Difference between input flow rate and result flow rate

[HyeonWoo Park]

I'm studying aortic blood flow through Simvascular.
To do this, an issue occurred in the flow rate during several tests through the cylinder model.
I'm aiming for the flow rate I entered to match exactly in the results as well.
However, I tested several flow conditions and found that the flow rate was different.
Is there a way to solve this? I'd appreciate your help.

[David Parker]

Hello,

Finite elements is a discrete approximation to a physical process, the finer the mesh the better the approximation. Try increasing the mesh density to get a better approximation.

Cheers,
Dave

[Vijay Vedula]

You have a very steep rise and fall in the flow rates in your input file. The code uses Fourier modes internally to interpolate this prescribed flow rate at each time step. However, Fourier modes cannot capture such sharp gradients and will create a smoother profile, which is sometimes oscillatory. Increasing the number of Fourier modes will help, but it still cannot match such sharp gradients.

If you still want an exact match with the input profile, you may use General BC and prescribe the flow rate over the entire face (similar to a bct.dat file or an ASCII as described in https://github.com/SimVascular/svFSI/bl ... r.inp#L890). For General BC, the code uses linear interpolation in time and will be able to capture these sharp gradients.

-- Vijay

[HyeonWoo Park]

Hello,

Finite elements is a discrete approximation to a physical process, the finer the mesh the better the approximation. Try increasing the mesh density to get a better approximation.

Cheers,
Dave

I'll try. Thank you Dave

[HyeonWoo Park]

You have a very steep rise and fall in the flow rates in your input file. The code uses Fourier modes internally to interpolate this prescribed flow rate at each time step. However, Fourier modes cannot capture such sharp gradients and will create a smoother profile, which is sometimes oscillatory. Increasing the number of Fourier modes will help, but it still cannot match such sharp gradients.

If you still want an exact match with the input profile, you may use General BC and prescribe the flow rate over the entire face (similar to a bct.dat file or an ASCII as described in https://github.com/SimVascular/svFSI/bl ... r.inp#L890). For General BC, the code uses linear interpolation in time and will be able to capture these sharp gradients.

-- Vijay

Vijay, thank you.
Then, would it be helpful to use the same method you suggested when there are many points in the waveform, such as the attachment file?

[Vijay Vedula]

In principle, yes. You can apply General BC even if you have many data points, but the input file size will be large. Unless there is some reason for you to apply the exact profile, you may create a smooth interpolant that is periodic, e.g., by using Fourier modes. This will also reduce the total number of points. Your current waveform still has sharp variations in the flow.

flow.png (29.42 KiB) Viewed 792 times