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

Re: ASD vs. LRFD

[Subject Prev][Subject Next][Thread Prev][Thread Next]
>I've held my tongue on this thread until the teeth marks are deep and painful 
>and I've got to let go.

I am in a similar situation, but I am also very new to the engineering
profession. So I like to sit back, watch and learned from the experience
of the people on this list.  However, some of your comments do 
demonstrate a small lacking of understanding in LRFD design.

I use LRFD and its not because I am straight out of college (graduated
2 months ago), and I will try to explain why.

>1.    *Unless* the dead load *and* other loads are applied in the same manner 
>and at each and every location, the different load factors can give very bad 
>results.  In other words, wherever the dead load is distributed, the live 
>load, wind load, seismic load, etc., *must* also be distributed at the same 

LRFD does not profess to give an "exact" answer for all loading
conditions. Instead it is based on the assumption of an 
"envelope" answer.  i.e.: we all typically assume shear to be zero 
in the middle of a simple supported beam span, and never check 
the shear there.  However, by LRFD design (and I believe ASD 
as well?) if you load only half the span, shear is not zero
at that point.  Furthermore, you are not supposed to group loads
together until you check forces at a point.  You need to 
factor your forces differently at each point to determine 
the governing loading condition.  Therefor you don't have
to always distribute forces each force to each point.  If there is
no dead load, or live load - then don't factor it!

And yes, you will get different inflection point and maximum moments and
shears.  This is part of that "envelope" design discussed above.  So if
you only check one load case at you cover plates, you very well could
an inadequate section.  Following this, LRFD does not address deflection
(service) criteria - the governing code does (UBC).  So if your section
does not meet service loading conditions, then make it stiffer.  Again
think envelope here.  Service loads are will often govern LRFD designs,
so it is an applicable limit state.

>2.    In ASD, the factor of safety is with respect to first yielding; in LRFD 
>the factor of safety is with respect to a fully plastic (yielded) section.  
>Therefore with ASD, you have an additional safety margin between first 
>yielding and when the section becomes fully plastic.  The LRFD fudge factor 
>for material understrength is ridiculous.  What I am interested in is what 
>the safety factor is to failure (and I will define failure later), regardless 
>of whether it is caused by overloads or by material understrength or a 
>combination of both.

That "Fudge Factor" for material understrength is one accounts for
material variability such as low yielding stress. So yes, if you 
leave it out your section may yield (very, very, unlikely). 
a high safety factor is not always the best answer.  Yes it is the 
most conservative, show me a client that will pay you to intentionally 
overdesign his project.  If you client is straight with you on loading 
conditions, then you will be able to meet his criteria without yielding.
Furthermore (as most of this discussion is on beams) LRFD gravity beam 
design is not based upon plastic sections. Last time I checked Fy was 
the yield stress, not the ultimate stress. If you want to design a 
fully plastic section, the seismic provisions help to account
for the true strength of a section. (As any modern provisions
should.....) Even the current AISC seismic provisions for ASD have 
plastic criteria in them - or didn't you know?

>3.    The structure has to be analyzed under (unfactored) service loads to 
>determine if it complies with serviceability (deflection, vibration, etc.) 
>requirements.  If serviceability requirements control the design, then 
>strength is irrelevant, ergo, LRFD is not necessary.  If a structure does not 
>meet the serviceability requirements, then it has FAILED, just as much as if 
>the members fractured under load.  (Failure is when a structure can not 
>function as originally intended.)

Here we go again....There are cases were a service controlled beam
design will work in LRFD and not ASD.  If the two failure modes are 
close, then you section will work in LRFD and not ASD.  Let me guess 
you want to skip the LRFD approach because ASD is faster? Yes for 
people that initially learned ASD and do not regularly used LRFD.  
And there are occasions when you are required to verify multiple 
load combinations to locate the worse one. This is very time consuming.  
However, I will save my client money.  And isn't that why we are hired?  
To give them a safe and economical structure?

>4.    If, under service loads, yielding takes place, and service load 
>conditions were not investigated under LRFD, then deformation is going to 
>occur and grow.

If you don't check service conditions you failed in your job. Or don't
have to check deflections when using ASD?

>5.    Probability is for gamblers and abstract mathematicians.

You are in the wrong business.  Even your precious ASD criteria is based
upon probability.  (Read a little closer into its development) 
wind design - probability, seismic - probability, live loads -
dead loads - probability, need I continue?

>Yes, I use USD for concrete design, but I use it much more carefully and with 
>more skepticism than I did 40 years ago.  

>A. Roger Turk, P.E.(Structural)
>Tucson, Arizona

OK, so if LRFD (or USD) is so bad why are you using it in concrete? 
doesn't make any sense.  I will agree with you comments on 
cutting edge however, especially when related to technology.  Yes, 
in the lab one aspect is tested - this is called a controlled 
experiment to test a hypothesis.  If you know how to test multiple
items at the same, you should patent it - you would make a fortune. 
Furthermore, these tests are performed on entire systems to see how 
the whole system works together (see the SAC page - you might learn 
something).  In closing, it seems to me that in Northridge your precious 
ASD didn't perform any better (higher safety factor?) than LRFD

Jake Watson, E.I.T.
Salt Lake City, Utah

P.S. Normally I do respect people who have practiced longer than me. 
But I do feel that like me, everyone has a lot to learn about reality 
and the way our systems will really function when tested.  LRFD 
represents the most current and up to date information.  By ignoring it, 
you are simply hiding in a cave and cheating you client.  I am not 
proposing that everyone use LRFD for everything.  However, I do 
propose that like me (trying anyway) everyone at least have a 
background knowledge of both systems.  ASD has many advantages 
over LRFD in the design office.  But has very few if any in the field.