Responding to Plan Checker's Comment (the polite method) 35

[SE2607]

I have a plan check comment that one of my beams is 2% overstressed. While I always try not to exceed a CSR ratio of 1.0, it sometimes happens (putting a bigger beam is expensive, not available, etc.). This is a wood-framed residential remodel, which should be as safe as any other structure, I find this ridiculous. I could reduce the assumed dead loads enough to make this work, but I'm wondering if there are any provisions in the code which allow me to do this.

TIA

Sorry for the double post.

 

[bugbus]

I wouldn't bat an eye at 2%, it's pedantic and finicky in my opinion, and arguably bordering on arrogance/ignorance to think that we can possibly know any of this stuff to such a fine degree of precision. All the loads given in the codes, load and reduction factors, material properties, etc. etc. are only ever given or knowable to about 2 significant figures of precision anyway. Structures don't fail because the utilisation was 2% more than it should have been. It's usually something that was missed completely, which should have been glaringly obvious in hindsight.

Another point, which I've always strongly believed in, is that structures almost always have more capacity and resilience than we (and design codes) give them credit for. E.g., it is common to ignore load redistribution, plastic hinge formation, alternative load paths, (etc. etc. whatever it may be), as overload is approached, simply because it is easier and more convenient to do so. I would sleep pretty comfortably an night if the design is a fraction over 100% utilisation (on paper).

Agree with previous comments on significant figures. I don't know how some professional engineers still can't grasp this. I think we covered it in first year high school maths. In the past I've had to review reports unironically presenting calculations with up to 7 or 8 significant figures.

 

[GeorgeTheCivilEngineer]

I would have included a note alongside the 1.02 to explain why it's OK - not just leave it as 1.02 / failed without comment.

Agree - checker is right to comment on it.

Agree - in most situations, 1.02 is acceptable. This is based on situational engineering judgement, knowledge of factors of safety, probability of exceedance and the significant figures.

I would try to justify on that basis it before spending more time and money on rerunning unnecessary calculations, but that is an option open, as is a bigger beam.

 

[XR250]

There is a PEMB document about significant figures out there that is used by the PEMB industry for 103% utilization rounding down to 100%

Given the lack of fat in PEMB designs and other sketchy-ness, honestly, they should be rounding up to the nearest 10%.

 

[phamENG]

Most live loads have 1 sig fig, wind is presented with up to 3 significant figures, seismic is calculated with 3. The two materials I use most - wood and steel - have material properties presented in 2 to 3 significant figures. Section properties presented in the in the steel manual and the NDS are 3 to 4 significant figures. Load factors are 1 or 2 significant figures.

Two questions (mostly rhetorical in nature):
1) Based on the academic consideration of significant figures where you define the precision of your result with the least precise input, that means our calculations should result in 1 significant figure, right?

2) If my calculation without truncation shows that the bending moment is 132.4689 kip-ft, I should design for 100 kip-ft, right? Or maybe 200 kip-ft?

Obviously, the answers are no and neither of them. Do we need to carry our calcs out to 8 or 9 significant figures? No. I was taught to default to 3 unless the situation dictated otherwise, and I like that. It's consistent with the precision of most of the inputs. And besides, conflating accuracy and precision has very little to do with significant figures. I know how (in)accurate our calcs are, which is precisely why I choose not to push my designs over "the edge" - even if it is only in appearance.

 

[Eng16080]

I agree with with what I think phamENG is saying.

In my mind, a calculation will use values that are not certain, meaning that the values have some tolerance. The values we choose are out best guess at what the actual values might be. So, we take these values and calculate a result. That result is then our best guess at what the answer might be. So, in the example above, that best guess is 132.4689. Now, due to the cumulative tolerances of the values we used in arriving at this number, maybe the tolerance is plus or minus 50 (just picking a number for the example), so we might say the true value is somewhere between 82.4689 and 182.4689. At this point, given this range of potential values, I think the most reasonable value to use is 132.4689. Rounding down or up both seem incorrect to me. Using 100, for instance, doesn't seem right because the true value is most likely greater.

As far as significant figures go, I'll admit it's never fully made sense to me. Sig figs seem like a lightweight way of indicating a value with a tolerance.

For example, suppose we have a simple calculation: A = B x C. Both B and C are given as 1.0, meaning that they each have 2 sig figs, which (I assume) is generally accepted to mean that:
0.95 >= B < 1.05 and
0.95 >= C < 1.05

In multiplying these values to get A, in using sig figs we find that A = 1.0, meaning that:
0.95 >= A < 1.05

But if instead of using sig figs, we multiplied the ranges given above for both B and C, we'd get a different answer because at the low end, we get 0.95 x 0.95 and at the high end, 1.05 x 1.05, so:
0.9025 >= A < 1.1025

Using sig figs in the example above implies a tolerance in the answer which is too low.

I see no point in really using sig figs at all. In terms of presenting an answer, I'll still round a number to something that isn't super annoying to look at (usually 3 or 4 non zero digits), but internally, in excel or some other software, I don't think the computer really cares.

 

[kipfoot]

I highlighted it and then added an explanation about where I felt I'd been overly conservative with my applied loads, unbraced lengths or such.

I'd guess that this is what the reviewer was looking for. To me, this is not a question about whether 102% is just as good as 100% given the circumstances, but rather who is the person who must make that determination.

That's not a judgement for the reviewer to make; it's the responsibility of the engineer.

 

[lexpatrie]

I'm not going to agree that load factors have significant figures. They have two digits typically, sure, but I wouldn't go about truncating things to trick the numbers lower and make something work.

The significant figures should be based on the input data, so if you are placing rebar to the 1/4", there's 3 significant figures, at most. It is probably typical that things have 2 significant figures but taking a 132 and making it 100 seems really wrong to me. I could see 130.

The thing with tolerances is they aren't the same as significant figures. They don't work the same way, and capacity checks aren't tolerances. The load and the resistance have a variance to them, and the checks are intended to produce a reasonably low probability of failure where failure may not necessarily mean collapse or failure in the most traditional sense (sill plate crushing, deformation at a bolt hole, for example, slipping into bearing, "excessive" deflection, etc.

Justifiable Precision and Accuracy in Structural and Civil Engineering Computations, Sputo, this is perhaps what was published in Practice Periodical on Structural Design and Construction in August 2005, though that has four contributors... including Thomas Sputo.

Personally I'd not have submitted an inadequate design, regardless of if there's a calculation package, particularly when it comes to bending or shear stress versus, say, deflection, or bearing stress/plate crushing which isn't as recognized as a life-safety issue, I suggest you re-size the beam.

A lot of these other suggestions sound like point shaving. Trying to manipulate the significant figures in favor of a deficient design.

There may be nuances in the design that would make it pass if your design software isn't quite right, definition of span, etc. Not taking shear d from the face of a wood beam's bearing condition and so forth, that's not the same thing as trying to put your thumb on the scale to make it work by putting down 12 psf dead load when the rest of the building is designed for 15 psf.

As a point of reference, Florida does have specific rules for this king of thing, where you deviate from the design standard and explain how and why, so you need to explicitly disclose such a deficiency.

https://www.fdot.gov/docs/default-source/roadway/fdm/current/2018fdm122varexcept.pdf

"All deviations from Department criteria and standards must be uniquely
identified, located, and justified; no blanket approvals are given."

In the context of design engineering, I think it is best that these sort of plans where you obscure the engineering design criteria should be handled as "NOT FOR CONSTRUCTION" sets, but, having been involved in a fair number of projects that did get built off of "NOT FOR CONSTRUCTION/FOR PERMIT ONLY" drawings, well, this sort of thing is not actually within your control in the absolute sense. I don't endorse doing this.

https://fbpe.org/understanding-the-responsibility-rules/

This disciplinary ruling isn't all that on point, but it does reference FL 61G15-30.
ABBAS BORUJERDI, PE, PE No. 38553 – REPRIMAND, Case No. 2020010497

ROBERT T. HAUG, PE, PE No. 24575 – PROBATION, Case No. 2021021749 & 2020055905

 

[GeorgeTheCivilEngineer]

I'm not going to agree that load factors have significant figures. They have two digits typically, sure, but I wouldn't go about truncating things to trick the numbers lower and make something work.

That's not what I took it to mean. I took it if you've been using rounded up values in the calculations then by the time you get to a utilisation of 1.02, it may have little significance.

 

[lexpatrie]

I suppose for a beam it's a "pure" overstress because there's no second order or stability effects, but then again, maybe there is if it's got fairly long unbraced top flange length.

 

[Tomfh]

I'm not going to agree that load factors have significant figures....The thing with tolerances is they aren't the same as significant figures.

Agree. Yes they are written 1.5 etc, however factors like these are infinitely precise. They aren't numbers we read off a tape measure.

 

[SE2607]

The reason I haven't responded is that, IMO, this thread has gotten ridiculous. Where are all the folks who, by using "engineering judgement", call a fixed based connection (4 anchor bolts) a pinned based connection? Sheesh!

To give you an update, I was able to shave more than enough off the roof dead load for the member to work.

What I took issue with is that the plan reviewer was not accepting 2% overstress. The member in question is a wood beam. Wood has a high variability of strength and the grading agencies take this into account by providing very conservative, in most cases, allowable stresses. Roof live load of 20 PSF really doesn't have a scientific basis. 20 PSF is multiple 160 pound people standing 30 inches apart over the whole roof (assuming less than 200 SF, of course) or 24 inches apart on a floor. Etcetera. I would be ok for the reviewer to point it out to me, just in case I missed the issue, but not passing it because of 2% overstress is ridiculous, notwithstanding all the blather in a lot of the posts above. I've been doing this a while and I have NEVER had a plan checker refuse to approve my design for such a transgression of, OMG, being 2% overstressed.

I'd guess that this is what the reviewer was looking for. To me, this is not a question about whether 102% is just as good as 100% given the circumstances, but rather who is the person who must make that determination.

That's not a judgement for the reviewer to make; it's the responsibility of the engineer.

Exactly!

Thank you.

 

[Tomfh]

While a 2% overstress might seem acceptable to you, it's worth remembering that peer review evaluates designs based on broader consensus rather than your own personal standards. The opinions expressed here suggest that many of your peers deem overstress unacceptable without proper justification. Relying solely on conservative codes isn't sufficient justification in this context.

 

[lexpatrie]

What bothers me most about this is your attitude.

You submitted something that failed and the plan checker asked about it. You should appreciate that the plan checker put effort into reviewing your submittal, not resentful. They probably thought it was an oversight, and rather than escalating it, they merely asked about it. Now you've gone out on the internet, looked for a supportive community to say "tell me he was wrong and take my side" and you're not getting that result.

There's no surprise here, really, you should have expected this result from this forum. Life Safety, Health and Welfare.

I'll skip over the shady ethics of putting something that's overstressed into your design, willfully and intentionally, as you describe it, since you're not receptive to that discussion either. By your own admission this was not an oversight or a mistake. You deliberately chose to put it into your designa and your calculation package. You didn't conceal it, you provided calculations, that's true, but you didn't revise it (this is wood, just switch it to LVL, add a ply, increase the depth of the element (2x8 to 2x10), etc. done, this is not some insurmountable difficulty where it's impossible to do it correctly, existing condition, inaccessible, .... the correct approach is simple, feasible, and at hand.), and you didn't highlight the deviation from the accepted engineering practice, either.

I think you're fortunate the building official didn't send your work to the appropriate board. This little matter could well have blown up considerably.

Regarding the "pinned" base comment, as you should be aware, this is based on research, and from a stability standpoint, considering a base pinned is conservative compared to reality, You are using a generally accepted conservative design convention in service of allowing an overstress, the two are completely opposite from a safety standpoint, the design convention of the pinned base creates a conservative design, i.e. safer than the calculated value. (hence the recommendation that Gb = infinity which is theoretically correct for a pin be replaced with Gb = 10 in the old K factor commentary in the AISC). "Fixed" bases are rarely achieved in practice, hence the guidance that Gb = 1.0 be used in K factor determination as well. So that's not a valid comparison.

The plan reviewer, quite frankly, is absolutely not in a position to "approve" your overstressed beam design.

They did the right thing.

 

[Eng16080]

The post above by lexpatrie is spot on. This whole thing is honestly ridiculous. If wildly conservative assumptions were made in arriving at a 2% overstress, as OP implies, then just make a simple revision to the calcs using more realistic values, resubmit to the reviewer and be done with it! It shouldn't be that difficult. I don't understand anybody asserting that the reviewer is in any way out of line here. The submitted calculation probably has the word "FAIL" written right on it. What is he/she supposed to do?

 

[jerseyshore]

Imagine if something happened on this job and OP's FAILED calc is plastered all over the Engineering Failures & Disasters page. And you thought you were getting a lot of heat in this thread...

 

[IceNine]

I personally am going to cut the OP some slack. While I agree that we shouldn't exceed 1.0 and it is easy to correct for new construction, it gives me exactly zero heartburn, especially for wood construction. I've evaluated enough existing wood structures that are 200% 'overstressed' and are still working after 80 years to know this is not much of a concern.

 

[canwesteng]

There's no way a 1.02 utilization leads to failure, and there's no way anyone would seriously consider reporting you to the board for this, that is just ridiculous.

 

[milkshakelake]

2% overstress will not fail. That's showing a disregard or misunderstanding of how the stress is calculated and how live loading was "invented," whether it's ASD or LRFD. If it does fail, OP can say they added a few psf of load, or there could've been a bunch of other things mitigating the overstress.

When you have a failure, it's usually a combination of a design error and a construction/maintenance error. You can have one of those sometimes, if it's particularly egregious. A 2% overstress is not that. In the Hyatt Regency walkway collapse, the overstress was about 200%.

I get people hammering OP about dismissing the plan examiner's claims, and I agree with it, but the criticism is over the top. I don't think it's deserved.

 

[Eng16080]

2% overstress will not fail.

I know this thread is getting a bit ridiculous, but I don't even agree with this statement as an absolute certainty. The risk of failure in this case is exceedingly small, of course, but suppose the beam is a wood member at the bottom 5 percentile strength, suppose the live load supported by the beam exceeds the minimum required design load per code, and suppose the contractor made significant errors installing the beam (which was never noticed). In that case, perhaps the beam fails. The design check is trying to account for all these risks to some extent, in a way which should give a reliable solution.

If nothing else, per this example, if the beam were to fail, even if the root cause was a construction error, it certainly looks bad if it comes to light that the calculation done by the design engineer shows a 2% overstress. Good luck convincing a group of jurors (non-engineers) that it was actually ok. If I was one of those jurors I would certainly argue in defense of the engineer, making a lot of the points mentioned here, but I really doubt others would see it that way. It would of course be presented to the jury as a black and white matter, and people like to see things as black and white.

Purposely providing a design check with any amount of overstress is certainly not wise from a risk management perspective.

 

[canwesteng]

The risk of failure is negligibly higher than if it was at 1.0. I don't even agree with leaving 1.02 in as a ratio for new construction, and I would do the same as the plan reviewed, but whatever caused a beam at 1.02 to fail would surely do the same for a beam at 1.0