guess based on the CA, right or wrong. Does the CA match the root cause, is many times the verification.

Just back from a training course (on safety mgt.) given by a well-known registrar. The trainer (who has been used by this registrar for consulting work as well as training for a number of years) did not know the definition of "corrective action" as given in ISO 9000:2000, and argued that CA is not to eliminate the cause(s) of nonconformance but simply to reduce them to acceptable levels. While I can accept that the end result of CA typically achieves acceptable levels, the overall goal and definition of CA is to eliminate the cause. He would not believe me and argued voraciously against it. I told him to argue it with the ISO folks because that's the definition provided. He still refused to believe it and said he guessed he would have to "brush up" on the subject.

Scary. You'd think a trainer would know these things wouldn't you? He lost a bit of credibility in my eyes at that point. BTW, I didn't embarrass him in front of the class - the conversation took place on a break.

Sounds like the understanding of common causes of variation and special causes of variation is somewhat missing here.

When a complaint is received investigation should be undertaken to establish the root cause. It should then be determined if the root cause is a common cause, or a special cause. Common cause and special cause relate to the statistical performance of the process. Typically a special cause would require that you take action on that specific incident in order to remove it, common cause would require you to take action on the system as a whole if it was perceived that the process performance was statistically above acceptable limits.

Your argument Lucinda appears to be taken with the consideration that all causes of non-conformance are special causes, and similarly the training guys argument was taken with the consideration that all causes are common causes.

I think perhaps you caould both learn from each other on this one.

No Martin, I simply was applying the definition of corrective action. . That was the basis of argument. He did not know the definition of corrective action.

Although practice in applying a corrective action system may be to differentiate between root causes of nonconformance and take different approaches in the handling of them, there is nothing in ISO 9001 that outlines this practice. Be careful when you say that such and such is the correct way for doing things, even when the practice is common. Each company will employ their own way.

The definition of corrective action is provided in ISO 9000 and it simply states that corrective action is taken to eliminate the cause of nonconformance to prevent its recurrence. I don't recall anything that qualifies this definition.

Fine Lucinda

If you choose to rely on ISO9000 as being the sole source of knowledge regarding quality management then so be it. Unfortunately I think you will find your system wanting.

If on the other hand you really want to learn about quality I would a good place to start would be to read 'Out of the Crisis' by a Mr E Deming. This book will teach you to understand processes, and causes of variation (non-conformance).

Your simplistic view will be blown out of the water.

Regards

Martin

Lets be careful here. There can be major differences in Corrective Actions and Preventive Actions. A lot of people tend to get these confused. I understand that they could be the same, but the definitions are different. Plus, it will depend on what type of system you're using (i.e. 8D, CAR, etc.). In some instances you'll see that a CA is nothing more than something to protect the customer and no PA is ever done. I've seen the steps as Immediate Corrective Action, Corrective Action and Permanent Corrective Action. It's all basically the same, the bottom line is the understanding of true Root Cause Analysis. Call it what you will, simplistic or over loaded with statistics. If you don't know how to determine the Root Cause you'll either paint over it or baffle 'em with your BS. I like the KISS methodology myself and if you follow Dr. D's teachings you'll see that breaking down processes into their simplist form is how you understand what effect the variation has.

Here is another slant.

Consider a manufacturing process that has a process capability of 3 parts per million. That is, we understand our process performance from statistical analysis that it will produce 3 defective parts in every million produced.

Now the achievement of this level of defects is considered globally to be a measure of near perfection, and excellence of process performance. Acceptance of this fact is acceptance and understanding that all processes by their very nature vary, and we 'accept' that 3 parts in a million will be defective. Therefore if I identify a defect rate that pushes my mean process capability to say 6 parts per million then I would need to take corrective action on the system to bring the defect rate back to the measure of excellence that is 3 parts per million. Therefore my corrective action has not actually eliminated the non-conformances, it has just reduced them back to an acceptable level.

One might consider in this case the defenition of non-conformance to be a process performance with a mean value above 3 parts per million, and conformance to be a process operating within 3 parts per million. The defects still exist, but they are within acceptable limits.

This understanding is fundemental to quality. We naturally accept it with tolerances given on drawing dimensions. The designer would ideally like parts manufactured exactly to size, but he accepts that the manufacturing process has an inherent capability, and thus applies a tolerance to a dimension. This in theory is accepting (or tollerating) parts that do not meet the ideal value, and this is exactly the same in the above example of process performance on numbers of defective items.

I don't disagree with either of you.

One might consider in this case the defenition of non-conformance to be a process performance with a mean value above 3 parts per million, and conformance to be a process operating within 3 parts per million. The defects still exist, but they are within acceptable limits.

The only people who would argue 3PPM being acceptable would be the ones that received those straps on their parachute!!!!!:biglaugh:

In order to get your process to where the example is at, takes a lot of "drill down" into the basics of the process. Let's look at it like this: Occurance of the variable happens first. Defect is caused by the occurance and rejections are a result of the defect. We can limit the rejections by monitoring for the defect, we can limit the defects by monitoring the occurances, but we will not be able to get to a 3 PPM until we evaluate the occurances.
:agree:

Martin,

I don't disagree at all that the result of "corrective action" is typically a reduction in the occurence of nonconformances to an acceptable level. See my first posting.

However, despite what the overall result is, or what a company defines as acceptable, the fact is that the term "corrective action" is paired with a goal of "elimination" of the cause of the nonconformance. Nowhere did I say, or does ISO 9000 dictate, that the nonconformances actually go away. Merely that the intent of corrective action is to remove the identified cause. Eliminate it.

Perhaps elimination isn't possible - and that is where we measure the effectiveness of the CA. We draw a conclusion that the cause of the nonconformance has not been eliminated because it continues to result. But we may also say that we are happy to accept the amount of nonconformance and will not attempt corrective action any further.

Anyway,.... you can define corrective action however you like. The ISO definition is the one that has been provided as the internationally consistent definition. And I found it amazing that a registrar representative did not know it.

Martin, no need for cheap digs. You would be hard-pressed to find anyone who takes this task more seriously than I do and you should have noticed that by now.:p

Lucinda

There were certainly no intentional cheap digs in my posts. I beleive everyone in this forum has the best intentions, I was just perhaps trying to expand your concept of corrective action.

Perhaps we infact have to think of how we define non-conformance.

If we accept that a corrective action is something done to prevent non-conformance re-occurring then perhaps we need think of non-conformance being something whose variance is outside of accepted limits.

Therefore our prevention of re-occurrence of non-conformance could reasonably be to bring the variance back into acceptable limits.

For attribute data on a process this might be numbers of defects. For variable data on dimensional sizes this might be bringing size back into tolerance.

So from your disagreement with your tutor you are either both right, or you are both wrong (or have partial understanding), and maybe have to think again how you define 'non-conformance' as well as 'corrective action' in light of knowledge on how processes perform.

Does this make any sense ?

I am trying to help :bigwave:

Martin,

I agree with your comment regarding the need to define what constitutes a "nonconformance". That is always the crux of a corrective action system. But I don't feel a need to define "corrective action" any further than the definition provided by the ISO group. How that definition of corrective action is employed within our CA system is what I am at liberty to define.

Let's get back to the thread. You obviously disagree with ISO's definition of CA (at least you recognize that it exists - the "tutor" had no knowledge of it) while I have no argument with it.

Lucinda

The ISO definition is fine by me, so long as we understand 'nonconformity' and what the 'requirements' are, and what causes processes to vary.

M Greenaway said:

Fine Lucinda

If you choose to rely on ISO9000 as being the sole source of knowledge regarding quality management then so be it. Unfortunately I think you will find your system wanting.

If on the other hand you really want to learn about quality I would a good place to start would be to read 'Out of the Crisis' by a Mr E Deming. This book will teach you to understand processes, and causes of variation (non-conformance).

Your simplistic view will be blown out of the water.
We may be in different worlds. I don't interpret Lucinda's response as a 'sole source', but rather it appears she is saying that ISO has a definition (one of many) and that if you're 'doing ISO' or training it, one should know ISO's definition.

We all know that zero defects is possible, but (depending on the industry and processes) at what cost? ISO says that a corrective action should eliminate a defect cause. Sometimes reality is you can, in fact do that. But as often as not totally eliminating a problem does not happen. Look to an FMEA to rank and evaluate.

As a last comment, as Jim says we should all remember that a true root cause is seldom found. In addition, there may well be multiple root causes. Don't over simplify reality.

By the way, elimination of defects is the ISO definition, but I don't see where it dictates that you have to totally eliminate the (a) problem.

Marc

You say that 'we all know zero defects is possible...'

I would say that those that understand processes know that zero defects is impossible.

Also can you clarify the difference between eliminating a defect and eliminating a problem ?

I guess you are talking about correction as opposed to corrective action ??

Jim

That explanation suggests that you can eliminate a root cause but not eliminate a problem.

If it wasnt the root cause that was causing the problem, then you havent correctly identified the root cause have you ?

Then by definition you havent conducted 'corrective action' have you ?

Or can you say that you have, but it was totally ineffective ?

Right Jim

Does ISO9004 mention how you determine root cause, i.e. fault trees, Ishikawa diagrams, 5 Whys, or any other method ?

Jim

So the ISO9000 series of standards tells us only what we have to achieve, without actual providing the methods of HOW to achieve it ?

As such the ISO9000 experts out there cannot actually improve our business because they cannot tell us HOW to achieve the wonderful things in ISO9000. Which is perhaps as much use as being told you have to fly to the moon !

(can you see where this is leading ;) )

Just a project I am considering ...........