Exceeding allowable stresses 3

[pcbtmr2]

Can someone offer a logical explanation for the common practice of slightly exceeding allowable stresses. I am having to work against the argument, "The code allows 100% of allowable stress, not 100+%". Forces are generated by wind and gravity in steel (TIA-222, AISC ASD).

Many Thanks.

 

[Tomfh]

Everyone does it, including me, but it really really annoys me. It's like the Spinal Tap amps which "go to eleven". Can't we just make the allowable stress a bit larger and call that 100%?

 

[msquared48]

Yup. Just a call by a committee of men. Arbitrary value based on test data.

Mike McCann, PE, SE (WA)

 

[JStephen]

A couple of older threads on the topic:

http://www.eng-tips.com/viewthread.cfm?qid=220274

http://www.eng-tips.com/viewthread.cfm?qid=52511

 

[MrHershey]

A few things for me:
1. I've typically made at least a half dozen conservative assumptions in my design by the time I get to the end. If I go back and revisit even one of them I could likely bring that number down under 1. We're talking a couple percent here. If I rounded 9.7 up to 10 in my calculations anywhere then I can bring 1.03 down to 1.0.
2. We typically carry our design checks to a fair bit more significant digits than is really warranted given all the assumptions we make. There's a reason most of our inputs in the code stop at two. Some even stop at one (R factors, some live loads). If we're doing things as we were taught way back in middle/high school, we should be doing all of our calculations to the lowest amount of significant digits for any of our inputs. If going to two significant digits, 1.03 = 1.0.
3. Material overstrength. At least with steel, I don't remember the last time I had A36 steel yield strength come in anywhere near 38 ksi (~5% overstrength), for instance. Typically mill reports come back in the 40s. Similarly for A992, don't recall the last time I had anything come in at 53 ksi (again ~5% overstrength). Typically it's at least in the high 50s. Concrete I see come in low all the time so maybe I'd be more likely to not take the 5% there, though rebar is similar to steel in that I don't recall the last 60 ksi mill report that came in at 63 ksi or lower.
4. At least with wind/earthquakes, keep in mind the magnitude and rarity of the event we're designing for. The design wind event in US codes is a 700 year return period, almost three times the age of the United States. Design basis earthquake is roughly a 500 year return, about twice the age of the United States. If you're going further and designing for full MCE the return period is closer to 2500 years, literally a biblical event. If our behavior during such rare events is actually within 5% of what we predict (up or down), I'd be pretty ecstatic.

 

[dhengr]

Pcbtmr2:
It seems to me that this particular topic needs much more common sense and engineering judgement, and much less code b.s. and pseudo justification and sometimes permitting. Certainly no one suggests you can do this 5% thing all the time, on every member. The idea that a group of people, however knowledgeable, or not, sets an allowable stress, not to be exceeded, except by 5%, doesn’t make much sense. Why not just set the allowable stress at 1.05(the former value), and be done with it, as others have suggested?

I’ve been at this Structural Engineering thing for many years, I’ve done a lot of buildings, non-building structures, equipment, machines, etc., and a lot of forensic engineering on problems and failures on all of these. And, I’m not sure that I can really point to an instance where a 5 or 10% overstress, w.r.t. normal design loads, was the cause of a failure, or I dare say even a major problem cause, when all other things were o.k. We are fooling ourselves if we think we know most structures that well, their loadings, the way they actually act under all the loadings, etc. And then, the idea that if it doesn’t check using ASD..., use LRFD and it’ll be o.k., or visa-versa, when the structure hasn’t actually changed one iota, is absolutely crazy. Who do we think writes these codes, God his/herself, following perfect logic and all the laws of physics and nature, infallibly? The only bigger fools are the reviewers or plan checkers, who know nothing except the code verbiage, verbatim. They don’t know the intent or history of the code or that paragraph, but they can recite the whole damn code section word for word. But then, you see hundreds of details and conditions where it is obvious that the detailer/designer/engineers hasn’t the foggiest idea how that detail really acts or works, how it is fabricated or how well, and why it is an inferior detail for that condition.

The way they are complexifying, finessing and continuously changing the codes these days, all for the publishing dollars and to keep a bunch of non-productive people employed as code diddlers is stupid. What they should be doing is making the codes cleaner and clearer, pounding home the intent of the codes, working to make them cleaner and simpler (as best they can); this, instead of more complexity to try to cover ever condition and detail in the universe, and pretending that we are being more exact, and complete, and perfect, in our approach.

 

[SAIL3]

dhengr....finally some engineering logic and common sense.....the engineers that parse the ever-increasing code for every nit-picking change and nuance would do well to read the blurb at the beginning of the AISC code which essentially places the final responsibility for design in the hands of the engineer and engineering judgement...

 

[IDS]

The problem of imposing precise limits on continuous variables occurs everywhere. The solution is not to second guess what the governing authorities will let you get away with (should they ever check). The solution is to decide a reasonable margin (say 5%) and then aim to be as close as possible to that margin below the limit. If you then hit 101% you either review the calcs to get that number below 100, or change the design.

The reason for doing that is not to reduce the probability of a major problem (it will do that, but only by a very small amount). It is because when a major problem does occur (perhaps because the contractor left out half the reinforcement), the design will be examined in minute detail, and if the contractor's defence team finds a stress of 101% of the specified limit, they will find that much more useful than finding a maximum stress of 100%.

Regarding the probability of extreme events; if you design 1000 structures over a 50 year career, then the probability that at least one of them will be subject to a "1 in 500" year event is close to 100%. Working on the basis that extreme events are so rare that you don't need to worry about them makes no sense.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[JStephen]

Long ago- company had some old textbook that said "In allowable soil bearing, 5% really doesn't make much difference" or words to that effect- basically, some of the lines of reasoning as used above. Consequence: Every project, size foundations for 105% of the recommended maximum allowable soil bearing.

 

[CANPRO]

I had a great boss/mentor straight out of school. He would always say that the code(s)are guides only, BUT you have better have a damn good reason for working outside of the recommendations of the code. And that goes both ways, you may follow the code to the letter and at the end feel like something is unecessarily large, or not as strong as it should be. I learned from him that after all the numbers are crunched and code checks are made, you still have to use your own judgement and ask if this makes sense.

 

[ajh1]

Speaking of the 500 year event, in the winter of 1996-97 Yakima, WA had a 500 year snow storm on Christmas Eve, then had a second 500 year snow storm on New Year's Eve one week later. So much for recurrence intervals, and yes, quite a number of buildings didn't survive for one reason or another although in my limited inspections most came through just fine if something had not either been missed during design, adjacent higher building or something changed after erection, "we don't need all these braces in the walls, they are just in the way of where we want to put the door."

 

[klaus.]

The German Steel design code allows 10% overstress in 'small' areas
Whatever small areas are...usually local areas like connections ....
But at the end it is the responsibility of the engineer...



best regards
Klaus

 

[OldBldgGuy]

All design is an analysis of probabilities. That's what you're paid for: to analyze the probabilities & tell someone what to do to make them work for them. Do you honestly think that any set of load calculations is actually 100% accurate? Do you honestly think that any analysis of capacity is 100% accurate? Who produced the material? Is it a tree that grew in an open field, a plantation, or a virgin forest? Is it steel produced in a new plant, an old plant, with scrap, with raw ore, with new equipment, on a cold day, or a hot day? Who mixed the concrete? Who produced the portland cement? Where did the aggregate come from? You have to be kidding if you think any number you produce is specifically and significantly more accurate than any other if it is based on values in a code or a set of tables; step back & look at what we're really doing here. Generally speaking, we know where the lines are drawn, we stay within them, and we crowd them or cross them when we feel comfortable about it.