O'Reilly logo

Lying by Approximation by Paul D. Gessler, Christopher Papadopoulos, Vincent C. Prantil

Stay ahead with the world's most comprehensive technology and business learning platform.

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

Start Free Trial

No credit card required

1.5. PUTTING IT TOGETHER: TOWARD A NEW FEA PEDAGOGY 7
With the advent of high speed computers, it is clear that the machine wins in the battle
of raw speed and avoidance of computational error. However, while speed and computational
accuracy are necessary, they are not sufficient—and not even most important—for producing good
FEA results. e machine cannot provide the intellect, strategy, and judgement of the human
mind, all of which are crucial to perform good analysis.
e myth that the “computer is always right comes, in part, from the truth that yes, most
commercial finite element software has been sufficiently debugged, removing most or all internal
programming errors. Studies by Jeremić [2009] show that programming errors in commercial
codes persist only in a very small percentage of cases. In short, the computer, while working fast,
also works nearly flawlessly. It can therefore do the heavy lifting required to analyze complex
problems that lead to the solution of problems with thousands and even millions of degrees of
freedom.
But most errors encountered in finite element analysis are either due to incorrect user input,
i. e., garbage in—garbage out, or due to lack of prudent judgement regarding dimensional approx-
imations, active degrees of freedom, loading strategy, sensitivity to boundary conditions, or the
nature of the correct theoretical solution. at is, they can often be traced to one of two causes:
incorrect understanding of finite element modeling, or poor application of strength of materials,
and often both to varying degrees.
In most cases, therefore, it is operator error to blame for all of the top ten mistakes [Chalice
Engineering, LLC, 2009]. To correct these mistakes, the analyst must look for cause and effect.
And as remarked, most often, the code is not the cause, although sometimes the user should inves-
tigate if the model programmed in the algorithm is, in fact, the correct model for the application
at hand.
us, when the task at hand can be described in an efficient and robust algorithmic form,
the task should be owned by the machine. In those instances where the task requires judgement
and/or compromise, the mind trumps the processor. And this is where the practice of numerical
analysis most often goes awry. It perhaps comes as no surprise that the ten most commonly made
mistakes are found only in the procedural steps performed by the analyst and none involve the steps
performed algorithmically by the computer. is glaring reality is the driving force behind our novel
approach to learning the finite element method wherein we focus on user behaviors rather than
on derivations of algorithms.
1.5 PUTTING IT TOGETHER: TOWARD A NEW FEA
PEDAGOGY
We have reviewed common errors and standard procedure, in which we emphasize the need for
the analyst to be skeptical and to take responsibility for making good judgements. Recalling our
overall pedagogical philosophy based on constructivism and encounter of misconceptions, we now
outline our vision of a new FEA pedagogy that prioritizes user behaviors. We draw from our own
notes and examples to provide a set of exercises and case studies in which students can encounter

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, interactive tutorials, and more.

Start Free Trial

No credit card required