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

RE: Let's Talk STAAD-PRO

[Subject Prev][Subject Next][Thread Prev][Thread Next]
> -----Original Message-----
> From: Christopher Wright [mailto:chrisw(--nospam--at)]
> Sent: Monday, April 26, 1999 12:40 PM
> To: SEAOC Newsletter
> Subject: RE: Let's Talk STAAD-PRO
> now one engineer intern can run numbers fast enough to
> keep a whole office busy figuring out if they're trustworthy. Someday
> computers will be so fast that one engineer will be able to
> do all the analysis in the world, and the rest of us will be checking
> the results for programming errors.

The basic problem with computer modeling is that you can obtain results that
look great, that seem reasonable on the surface, but which are based on
faulty premises and even input error.

Now, this isn't the fault of the program, or the programmer, per se, but it
is an indication of why engineers aren't going to be made outmoded by
computer software.

I'll never forget the time that we had an entire pipe rack structure in a
chemical plant designed based on an analysis by a guy who was actually an
architect by training and had never even passed the EIT (it was, as a matter
of interest, a STAAD analysis).

New on the job, I was getting familiar with it by reviewing the work that
had been done when I noticed something interesting:  He was doing an AISC
code check on the members, but had NO k-factors input for the unbraced frame
to check stability including sidesway.

When I pointed this out to him he answered "oh, the program figures all that
out for you."

I did a few hand calcs using the forces and moments from the analysis, and
it showed the member sizes WAY under what was required.

Problem was: This was an EPC job and it had already gone to fabrication (and
some to the field)!

We had to explain to the client that his fabricated steel already on site
had to be "beefed up," and the stuff not yet shipped had to be beefed up or
scrapped. That was not pretty.