Implementing FEA at Your Company
I’ve been spending a lot of time with designers and engineers the last few months. Some have been to visit for the beta tests. I’ve been visiting customer facilities. I’ve done a few Teamcenter Appshare’s and phone calls.
I’ve tried to hit a good cross section of our base… places with just a few Solid Edge licenses, places with hundreds. Some just now thinking about doing Finite Element Analysis, some with huge investments in software for doing engineering analysis.
The decision of a company to start doing FEA is a hard. With CAD, you can design as accurately as you want although real world machining probably will not be able to match that. If you design a part to be one meter long and it is 1.1 meters in the shop, chances are the problem is in the shop and not with how the CAD model was setup.
FEA is a bit different (see my last blog). Despite all the mathematics involved, you are always dealing with fuzzy numbers. If a part deflects 4.35 mm in the analysis, what does it mean that the number you measure in the field was 4.95mm? In the CAD world, an error of 10% is almost always bad. In the FEA would, sometimes it is sometimes it is not. The value of a good FEA analyst is the engineering judgment for making these decisions.
It’s important for engineering managers to understand FEA is not CAD. If you want good answers, it will come much more from the person interpreting the software result than directly from the software or some pretty stress plot. It is good to do some initial FEA models that you can verify later. This will help the designer analyst to refine his models for the next time and give the company a good feel for what the errors might be in your particular situation. Allow ramp up time for building engineering judgment. Get training on FEA (beyond using the tools) or arrange for a mentor.
Organize your FEA problems by expertise. I’ve seen problems that require a PhD from MIT to solve and I have FEA problems that my kids have can solve. There is a large range of problems that a CAD designer can do but know the limits and when to elevate the problem.
