Learning Forum

[Farah]

 

I am running a two-way FSI involving blood and vessel, but I encounter excessive thickness change problem. I can only obtain results when using a very high Young's modulus (10,000 MPa). Despite this, the simulation does not converge consistently; sometimes it converges, and other times it does not. (time step= 0.001)

I want to perform the simulation using actual values for the vessel, which means using a lower Young's modulus, around 0.05 MPa. However, when I do this, the simulation crashes immediately with the following error message: "Element xxx has excessive thickness change".

I am confident that the fluid part of the simulation (Fluent) works well even without FSI, but the structural part keeps crashing when I use the lower Young's modulus. I know that people are simulating blood flow in vessels without any errors, but I am unsure how to fix this issue. 

I have seen this question frequently asked by users, but no answer has been provided, so I hope someone can help me resolve this problem. Thanks!

 

Edit (Rob) - cleaning up posts. 

I am running a two-way FSI  involving blood and vessel, but I encounter this error "excessive thickness change".

I can only obtain results when using a very high Young's modulus (10,000 MPa). Despite this, the simulation does not converge consistently; sometimes it converges, and other times it does not.

I want to perform the simulation using actual values for the vessel, which means using a lower Young's modulus, around E_Vessel=0.05 MPa. However, when I do this, the simulation crashes immediately with the following error message: "(DP 0) Participant Solution encountered a fatal error. Message: Element 248 has excessive thickness change. Please do not save the project if you would like to recover to the last saved state." 

I am confident that the fluid part of the simulation (Fluent) works well even without FSI, but the structural part keeps crashing when I use the lower Young's modulus. I know that people are simulating blood flow in vessels without any errors, but I am unsure how to fix this issue.

[Rahul]

Hello

You might need to refine your mesh or adjust the material properties to ensure they are consistent throughout the simulation.

If you have not already done so, I recommend checking the mesh quality and looking for any abrupt changes in thickness that could be causing the error. 

[Rahul]

Also, refer to innovationspace.ansys.com/courses/courses/topics-in-hyperelasticity-using-ansys-mechanical/lessons/how-to-handle-element-distortion-errors-in-hyperelastic-materials-lesson-3/ 

For certain two-way FSI coupled analyses, convergence instabilities can be addressed with System Coupling's Quasi-Newton solution stabilization and acceleration method. Improving Coupled Analysis Stability (ansys.com)

[Farah]

 

Hi Rahul,

I have tried different elements that are less susceptible to distortion. I checked the materials, and in the papers, they have all defined these properties for the wall, but I still receive an error.

 *** ERROR ***                           CP =      74.062   TIME= 10:32:17
 Element 3909 has excessive thickness change. 

There are four warnings that I will list below, but honestly, I don’t know which part I’m missing. I am just trying a simple FEA to see if it can pass.

*** WARNING ***                         CP =       1.328   TIME= 10:31:05
 Element shape checking is currently inactive.  Issue SHPP,ON or         
 SHPP,WARN to reactivate, if desired. 

*** WARNING ***                         CP =       1.578   TIME= 10:31:05
 Material number 2 (used by element 8668) should normally have at least  
 one MP or one TB type command associated with it.  Output of energy by  
 material may not be available.                                          
*** WARNING ***                         CP =       4.875   TIME= 10:31:07
 A reference moment value times the tolerance is used by the             
 Newton-Raphson method for checking convergence.  The calculated         
 reference MOMENT CONVERGENCE VALUE = 2.228180446E-06 is less than a     
 threshold.  This threshold is internally calculated.  You can           
 overwrite it by specifying MINREF on the CNVTOL command.  Check         
 results carefully.                                                      
     DISP CONVERGENCE VALUE   =  0.2255E-05  CRITERION=  0.5135E-05 <<< CONVERGED
    EQUIL ITER   2 COMPLETED.  NEW TRIANG MATRIX.  MAX DOF INC=  0.3668E-03
     FORCE CONVERGENCE VALUE  =  0.8002E-06  CRITERION=  0.5000E-04 <<< CONVERGED
     MOMENT CONVERGENCE VALUE =  0.1027E-08  CRITERION=  0.5000E-04 <<< CONVERGED
    >>> SOLUTION CONVERGED AFTER EQUILIBRIUM ITERATION   2

 *** WARNING ***                         CP =      74.062   TIME= 10:32:17
 The unconverged solution (identified as time 1 substep 999999) is       
 output for analysis debug purposes.  Results should not be used for     
 any other purpose.

[Farah]

Hi Rahul, 

I have tried different elements that are less susceptible to distortion. I checked the materials, and in the papers, they have all defined these properties for the wall, but I still receive an error.

*** ERROR ***                           CP =      74.062   TIME= 10:32:17
 Element 3909 has excessive thickness change.

There are four warnings that I will list below, but honestly, I don't know which part I'm missing. I am just trying a simple FEA to see if it can pass.

*** WARNING ***                         CP =       1.328   TIME= 10:31:05
 Element shape checking is currently inactive.  Issue SHPP,ON or         
 SHPP,WARN to reactivate, if desired. 

 

*** WARNING ***                         CP =       1.578   TIME= 10:31:05
 Material number 2 (used by element 8668) should normally have at least  
 one MP or one TB type command associated with it.  Output of energy by  
 material may not be available.                                          

*** WARNING ***                         CP =       4.875   TIME= 10:31:07
 A reference moment value times the tolerance is used by the             
 Newton-Raphson method for checking convergence.  The calculated         
 reference MOMENT CONVERGENCE VALUE = 2.228180446E-06 is less than a     
 threshold.  This threshold is internally calculated.  You can           
 overwrite it by specifying MINREF on the CNVTOL command.  Check         
 results carefully.                                                      
     DISP CONVERGENCE VALUE   =  0.2255E-05  CRITERION=  0.5135E-05 <<< CONVERGED
    EQUIL ITER   2 COMPLETED.  NEW TRIANG MATRIX.  MAX DOF INC=  0.3668E-03
     FORCE CONVERGENCE VALUE  =  0.8002E-06  CRITERION=  0.5000E-04 <<< CONVERGED
     MOMENT CONVERGENCE VALUE =  0.1027E-08  CRITERION=  0.5000E-04 <<< CONVERGED
    >>> SOLUTION CONVERGED AFTER EQUILIBRIUM ITERATION   2

 

 *** WARNING ***                         CP =      74.062   TIME= 10:32:17
 The unconverged solution (identified as time 1 substep 999999) is       
 output for analysis debug purposes.  Results should not be used for     
 any other purpose.
 

[Farah]

 

I've also tried Quasi and Explicit technique but encountere with the same error! 

 

[Ashish Khemka]

Hello Farah,

Please see if the following forum thread helps you: Struggling with Convergence in Structural Analysis. Force drop! (ansys.com)

Regards,

Ashish Khemka

[Farah]

*** Update ***

I finally figured out that I needed to address "Stabilization" I've been working on convergence stabilization by adjusting the:

1. Under-Relaxation Factor (URF)
2. RMS Convergence Target
3. Using different structural solvers (mechanical, explicit) 
4. Step size
5. Number of maximum iterations
6. Mesh size

Although it's taking a lot of time for the computer to run (like a day or two), I hope this helps you focus on these specific areas rather than trying out other things. I should try to figure out how to make it more efficient.

 

 

[Ashish Khemka]

Thanks a lot, Farah, for sharing the update!

 

Regards,

Ashish Khemka