ASD: 1.03 Allowable Overstress 6

[SmithJ]

I have been used to allowing a 3 percent overstress when working with the AISC code. However, I recently realized that I can not adequately identify the basis for this overstress allowance. Does anyone know the origin of this 1.03 or where I can find information that would explain it.

Thanks.

 

[ChuckerD]

As a metal building engineer I can confirm that the industry uses the 1.03 allowable stress ratio as the staple of our design diet. There are probably two underlying factors why this is the case:

1) The metal building business relies on competitive design - the amount of steel in our preliminary designs wins or loses jobs. "Bump it up to the next size" costs us profit margin and/or contracts, and improves my chances of looking for a new job. This is not the case for the consulting engineer who bids for a contract based on engineering time and does not shoulder any financial impacts based on the steel used in the design. So the metal building industry goes to great lengths to find a competitive advantage. Voila - our death grip on using the 1.03 ratio.

2) A metal building engineer will not bear any financial burden if a lawsuit occurs - the company will bear it.

How many of you have been involved in the oft-invoked "Lawsuit" that seems to govern design decisions? Would you consider a PE negligent that allows a 1.03 stress ratio on a member where all loads and conditions have been considered properly? Seems that jury behavior is a bigger factor in design than the structural behavior nowadays.

 

[broekie]

ChuckerD, just a question. If it is all about being competitive, why not use 1.05? 1.07? Why stop at 1.03? Or is 1.03 just one of those unwritten industry standards, i.e. "the way we have always done it"?

 

[UcfSE]

broekie has a good point in my opinion. You have to draw the line some where. Why not 1.04? That's only 1% over the "allowable" 3% and 1% doesn't matter right? You have to draw the line some where, again. The only time I go over is when I know I have been conservative in my estimations. How often are we that close, who knows? How many times have we approximated point loads as uniform? That will give you a similar moment but only for many point loads. Let's say I am designing a beam supporting 4 joists evenly spaced. The moment considering the joists as point loads is 20% higher than that used by assuming the loads are uniformly distributed. Now I started off 20% over my limit though the calcs won't reflect that because of the assumptions made. This is just an example to illustrate how we make simplifying assumptions that may have already used up any allowance we may have. It isn't hard to assume your way out of your 3% simply in the model you use without even touching on sig figs and accuracy and so forth. I think if you take all this into proper account and have everything dead on then 3% isn't going to hurt. You still have to draw the line somewhere though. It also depends on what it is that I am overstressing. I'm not worried much about overstressing a wide flange in flexure by 3%, but wouldn't consider it for, say, tapcons or expansion bolts. Those never go in right in the first place.

 

[ChuckerD]

Very good question broekie. As a new engineer fresh out of school a 1.03 allowable stress ratio is the "rule" that I was taught to be a self-evident truth. It's definitely an unwritten rule - doesn't appear in the MBMA industry bible (Metal Building Systems Manual) in any fashion.

I faintly remember a presentation where the origin of the magic 1.03 came up, but for the life of me don't recall the discussion. But it seems to be a line that somebody has drawn in permanent marker because it is a pretty widely used/known value, even outside the metal building industry as evidenced by this thread.

I understand your firm's hard-line approach to the 1.0 ratio, especially in today's litigious atmosphere. My engineering judgment - and surely all of those here would agree - tells me that a member calc'd to a 1.03 stress ratio won't fail if loaded similar to what is designed for. It's a shame though, "Lawsuit" has made engineering judgment a thing of the past it seems.

 

[ChuckerD]

From Preface to Part 5 of the ASD spec (top of Pg 5-13)

"The reader is cautioned that independent professional judgment must be exercised when data or recommendations set forth in this Specification are applied. The publication of the material contained herein is not intended as a representation or warranty on the part of the American Institue of Steel Construction, Inc. - or any other person named herein - that this information is suitable for general or particular use, ..."

So the ASD spec tells us specifically that we MUST (not Should or May) use independent professional judgment. Seems an adequate defense for "Lawsuit". They also disavow any statement of the suitability of the material in the spec. So if Engineer designs to a stress ratio of 0.80 and the member fails, Engineer has to defend whether the spec is even correct and why we chose to use it?

 

[VirtualEngineer]

When I first got out of school, I occasionally followed the old convention "less than 5% overstressed O.K.", mainly because the other "old school" engineers I worked with followed this convention.

However, when I went to an AISC seminar, one of the speakers stated that this convention is not considered good engineering practice.

One consideration worth considering is someone in the fabrication or construction of a member you've designed, may not do their job correctly. If the member fails, you will have a tough time avoiding blame.

If you're calculations show you're overstressed, especially if there is no way to relieve that overstress by another iteration, I don't see how you can avoid getting splattered by the blame of someone else's mistake.

Remember that the "5% over say O.K." rule came about in a different era when running another iteration was a time consuming process. Also design methods were deliberately conservative.

In these days, it is easy to design members very precisely and run another iteration if needed. Our design methods yield results that accurately model actual conditions. Design codes have shaved most of the fat out of the design process.

I certainly don't think anyone should start out by setting the stress limit in their computer analysis to 103%.

Perhaps the metal building industry needs to reign back their practices, and keep stress below the allowable limit.


Unfortunately, one firm cannot do it alone. It may take a serious lawsuit to cause changes. I for one don't want to be on the wrong end of that suit.

I guess I have to disagree with anyone who thinks it is OK to design to a stress limit over and above the allowable.

Regards to All

JPJ

 

[SacreBleu]

I don't even work for prefab building manufacturers, and I agree with their 3% "overstress" tolerance. After all, their procedures and documentation are very reliable, unlike the calculations of someone outside of that field of expertise.
As a structural engineer, to reject their calculations based on 1.03 seems to me, to be very petty and nit-picking. I worry about things more worthy of a 1.03 stress condition in a prefab metal building.
I can recall plenty of instances of inexperienced engineers in other types of construction inadvertently making errors far exceeding 1.03. If we as, an engineering community, stop debating this "1.03/fear of lawyers" issue, and spend more time mentoring new engineers, lawyers would not have so many "victims". It amazes me that there is so little mentoring happening in our industry. If hospitals treated their medical staff likewise, I would be reluctant to enter any hospital.

 

[IFRs]

Of course, in steel design, failure is not brittle in nature. If you were designing a glass structure, you might do it differently. Steel also age hardens, has tolerances on thickness, composition, hardness, ductility, erection, etc. My goal is to design to 1.0 of the allowable. If I am 3% over and there is some redundancy in the structure or conservativeness in my assumptions of boundary conditions, then I figure that if push came to shove I could re-calc and get below the magic 1.0 number, for the benefit of some idiot non-engineer who does not believe in "good engineering judgement". I have to remind myself that designing to 1.0 does not relieve me of the need to still exersize "good engineering judgement".

 

[SacreBleu]

IFR- I totally agree. There is a recent-grad engineer who makes sure that he doesn't exceed 1.00, however, he is totally unaware that using a uniform load on a girder beam produces much less bending moment than modeling as point loads (girders with 2,3, or 4 equally spaced point loads). I brought this to his attention, and his attitude was, his method is more productive. The funny thing is, he is really an advocate of LRFD, yet doesn't understand basic common sense.

 

[dig1]

I'm going to weigh in with a different perspective. The company I work for is an OEM of industrial equipment and we perform EPC projects internationally. We also get involved in upgrades and modifications of existing facilities where our equipment is installed.

These upgrades almost always require larger and heavier equipment. If we were to start at 1.03 and have to add something larger in the future we would be looking at 1.1+ real quick.

Also, I can't tell you how many times I've been through an older mill and cross bracing has been removed. You can tell it has been and no one knows when/why. Another area we have concerns about is corrosion. Painted, galvanized, and both. After a certain period of time the steel is becoming heavily corroded and the customer wants to put heavier equipment in.

We do not go above 1.0 and sometimes we are even on the other extreme because our "good engineering judgement" says there will be more cost down the road than saving some structural steel up front. This is built into our lump sum estimate.

I realize and appreciate that pre-fab metal building designer/supplier is working in a different environment than I am.

I am definitely not arguing about economical use of materials and in many applications, including many buildings where current and future loads are relatively well defined, I understand why people use 1.03 and this discussion is being made.

It is just different than the situations I normally encounter. There have been more times than not when I have been very thankful to the engineer in the 1960s or 1970s that didn't cut everything very tight.

 

[DRC1]

I have comments on both sides of the issue. First in construction 10% design overstress is not uncommon. I know guys who go 20%. Second, It really is not an overstress. When Fb=Fy the steel at the outer fiber goes plastic (in theory). The rest of the section can still accept moment until the entire section has gone plastic. So even at fb=Fy, the section can still take additional moment. Third the definition of Fy is somewhat arbitrary, the yield stress of steel varies through out the cross section and the certified test reports generally show yields consistently higher than the required Fy. Fourth, It depends on what loads are causing the overstress, how well you know the loads, how likely is an overstressing load, the duration of the load, what happens if the member yields. I have spent a lot of time explaining engineering to oposing attorneys. As long as you know what you did and why, you will be okay. I once explained that engineering is a human process and that if all we did was to blindly follow a code, we would not need inteligent trained engineers. The arbitrators seemed to agree.
On the other hand the material cost is the cheapest part of the construction process. I was taught that no one will ever know if you had 10% too much steel, but everyone will know if you had 1% too little steel.

 

[miecz]

This 3% "rule" should not be confused with building addition work or construction condition work.

Overstress for rehab work has been written into codes and is therefore defensible in court. Our model code, BOCA Article 16.14.2, reads "The addition to an existing structure shall not increase the forces in any structural element of the existing structure by more than 5%, unless the increased forces on the element are still in compliance with this code for new structures."

Under construction conditions, the area is generally closed to the public, and the contractor, who will bear the cost of any failure, is responsible for the design.

The 3% rule is nothing more than a way to gain a competitive edge over those who don't follow it, namely, the rest of us who don't design prefabricated steel buildings. There is no justification for it, only rationalization.

 

[SacreBleu]

jmiec-
I totally disagree. If I can't spec a W24 beam because it is too deep to fit, then a W21 with 3% overstress is perfectly fine. Since I know how to do an accurate tabulation of the loads, etc, my beam design is more accurate and proper than the one-significant digit guy who designs girders with 3 point loads as a uniform load. The Factor of Safety includes allowance for engineering error such as that guy incurs.

 

[miecz]

Sacrebleu,

As you say, the Code safety factors include allowance for inadvertent errors. However, I just don't see increasing allowable stresses because you don't make a particular error. What if you make a different error? By the way, when I calculate the bending moment for a beam with 3 point loads (uniform spacing at quarter points), I get the same moment as for a uniform load. Am I missing something?

 

[SacreBleu]

jmiec,
Since the cost of steel is greatly increased, and loads in commercial building construction are accurately and easily computed, I allow 3% over as a routine. When I do a custom wood-framed house, the loads are much more difficult to estimate. In that case, I am very conservative with my beam design.
Try this example:
Girder 30' long. 3 point loads, at 5', 15', 25' from one end.
Point loads due to beams with 40' trib, total floor load = 80 psf. Therefore, point loads are 32 kips each. Girder end reactions = 48 kips. Analyzed as 3 point loads, bending moment in girder = 400 ft-kips. Analyzed as uniform load = 80 psf x 40' trib = 3.2 klf, bending moment = 360 ft-k.

 

[whyun]

Modern Steel Construction had a few articles a while ago about this topic. One in October 2003 and other in October 2004. Here are the links:

http://www.aisc.org/MSCTemplate.cfm...Management/ContentDisplay.cfm&ContentID=24932

http://www.aisc.org/MSCTemplate.cfm...Management/ContentDisplay.cfm&ContentID=26812

Hopefully these links work... If they dont, visit

http://www.aisc.org,

find modern steel construction and look in the back issues for the pdfs.

Cheers

 

[HgTX]

I think that's a different topic.

Hg

Eng-Tips guidelines: faq731-376

 

[whyun]

HgTX, You are right. It's been a while since I read this entire thread.

As for the "arbitrary" 3% rule, I would find use whne checking existing element stresses after remodel such as live load increase, new equipment, etc. In fact, there is no document that allows 3% overstress. As far as I know, it is at the discretion of the engineer. I'm used to allowing 5% overstress...

 

[miecz]

SacreBleu,

You're right, the uniform load produces a smaller bending moment for this layout. Hard to imagine anyone would appoximate this solution with a uniform load. On second thought, it's not so hard to imagine.

 

[TPNY]

So how many of you have actually had to defend your calculations in court? On top of that, when did your design fail because of a 3% over-stress and not a bad connection, or construction fault, etc.? In defense of a little approximation for sig. fig.s and conservative uncertainties, who can easily determine when something fails in the field because it was load at 103 k, when it was designed to carry only 100 k?

I don't usually use an over-stress factor in most cases, but when I do, it's about 3-5% or less, and that's because I know my structure... where I'm conservative, my load paths, etc. I won't if I can't justify it.