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


[Subject Prev][Subject Next][Thread Prev][Thread Next]
On Thu, 27 Apr 2000, Christopher Wright wrote:

> Imagine someone walking into a fluid mechanics 
> class and announcing that viscosity is an obsolete term and thereafter 
> shearing stress is actually the velocity gradient divided by a better, 
> more up-to-date physical property called 'slipidity.' And never mind that 
> the physics hasn't changed--in five years viscous drag will be completely 
> replaced by non-slipitude, making the world a better place in which to 
> live. ;-> 

Except that you are here confusing "first principles" with "derived method".

The "first principles" of structural design--physical principles embodied in
statics and dynamics, for example--aren't being "changed" or even "looked at
in a new way". What is changing is the synthetic derivations we use to apply

Engineering design in general, and certainly structural design in
particular, must use simplifying assumptions to model system behavior, so
that those systems become susceptible to design using real world materials,
real world fabrication, real world construction, etc. The simplifying
assumptions are often arbitrary. In fact, the "coarseness" of such
assumptions due to the lesser level of understanding in former times can
seem rather quaint to us today, in light of current knowledge.

For example, once reinforced concrete "two way slabs" were designed assuming
that, for some reason, statics didn't apply. There were actually "patented
systems" sold to building owners, designed on the basis of some sort of
suspension of the laws of statics. The assumptions WORKED; the slabs were
designed and built and those buildings lasted decades. But the "simplifying
assumptions" were wrong, nevertheless. Subsequent formulations of the
theories of behavior of such slabs eventually made such design methods

With LRFD, no one is suggesting that "first principles" have changed. What
is suggested is a closer approximation to real structural behavior, and that
of individual elements.

Mc/I is a sacred "simplifying assumption" used as the basis for beam and
beam-column design under ASD. But any grad student who has fooled around
with fine analysis of structural sections knows that it is "invalid" when
you look at it on a finer level. The sections really DON'T behave that way,
but assuming they do makes them susceptible to design, with appropriate
factors of safety. "Good enough" is "good enough," I suppose.

But when someone says, "there's really no major difference between the two
methods as far as results are concerned," that's not something to be
dismissed. Getting from here to there, using LRFD, is done with asssumptions
that fit more closely to REAL behavior. And in some cases, such insights can
save time, money, and add to the overall reliability (read "safety") of a
given structure.

So, we're not inventing a "new statics" when we do LRFD design. The "old
statics" still pertains, but the behaviors of the materials and elements are
modeled more closely to reality.