Staged Analysis Fails

[nsharmanishant]

Hello,
I am trying to carry out staged gravity analysis of a single 2d soil element(linear elastic). The bottom two nodes of the element are restrained in both the horizontal and vertical direction. The top two nodes are restrained in only the horizontal direction. In the 1st stage, gravity analysis is done and the nodal reactions developed at the four nodes are noted. In the second stage the nodal reactions developed are applied at the corresponding nodes. In the next stages the horizontal restraints of each node is removed one by one. The program is able to remove the horizontal constraints of three nodes. However, when the horizontal restraint of the fourth node is removed the program shows the following error and in the last stage of analysis the program does not iterate even once.

"ProfileSPDLinDirectSolver::solve<> - aii< 0 < <i, aii>: <0,0>
WARNING NewtonLineSearch::solveCurrentStep<> -the LinearSysOfEqn failed in solve<>
StaticAnalysis::analyze<> - the Algorithm failed at iteration: 0 with domain at load factor 6
OpenSees > analyze failed, returned: -3 error flag"

Could you kindly share your thoughts on it. Your help will be of great value.

Thank You
Nishant

[akisingh]

Can you tell me more details with full error description with code,
i try to solve your problem.

[Jhno]

I did run into the same kind of trouble with my soil model. I had issue when updating the material from elastic to plastic. Although I did have some convergence issue in the linear domain, it was mostly the result of a ill defined model.
For me, the error came from the following issue : 1- bad fix condition for the model ; 2-wrong application of the loading (+ or - sign of the applied forces or displacement).
The error you get, "ProfileSPDLinDirectSolver::solve<> - aii< 0 < <i, aii>: <0,0>" shows that your model is ill defined. You should post your code here so we can help you better or you can re-check everything you done. I would suggest reading the following example, if not already done.
http://cyclic.ucsd.edu/opensees/Elastic ... ic/1x1.txt

[nsharmanishant]

akisingh wrote:
> Can you tell me more details with full error description with code,
> i try to solve your problem.


Dear akisingh,
ThankYou for your reply.Following is my code:

wipe;
puts "Creating domain nodes";

model BasicBuilder -ndm 2 -ndf 2

node 1 0.00 0.00;
node 2 2.00 0.00;
node 3 0.00 2.00;
node 4 2.00 2.00;


fix 1 1 1;
fix 2 1 1;
fix 3 1 0;
fix 4 1 0;

puts "Creating soil material";
nDMaterial PressureDependMultiYield 2 2 1660 7.5e4 2.0e5 33 0.1 80 0.5 27 0.07 0.4 2 10 0.01 1 0.7;

puts "Creating element connectivity";
element quad 1 1 2 4 3 1.0 "PlaneStrain" 2 0.0 1660 0.0 -16284.60;

file mkdir NDomainSize;

puts "Creating analysis recorders";

recorder Node -file NDomainSize/NodalReactions.out -time -nodeRange 1 4 -dof 1 2 reaction;
recorder Node -file NDomainSize/NodalDisplacemnts.out -time -nodeRange 1 4 -dof 1 2 disp;
recorder Element -file NDomainSize/ElementStress.out -time -dT 0.1 -ele 1 material 1 stress;

updateMaterialStage -material 2 -stage 0

puts "Material set to linear elastic stage..."

#-------Stage 1----------------
constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm NewtonLineSearch 0.5
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

puts "Finished Linear Gravity analysis..."

#--------Stage 2--------------
pattern Plain 1 Constant {
load 1 8142.3 0.0;
load 2 -8142.3 0.0
load 3 8142.3 0.0
load 4 -8142.3 0.0
}

constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm Newton
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

#------Stage 3-----------

remove sp 1 1;

constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm Newton
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

#------Stage 4-----------

remove sp 2 1;

constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm Newton
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

#------Stage 5-----------

remove sp 3 1;

constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm Newton
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

#------Stage 6-----------

remove sp 4 1;

constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm Newton
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

wipe;

#----End of Script------



Following is the Error message that the program shows

OpenSees > source QueryPosted.tcl
Creating domain nodes
Creating soil material
WARNING:PressureDependMultiYield:: numberOfSurfaces <= 0
Will use 10 yield surfaces.
Creating element connectivity
Creating analysis recorders
Material set to linear elastic stage...
CTestNormDispIncr::test() - iteration: 1 current Norm: 0.153533 (max: 1e-012, No
rm deltaR: 4.89778e-012)
CTestNormDispIncr::test() - iteration: 2 current Norm: 8.89273e-018 (max: 1e-012
, Norm deltaR: 4.06738e-012)

Finished Linear Gravity analysis...
CTestNormDispIncr::test() - iteration: 1 current Norm: 2.87607e-017 (max: 1e-012
, Norm deltaR: 1.81899e-012)

CTestNormDispIncr::test() - iteration: 1 current Norm: 1.89278e-017 (max: 1e-012
, Norm deltaR: 4.16783e-012)

CTestNormDispIncr::test() - iteration: 1 current Norm: 6.26136e-017 (max: 1e-012
, Norm deltaR: 1.28622e-012)

CTestNormDispIncr::test() - iteration: 1 current Norm: 1.27826e-017 (max: 1e-012
, Norm deltaR: 1.28622e-012)

ProfileSPDLinDirectSolver::solve() - aii < 0 (i, aii): (5, 0)
WARNING NewtonRaphson::solveCurrentStep() -the LinearSysOfEqn failed in solve()
StaticAnalysis::analyze() - the Algorithm failed at iteration: 0 with domain at
load factor 6
OpenSees > analyze failed, returned: -3 error flag


The Stress in the elements are maintained constant for the 5 stages. However in the 6th stage the analysis does not iterate even once.


Thank you for your help

[nsharmanishant]

Jhno wrote:
> I did run into the same kind of trouble with my soil model. I had issue when updating
> the material from elastic to plastic. Although I did have some convergence issue in
> the linear domain, it was mostly the result of a ill defined model.
> For me, the error came from the following issue : 1- bad fix condition for the model
> ; 2-wrong application of the loading (+ or - sign of the applied forces or
> displacement).
> The error you get, "ProfileSPDLinDirectSolver::solve<> - aii< 0 <
> <i, aii>: <0,0>" shows that your model is ill defined. You should
> post your code here so we can help you better or you can re-check everything you
> done. I would suggest reading the following example, if not already done.
>
> http://cyclic.ucsd.edu/opensees/Elastic ... ic/1x1.txt


Dear Jhno,
Thank You for your reply. I have attached my code along with the error in the previous post in this topic. Kindly suggest me how to overcome this difficulty.

Thank You

[Jhno]

I would try to add a "wipeAnalysis" command.
You did write your stage 1 with the following :
constraints Transformation
test NormDispIncr 1e-12 25 1
numberer RCM
system ProfileSPD
algorithm NewtonLineSearch 0.5
integrator LoadControl 1 1 1 1
analysis Static
analyze 1

Then you wrote stage 2 but you did not include a wipe analysis in between.
http://opensees.berkeley.edu/wiki/index ... is_Command

Also, when you remove the constraint at node, shouldn't you do it symmetrically?

Hope it helps,
Good luck.

[nsharmanishant]

Dear Jhno,
Thank You for your reply. I have incorporated wipeAnalysis in my script and have removed the constraints symmetrically in each stage. However, even now the program shows the same message. Kindly suggest.

Thank You
Nishant

[Jeena]

Hi,

I tried to change the last analysis part of your code as follows and it completed the analysis. I am not sure if the results are what you want, but the analysis was completed without errors.

I changed the numberer command to 'Plain' instead of 'RCM' and system command to 'BandGeneral' ...

remove sp 4 1;

constraints Transformation
test NormDispIncr 1e-5 10000
numberer Plain
system BandGeneral
algorithm Newton
integrator LoadControl 1 10000
analysis Static
analyze 1

Thanks,
Jeena

[nsharmanishant]

Dear Jeena,

Thank you for your reply. The tcl script is now able to run successfully after incorporating the changes suggested by you. Regarding the results, I shall let you know in this forum if they are turning out to be as desired. Thank you for your time and attention.

Regards,
Nishant