Solver Error when using the Direct Analysis Method 1

[dozer]

As anyone getting errors similar to this when using DAM?

***ERROR - AT JOINT 11 DIRECTION = MY
PROBABLE CAUSE NEGATIVE L-DIAGONAL
K-MATRIX DIAG= 6.5431013E+03 L-MATRIX DIAG= -1.9110223E+01 EQN NO 17

If I run using PERFORM ANALYSIS instead of PERFORM DIRECT ANALYSIS it works. Also I found that if I decrease the offending load below a certain level it elimates the error. I've already submitted this to Bentley but I'm just curious if anybody else has come across this because they always make me feel like I'm the only person in the entire world that ever has a problem.

Oh, and to save some of you time. Yes, I'm sure I've got the boundary conditions correct. Yes, I'm sure I didn't get my units fouled up. Yes, I'm sure I'm holding my mouth right. Actually, these problems were brought to me by other people in the company so I've already asked them these things.

BTW, what's an L-matrix?

 

[JoshPlumSE]

When STAAD does the direct analysis method, are they adjusting the Tau_b factor based on the axial load in the member? If so, then my guess is that their programers did not include a check to make sure that the stiffness reduction did not exceed a maximum value.

My guess is that Tau_b is approaching zero or has become negative. That's what's causing the "negative L diagonal" message. If you have a value that is zero (or less than zero) in the main diagonal of the stiffness matrix, then that would probably be bad.

The above is really just a guess based on what I know about the direct analysis method and the description you gave. I'm not really a STAAD expert.... I usually use another analysis package.

 

[dozer]

The value of the load is no where near enough to cause Tau_b to be less than one, however, in the name of thoroughness, I instructed the program to not compute Tau_b, just leave it at 1. It still crapped out. Thanks for playing.

 

[Gumpmaster]

My impression is that STAAD's implementation of the DAM is pretty half baked. Have you tried running it with the advanced analysis engine? We had lots of problems while using the normal engine.

Are you using 2007 build 5?

 

[dozer]

Gumpmaster,
Unfortunately, we don't have the advanced analysis engine. I've used it before and its pretty nice, though I feel it's a bit shady on their part to say, "Oh, by the way, if you want a quality solver, you have to pay us another $3k".

Anyway, good question about build 5. Yes, that's what I'm using. Build 4 (and I assume 3, though we skipped that one) has known problems.

 

[Gumpmaster]

I'm not sure if build 5 fixed the DAM not working with the regular solver or not.

If you have an enterprise license, they're offering the advanced solver free now. See the notice here:

http://feeds.rapidfeeds.com/8131/

Supposedly build 6 is coming out soon, so maybe that will fix it.

Good luck

 

[accountclosed]

Check the same model on Build 6. My guess is that this is probably no longer an issue.

Also, be sure to run your Direct Analysis on REPEAT LOADs that represent realistic loading conditions. Try to avoid running it on individual load cases that don't represent realistic loading conditions in and of themselves.

 

[Sye123]

The basic solver in STAAD was not equipped to handle negatives in the diagonal which led to these errors. As suggested by some of you, Advanced solver could handle that situation which led to successful analysis completion. However the good news is that the issue has been addressed by STAAD development team and a new build ( 20.07.07... ), which is expected to be released in couple of weeks, should fix the issue you had encountered with the basic solver.