Need a book? Engineering books recommendations...

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

Re: UBC97-- Redundancy Factor

[Subject Prev][Subject Next][Thread Prev][Thread Next]
Thanks Charles

I have observed that a number of code provisions have been adopted with
little technical data backing them up.  The SEAOC Seismology committee has
generally done a good job but ultimately the individuals find themselves
having to make decisions on subjects that they have not seriously
researched.  In this context people use their  best judgement.  I believe
that the provisions on redundancy probably fell in this catagory. The
problem is that what may seem rational occasionally isn't.  This is
especially true when few engineers have a good background in probability.

Redundancy, like ductility, has  become part of the mantra that we accept
without question.  One aspect of the redundancy debate has been the
acceptance that redundancy will protect the structure no matter what
defects are present.  

For example if you have a number of frames in a structure, each with a
defect that results in a brittle failure that fails at the same load then
how does redundancy help.  I believe that this is refered to as a common
mode failure.  We almost lost a nuclear power plant "Browns Ferry" because
of a comon mode failure.  They believed that because they had several
redundant systems they had good reliability.  The problem is that they
hadn't counted on a fire occuring at the location where the control wiring
for all of the systems came together and as a result they had no redundancy
with regards to that failure mode.

I believe that the pre-northridge connection comes close to being this
brittle  element and as a result  more members do not help as long as we
ignore issues of reliability and ductility.  On the other hand once we
address these issues the need for redundancy is less critical.  I am not
saying that there is no need for redundancy, but rather that the code
provisions regarding redundancy are in many instances probably overkill. 
In addition the requirement that systems with less ductility be designed
for higher forces may help make the non-redundant systems perform better,
because system overstrength apparently does more for system reliability
than redundancy in many instances.  

For me the lesson from northridge is that we must seriously address the
issues of reliability and ductility.  We have made a lot of progress with
moment connections but I believe that we still have a ways to go especially
with other systems.

Part of the reason I find this article by Bertero & Berter, interesting, is
that it provides a theoretical basis for these concepts.  Like most
engineers  my background in probability theory is weak and it is tempting
to accept something that justifies ones beliefs.   As a result I would be
interested in hearing from anybody who has read the article and is
knowlegable regarding probability theory.

When I graduated from Cal Poly  the emphasis was on the practical and the
theory was de-emphasized.  I have since come to the conclusion that there
is nothing more practical than a theory that works.  In this  context I
would suggest  that there is benifet in modifying the code provisions on
redundancy to be consistent with the theory of probability.

Mark Gilligan

>From Charles Greenlaw:

>Has either of the individuals that responded to my message read the
>article?       [In a recent edition of the ASCE jourlan of Structural
>Engineering there was an interesting article by Bertero & Bertero on the
>advantages of redundancy.]
>Mark Gilligan
        No, but your summary of it made sense immediately on its face. I
responded with approval, using an example from early day aeronautics, with
the conditions carefully qualified to be analogous. 

If the remaining frames are incapable of the job if any frame fails, then
multiple frames, like multiple engines in my example, are a disadvantage.

If the remaining frames can still do the job, like the remaining engines on
a returning bomber or airliner in Christopher Wright's example, then the
redundancy helps, assuming the probability of any one frame failing hasn't
become too much worse.

Your summary left the impression that if the members are loaded well into
inelastic conditions, then total inelasticity could result if one member
failed. That is, you can't still fly on the survivors.

As for interpreting the article, I know you, and if you misunderstood the
article then the article was not well written. I should expect to do no
better with it.

The caveat is: If we do have redundancy, are we getting cavalier or
as to the reliability of any one element, or as to the survivors being
sufficient if one fails?   Like with flying, one must be aware of what one
is relying on, and its limitations, because mistakes make wreckage.

Charles O. Greenlaw  SE    Sacramento  CA