First let me thank all of the Elsmar Cove Forums users, you have helped me so many times.

I work for a low volume auto manufacturer and while working on a PFMEA for for assemble steps i realized the the "standard" AIAG occurrence scale of .01 per thousand thru 100 per thousand did not work well when I look at our planned yearly volume of less than 500 units.

Have any of you "adjusted" the occurrence criteria and what was your justification for doing so?

kdee122

What has your history of failure been? If you have some history of inspection rejects that would help to calculate your new RPN.
Andy

It is not unusual for organizations to create their own ranking number scales based on the uniqueness of their product/process. Customer will look at as part of PPAP submittal.

I never tried that while I was still in the automotive industry, but after leaving, I tailored all three scales to be more relevant to the particular product line.

We found that if you specified your particular situation well enough, the arguments over what rating to give Severity and Detection situations virtually ceased making a FMEA proceed much faster.

I changed the Occurence ratings to something an operator had a better feel for such as 1/hour, 1/shift, 1/week, 1/month, etc. This works very well if your products fall within a "product family" of consistent production, but would not if your product lines are diverse (e.g., different processes, different failure modes) and production rates are not even for each.

You might be able to develop a scale based on the occurrences per order, and you are not locked into a 1 to 10 scale. Use a 1 to 5 scale if it fits your needs better.

You need to document and consistently apply these new scales for ISO/TS as well as the RPN threshold for corrective actions, but your customer may have the final approval.

We are setting up a brand new 'plant" with a new design, so there isn't really much history to go by. I have dug up a few generic failure rates as a guide.

I have set my scale as follows using information from qualitytrainingportal.com

10=more than 1 occurrence per day
5=1 occurrence ever six months
1= 1 occurrence every 5 years or more

What do you think of something like that? The good thing is that as a OEM we have lots of flexibility.

kdee

The scale should be whatever works best for you. If you feel that this scale will work for you then try it out. Remember that the real purpose of an FMEA is to identify potential risks and prioritize them for improvement. The absolute value of an RPN should not be the issue. The relative risks of the potential failure modes should be.

The bigger issue for you is the new design issue. If you have experience with similar designs/processes, you can use the occurrence of those designs/processes to determine your occurrence rating. Most of the opportunity for an occurrence is tied to the process itself. The design can modify that occurrence up or down by its manufacturability. If the new plant uses an existing process from another plant, try to involve operators from the other plant.

Tread lightly. The OEMs and many Tier 1s either have their own rating tables or expect you to follow the AIAG suggested tables. If you modify one or the other, you will need to be very well prepared to sway your SQE into accepting your modifications. There is still the attitude in automotive that the customer knows best - even though you produce for them because you are the "expert" - and they want to see the PFMEA in a way they are "comfortable" with. There are still too many SQEs who want the PFMEA to be "their" document as opposed to being a useful tool for you organization. I'm not saying it can't be done, but it may take some selling on your part.

I don't need to worry about that since we are the "customer". It's nice to finally not be a supplier.....:)

Looking at other internal FMEA's I see a large variation in occurance number for common process, it's very hard to use the "normal" scale when your yearly volume is a few hundred pieces.

kdee122

If you have current data on inspection & test failures, you can use that to generate your ppm numbers.
Andy