Need a book? Engineering books recommendations...

Return to index: [Subject] [Thread] [Date] [Author]

Re: Non Professionals doing Engineering

[Subject Prev][Subject Next][Thread Prev][Thread Next]
Paul and others:
Okay, I think we have beat this one to death.  It is all pretty much a
matter of semantics I guess.  KNOWING the answer and having a FEEL for
the answer are two different things in my opinion.  The original
statement I responded to stated that you must KNOW the answer prior to
running a computer analysis.

Designing simple beams and slabs is one thing.  If you are running a
more complex 3D finite element Dynamic analysis on an irregular
structure, I have been surprised by the answers at times.

Especially when I am surprised (meaning I really did NOT have the
ability to "know" the answer ahead of time, or even have a "feel" for
the answer), I start to check the model and loads.  I run the model for
loading conditions that I CAN predict the answers for, and see if the
model gives the right answer.  Do members that are supposed to have
pinned ends have moments at the ends?  Are negative and positive moments
in slabs located where they are supposed to be?  If I run a self weight
only loading condition is the answer correct?  Once I feel the model is
correct, you can start applying very simple load cases, and build on it,
etc., etc.  This is what I mean by applying basic engineering checks to
the computer model.  

Some of you may feel that I had no business even attempting this
analysis because I failed to know the answer ahead of time.  It has
turned out that I really did not even have a "feel" for the right
answer.  But after checking and re-checking, I discovered the true
nature of the structure and eventually came to a true understanding of
how the structure worked.  All thanks to the computer.


Paul Feather wrote:
> Lynn
> How do you check the results to make sure they conform to basic principles?
> To check the results you must have something to measure them against.  No
> matter how "simplified" an approximate analysis you perform, you must still
> compare the results to expectations.
> I don't believe that Phil's original intent was for you to know the specific
> and final result before using the computer, but rather that you have an
> expectation of the realm where-in your results should be. Performing a
> couple of simplified model upper and lower bound checks before analyzing a
> complex problem can at least provide an envelope of expectation for
> comparison purposes.
> Paul