Eng. practice of allowing 5% overstress 33

[radair]

It seems to be standard structural engineering practice to allow up to 5 percent overstress in structural design. It's been this way since I graduated college in 1980 and I've seen the practice commonly used in the tower analysis field for the last 15 years.

I've been asked my opinion by a government agency as to why this is a safe and acceptable engineering practice, including citing any relevant structural codes. They are not questioning my work but are asking me for a signed & sealed letter of opinion. It seems to me that this would be a better question for their state engineering board of licensure.

Can any of you help?

 

[HgTX]

The codes themselves are full of conservative estimates. That's the point. Even in new construction, if we have to go up a whole beam size for 1% "overstress", that's just plain silly. What makes an engineer more than just a software operator and number cruncher is that abstract thing known as "engineering judgement".

I suppose one could bury that 1% (or 5% or whatever) by playing games with rounding and significant figures along the way. But the only reason to do that would be to make the lawyers happy, not for any real structural reason. This is NOT a safety issue. It's not even an ethical issue. It's purely a legal issue. (Moral<>ethical<>legal, though there's overlap.)

Hg


Eng-Tips policies: faq731-376

 

[frv]

Very well phrased, HgTX!

I've attempted (and failed) to explain to our piers that we are not simple number crunchers.

The reason we go to school and the reason they teach us the theory of structural and material behavior is to give us the tools with which to exercise judgment. Our classes are intended to teach us why code provisions are written the way they are. This in and of itself arms us with the knowledge to determine when and why exceptions can be made.

If the idea is to rigorously follow every code provision without the possibility of using judgment, then I'm sure a technical Associates degree would suffice to practice engineering.

Read the disclaimers at the beginning of all the codes; invariably they state something to the effect that the information is though to be suitable, but that it should not be used without "competent professional examination and verification of its accuracy, suitability, and applicability..." (from AISC). In other words, USE YOUR JUDGMENT

 

[JaredS]

MechEng

It is true that if there is loading greater than anticipated on a structure that the supporting elements are overstressed. However, it is not true that the supporting elements will be stressed to failure. Probabilistically, the overlapping regions between the statistical capacity and the statistical loading (assuming normal distribution for both) is typically outside of 5 standard deviations. This indicates a confidence interval for the design on the order of 99.9999994%. In other words the design load would need to be significantly greater than considered, such as replacing that SUV with 3 or 4 SUVs, or the element would need to be critically weakened or crippled, in order to cause failure. Because of this it is conservative to allow loading greater than 5% of capacity.

Jared Stewart

 

[msquared48]

Hokie and JAE:

"And I think it is spurious to argue that old designs are more conservative than new ones because the codes have changed"

I have to take issue with this statement in one particular, major, and far reaching case here. I am running into a lot of problems in the foundation loads on remodels where the allowable soil pressure without a soils test was 2000 psf. It is now 1500 psf without a soils test. In this case, based on the new IBC, the old UBC value is NOT conservative, but a real problem for extensive remodels, particularly where a flat, stick framed roof is changed to a trussed roof.

I don't think that you can make that blanket statement JAE. I'm sure there are other code examples too.

Mike McCann
MMC Engineering

 

[frv]

Mike-

I don't think that's so much of a code issue as it is a "standard of care" issue.

I've always required a soils report. This touches precisely on what many of us have been arguing, as this is a perfect example of when you should be hesitant to allow "over stress". In this case you DO NOT have the information required to make an intelligent decision.

Quite honestly, if I were asked to design something without a soils report, I would be more conservative than the IBC in the design of the foundation system.

 

[whyun]

Overstress does not mean collapse. An analytical solution that results in ANY overstress is a "violation" of code; however, calculations can be easily "adjusted" to reduce the demand-capacity ratio.

The notion to "allow" up to 5% overstress is an incorrect one. Final design, even with 5.1% or 10%, may satisfy the performance requirement without failure. The safety of the end user depends on the expertise of the engineer making the judgment, not whether a member or a connection is numerically determined to be overstressed.

Though calculations are typically not a part of the contract document, any indication of accepting member that is overstressed can be used against you as you have no control over how many "copies" of calculations are floating around. To be on the safe side, legally, avoid ANY overstress in the final calculations that leave your office. If this can not be avoided, where acceptable amount of overstress is shown on the calculations, state the reason for accepting the overstress without increase in size.

 

[JAE]

Mike,

Well, I didn't provide that quote - hokie did.

 

[hokie66]

Yes, that was my statement, but I don't know why Mike says he takes issue with it, because it agrees with his point.

 

[cvg]

whyun
good answer and I agree with almost everything - except - I require (and my clients generally do also) calculations be submitted for all projects. Design requirements should be stated on the calcs. Anything exceeding the requirements is not allowed and I check it myself. Any overstress would need to be thoroughly justified before the plans are considered final. Even though I am not sealing the structural calcs, I am sealing the plans for which the structural subconsultant has prepared those calcs.

 

[JAE]

Well, as far as submitting calculations on projects, read this: thread507-89434 and your head will spin.

 

[DonPhillips]

I read that thread, JAE, and I agree, when applying for permits, calcs. are not needed - unless there is no seal. I ask owners, tenants, and/or contractors (residential and commericial) to submit calcs or have the design sealed by a design professional. I copy the Chapter 16 requirements for the design loads to be on the drawings and I usually get glazed looks. I cannot remember the last time I had calcs submitted - perhaps for a beam for a deck or something.

Don Phillips

http://worthingtonengineering.com

 

[swearingen]

I admit I have accepted overstress in designs before, but the justfication was repitition of similar members. If one roof member has a little more load on it (due to some local mechanical equipment, for example) and comes in a few percent overstressed, I'll keep it the same as size as the others. It truly falls under the engineering judgement issue. In a court of law, the opposing council could blast me all he wants (I've had bouts with them before) and I would stand by the decision. I could also line up a slew of established, respected engineers that would back me up.

To those that give the "slippery slope" argument, that's hogwash. Engineering judgement will tell you that a tiny bit is OK by inspection without even looking at it - anything more requires closer inspection. There is no slippery slope - the wall of engineering judgement is very near the top of the slope to stop you from falling further. Those that don't have a wall, or if the wall is too far down simply don't have good engineering judgement and those folks stand to get burned from it...



If you "heard" it on the internet, it's guilty until proven innocent. - DCS

 

[miecz]

Dinosaur-

It's not clear to me what you meant by

I design new to within 5% of allowable.

Does that mean that you design to a 5% "overstress"?

 

[JStephen]

Swearingen, do you agree with the metal-building idea expressed earlier of designing everything for 103% of allowable stress?

 

[frv]

The metal building industry is an entirely different proposition all together.

They are notorious for using any and every code "loophole" to get loads down, even if the intent of the code is not there.

This is an example of where you know they are not using conservative load estimates, they probably always use exposure B for wind, the connections are done as cheaply as possible, nearly every member is built-up with no "fat".

I think this is a perfect example of not exercising engineering judgment in the other direction. I'm willing to bet that their buildings do, technically, meet code (i.e. they are not overstressed); however, every time there is a big wind storm one or two of those things fall down.

 

[msquared48]

Hokie:

Yea... you're right. I guess I was 180 degrees out of phase with the conversation. Cheers...

Mike McCann
MMC Engineering

 

[MechEng2005]

JaredS-
I understand that "overstessed" does not necessarily mean failure. However, the values calculated give a certain confidence interval. You stated this is very high. My issue is that there is a reason for the high confidence interval. I don't think you can ever completely prevent failure. Even if you design with a factor of safety of 200, maybe an ice age comes along and a glacier sits on the house. It will most likely fail. The point is that the confidence interval is supposed to be very high, and I question when it is ok to reduce it. What is an acceptable failure rate? 1 in a million? 1 in a billion? By using the calculated values as a limit, you have a well defined value. I agree that engineering judgement and is important. I also agree with Swearingen that the "slippery slope" argument is, to some degree "hogwash" for the reasons he stated. However, engineering judgement leaves a gray area in what the limit is, and I would much rather be safe and have a concrete value. I completely understand that this is more a personal preference than something that is preventing failure.

Swearingen-
In the case of repeated members, I agree that an overstress in a single member can be accepted because multiple members will support the load. However, I think that to completely justify that assumption there should be a calculation. Perhaps the max stress of the members was determined to be 15, when it has been determined that actual values that beam will allow a stress in the range of 14.9-20. It is very likely that repeated memebers with an stress of 16 (overstressed) will be less likely to fail than a single member with a stress of 15 ("acceptable"). However, I feel that this method should be justified. Experience is certainly an option as a way to justify it, if the engineer is very confident that it will not be a problem. However, I think that when the overstess is calculated, it should at the least be noted why it was determined acceptable. In an ideal world, the load would be calculated and statistical analysis of the beams could be performed to show that the many overstessed beams has a greater confidence interval than a single "acceptable" beam. I understand this isn't always possible. Engineering judgement does have a role.

I guess my comments about not accepting overstress were more based on an ideal world, in which every possible item can be calculated (with statistical confidence values where appropriate). I understand that this isn't possible or reasonable for all cases. However, I do think that a reasonable effort should be made to shoot for the ideal case when possible and reasonable. This is where engineering judgement comes into play, What needs to be calculated, what needs to be estimated, and what can be shrugged off as acceptable by experience without even needing anything written down.

-MechEng2005

 

[whyun]

Full compliance with the Building Code and every member and connection designed within code allowed stress levels do not guarantee collapse will not occur. Full code compliance provides only a reasonable level of confidence that the finished structure will perform.

Building design and consumer product design are different in many aspects. One is that each building is considered unique (even identical buildings side by side) whereas consumer products are not. The other is that manufacturers set the acceptable failure rate for their product (driven by finance) whereas in building design, the building codes establish factors (determined based on probability of occurance, variability in material properties, system's ability to dissipate energy, etc) to prevent failure.

 

[JaredS]

MechEng

The problem is that the code was written by men, not some higher authority. The problem is that the code is only good for ideal cases. The problem is that there is nothing ideal in or about our world. Just as you said a glacier !surprise! could come out of nowhere and squish that house. The solution is laying your foundation on the code but building out with knowledge. I may not have a lot of experience, I only just graduated, but I did learn a lot and I can tell you that the code is not just conservative. It is very conservative. I think it is too conservative in fact, however, because the code takes for granted a few things, like loads, sizes, connection fixities etc, it is a good thing that the code is so conservative. However, because the code is so conservative it is very safe to design inside 5% overburden provided that the design is not too wild.

For really wild things like this

http://www.sci-denver.com/images/technologymain.jpg

the code is of really limited use. It is in these applications where I would base my design off a yield theory, and not simply what is stated in the code, where I would be wary of using a 5% overburden. However, this level of design is not required for things that are in the code and fortunately I am comfortable with allowing to slip a little past what the code deems as safe.

 

[RWF7437]

“This is what The Dalai Lama has to say on the millennium. Instructions for Life in the ... Learn the rules so you know how to break them properly. ...”