Mill reports for redesign 3

[Samwise Gamgee]

On a project, due to some miscommunication, one of the areas was designed for lower occupancy. The project is all constructed and about to be occupied. When I ran numbers for the higher occupancy, a total of 4 beams and 3 connections are overstressed.

Beams : 2 have a strength DCR of8-10%, 2 have deflection DCR of L/190 instead of L/240).
Connections : 2 are overstressed by 3% and 1 is overstressed by 9%.

For the beams which are overstressed for strength, I got the mill report for the exact member and they have a higher Fy which brings down the DCR to 1.0
For the beams which are deflection more, as its a serviceability issue are there any additional concerns I need to consider ?

For connections, I asked a mill report for the shear plates which are overstressed due to shear rupture. The contractor said, they can only provide representative samples of the shear plates which were used. I used the one with lowest strength which was still higher than Fu of 58 and can keep the DCR 1.0

My question is for the beams as we have exact mill reports, can we justify the design to be well within safe limits using higher Fy. For connections, it is reasonable to consider the lowest strengths of Fu from the representative samples and be sure its safe ?

Also, the building has been constructed for more than a year and was occupied by construction personnel and their equipment. Haven't heard any complaints about behavior or discomfort.

 

[Celt83]

For the beams you can investigate cover plates to achieve strength and serviceability.

For the connections I wouldn't mess around and determine what a retrofit connection would look like.

 

[Lomarandil]

I haven't used it myself, but recall mention here of a clause (IBC?) that prohibits the application of mill certs to get capacities/properties better than the material specs.

(But if you can identify that steel is actually A992 instead of A36, that's OK)

You might search for that.

 

[dold]

I actually stumbled upon this exact scenario when browsing this recent thread: Link

See lexpatrie's mention of the Station Square collapse

Steel strength
The MSS Group recommended strengthening the beam above the column by adding a beam bracket with web stiffeners. This change was not implemented when Tamm Tacy advised that mill certificates obtained from the steel sub-contractor indicated to be 25% higher strength than expected. It was later determined that this type of data is not a truly representative sampling of the steel as fabricated and so is not a valid measure of actual strength.

See this link for the final report for that failure: http://www.llbc.leg.bc.ca/public/pubdocs/bcdocs/43459/commissionerinquirystationsquare.pdf]Link[/url]

So, I'd agree with others here. Don't rely on mill certs to "increase" available strength of your beams or connections.

 

[canwesteng]

One thing I strongly disagree with in the report above and might help you is using the column face to column face distance and not CL to CL distance, provided you have properly accounted for the moment in the columns (which I would argue perhaps you already should have).

EDIT - upon looking at the actual layout, using col face to col face in that mall was certainly an error. But for simple beams between columns I think it is not.

 

[dold]

@canwest
Yeah this Tammy Tacy was trying to squeeze every last bit out of this structure. Somebody somewhere along the way definitely knew better but pushed this crap through anyway. And this was the result, Although not necessarily the root cause in the case of the beam flexural overstress pencil sharpening. Sounds like it was generally a stability issue.

 

[JLNJ]

I am curious. What was the "lower" occupancy in the original design and what is the "higher" occupancy now?

Maybe you can sharpen your pencil on the load side rather than the resistance side.

 

[ETX]

The AISC code is clear that the design strength is based upon the specified minimum yield stress of the steel type used. The safety implication of using CMTR values is the reduction in the AISC factor of safety or safety margin.

 

[bridgebuster]

I used mill certs to reanalyze a temporary railroad structure. I inherited a project during construction and stumbled upon a major design error. I went the mill certs route and fortunately the yield strength was much higher.

 

[Heaviside1925]

A mechanical's viewpoint, so take it or leave it. MTRs are a validation of the ASTM standard and should exceed the ASTM standard. Fabrication modifies the materials properties. If proper fabrication procedures are followed, then material properties should not fall below that as prescribed in ASTM. So, unless you are using the material that hasn't been modified from the mill, the material properties realistically fall somewhere between ASTM and the MTRs or even may exceed but there is no way of validating this unless the actual fabricated piece is tested and validated.

 

[TLHS]

On mill reports, it depends what they're saying. Do they dual qualify for a stronger grade? If they do then it's super easy to just change your specified steel to that. If the cert has specifications you aren't familiar with as alternate grades, research them. If it's qualified as A36 and also CSA G40-300W on the plate, for instance, you could probably justify using the stronger Canadian grade even if you're in the US, for this specific situation. Here, steel coming in from China can often have a two or three different families of standards on them. If you're just some moderate amount above what you specified but not qualified for anything else on the report then that's a problem.

Here, the Canadian steel code specifically says you can't use the mill test values for design. There would likely be an argument for looking at it to justify an existing structure, but the Canadian code is also clear that you aren't allowed to use the existing structure provisions to justify issues in new construction. So the answer in Canada would be that, no, you can't unless you want to put your neck out and justify it for a very specific scenario using the general 'the code doesn't cover everything and the engineer may have to do their own thing' clauses we have have at the beginning of the codes.

You may not be able to say who's fault the 'miscommunication' was, for legal reasons. If it's not your fault, I wouldn't be initially pushing to be aggressive. I would start by looking to see how bad reinforcing may be. If it's reasonable, it may be easier for everyone to just go down that route.

If it's potentially your fault and you start getting aggressive in a way that people may not agree with in the future it may not be the worst thing in the world to get a review by an outside person or an uninvolved colleague. Not that you necessarily release it to anyone, but as a check on yourself. Basically, if you can make a mechanism to look over your shoulder a least a little bit and make sure you're keeping yourself honest it isn't the worst idea in the world. Not because you're up to anything, but because we can start pushing ourselves in ways we aren't realizing in these situations.

That being said, I'd start in other areas regardless.

The issue is likely the 9% connection and the two beams. Have you gone through all the demand side calcs to see if there's stuff you can play with? If your beams are fully restrained against LTB and you're basically looking at yield there may not be a lot of play, but if there's LTB there are a fair number of factors that you can potentially sharpen pencils on. Can you reasonably shift loads to other members based on floor geometry (i.e. is your tributary area actually your tributary area)? Can you play with your dead load at all, there tends to be more discretion there? These sound simply supported, but if not can you look at plastic analysis of some sort? Can you justify any composite action?

If the connection is a shear block failure only, it's reasonably accessible, and there are contactors mobilized, it's likely cheaper to just have them go replace the bolts one at a time, slip a doubler plate in and then weld the doubler plate. That is faster and less work that spending a bunch of time problem solving the engineering.

With beams, if it's a short span it's often easier to just put a parallel beam next to what you've got, a couple of feet away. With long spans, if you end up comfortable with deflection, reinforcing the center part of the beam is likely economical.

 

[Tomfh]

For connections, it is reasonable to consider the lowest strengths of Fu from the representative samples and be sure its safe ?

No, because it's very unlikely that you tested the lowest strength material.

 

[Tomfh]

Although not necessarily the root cause in the case of the beam flexural overstress pencil sharpening. Sounds like it was generally a stability issue.

Yes that definitely looked like a stability failure. The beam rolling off the column, like a rolled ankle.

 

[SWComposites]

There are statistical methods for analyzing a set of sample strength data to estimate a lower bound strength, for a specified level of reliability and confidence. We do this all the time in aerospace to get material strength allowables. I seem to recall seeing similar methods published for use with civil structures materials. Just using the lowest test value is usually not acceptable, unless your sample size is very large.

 

[EngDM]

I haven't used it myself, but recall mention here of a clause (IBC?) that prohibits the application of mill certs to get capacities/properties better than the material specs.

CSA S16 has this as well.

 

[lexpatrie]

Gold Star for Dold!

I'm not caught up, so if this has been covered already, sorry.

The objection (as I recall) with the mill cert is it was for the full heat, i.e. a lot of the steel, meaning it was a point sample that wasn't statistically significant (i.e. less than 30 tests), would be my comment there, I think that's the objection the committee and various investigation folks had as well.

I'm not sure you can get a mill cert for a specific piece of steel. That sounds like a misunderstanding of what the mill cert was/is.

If, maybe, you can rule out the other forty eight things that went wrong with Station Square, maybe this degree of overstress (which is "on paper" at this point), maybe that degree of overstress is more tolerable, but you'd really want to recheck everything surrounding the calculation, i.e. is there any extra dead load that was missed, was the beam section transcribed wrong, did the contractor add any extra weight, installed with natural camber downward (Does the bolting even permit that?), etc. We tend to design steel beams for full width of dead and live load, rather than having some of the load at the ends going into the girder, which is at least a potential secondary load path, if not how they actually behave. (I just got a hold of Fling's concrete book from the local library and that's why I mention this).

 

[PersonalProfile]

The following may be of interest:

he compliant mill (Figure 1(a)) targets a mean yield strength value well above the minimum requirement of
AS/NZS 3679.1 300 Grade. Obviously some of this material will be above the average and some below the
targeted average, however all batch test results will be above the 300 MPa minimum.
The (rogue) mill manufactures to a generic Standard with grade 235 MPa product compared with the
compliant mill producing the Australian Standard 300 grade. Even though the rogue mill is targeting 235
MPa (Figure 1(b)) it will produce a certain percentage of product which when batch tested will return values
in excess of the required 300 MPa for AS/NZS 3679.1 Grade 300 product. The rogue mill/distributor will
market the material that exceeds the 300MPa tensile test result as AS/NZS 3679.1 Grade 300 product with
mill certificates supporting that claim.

Why grading by selection does not produce compliant product?
Even though the batch testing of the product returns results that meet the strength requirements of AS/NZS
3679.1, the variation within a batch is not taken into account by this grading process. AS/NZS 3679.1
requires that one test is taken for a batch size of under 50 tonnes and two tests are required if the batch
exceeds 50 tonnes. For a typical 120 tonne batch of 310UB40 beams or 3000 metres of beam, just two
sections of approximately 0.5m in length will be taken for testing. Although steel is a relatively homogeneous
material there are still variations in a batch which are not identified with just two samples.

Ex:

https://www.steel.org.au/ASI/media/Australian-Steel-Institute/General Imagery/32_Ng_Syam_Taylor.pdf