Should Analysis Be Left to the Specialists?

I had the opportunity to speak to the Kansas City SolidWorks User Group on Tuesday. We had a great turnout to discuss the role of analysis in design.  The premise of this talk was this:  If you can master a full featured CAD tool like SolidWorks and are comfortable with basic stress calculations, there is NO REASON you should be avoiding design analysis.  We discussed the common objections and spent some time talking about one of the more prevalent ones, "Only specialists should be doing FEA." There has been a recent discussion on this in Roopinder Tara's TenLinks blog. Having, in a previous life, been one of those specialists who regularly stated concern over the use of FEA by people who haven't been "checked out", let me share with you some of the revelations that led me to join the COSMOS team...

1.  Designers and part-timers ARE going to be doing analysis. They want to, their companies want them to, tools like ours give them the confidence to, and there really is no way to police it or stop it...should one wish to.

2.  Analysis WILL become a mainstream task as tools get easier and more accessible (read 'cheaper').  To realize its true potential, people making front-line design decisions will need to be using analysis before the baseline design is committed. Analysis will not become a design driver if the task has to be handed to specialists and it is unlikely that analysis specialists will be making baseline design decisions. The only logical option is that designers will be doing FEA...albeit some better than others.

3.  What is the real danger of a less knowledgeable user doing their own FEA?  I ran this past my long-time friend and design analysis pundit, Rich Bothmann at IMPACT Engineering and got him to agree, grudgingly, that the only real danger, in nearly all companies who don't employ analysis specialists, is lost time in the process as work is spent doing analysis that might not yield any useful information.  If this is true, why are there so many passionate statements by people likening this to giving a child a loaded gun?  Historically, the concern was that as users with less knowledge on the finite element method, or structural mechanics for that matter, start doing more design analysis, the market will be flooded with time-bombs resulting from flawed design decisions. 

4.  However, the fact remains that for companies who haven’t committed to analysis as a way of life, thus acquiring analysis specialists, and even some who do, the “proof is in the prototype”. Good analysis or bad, the testing is the deciding factor on whether a product is shipped or not. If a company tests the same way they did before employing analysis prior to the prototype stage, where’s the danger?

5.    Along those lines, the real issue is then that a designer will determine a design is OK when it wasn’t. We aren’t too concerned about a conclusion that a design is bad based on FEA results since it will then get even more scrutiny. So, if a designer is prohibited from using an analysis because he might say a design is good when it isn’t, what is the alternative?  He or she will simply by-pass analysis and take that same design to prototype…which is what they would have done after the analysis anyway.

6.    Taking all this into account, the benefits of gaining unexpected insight or making design improvements using trend analyses and trade studies far outweigh the risks of making erroneous pass/fail decisions.

That said, the benefit of analysis in the design process can be boiled down to one concept…risk mitigation. If you can catch problems earlier or improve performance or cost when it is less costly to make changes, you are taking some risk out of the design process. However, if a user does not try to learn all he or she can about the analysis tool, the analysis method, the product in question, and the engineering mechanics relevant to the problem, the impact of design analysis will be limited.  Still worthwhile to try, just in case a revelation jumps out of the process, but potentially ineffective. Consequently, we at the COSMOS team and me personally, believe that our users should do their best to ask questions, talk to their VAR tech support, read up on relevant engineering concepts, attend user groups and COSMOS Companion webcasts, and any other activity that might improve their use of this important design tool.

On a closing note, I’d like to thank Wayne Tiffany at Automatic Systems, Inc. in Kansas City for his hospitality and support in setting up the SWUG as well as the guys at Tridaq-KC for bending over backwards to accommodate my schedule on Wednesday, including last minute chauffer services!

-- Vince