Occurrence and Detection Ratings
Your questions are very good ones. Let me deal with each one individually.
Question #1:
The Occurrence (O) is the frequency of failure mode or the frequency of failure cause?
Answer #1:
The occurrence rating is the probability that the the failure mode will occur due to the cause. In order to calculate this rating, you must multiply the probability of the cause occurring times the probability of the failure mode occuring when you assume the cause is present. This may sound a little complicated so let me explain. If you ask most companies if they ever run with worn tooling, they would say yes. If you then ask them if the worn tooling always results in a defect or failure mode, they will say no. Consequently, if you want to know the true probablity of the defect or failure mode happening due to a particular cause, you must perform the calculation above.
Unfortunately, in 24 years of doing Process FMEAs I have never seen any company who has the required data. The lack of this data leads to countless arguments that are a tremendous waste of time. There is good news in that there is a way to overcome this problem and I will begin to provide the solution by answering your second question.
Question #2:
If the O) rating is for the frequency of failure casue, how can we use the statistic data to count it because normally, we monitor ppm or % defect for only failure mode.
Answer #2:
As described in the answer to Question #1, the occurrence is the probability of the failure mode due to the cause. The ppm or %defect data that you have represents the probability of the failure mode due to all of the causes together. Although this is not what the Process
FMEA is asking for, you can use the ppm and %defect data that you have to establish a maximum occurrence rating for any cause linked to the failure mode.
Question #3:
Also, for the Detection (D), this is for detection of failure mode or detection of failure cause?
Answer #3: It depends on the type of control that you are talking about. If it is a prevention control, it is about detection of the cause. If it is a detection control, it is about detection of the failure mode.
Here in lies the problem with most of the literature published about Process FMEAs today. They tell you to concentrate on Type 1 controls or prevention controls. However, when you look at their detection rating tables they only talk about detecting failure modes. They do not tell you how to rate prevention controls. Some books go as far as telling you that if you do not have a detection control you must put a 10 in the detection rating column. This simply does not make sense if you take the time to consider that if you ever created the perfect process were all sources of variation were controlled through prevention controls, you would never inspect the product. However if you follow the majority of the current Process FMEA publications in print, you would have to put a 10 on every row thus sending your RPNs through the roof.
I want to congratulate you on your excellent questions. I have seen literally hundreds of people struggle with them over the years. There is way too much time wasted when people try to assign ratings to their Process FMEAs.
Rich Harpster
Harpco Systems, Inc.