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.

 

[CANPRO]

I'm sure the logic changes from person to person - for me, the amount of overstress I'm comfortable with depends on the component that I'm designing (floor beam, fall arrest anchor, lifting lug, etc...), the nature of the loading, and the assumptions made in the design, among other factors.

If I'm running a quick calculation and making a lot of conservative assumptions along the way and I end up with 5% overstress, I will recognize that I could re-visit my assumptions and get the allowable stress back to 100%. Again, depending on the nature of the design and the assumptions I have made, I might say 5% over is ok, or I might re-run the analysis with more accurate info. If its a life safety issue (fall arrest anchor or lifting lug for example) I'm likely going to be very rigid at 100%. If I can bump up my beam size, plate thickness, etc... without any issues then there is no reason to exceed 100%.

 

[EngineeringEric]

I was once told that PEMB often design to 1.039. and the reason given during the conference was that 1.04 was too high.

We laughed, but I did not take it was an exaggeration or lie.

 

[jayrod12]

When it's new design, you should not be exceeding allowable stresses. However when checking an existing structural member, slight overages are generally acceptable. A few codes have sections dedicated to existing building elements and seem to indicated up to a 5% overage is typically acceptable.

 

[pcbtmr2]

Jayrod, any way you could point me to those code sections?
Thanks.

 

[JedClampett]

pcbtmr2, if you're looking for something in code to allow you to exceed allowables, good luck. This is tribal knowledge, passed on from generation to generation. Every engineer does it, but no one would want to defend it in court.

 

[JStructsteel]

I think as engineers we know that allowables are not ultimates. 5% overload on a beam in not going to bring it down. You can always say the dead load is not right, who uses 150lbs/ft^3 for concrete still? Could it be 145, could it be 155?

 

[MotorCity]

Simple logic. Despite our best efforts, we cannot possibly predict the loads a structure will see to within a 5% accuracy. Hence the reason for safety factors in ASD or load factors in LRFD. They are merely fudge factors to account for variations in design vs. actual loads, design vs. actual material properties, etc. and are based on statistical models and probability.

 

[pcbtmr2]

MotorCity, I have used your rationale before and "they" didn't accept it. If there was something code-based that says something similar and thus an over-stress of 5% is typical and acceptable, they would buy it.

 

[JAE]

pcbtmr2,
Look in the IBC - chapter 34. Overstress allowances for existing structures are outlined there.

Check out Eng-Tips Forum's Policies here:
faq731-376

 

[WARose]

I think a lot of people are comfortable with it because in most cases we are talking a very short duration load causing the overstress. Steel (for example) shows a higher yield strength and buckling value (under some circumstances) under rapid loading than under long-term loading. There is also the probability issue (i.e. everything in the load combination happening at once).

Speaking of that, the fact is: some of this is historical: in the older codes, you use to be able to take the 1/3 stress increase for load combinations involving wind and seismic. The West coast guys (among others) fought against that for seismic......and eventually it went away. So some older guys still have that in the back of their minds and try to justify it that way. The 5% has been a standard in a lot of offices I have worked in. (Depending on the situation.)

If you have access to AISC's Journal (archives) there is an interesting article that develops the history of the 1/3 increase and mentions some of the points I make here. The article is:

'The Mysterious 1/3 Stress Increase', by: Duane S. Ellifritt, 4th Quarter 1977

 

[EdStainless]

I think part of what MotorCity is getting to is that you don't know many of the factors well enough to say that it is actually over stressed. So we develop reliable estimation methods. In cases that I have seen there was more than enough 'slop' in some of the load assumptions to make up for the supposed 'over stress'. When things get padded at every step you end up with lots of margin.
Now if you are dealing with people that believe that design, fabrication, and use are fully deterministic then you have another problem.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[ajh1]

Various arguments are as follows:
1) 3% is all the closer a slide rule was ever expected to provide accuracy, so anything tighter than that is not required.
2) We really don't know actual loadings more precisely than 3% or so.
3) CSR equation (AISC H1-1) is expressed as less than 1.0. Given normal rounding rules anything less than 1.05 would round to 1.0. (I find this rationale stupid at best, but I've heard it a number of times.)
4) Common industry practice, probably not only for the metal building industry but in general. Our certification auditors have never had a problem with the 1.03. I think Jim Fisher is comfortable with it as well and he is a widely recognized expert.
5) A number of textbooks in their examples allow slight overstresses (say OK), although I don't know that anyone has explicitly stated that 3% is okay, but 4% is not, etc.
6) Our steel will generally have higher yield and tensile values than the minimum allowables That is not a particularly good rationale for initial design, however can be of some use in litigation and failure analysis.
7) Depending on the structure, what is the range or pattern of the overstress. If it is at only one point, it is not really a problem from a practical point of view as the forces will redistribute somewhat if necessary. This is particularly true if it is at a point say for example directly over an interior column. While the mathematical model might generate an overstress, when the physical structure is examined and the depths of members and connection areas are taken into account it is unlikely that the mathematical stick model will truly represent the physical at that location. A point out at the middle of a span would be interpreted differently. If one point is 1.03, and the points 5' to either side are in the 0.80 range (as an example) it is unlikely that there will be a long-term problem. Other items like partial base fixity, bearing lengths on purlins, etc. could show that the overstress doesn't really exist if you are willing to do some heavy duty analysis and model the structure in a higher level of detail.
8) Evaluate the load combination causing the overstress, is the combination one that is likely to happen, for example 12 psf LL in a no-snow county. Pretty unlikely that the roof is going to see a uniform 12 psf over its entirety once construction is complete. On the other hand, I get real conservative in the area of a step snow drift, where there is a history of design or higher loads in actual occurrences.

Bottom line:
There is no specific allowance for using 1.03, but most everyone does it. The checker is technically within his rights to criticize a 1.029 ratio, although in general I would say he could find something better to be uncomfortable with.

 

[canwesteng]

You don't know anything about the building (except geometry) to a precision beyond 2 sig figs, if that even. How can you say it's 3% overstressed? That's my rational anyway, and load factors/resistance factors are only ever given with 2 digits.

 

[Leftwow]

I like CANPRO'S approach to your question because as an engineer I feel if something is overstressed, then more analysis should be done, and load magnitudes looked at more in-depth. So his comment along the lines of - if I'm doing a quick analysis of something I generally take into account any assumptions and over 'calced' situation then I'll give it +5%.

Instead of that, I'll put more time into honing down my analyses, not particularly to try to fudge the numbers but, looking at more exact data, trying other combinations such as LFRD.

One job in particular I found wind bracing failing, eventually I learned that if the braces are connected in the middle, then you can take half the L value for critical buckling stress formulas... there's a lot of these little hidden gems in the steel book that took me a while to learn.

 

[MotorCity]

We've all heard it, but its one of my favorites and further reinforces (no pun intended) why a slight overstress is ok....."structural engineering is the art of molding materials we do not wholly understand into shapes we cannot precisely analyze, so as to withstand forces we cannot really assess, in such a way that the community at large has no reason to suspect the extent of our ignorance.”

 

[JLNJ]

So if it's a new member it can't go over 100% but if you check it a day after it was installed the existing building code (Chapter 34 as mentioned) let's you go up to 5% over without question...

Sometimes I look at the design and upsize certain members and/or connections just because they don't look right to me, especially in industrial facilities where any member might live a hard life and we just don't know what forces it might be subjected to.

I guess that's why it's called engineering judgment and why I like to be the EOR on projects.

 

[cliff234]

You mentioned AISC ASD. If you check your design using LRFD you might be able to get the design to "work". ASD and LRFD will give the exact same answer when the average load factor is 1.5. If you calculate your factored loads and then divide by the service level, that will give you your average load factor ("ALF"). If your ALF is < 1.5 then you'll see a slightly more economical design using LRFD. (Wind loads are the wild card because they get a 1.6 factor, so hopefully you have a lot of dead load, and not very much live load and wind load.)

 

[IFRs]

https://www.aisc.org/globalassets/modern-steel/archives/2004/10/2004v10_in_the_beginning.pdf

 

[CURVEB]

I've never read IBC Chapter 34 to mean that you can exceed the capacity of a member per se. An increase in force of up to 5% might be under the assumption that the member was designed to 95% of its capacity originally. Of course there is nothing that prevents the original designer to utilize 100% of the original capacity, but if you are looking for a reason to not permit an overstress, that might be the language you need.