Capturing the step of analysis that a FBE fails

[Tas]

So, the version i am using it isn't the latest... I will try to work with the (real) latest and i will let you know (probably tomorrow because here is late midnight right now...). Thank you

[Tas]

I run the script in the latest version (2.4.4 rev 5730) and it is the same thing as before. You should watch the process of the analysis (when the steps are keep running) and the 'failure"(?) occurs somewhere between the middle and the end.

[fmk]

are we running the same script. SPEAR.txt

when i run it (i removed the 1 at the end of the convergence test to see if any error messages all i get on my mac and windows 64 is:

Script> OpenSees SPEAR.txt


OpenSees -- Open System For Earthquake Engineering Simulation
Pacific Earthquake Engineering Research Center -- 2.4.3 (rev 5695)

(c) Copyright 1999-2013 The Regents of the University of California
All Rights Reserved
(Copyright and Disclaimer @ http://www.berkeley.edu/OpenSees/copyright.html)


29.87 31.215 32.56 0.0034 0.0238 0.0442
60.71 61.61 62.51 0.0035 0.02375 0.044
30.37 31.055 31.74 0.0035 0.02385 0.0442
30.09 31.545 33 0.0033 0.02365 0.044
60.71 61.61 62.51 0.0035 0.02375 0.044
30.4 31.585 32.77 0.0034 0.02365 0.0439
30.27 31.525 32.78 0.0034 0.0236 0.0438
60.71 61.61 62.51 0.0035 0.02375 0.044
30.27 31.525 32.78 0.0034 0.0236 0.0438
124.45 127.28 130.11 0.0036 0.0241 0.0446
205.42 208.73 212.04 0.0038 0.02435 0.0449
288.4 291.06 293.72 0.004 0.02465 0.0453
206.38 208.895 211.41 0.0038 0.02435 0.0449
124.9 126.485 128.07 0.0037 0.02405 0.0444
124.34 125.925 127.51 0.0038 0.0244 0.045
30.65 30.925 31.2 0.0035 0.0238 0.0441
60.71 61.61 62.51 0.0035 0.02375 0.044
30.65 30.925 31.2 0.0035 0.0238 0.0441
30.27 31.525 32.78 0.0034 0.0236 0.0438
60.71 61.61 62.51 0.0035 0.02375 0.044
30.27 31.525 32.78 0.0034 0.0236 0.0438
31.23 33.615 36 0.0034 0.0236 0.0438
205.82 208.215 210.61 0.0039 0.02445 0.045
30.44 32.54 34.64 0.0034 0.02375 0.0441
30.4 31.585 32.77 0.0034 0.02365 0.0439
60.71 61.61 62.51 0.0035 0.02375 0.044
30.4 31.585 32.77 0.0034 0.02365 0.0439
83.53 85.37 87.21 0.0036 0.0239 0.0442
165.64 167.39 169.14 0.0038 0.02415 0.0445
83.38 85.455 87.53 0.0035 0.0238 0.0441
30.4 31.585 32.77 0.0034 0.02365 0.0439
60.71 61.61 62.51 0.0035 0.02375 0.044
29.99 32.17 34.35 0.0033 0.02375 0.0442
30.59 32.51 34.43 0.0034 0.0237 0.044
30.91 31.685 32.46 0.0034 0.02365 0.0439
60.71 61.61 62.51 0.0035 0.02375 0.044
30.65 30.925 31.2 0.0035 0.0238 0.0441
30.65 30.925 31.2 0.0035 0.0238 0.0441
60.71 61.61 62.51 0.0035 0.02375 0.044
30.09 31.545 33 0.0033 0.02365 0.044
1556.93 1556.93 1556.93 0.00345753 0.022474 0.0414904
1556.93 1556.93 1556.93 0.00345753 0.022474 0.0414904
1556.93 1556.93 1556.93 0.00438563 0.0285066 0.0526276
1556.93 1556.93 1556.93 0.00522325 0.0339512 0.0626791
number of steps: 306
Analysis Type: Pushover+Y
*** End of model generation ***
Script>

[Tas]

As i said, the analysis procedure finally ends succesfully but before it ends the warning message that i wrote in a previous post appears on the screen (while all the steps are appearing one by one on the screen, in one of them this warning message appears and after that step the procedure continues untill the end), so the question is if this message is something serious or not.

[fmk]

given the dW in your first message ... NO i would not say it is serious as the ele converges in the update.

BUT

what is puzzling is that i do not see the error messages you see! as i said, i removed the 1 at the end of the test so the screen was not filled with the convergence info AND there is no warning message popping up from any element anywhere when i run the script .. the results i get are shown above from start to finish.

[Tas]

Can you keep the 1 at the end of the test so you can view what i am saying in the screen in order to tell me if everything is o.k indeed ?
Maybe this warning message concerns only one iteration of the analysis step (in which it appears) and the "failure" finally doesn't take place at the end ofthe step but just in one iteration of it?

[fmk]

my point is that even with the 1 i see nothing from an element (the error messages from the element and the output from the test are seperate) .. i just took out the 1 to show you what i see and to make sure i was not missing it.

[Tas]

So the warning message can't be reproduced..!
Anyway, if a "capture" of what is appearing on the screen is what i wrote in my 3rd post:
...
CTestNormDispIncr::test() – iteration: 1 current Norm: 0.00218257 (max: 0.01, Norm deltaR: 542.135)

WARNING – ForceBeamColumn3d::update – failed to get compatible element forces & deformations for element: 8(dW: << 3.11392e-007, dW0: 1.90313e-005)
CTestNormDispIncr::test() – iteration: 1 current Norm: 0.0021708 (max: 0.01, Norm deltaR: 582.041)

CTestNormDispIncr::test() – iteration: 1 current Norm: 0.0020197 (max: 0.01, Norm deltaR: 540.048)
...

an explanation to this can it be what i said before: "Maybe this warning message concerns only one iteration of the analysis step (in which it appears) and the "failure" finally doesn't take place at the end of the step but just in one iteration of it?"

[fmk]

not by me . yes it is for 1 trial step (the predictor step)

[Tas]

When would such a message about ForceBeamColumn be a problem, only if there is no convergence at the end of the analysis procedure? If such warning messages appear for ForceBeamColumn elements, but at the end of the analysis procedure convergence is reached, is it something that we should worry about?

[fmk]

in every nonlinear analysis there is an initial prediction step in which the analysis increments the loads and may specify some initial guess at displacements for the analysis step, and then correction steps in which the algorithm alters the displacements to get the model into a converged state for loading and displacements set.
that message you showed should also be accompanied by the analysis step failing and control coming back to the script to try something else. what was happpening was that your script showed that if the element failed in the prediction, the analysis went on. I put a fix in the code for that, but cannot test it with your script as it does not show on any of my platforms.

if it fails in the trial steps, that error does stop the analysis. so in your case what was happeing is that 1 element was using the restoring forces and stiffness from the last converged step at the first step in the correction phase whereas the others were using the states at end of prediction state. at subsequent steps the problem element was able to achieve force-defo equilibrium and so did not stop the analysis. as mentioned, looking at the dW that were spit out when the problem occurred the ele was close to achieving convergence in that prediction step.

[Tas]

Thanks for the answers.
You said: "I put a fix in the code for that.." Does it mean that if i download again the latest version i will be able to try it in my script and test if the warning message still appears?
You also said: "..looking at the dW that were spit out when the problem occurred the ele was close to achieving convergence in that prediction step." That comes out by comparing the dW<<3.11392e-007 with dW0:1.90313e-005 in the warning message?

[fmk]

yes and NO .. I have not updated the .exe since i last mentioned in this topic .. I don't get the message when i run your script with that .exe, you say you do .. so if you did before you will again.

yes, e-10 is convergence .. your element was converging just not getting there .. most times it just diverges.