Monday, August 18, 2008

Finite Element Training Classes

In a comment to Paris's post about Abaqus features, Fern noted that, "The ABAQUS training programs were fantastic in that they didn't shy away from theory at all." This brings up a very good point, "How much and what kind of theory should be presented in an FEA training class?".

As I see it, there are 2 branches of theory that can be dealt with in the training class. First are the engineering principles that are represented and how they are implemented/accessed activated within the analysis code. Things like plasticity models, large deflection, thermal radiation, random vibration and piezoelectrics fall into this category.

The second branch of theory which can be dealt with is the computational/programming aspect of the analysis code. This deals with items such as element formulations, matrix solvers and user
programmable features.

There is also significant gray area between these to branches. This touches more on the art of FEA. Things like determining the correct damping to use, Hertzian stresses, choosing the correct turbulence flow models. It soon becomes apparent that developer must pick and choose which topics to cover in the typical limited 2 to 3 day training classes they offer.

From my experience in giving training classes, most students tend to lose focus (that’s code for falling asleep) when a lot of theory is presented. Most students want to know the just the basic
commands to get their analyses running and worry about the quality of their results after the fact. This is where qualities brought by the experienced engineer really shine. It is their upfront thinking about the results they want and knowledge of the theory that they consider BEFORE they start building their models.

Should theory be presented in all classes or should there be follow-up classes in theory made available to those who are interested?

7 comments:

Unknown said...

According to me knowing the theory behind the analysis is of great importance. However, I feel that it will be very difficult to cover all the pertinent theoretical aspects of the analysis during the 2-3 day FEA training class as it will make the class duration much longer. Hence the optimum way could be to cover the introductory part of the theory during the FEA training class & stress the importance of knowing the theory while performing the analysis. A curious engineer will always go back and will try to learn more about the other theoretical aspects of the analysis by reading the FEA package documentation, text-books etc.

Just an opinion....

Paris Altidis said...

When I learned the ropes in ABAQUS, my instructors at the now ABAQUS Midwest (was AC Engineering then) had a way of explaining the most complex topics in simple terms that kept the entire class attentive and not getting into hybernation (aka sleep) mode. A real joy !!!
A training class is as good as the instructor and how captivating his/her lecture can be. Everyone can read the slides and follow the workshop notes. The intricate parts of the do's and dont's are what I'm looking for.
In my tenure as an instructor, I've always had one role Model: Den Hartog and his famous lectures where he would explain the entire topic without writing a single equation on the board for hours. That alone, in my book is enough stimulus for the audience to delve deeper into lecture and reference material after they leave the training place.

Paris Altidis said...
This comment has been removed by the author.
Raghu said...

I believe "application" of FEA is extension of the theory.Hence it will be a welcome sign to know the basics right before implementing something using the commercial packages.

Keith_DiRienz said...

I think the main issue I wanted to put up for discussion is, do you think it is incumbent upon the developers of FEA software to educate their users in engineering fundamentals or should they expect that level of proficiency from the students? Is it akin to the developers of a video editing software teaching their users the fundamentals of optics?

Anonymous said...

@Keith,
"Is it akin to the developers of a video editing software teaching their users the fundamentals of optics?"

akin? Not really, unless, we are year 10 passouts using the video editing software and if the software does not work (?), he would call the software help desk and where the support desk staff would ask the question "did you try switching the computer off and on again". As ANSYS users, we dont call the ANSYS support and ask why the cantilever beam bending problem would not solve? This is just my honest opinion. Moreover, to your question whether FE theory should necessarily to be taught? yes, I would go with Nimish, a bit of basic theory (preferably morning session :) ) to know what will go on behind the black screen the remaining day.

Unknown said...

When you say "FEA training class" - what do you mean? If this is the academic course in university, the answer is quite clear - it should cover the general theory of FE with computational/programming aspects. If you mean a software class like Ansys or any other, then it depends on your audience and their expectations. For an introductory course - only gui, basic commands and capabilities. for advanced course(s) - a combination of implementation specifics together with what you've said the 'gray area'. In any case I believe that software class is the wrong place to teach/learn the engineering principles.