Okay, I live in a mobile home. We all know they are magnets for tornados, so I develop a preventive action, in which I duct tape empty Spam cans to the home’s exterior. How could I prove the effectiveness of the preventive action? How can any preventive action be “proven”?

Point well taken. I can't argue against it. I sure would love to hear some examples of proof of effectiveness of a preventive action. This could turn out to be an interesting thread. :thedeal:

The only "proof" of effectiveness in a preventive action is that it never happens.

In the example, if a tornado tears through the area and the mobile home shows no ill effects, that would be enough "proof" enough for me.

Dave

I agree with D Scott. If it has not happened,it may have been effective. If the PA is a result of a trend, then you would see the trend reverse. That is, if you see trends as a source for PA

If you are into this kind of thing, FMEAs can help evaluate the preventive action by showing a reduced RPN.

I wouldn't accept it as "preventive action" just because the tornado didn't hit it unless you could somehow prove that the action taken was thought out, planned, and had a basis in actual liklihood of success.

If you had done research that showed that metal attached to the mobile home would reduce tornado strikes, and that spam cans had the particular shape that was most likely to be effective, and then in a situation with potential for a tornado to hit the trailer (as Dave suggested) the tornado avoided the trailer, ....Then I would accept it as preventive action.

Preventive action must somehow be tied to the event. I can't say that having a box of tissues on my desk is preventive action for getting neck strain reading this little laptop screen. They are unrelated.

First prove the relationship. Then take the action. Then measure success by the event never occurring or being less severe than typical.

So this would be FMEA, as SteelMaiden suggests, right?

-Trend reversal I would see more as a corrective action since you would be changing something already in existence.

If it hasn't happened, how do you know you need it.:rolleyes:

Sam,

We all have seen those looming "accidents waiting to happen". I hear people coming up with PAs all the time. "ya know, if we decrease xxxxxx at this point, we can reduce our risk of having yyyyy happen" The definition of preventive is to keep it from happening, therefore you are dealing with possibilities not actualities. Will we ever know for sure that the situation would or would not have happened? nope, all we can do is assess the posibilities through statistics, knowledge and previous experience, or the phase of the moon. Whatever we use, we just need to have our ducks in a row and be able explain why.

Something like being on the debate team in high school, you never know if you'll be arguing pro or con.

Have a good one!

I use my “mobile home” scenario in classes all of the time. My contention is that you cannot “prove” preventive actions, just because the event you are trying to prevent doesn’t occur. If you correlate the lack of event occurrence with the opportunities for occurrence, then you are closer. I agree with Lucinda and Steel Maiden. It does bring out some good discussions though!

Now, if I could only get roadkill in the scenario.....:biglaugh:

The assumption thus far is that a preventive action leads to a reduced negative event (lower RPN) or a nuetral event (never happened but as gpainter points out, do we really need it?). Can a Preventive Action lead to improvement and can that improvement be measured?

Anyone?

Regards,

Kevin

You sure know how to pose them Kevin. :ko:

I'll jump right in and get the ball rolling.

No - preventive action cannot lead to improvement. If it did, it would have to improve something from a less deseriable position (corrective action). We all have been through the grinder on why corrective actions can not be considered preventive.

I disagree that a liklihood of success has to be proven to judge the effectiveness of a prevention. I certainly see your logic and in an auditing sense it would certainly establish effectiveness in my mind. But must you see and understand the reasoning for an action if it prevents a non-existant event from occurring? The fact that it didn't happen has to stand as evidence until it is physically disproved by occurance of the event. NOW you can say it was ineffective. The responsibility of establishing effectiveness is satisfied by the non-event. The responsibility of disproving the effectiveness still remains open (with the auditor in this discussion). To judge the reasoning and application of the action goes past objective evidence as found and becomes judgemental. If you can't accept it unless it is tied to research and science, better stay clear of Haiti. :ko:

Dave

Yes, Kev, I think that a preventive action can lead to improvement.

While i understand what Dave has said regarding improving something from a less desirable position, I don't think that all improvements are the result of corrective action.

Let's remember what Corrective Action is: correcting a nonconformance.

There are many occasions where you see the opportunity for improvement prior to a nonconformance happening. That is what Preventive Action is all about. In preventing a noncon you can reap rewards in other, measurable ways.

Simple example: Take a vitamin C to prevent catching a cold. Never catch a cold = preventive action. Improvements as a side effect: Clear skin, better attendance at work, sound sleep, and reduced free radicals wreaking havoc in your body.

P.S. Isn't preventive action the part we are supposed to feed into the continual improvement loop??

If we through in the concept of variation, would corrective action be taken on "special causes" and preventive action taken on "commen cause." This of course would be common cause variation for a stable in spec process that we are trying to make better. Then it is measurable. But what are we "preventing?" (Thinking out loud now) I suppose we are preventing potential problems? Dunno. I was just doing some training slides on common and special when I read this thread.

Others?

Hmmm;)

The way I see it preventive action is effective in two respects.

If you are using preventive actions to prevent a non-conformance from happeningagain then you can measure the effectiveness if the non-conformance doesn't happen again.

Next point, in order to prevent something from happening at the outset you need to conduct a risk assessment including actions that can be taken and likely outcomes. Also you must include the 'do nothing' option and its likely effect.

When you have come upon what you believe is the most effective action, then you implement (this is the only way to test its effectiveness!:D )

Having implemented and embedded the PA into your system two things will happen.

1. The effectiveness of the preventive action will become glaringly obvious when what you were trying to prevent happens...oops!

2. What you were trying to prevent doesn't happen. Your objective has been achieved.

Remeber that what you were trying to prevent has been outlined in your risk assessment. If what happens matches your likely outcomes as defined in your risk assessment. Surely, your preventive action has been effective.

IMHO, this is basic PDCA!

Is this making sense?

Laura

How would you know that your action taken to prevent something happening actually was the factor that stopped the thing happening.

I.e. if you hadnt taken the action would the problem actually have occurred ?

The only way to determine this is to selectively remove the preventive actions to see if the problem occurrs - which is a bit of a painful waste of time.

Also the ISO defenition of preventive action can never be applied to stopping a non-conformance from happening again - this is corrective action.

Personally I think taking preventive action is impossible if you use the ISO description. Even with FMEA you are using your experience of things actiually going wrong - hence your action is based on occurrences.

While I respect the other opinions and see good logic in what has been said, I am choosing the "con" side of this on the premise that a Utopian situation cannot be improved. If by definition prevention cannot be based on an event that has happened (otherwise a corrective action) then it would by the same definition be impossible to cause an improvement. Can perfection be improved?

Lucinda makes a great argument with the cold scenario but because the resulting "improvements" are known to have happened and therefore caused a prior "nonconformance" (or undesireable situation), taking the vitamin C cannot be considered a preventive action. It must therefore be a corrective action leading to continuous improvement.

We had a problem with contamination in open bins. Our corrective action was to cover the bins with cardboard. It works great. We can't now claim covering the bins (taking vitamin C) as a preventive action with resultant improvements of no mixed parts, no dirt, etc. (clear skin, attendance, etc.).

Once the problem you intend to prevent has occurred you lose the right to claim it as a prevention and must be classed as a corrective action according to the rules.

That's my story and I'm sticking to it.

Dave

MG

If you are documenting a process, conduct a risk assessment on each section of the process.

When you look at each section in turn you will be able to identify where areas of non-conformance may occur.

(If we do this bit of the process what could happen? Could we non-conform? Are there any holes in this section of the process which may lead to non-conformance).

By this identification you are recognising a potential non-conformance.

From this you look at ways of preventing this non-conformance from ever happening. Isn't that the aim after all?

Also, as you have conducted this check on your process, you could document this activity and this would be evidence that you have identified and prevented a potential non-conformance.

Also, your point about corrective actions. COrrective action can also be preventive action, by its very nature.

I can also give you an example of preventive action.....a friend of mine used to run a post room. One morning the post didn't arrive. The business would not have been able to proceed without the post which would have meant a non-conformance. So my friend made enquiries with the post office and arranged for him to collect the post. I believe that this action prevented a non-conformance.

A simplistic example, but then isn't is easier to keep it that way?

:thedeal:

Lau.

Dave - you get my point exactly.

Laura - I think you are drifting from the ISO definitions of Preventive action, as opposed to Corrective action.

Even a risk assessment of a new process is based on the experience of troubles in a similar process, hence can it really be considered as Preventive action as defined by ISO ?

Just because you are now foolproofing a new system you have somewhere in the past experienced the problems that you now foolproof against. That is how you make a considered judgement on the rpn. Hence you are performing corrective action arent you ?

Any action taken on past bad experiences must be Corrective.

A better example of preventive action, and the only example which I can think fits the ISO description, is when an operator adjusts a machine based on a trend line given on a control chart that hasnt reached the upper or lower control limit.

Martin, I know that your example of an operator adjusting a machine is just an example, but if he/she is making the adjustment due to a trend, isn't that corrective in nature?

My original scenario involving a tornado and a mobile home…..according to Dave (Scott) would also be corrective because I’m basing the action on the history of tornados and mobile homes (even if I’ve never had a tornado hit my home before). Hmmmm, me thinks I need to ponder this a while. Thank’s Dave, you might have ruined one of my favorite teaching scenarios. :( :frust:

I hate it when I have to think!:smokin:

I think that the action taken by the operator has occurred before he has actually produced a defective item - hence it is preventive.

But you might argue that the control chart process has been put in place due to the fact that people made defective products in the past - hence it is corrective action.

The answer is - there aint no answer, it is (dare I say) another stupid clause.

Or maybe I am taking the interpretation too far.

But following the argument through it would actually be impossible to perform a true pure Preventive action.

Dave, Excellent points. I have chosen the 'con" side also.

Let me use your contamination solution as an example:

"We had a problem with contamination in open bins. Our corrective action was to cover the bins with cardboard. It works great."

"As I walk through your shop at lunchtime I see some people playing cards on the cardboard covered bin; I think to myself, they might damage the cardboard and allow contamination to enter the bin. So I make the suggestion that you might want to change something to PREVENT this from happening.
You run back to your office and issue a change to replace all the cardboard with 1/4" Lexan."

Is this a preventive action: Yes, based on your observation that there is a possibility the cover could be damaged.

Is this an improvement? No, you did not improve the method keeping contamination out of the bin.

Is this a corrective action? Only if the cardboard had been damaged.

IMHO, of course

Good grief. No one ever said that the event has to have never before happened in the history of man!!!

Only that it isn't something that has already occurred in your process!!

You are certainly allowed, no, you are expected to use reasonable thought and judgement and any other tool (past experience, test results, etc. )you can put your finger on to make the preventive action decision.

If it is something that you suspect may happen based on X and Y and Z and you reason that by doing A the risks of it happening would be reduced or eliminated, then you have done your job.

Surely you have all used fishbones...

How do you know your action is what prevented it from occurring? You only have the logic that you used before and as long as the event doesn't occur, you can happily assume that what you did was correct.

Sam - going back to earlier posts - how would you prove the effectiveness of your preventive action on the bin lids ? Would the contamination actually of occurred ?

Lucinda - absolutely right in practice - again its the definition of the standard that is ludicrous.

I agree completely (with the exception of the duration of this discussion) with your logic and conceede that we all, as good quality professionals, use the strategy you outline to recognize and improve our process.

I am on one hand feeling the fresh breath of a "great debate" and on the other worried that some people who are really looking for knowledge will be even more confused. For anyone in that catagory, please look to Lucinda and the other side for the "right way" to determine preventive actions.

That said - IMHO we distinguish between the two because our auditor said there has to be a distinct difference. There is, in reality, no difference. The two go hand in hand. If a corrective action didn't "prevent" something from happening, it would be considered ineffective. When we do something to prevent anything from happening, it is a corrective action to "something that happened in the history of man". Can they really be seperated? I don't think so, unless you want to split hairs and create a whole list of "exceptions" and "ifs".

When the original requirement was written, if memory serves (which is not always the case), corrective and preventive action were addressed in the same paragraph. Now they are seperate.

:rolleyes: Supposition Follows: :rolleyes: The intent was to ensure we looked at and corrected our problem areas including looking at the possibility of other things going wrong which we didn't necessarily expect. We all did this and when we corrected a problem on one process, we looked at the process next to it and said do that one too - Presto! - instant prevention. That is all too easy; so because it was initiated by a prior occurrance, it doesn't count.

We are even instructed by the QS-9000 manual to use the same drivers used for CA in determining preventive actions. The only difference is the word "potential" which is overused by auditors. The simple fact is that ALL future nonconformance must by definition be "potential". It doesn't matter if it happened one million times in the past, all future events are "potential".

Just as MG states, it is the tangled and twisted definitions of the standard that even create the distinction.

But it sure makes a great discussion topic :ko:

Your turn.

Dave

Nice contributions here folks!! Lots of good discussion and dialogue.

Laura M. was thinking along the lines I was in regards Common and Special Causes and how one might measure progress. I also agree with Lucinda’s view that we must narrow the focus to include things relative to our process/product under concern. Our valuable experience in one area, old job to new job, might be applicable in another situation. Provided the item under consideration hasn’t occurred in the new playing field, I would have to agree actions taken are preventive in nature. The FMEA example is a good one for thought: some time ago, this example was raised with the same logic being presented. I think that by taking Lucinda’s comments to mind when considering whether the FMEA is driven by past problems (corrective in nature), one should consider the new playing field that they are on. Just an opinion.

Dave usually buckles my knees with his responses (thanks Dave!). Dave posted, “…by definition prevention cannot be based on an event that has happened (otherwise a corrective action) then it would by the same definition be impossible to cause an improvement. Can perfection be improved?” This caused me great pause. Here is my next spin: perfection can be improved. Consider the implication of future state of any process or good. B&W televisions may have been perfect in their day, but when color came along, the paradigm shifted. As the paradigm shifts, so does the definition change. When I posed the question regarding the various states we tend to drift to, negative to neutral, I was thinking that we may have cut our search out to soon (I think we tend to do this too often).

In my example of the television, a preventive action might have been innovation, a means to delight the customer (a positive effect, perhaps measurable in the declined sales of B&W TVs). To prevent sales loss, we took steps to maintain/grow market share (or the market itself). Sales were stable and predictable. With the advent of an innovation, sales were maintained and eventually increased (trend analysis). I guess that one could argue that innovation is a special cause, and as such, argue that it is a corrective action. But I have a feeling that when an action is taken to maintain/increase the future state, one has taken a preventive action towards constancy of purpose.

Well everyone, have I had one to many sodas this afternoon (running on sugar)?

Regards,

Kevin

Dave (and others),

Good points. It is a good topic to discuss. I am constantly amazed that as I think that this might be put to rest, someone raises another interesting angle. From my perspective, this creates excitement (I should find a life I guess)!!

I read somewhere a contribution stating it this way: preventive or corrective in nature, it doesn’t matter so long as it drives improvement.

True enough, I suppose, but Sam’s post is rather important in regards making the distinction between change and improvement. Changing a 40 watt bulb to a 60 watt bulb might make things brighter, but is this change improvement? What were you trying to correct/prevent? Is it measurable?

My head is spinning!!

Kevin

Kevin can you expand on your ananlogy.

My question is what non-conformance have we avoided by creating colour TV ?

i.e. is innovation really a preventive action ?

Somehow I missed this thread when catching up on the last couple of days...so I'll jump in now with both feet to the already choppy water.

When analizing data and following negative trends which can indicate that a nonconformance can/will happen if the situation isn't corrected, then yes a preventive action (in that specific sense) can be taken and proven effective by an ensuing analysis showing a positive data trend.

Obviously if allowed to continue, the previous senario would have lead to a corrective action. However by stepping in (whether you are adjusting settings, changing a belt, cleaning a mold, etc...) you are preventing this occurance...saving yourself several types of waste (scrap, delay in order delivery, machine down-time, employee idleness, etc...) and maintaining a much more efficient process.

Are there cases of preventive actions where effecitveness cannot be proven?

A senario for this example would be (this is in a plastic heat seal operation):
A square platen (ID is etched on the top - mounted surface) with four corner bolts can be mounted incorrectly and create a bad seal. So far, this has not happened. Improvement suggestion received to alter platen by marking the front side of the platen with ID (facing out when mounting) to avoid a possible nonconformance and also provides a visual checkpoint to ensure the correct platen is mounted for that product.

Mgmt considers this worth the cost to put in place. Can the improvement be measured? Not in measurable sense other than it not being a cause for bad heat seals. But (IMO) yes... Operator isn't second guessing or wondering if they did it right, they are able to check without removing the platen (which would waste process time and affect takt time). You are avoiding possible scrap thereby making your process more efficient.

Lighter Side...I'm sure everyone joins me in hoping this never needs to be proven out, but for your consideration Dave:
Are Dave's spam cans an appropriate tornado alarm?
Have they been weighed down so only >? (example: >50 mph) wind will make them jostle? Can they be heard over the patter of heavy rain (and other known noises associated with tornados) in addition to TV or radio (not to mention conversatioin/kids/etc...) internal home noises?
Can this be tested by simulation? Yes
Is it worth the effort and costs to perform the test? That would be Dave's call but if he's thinks this will work and feels there is value...why not? Of course he could evaluate historical data as part of the process, but when the protection of our home and loved ones are concerned it may not be enough to sway us from our determined course to protect them. Some historical data could be: How often in the recorded history has a tornado occured within 1 miles, 2 mile, 5 mi., etc.. (convert to km if you like). What has been the frequency? Due to topographical configuration some areas are more likely to experience tornados than others but that does not guarantee that it won't happen.
Even if there has never been a tornado in recordable history to come within 20 miles of the home, can it happen? Yes...Can you estimate the likelihood? ??? I don't know.

Back to the daily grind for me...
Eileen

Eileen, I see what you are saying. If the trend line indicates a future failure, and you alter the course (make a correction?), and you can show the failure did not occur then the action taken could be considered preventive.

Okay, I can buy that. One thing to remember is we want to improve. Does it really matter if we call improvement preventive or corrective? (excluding ISO-speak).

BTW, Since my Spam can installation, my tv reception is amazing! Also, space aliens cannot scan my home. Now, Hormel (the manufacturers of Spam) want to buy my mobile home to put into the Spam museum! :biglaugh:

Dave,

Tell them nevermind about the museum...tell them you want to be featured in a series of commercials. As the new spokesperson, you can endorse their products of being more than just a satisfying meal...they will improve reception, warn of inclimate weather, and perhaps have other healing properties yet to be discovered by the public at large. Of course, you would be paid handsomely for your hearty endorsement!!

Great thoughts folks!! Keep them coming!!

Kevin

Eileen,

I agree wholeheartedly!

This is a wicked post, I am following this avidly!

Lau.

Great thread going here. I was really interested in the question, can PA be measured for effectiveness. I would really have to say no. If the event you were trying to prevent does occur, do you toss out the PA?

An example.....

We do machine maintence on a VERY regular basis. We have daily, weekly, monthly and yearly item that we perform. We just completed our yearly maintenance on the turrets for our punch presses. All of this to reduce the possiblity of machine downtime for repairs. But even through all our good efforts, we will loose machine time because of failure of a component beyond our control (i.e. a relay inside the control box, a bearing in the brake siezes) and we could loose up to a week of machine time (sometimes more). Do we claim our PA was ineffective because we had downtime. No, we do not. Do we take corrective measures to reduce downtime? Yes we do (i.e. keeping a stock of relays on hand). Do we now call this CA?

When does CA become PA, and the other way around? The lines can begin to blurr.

CarolX

Response to "If the event you were trying to prevent does occur, do you toss out the PA?"

Don't toss it out, but do update your records to reflect that the PA wasn't effective. Then reference that PA # in your CA to close the loop (and create not only a problem solving path but also an audit trail).

Carol you stated "loose machine time because of failure of a component beyond our control"

Does this mean that the component failings isn't an item covered under your PM system?

Example of failure of the equipment PM system (at my location) is when downtime indicates the frequency in which your are performing PM is not adequate (resulting in extra unplanned repair time).
This could also be when having a problem with component that isn't on the PM but is recurring.

Better explanation? Eileen

Okay Carol Let's try this. You replace the bearings because you know that they are towards the end of their useful life (predictive maintenance). Two weeks later, one of the new bearing seizes (defective part). Was your PM effective? Was your PM PA that AM? :confused:

One reason I started this thread is I too have trouble seeing the blurry line. The longer this thread lasts, the more confused I am (and I thought I reached my maximum confusion saturation point) :bonk:

Hmmm - It would seem to me that if your PA failed, it could no longer be classed as a PA - it would have to be a corrective action because of the failure. So the proof of an ineffective preventive action is actually a corrective action generator.

Would logic have it that in order to reduce corrective actions you should avoid preventive actions? After all the PA was a gamble in the first place trying to prevent the potential of something which has never happened from happening.

"Oh what a tangled web we weave ......."

Dave

This is probably going to just confuse things more, but that's never stopped me before...:vfunny: :biglaugh: :eek:

I look at our entire preventive maintenance as preventive action (notice that singular, it will make sense as I ramble on). I think to look at each and every pm as a pa lends itself to the situations described above. If you pm and have a defective part (no way for you to know that) it doesn't really reflect on your system. Now if every one of those parts that you change in the year are defective, it may point to some deficiency in your purchasing and SQA. Does this make sense?

I'd say if your PA failed it would no longer be a PA, it would be a 'waste of effort'.

Ineffective PA's dont generate CA's themselves, unless the PA actually caused the problem. In the case of the new bearing siezing it wasnt the PA of replacing the bearing that failed, it was the quality of the bearing itself. Your CA might be to purchase better bearings in future.

Hey if I say that I have taken a PA by painting the roof of my house white to avoid a direct meteorite hit is my PA so far successful ?

Yes it is ludicrous.

Greetings from one of the newest members to the Cove! Here's my 2 cents worth (that's 0.02 CDN, though)....

My organization has four possible reactions to a situation. The "index" for these four options is detailed in our Nonconformance and OFI Guidelines Matrix, which details the parameters for each of the options. When a situation occurs, we refer to the Matrix which will provide guidance on how much action is required. If there is more than one possible option to take (things aren't always black and white in the land of ISO, are they?), we take the "higher" route.

1) A "Nonconformance" is a simple discrepancy with the system, requires some form of corrective action, and life goes on. A quick-and-dirty fix to an issue easily addressed.

EXAMPLE: A small windstorm blows through your trailer park and results in a few garbage cans impaling themselves upon your mobile abode. Action? Tape some Spam Cans on your home - scientific studies have shown this to be an effective deterent against strong winds.

2) "Corrective Action" is required when there is a break down within the system or when the item is rather extensive and will require the use of various problem-solving techniques. We document the Root Cause, Action Plan (and Results), and the Verification Plan (and Results) to demonstrate that the discrepancy has been addressed appropriately.

EXAMPLE: A tornado rips through your trailer park and, for reasons unknown to us, Mother Nature decides that your palatial residence should be a riverboat instead, and places you in the middle of the nearby lake. You buy a new and improved trailer (this one already has the pink flamingos on the front "lawn"), and tape Spam Cans all over it, and even duct tape the perimeter of your "property". As we all know, nothing beats duct tape. Your verification plan is that no tornado will touch your home for a period of three months.

3) "Preventive Action" is implemented when there is an adverse trend. Our Key (Performance) Indicators are monitored on the stop-light system (red, yellow, green). Two consecutive yellows, followed by a red means that a PAR is required. This system was implemented before I arrived, and while I don't find it's the greatest, it's not an item I'm will to change at the moment. (bigger fish to fry within our QMS :) ) Root Cause and Action Plan is usually easy in this area. For the Verification Plan, as this based upon an adverse trend, we designate x-number of months as the time to monitor the issue and verify that our actions have reversed the negative trend.

EXAMPLE: The number of tornados that have ripped the walls off of your home and resulted in your fellow trailer park residents seeing you decked out in a mud mask and pink curlers has increased over the past several years. You decide to tape Spam Cans to your trailer, and will monitor if this adverse trend reverses.

4) And finally, we come to an "Opportunity for Improvement." This is for items where the system is working beautifully, but there may be another way to make it more effective and/or efficient. Root Cause is not possible here, neither is a Verification Plan. Action Plan and Results, though, are possible. The recipient of an OFI is to analyze it and determine if it is feasible/practical/appropriate and has some form of ROI. Depending upon the analysis, the OFI is either accepted or rejected. If accepted, an Action Plan is developed and the results documented. This is how we're addressing the Continual Improvement aspect of ISO 9001:2000.

EXAMPLE: You've survived countless tornados and ridicule from your neighbours (they just can't forget the sight of you in pink curlers)..but, hey, you're still alive. OFI here...move. And the local real estate agent has heard about some lovely condominium buildings being built in the Everglades.

Rather simplistic examples, I realize, but it's Sunday, the sun is shining, and there's plenty of snow to play in. Life doesn't get much better than this!

Boy I am glad were not all Lawyers.

Just got around to reading this thread, and keep getting hit by fractured cliches.
"How many ways can you split a hair." or
"How many PA's fit on the end of a pin.

IMHO:
PA's are mostly undertaken as a part of some sort of advanced planning. This is where we undertake to determine potential problems, and risks and to put in place tools to monitor output to prevent problems.

CA's follow only when the output of a process becomes non-conforming.

For example. My control plan calls for monitoring a process output. This is a prevenative action. But at some point during the run we produce bad parts. The failure is caught because we are monitoring. (good thing) However our origional assumption about the process was incorrect and we took readings too infrequently. (Bad thing) Now we institute a CA. We isolate the bad parts and increase the frequency of readings. At the new levels we find that we are able to detect when the process is trending off and are able to correct before non-conformance occurs.

Now, 6 months later I prepare a different job which must pass through the same process. Because of my new understanding of the process, monitoring is at the new, more frequent level. This prevents any non-conformances in this NEW PRODUCT. This is again a preventive action.

The difference between the two is that the first time, the PA was not effective. However, because of the CA taken at that time, the PA on this new job is effective.

It may not be the book definition, but it lets me sleep at night:cool:


James