Roger Turk wrote:
> "Reliability" is the compliment of "failure" or "error" and is a euphemism
> coined to make it sound better than to report that the analysis "... results
> in an x percent failure or error." People would much rather hear how correct
> or safe something is than hear how wrong or unsafe it is and 97 percent
> reliability sounds much better than saying 3 out of 100 will fail. (Even a
> reliability of 60 percent sounds safe, when it really isn't.)
What I actually meant by that was I have heard figures quoted like 1 in 100,000
rather than 3 in 100. Does anyone have any literature to offer on what the
statistics are (or what the academics intend them to be) for gravity loading?
> filling all seats double and all aisles and open space. The resulting live
> load, including furniture, was 41.7 psf. [Ref. Gaylord & Gaylord, "Design of
> Steel Structures," 1957] Does this justify designing a school for a live
> load of 8 or 10 or 16 psf? Would you do it and justify it by saying that
> "... there is little *probability* that the load will be exceeded."
Actually, that is the exact principle behind live load reductions. If a
multi-level structure had sufficient tributary area on a column, 16psf is exactly
what you could use for a 40psf LL area. This is allowed because the odds that
the loading is exceeded over many rooms on several levels is very low, indeed.
> require 200 pound people spaced every 1.4 feet on center. (Ever try to
> space 200 pound people 1.4 feet on center?) What is the probability that
> this will be exceeded? Zilch! What is the recurrence interval? Once!
On a lighter note... yes, I think I went to a party like that in college once.
Maybe dorm and fraternity corridor loads should be much, much higher than
It seems like 3 main arguments against LRFD have been presented on this
discussion list, and I feel that 2 of them are not well founded:
1) LARD uses dangerous probability while ASD uses sound factors of safety.
In general ASD and LRFD produce quite similar values. To characterize one as
gambler's probability and one as a sound use of engineering factor of safety
principles just doesn't seem justified to me. All of construction boils down to
probabilities. An unforeseen loading is always possible (even if unlikely).
Even the best designed building may be built incorrectly. ASD boils down to
probability of exceedance, too, it just isn't stated as directly. I like the
directness of LRFD, even if it's a bit of a mathematical, academic exercise.
2) LRFD requires lengthy deflection checks that a service load method like ASD
design doesn't need.
Neither ASD or LRFD methods directly address deflections or vibration. These
need to be checked for either method. For a design with a high LL/DL ratio, the
ASD beam is actually more likely to have a deflection problem.
3) The benefits of LRFD are not sufficient to justify the time and expense of
This is the only argument that "can't" be argued with because it is in the eye of
the beholder. I like the method, but it's certainly your right to disagree.
Most design methods that we use change over time as new experience and new
information are brought into the field. The seismic design in the UBC has
changed substantially 3 or 4 times in the past few decades as new research has
shown better approaches. I don't see why steel design should be different. I
don't doubt that LRFD will be modified some day, too.