CAD for FEA in the BAD old days
Once upon a time, I was a real Engineer. It was the mid-90s, and I was on loan from EDS to the GM, Delco, and Delphi plants in central Indiana.
Back then, there wasn’t just a line in the sand between the Engineering and CAD Designer roles… there was a deep trench. Engineers weren’t trained in UG (Unigraphics, now NX) and didn’t have the local workstations to run it. The Designers in the CAD department were responsible for that.
I remember the ECO (Engineering Change Order) sheets we had to submit to the CAD department to request a design change. You could change anything you wanted if you could describe it in 3 handwritten lines or less.
And, then you had to wait. And hope.
A couple of my Engineering buddies would have made great police sketch artists. They could whip out an isometric view of some new idea on the back of an envelope at their desk. I never really had that skill. My drawings look a lot like my handwriting. Be glad you don’t have to interpret either!
At one of my posts, I did briefly get my hands dirty with UG. I wasn’t doing detailed design or generating prints, though. My CAD skills were way too meager for that. I just remember being frustrated at all the error messages I’d get when I broke relationships or sliced through assemblies in my preferred butcher-brute-force methodology.
As I started getting into CAE with FEMAP, Hypermesh, Nastran, CFdesign, and STAR-CD, I increasingly needed to manhandle the geometry myself. The manufacturing-ready models from the CAD department were impossible to mesh and solve as-is. They required a lot of simplification work first.
At one point, I convinced my manager to buy me a license of an early version of SolidEdge. I could at least open the UG models in a friendlier environment, but I still couldn’t accomplish much on my own.
That’s when I set out to become a FEMAP expert.
FEMAP is in the “pre-processor” class of engineering software. It is an environment built with specific tools for CAE analysts to create meshes, apply boundary conditions, set material properties, and then serve all that up to an external “solver” for simulation. Many pre-processors also include “post-processing” functionality. That is, once the solver finishes its number crunching, the raw data gets sent back to the pre-processor to produce all that colorful, sexy visualization of stress, strain, temperature, pressure, etc.
While the Designers would have laughed at FEMAP’s bush-league drawing tools, it was a breath of fresh air for me. I was finally able to arbitrarily slice & dice geometry without worrying about violating the UG or SolidEdge feature tree! If I wanted to divide a complex part into several chunks to ease the meshing chore, I could go Boolean crazy!
There were a few problems with this approach, though:
- It was difficult to remove unwanted manufacturing details like rounds, tiny edges, fasteners, holes, and “noogies.”
- I still couldn’t generate realistic geometry on my own in FEMAP. I had to rely on the Design wizards to give me a good starting shape.
- Modifying my pre-processor geometry for additional “what-if” runs was challenging (at best).
Back then, this was the best approach for getting CAD geometry whipped into shape for FEA or CFD simulation. I became so thrilled with the idea of replacing physical prototyping and testing with CAE simulation, that I soon left “real Engineering” to join a small, upstart CFD company, CFdesign.
Over the next 9 years, I had a front row seat for the evolution of this CAD for CAE evolution.
Read the 2nd installment of this 3 part series:
CAD for FEA in the GOOD old days
