Error generating Mesh

[Sohrab Tehrani]

Hi David,

Could you please explain more? How should I create an artificial inlet? Attached, please find my inlet profiles.

Thanks

[David Parker]

Hi Sohrab,

I was thinking you could add two artificial nodes, one for an inlet and one defining a bifurcation to the actual inlets of your model (see image). The inlet flow would be the sum of the inlet profiles and the flow split to your original inlets could be determined by the vessel segment areas (maybe).

Cheers,
Dave

[Sohrab Tehrani]

Thanks so much. That's a great idea. How should I determine that area of the vessel cross section? I want to use just circular tool to segment the virtual branches.

[David Parker]

The cross section areas are given in the solver.inp file created from the centerlines. I suppose you could just use the same ones from the inlet segments? You will need to experiment to see if you get the correct flow split.

Cheers,
Dave

[Sohrab Tehrani]

Thanks David.

I created two virtual separate path and added to the two inlets from the bottom (please see the attached image) and chose the cross section areas based on their volume flow rate ratio (1.25). However, the SimVascular cannot extract one of the centerlines (please see the attached image). However, previously when I didn't have these virtual inlet branches (bottom ones), I could extract all the centerlines. I have also tried to remesh and smooth the geometry, but again I get the same things. Why is that? How should I fix it?

Best,
Sohrab

[Sohrab Tehrani]

Also, I constantly receive the error message that I have attached here when running 1D simulation. What does it mean?

Thanks,
Sohrab

[David Parker]

Hi Sohrab,

What I meant by adding two artificial inlets was to just modify the solver.in file, not modify the model geometry. Sorry for the miscommunication! What I thought you could do was to add a couple of NODE lines and modify JOINTINLET and add some new JOINT and SEGMENT lines.

If you'd rather modify the geometry then you will need to connect the artificial inlet geometry directly to the inlet geometry to get the centerlines to work and to create a valid 1D model.

The non-zero exit code message means that the 1D solver crashed. The solver writes out a log file named sv1dsolver.log in your project's Simulations1d/JOBNAME directory. That should give you more information about what is going on, like cross section area going to zero.

Cheers,
Dave

[Sohrab Tehrani]

Dear David,

Thanks, I got it. I modified the solver.in file according to the virtual segments and joints. However, the 1d solver crashes when I use the conservative form of the equations, even for very small time steps. But, the solver works fine even for larger time steps if I use the convective form. I know the main difference is the treatment of the boundary conditions. I was wondering how that affects the results? I mean the final results should be the same either we use the conservative or convective forms right?

Thank you very much.
Best,
Sohrab

[Sohrab Tehrani]

Hello David,

In my 1D simulation, I receive an error during the run time, and it is terminated. The error message is "Error: Cannot Solve Linear System". Attached, please find the "generate-1d-mesh", and the "solver.in" files. Please note that here I use the convective form of the equations, but when I use the conservative form the solver crashes due to the error that I have attached here and the simulation doesn't proceed at all? By the way, I have another question about the size of the vessel in the input file. Is it the vessel diameter or its actual area?

Could you please help me resolving the issue? What is my mistake?
Best,
Sohrab

[David Parker]

Hi Sohrab,

The error message is "Error: Cannot Solve Linear System" means that the system of equations for you 1D network is singular, maybe something to do with boundary conditions or possibly a bad connection between nodes?

From Melody Dong:

Advective form - non-linear term to RHS, explicit, depends on previous time-step, less stable, less time
Conservative form - nonlinear term stays in LHS, implicit, more stable, more time because you have to invert each time step

I think in general, we decided that we should just stick to the conservative form because then the simulation results would not rely on the time-step and was generally more stable for a wide range of simulations.

The generate-1d-mesh.log file contains several warnings

Code: Select all

[generate-1d-mesh] WARNING - Ain < Aout for group id 0, face name 'Out_RICA'
[generate-1d-mesh] WARNING - This will cause problems for the 1D solver so setting Ain = Aout
[generate-1d-mesh] WARNING - Ain < Aout for group id 7, face name 'Out_LICA'
[generate-1d-mesh] WARNING - This will cause problems for the 1D solver so setting Ain = Aout
[generate-1d-mesh] WARNING - Ain < Aout for group id 8, face name 'Out_RBasi'
[generate-1d-mesh] WARNING - This will cause problems for the 1D solver so setting Ain = Aout
[generate-1d-mesh] WARNING - Ain < Aout for group id 11, face name 'Out_LBasi'
[generate-1d-mesh] WARNING - This will cause problems for the 1D solver so setting Ain = Aout
[generate-1d-mesh] WARNING - Ain < Aout for group id 12, face name 'On_LICA'
[generate-1d-mesh] WARNING - This will cause problems for the 1D solver so setting Ain = Aout

Having vessel areas that are the same may be causing problems.

The vessel size in the solver input file is the vessel area.

Cheers,
Dave