Need a book? Engineering books recommendations...

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

asd vs lrfd

[Subject Prev][Subject Next][Thread Prev][Thread Next]
One of the real serious problems that I have run into in lrfd in
concrete is that the different load factors for dead and live load can
seriously distort critical locations in a member.  Inflection points
can move, or disappear entirely, maximum moments under factored dead
and live loading can be below the unfactored dead load, etc.

While the *concept* of different load factors for dead and live load
is correct, i.e., we can more accurately estimate the dead load on a
member than we can the live load, the end result can be disastrous.

One example that I like to give is in the design of a concrete footing
subject to overturning and partial uplift.  Unfactored loads have to
be used to size the footing and get the soil pressures.  But, when
load factors are applied to the loads, uplift on the footing will
frequently disappear!  This, in my opinion, does not result in a safe
design.

USD in concrete became popular because ASD did not give any indication
of the factor of safety due to the *non-linear* aspect of the
stress-strain curve of concrete.

Reasons not to use LRFD:

1.	Service load criteria, i.e., deflection, drift, vibration,
etc., frequently control the design, not stresses.

2.	When the material strength is so variable, as with wood, that
LRFD does not give any better indication of the factor of safety than
one would get with ASD.

3.	When live loads are of a different type of load than dead
loads (concentrated, varying, stepped, etc.) such that the different
load factors seriously affect the points of inflection, maximum
moments, etc.

4.	In foundations when subject to partial uplift.

I should also like to point out that the steel lrfd specs do not have
a requirement for cover plates, or flange plates on built-up members,
to extend beyond the theoretical points where they are not needed. 
The uncertainty of the location of these theoretical points was
recognized many, many years ago as being critical, even with ASD.

A problem that I see with *new* information these days is that the
testing is frequently done at one institution, then a computer model
is made, and precise predictions are carried out from the computer
model.  No confirmation of the validity of the testing is performed at
another institution.  One example that comes to mind is eccentric
loads on fastener groups and welded connections, Tables XI-XVIII and
XIX-XXVI in the AISC ASD manual.  These tables were developed from a
total of six tests each (see cited footnoted references the AISC ASD
manual) and were not verified by any other investigator or institution
that I am aware of.

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