From: Chris Willcox <cwillcox(--nospam--at)earthlink.net>
Date: Thu, 24 Jun 1999 09:36:18 -0700
Sorry Stan, I can't help myself. I've gone through the list for the first time
in a week and I've just got to make a few comments on this subject. They follow
in no particular order and with no particular connection.
Like most of the younger engineers who have responded, I learned LRFD in school
and ASD on the job. The way my steel professor explained it to us in 1991 was
that the department had decided to teach LRFD since ASD was easier to pick up on
the job. Not the most ringing endorsement. By the way, I 've been told that
about half of the civil engineering programs in the U.S. still teach ASD to
undergrads although most are giving serious thought to switching.
First off, is LRFD better than ASD? Certainly it's more precise, it includes
current research results, and it is more "correct" in that it deals explicitly
with specific limit states instead of burying them into one or two catchall
equations. Yet, in my mind that precision is compromised by the fact that AISC
calibrated LRFD's results to ASD. (AISC wasn't blatant about it as someone
pointed out , but it is right there at the beginning of the commentary to the
specification (LRFD, 2d Ed., p.6-162))
Whether all this makes it better I'm not so sure. (But, the explicit treatment
of limit states does make it the best method to teach in school in my opinion.)
In my experience LRFD does take more time and effort. My particular peeve is
having to keep track of factored loads for the strength limit states and
unfactored loads for the serviceability limit states--especially since
serviceability governs more often with LRFD. On top of this, with the
transition to Grade 50 steel as the standard, serviceability will govern even
more floor beams and girders leading to even less difference between ASD and
LRFD for floor framing. As far as designing for serviceability and checking
strength, I wouldn't be comfortable spot checking strengths, whereas I am much
more comfortable spot checking serviceability. Let's face it; in most cases
serviceability is about comfort and appearance, while strength is about safety.
Also, I can think of at least one case where the disconnect between factored
strength design and unfactored serviceability design can lead to a problem.
LRFD flexural design is based on the plastic moment of the section (Fy*Z) which
is a more accurate flexural STRENGTH limit state than the elastic moment (Fy*S)
which relates more to the serviceability limit state. For wide flange sections
with typical shape factors (Z/S) around 1.15 the load and resistance factors
pretty much guarantee that the allowable unfactored (service load) moment will
be well within the elastic range. However, for square tube sections, etc. with
shape factors around 1.5 (the max allowed by the specification), for some load
cases the LRFD allowable unfactored moment may actually be beyond the the
elastic moment (S*Fy). This can lead to some substantial inelastic deformations
that will not show up in a service load deflection check.
Due to the calibration of LRFD to ASD, most of the variation in steel weights
between ASD and LRFD is due to the load factors for dead and live loads (a very
worthwhile refinement in my opinion even if most live loads are already grossly
overstated) and this works to LRFD's advantage only when the dead loads are
large relative to the live loads (not the usual case in buildings) or there are
large live load reductions which is particularly true of multistory columns.
Thus the cost advantages for low rise buildings seem to be limited.
I'll admit that LRFD is probably the best thing to use for seismic design of
moment frames since it gives a way to relate b/t ratios for beams to plastic
rotation capacity prior to buckling. I'm not so convinced when it comes to
braced frames. However, what with the new 1997 UBC strength design earthquake
loads (another pet peeve) LRFD may be the way to go for all seismic design, but
not yet for me.
The wind uplift condition presented by Michael Valley is a valid case, but I
think that most ASD designers will check a reduced DL load plus wind combination
where the wind load is close to or greater than the dead load and load reversal
has the potential to be the governing case. My preference is to use the 2/3*DL
that is specified for primary lateral frames in 1994 UBC Section 1619.1. I
believe that this is a required working stress load case in the 1997 UBC (with a
.9 DL factor I think)
Finally, nobody here is saying that ASD is dangerous. In fact, AISC has
implicitly said just the opposite by using it to calibrate LRFD. Those who are
saying that it is "wrong" are basing that statement on a definition of "wrong =
not updated with the latest research." Now, I'm a bit of a Luddite at heart
which means I don't believe in technology for technology's sake; so for me to
switch to LRFD, it's got to offer me something tangible. Thus, in the end the
whole debate boils down to my time (= my company's money) vs. the client's
money. If the client is willing to bear my added cost for using LRFD that's
fine, but for small projects it just doesn't make sense. If the steel cost is
8% of the total building cost and LRFD saves 10% of the weight (a very high
assumption in my understanding) and the material cost is one third of the total
erected steel cost, then you've saved a grand total of 0.25% of the project
cost. Now for high rise projects where 0.25% is a lot of money and the actual
savings from LRFD might approach that 10% assumption it's probably worth it.
After all this of course, I realize I've changed absolutely no one's opinion,
but at least I feel better and isn't that what it's all about.
Chris Willcox, PE