Hello Everyone,

I am new to this Forum and new to FMEA. I have done some research and read some books on basic DFMEA, but examples always seem to be based on fairly simple scenarios. I am trying to figure out the best way to organize an FMEA for a complicated system, and can't decide if it should be by system, performance metric or part. I will use an automotive engine as an example (the product is not automotive, and for those of you in the automotive industry, please forgive the inaccuracies.)

Lets say you have a multi cylinder engine and one cylinder has low compression, and a twisted compression ring causes that low compression. So I can organize an FMEA (after the fact - realize I am not supposed to have the answers to start) three ways:

1.
A: Item/function=Engine;
B: Failure mode = poor fuel economy, rough running, higher pollution etc;
C: Potential effects = customer dissatisfaction;
D: Potential causes = poor combustion efficiency; high friction, bad engine controller

In this case, we never get to the "defective compression ring"

2.
A: Item/function=Combustion efficiency better than 75%;
B: Failure mode= poor combustion efficiency, high friction, bad controller etc;
C: Potential Effects: poor fuel economy, rough performance, and high pollution;
D: Potential causes (separated for each failure mode) = wrong piston, defective rings, bad valve; bad bearings, wrong oil... ; water damage, bad solder joint....

3.
A: Item/function= compression ring;
B: Failure mode: twisted; too large, etc
C: Potential effects: poor fuel economy, rough performance, and high pollution
D: Potential cases: Insufficient design margin, wrong process, and wrong material


While doing it by individual component seems to make the answer clear, it would also be a huge matrix to include every part. Doing it by performance metric seems to make sense because that allows a high enough view to not get lost in detail, but enough resolution to find the problem

I understand I am asking a somewhat vague and confusing question, so I appreciate you patience.

Thanks!

Dave

Welcome to the Cove, Dave. :agree1:

I think maybe the problem is that your view is too wide. The FMEAs for complex assemblies must be broken down into functional modules. It's clear that your example #3 is best at addressing the salient issues wrt piston ring performance, but piston rings are part of a bigger assembly (the pistons themselves), and the FMEA should be approached from that level.

The way to work is to start at the top level and break the work up according to the BOM structure. Evaluate each subassembly independently first. When all of that is done, you can examine the whole structure--and the individual FMEA documents--and see how it all fits together at the top.

Hi Jim,

Thanks very much for the feedback.:thanx: Your suggestion adds a lot of clarity to the process. I think the potential number of parts on which we needed to do a DFMEA overwhelmed me. Breaking it down by sub-assy makes sense. But in reality, it seems that we will eventually have to get down to the part level, and do the work needed to get there.

Thanks again,

Dave