RISA 3D 7.0 Code checking discprepancy

[cessna98j]

So we just received the latest version of RISA 3D, v7.0 which now includes code checking from the AISC 13th. I thought I would do some code check comparing and noticed a pretty substantial difference between the LRFD 3rd edition and AISC 13th (LRFD) code check when looking at WT sections in flexure. For example, a simply supported WT8x20 unbraced, with a 10K load at mid-span yields a unity check of around 1.43 based on LRFD 3rd ed. code in RISA. However, by just changing the code to AISC 13th (LRFD), the code check drops by nearly half to 0.7. The thing is, both code checks reference the exact same equation in their respective references (H1-1b, and hand calcs tend to agree more with the AISC 13th (LRFD) code check, but I would think RISA would show exactly the same code check because the equations are the same for singly symmetric members in both books. Seeing as how hand calcs tend to agree more with the AISC 13th ed. code check, it looks like the LRFD 3rd code checking for WT sections was perhaps not done correctly in the previous versions of RISA as well. Has anyone else had a chance to look at RISA 7.0, or can somebody confirm? Was thinking maybe I should send an email to RISA to let them know of the situation.

 

[JAE]

Definitely send them an email. This is the sort of thing that we as engineers should be very proactive about. The use (and mis-use) of software can be minimized if we always notify our software vendors of suspected problems right away. We've received 7.0 but not installed it yet so I can't check your numbers myself with 7.0.

 

[271828]

Definitely send an e-mail.

Great job verifying with a hand calc! It's scary how many people don't do that. If the 3rd Ed. implementation was wrong, that's REALLY scary considering how long ago RISA added that.

I know of one multi-zillion dollar fix that was required because an engineer did not check what another program was doing with WTs. The program's WT implementation was flat wrong and the trusses started to collapse when they started sitting the precast on it. Problem was that there were dozens of these trusses already up.

 

[JKStruct]

I agree with the above. Send the email (and tell us what they say). And assuming your hand calc shakes out correctly, kudos on the catch!

I really don't mean to hijack this thread, but 271828, do you have a reference, case study, or any further information about the failure you mentioned? I only asked because our office is currently designing a staggered truss building with WT sections as the diaphragm chords. We're detailing the precast plank to brace the trusses during construction, but I would be interested in reading more about that issue.

I wish they had a personal message feature on these forums, but I guess that would attract recruiters. Ah well...

 

[271828]

JKStruct, I can't say enough to ID the firm or project, but I can say is that it had to do with checking the WTs for bending. It was a large stadium and had the same trusses all the way around. At some point in the design, a walkway was added that chopped out part of the truss and made a vierendeel panel.

This was about 6 years ago, so that software isn't being used nowadays. It wasn't a Spec. problem. The software was just plain wrong, spitting out phiMn that weren't even close. I'm sure it's been fixed by now.

It comes down to the age old issue: The design engineer was in a big hurry like we all are and didn't verify the results. I've done that as I'm sure most here have, but he was unlucky.

I also wish we had a PM feature.

 

[cessna98j]

Well I emailed RISA and here is there response:



"The difference is likely related to the different ways that the two codes handle slender elements. It looks to me that the 13th edition doesn’t care about web slenderness for bending. In the 3rd edition, Chapter F specifically does not apply to member with slender elements. Therefore, the Qs value is taken into account.

I have little doubt that the 13th edition code check is more on par with what the code writers’ intentions were. However, baring a specific section of the code that deals with Slender sections, we cannot use those same equations for LRFD 3rd."



Haven't had time to research this yet, but I suppose it does sound reasonable - what do you guys think? I did notice that in the RISA output for both code checks, a Qs value is calculated but perhaps not taken into account with the 13th ed. code check.

 

[271828]

I would've hoped that RISA would KNOW why the answers are different not use language like "likely related"--oh well, I guess that's too much to ask.

The Commentary to F9 talks about this stuff.

I don't have my 3rd Ed. here, so I can't look at the difference between the two formulations.

 

[haynewp]

What is the length of the WT?

 

[cessna98j]

The length that I used was 10 feet, WT8x20.

 

[cessna98j]

Also, the WT is oriented such that the Loading is in the plane of symmetry, and the stem of the tee is in tension.

 

[miecz]

Looks as though RISA used a "conservative" interpretation of the 3rd Edition in Version 6. The opening paragraph of (3rd Ed.)Chapter "F" reads "for members with slender compression elements, see Appendix B5. If the neutral axis is in the stem, then technically, some part of the stem is a slender compression element. The 13 Ed. doesn't have this verbiage, at least not in F9.

 

[271828]

jmiec, the Commentary to F9 explains the situation with the stem in compression. So much for Specs always getting harder and never making life easier!

 

[haynewp]

1) I got a 0.83 stress ratio with LRFD 3rd with my hand calculation. I ran it in GTSTRUDL and got the same.

It appears that RISA is using the Qs factor since part of the stem (nearest the flange) is in compression but it still uses the 1.5My instead of 1.0My for the limit. This seems inconsistent to me at this point. But I bet what they are doing is still overly conservative since you say the AISC 13th gives a ratio of 0.7.

2) Next, I flipped the tee to where the tip of the stem is in compression. I calculated 2.149, and STRUDL gets 2.148. In this case, I used the Qs reduction with the 1.0My limit as it of course should be.

3) Next, I tried RAM Advanse and I got a stress ratio of 2.0 and it didn't matter whether the stem was oriented up or down.

 

[hokie66]

Doesn't give me much confidence in some of these programs. That is the point, you have to know yourself whether the program is giving correct results before you can depend on it.

 

[WillisV]

haynewp - what load type (D or L?) and combo (1.4D?) did you use for your check? Did you include SW? Did you calc CB?

 

[haynewp]

I applied the 10kip load as was given, I don't know if that already included a factor or not. No self weight and there is no Cb in the tee equation so 1.0.

 

[haynewp]

I also tried a slightly older version of RISA today. I got the same as the op with 1.43. Flipped the tee so the stem is up and got 2.149 with RISA which matched previous STRUDL and hand calc.

 

[miecz]

It seems to me that RISA 6.0 agrees with the 3rd Edition, and RISA 7.0 agrees with the 13th. I don't think the problem is so much with RISA as it is with AISC's 3rd Edition. The 13th Edition is worded differently than the 3rd edition with regard to Tees (See my above post). RISA made a correct, though perhaps conservative, interpretation of both codes.

 

[haynewp]

I think that is right. It looks like the Qs is excluded for tee stems in flexure in the 13th. On page 16.1-45 of Chapter F, it has "N/A" for web slenderness of tees. It appears the commentary supports this also.

"No limiting stem width-thickness ratio, ?r, is provided in this section to account for the local buckling of the stem when it is in compression. The reason for this omission is that the lateral-torsional buckling equations (Equations F9-4 and F9-5) also give the local buckling strength as Lb approaches zero"

What I am not convinced of is should the limit be 1.0My or 1.6My according to the 13th even if only a small part of the stem is in compression adjacent to the flange? It would seem to me as it would be 1.6My. Penalizing the entire section because only a small portion of the stem above the N.A. is in compression seems very conservative and would exclude most all tee sections from using the 1.6My limit for bending.

 

[WillisV]

Just as a matter of comparison, SAP 2000 is giving 0.83 for 3rd ed. check and 0.78 for 13th.