Results on simulation NaN

[Pilar Ortega Arévalo]

Hi,

I am trying to run a simulation using a 34000 element mesh. The boundary conditions I used is a steady flow of -2600000 mm^3/s and a resistance of 0.004615 Pa·s/mm^3.
When I obtain the results files, the pressure results are all NaN for all timesteps, as so the outflow results.

Captura de pantalla 2017-11-14 a las 12.01.21.png (28.62 KiB) Viewed 1875 times

I am using macOsSierra v. 10.12.6.

Do you know what can be wrong?

[Justin Tran]

Hi Pilar,

Thank you for your interest in SimVascular! A few follow up questions so that we can work through your issue:

1. The inflow rate you have specified is very high. -2600000 mm^3/s is equivalent to -2600 mL/s, which is much higher than a typical cardiac output of 80 mL/s. With an anatomical geometry and a flowrate that high, it is likely that the CFL number of the simulation is very high. The CFL number is defined as the ratio: (velocity)*(time_step) / (mesh_edge_length). In the worst case for an explicit time integration, the CFL number should be less than 1 for a stable simulation. SimVascular uses the generalized-alpha method for time integration, so it is not as strict. But can you confirm if this is the right inflow rate that you want to specify? What kind of model have you built (i.e. what vessels have you modeled)?

2. The outlet resistance you gave me corresponds to a pressure of 11999 Pa, which is 90 mmHg, which is a physiological value. Can you tell me how you have distributed this resistance to your three outlets (cap_Horse2, cap_Horse2_2, and cap_Horse2_outflow)?

3. Was the unit of measurement in your model mm? If the measurement system in your image data was in cm, then you will need to make sure the boundary conditions you apply in the simulations are consistent with that measurement system.

[Nathan Wilson]

Hello Pilar,

In addition to Justin's comments, can you please post an image of model and/or surface mesh near your outlets? 34,000 elements is a very, very small mesh density unless you are simulating a very simple geometry.

Nathan
SimVascular Development Team

[Pilar Ortega Arévalo]

Hi Nathan and Justin,

Thank you for your responses.
1. The reason for that big flow is that I am simulating a horse aorta
2. For the output resistance, there is a typo in my result file: I changed the name for cap_Horse2 and cap_Horse2_2 to cap_Horse2_outflow and cap_Horse2_inflow, respectively, and after the simulation both appeared in my file. So actually is the same data that appears repeated in the file (the output resistance is just applied to cap_Horse2_outflow, which is the same as cap_Horse2).
3. The unit of measure of my model was indeed mm.

Captura de pantalla 2017-11-16 a las 11.18.23.png (78.55 KiB) Viewed 1850 times

As an update, I was trying to run the simulation without the MPI. When I turned on again the MPI, the simulation worked perfectly. The reason why I did run the simulation without MPI is because I was getting an error message. For some reason, I only get that error message when I am connected to a wifi network that is not my university one. So this means I can only run the simulation connected to my university network, since if I want the simulation to run in other case I need to turn off MPI, and if I do that then I get NaN data.
Do you know why this happens? I just work now with a VPN so I do not have this problems anymore, but I am curious to know why, if SV does not need an internet connexion, is this happening.
Thank you a lot for all your help and online support!