[hjhjrh]

Hello,

I am conducting a CFD simulation of blood flow in the aorta and have encountered an issue where the MKE (Mean Kinetic Energy) equation does not hold.

The aortic model includes a 930 mm inlet extension to ensure a fully developed velocity profile, as this was also done in the experiment. I have verified that the simulated velocity profile matches the experimental data.

The GCI value is 0.1%, and the timestep size is set to 0.5 ms, determined based on the CFL condition.

Initially, I used the SST k-ω model, but since it could not resolve Reynolds stresses, I switched to the Reynolds Stress Model (RSM).

Since the velocity profile is uniform, I simplified the MKE equation (https://www.docsity.com/en/docs/mke-...-notes/409345/) by neglecting terms 3 and 6. The individual terms were computed as follows:

Term 1 (Material derivative of MKE) was computed by evaluating MKE at different time steps, using spline interpolation, and then differentiating.

Terms 2 and 4 were computed in CFD-Post, using manually defined expressions and the areaAVE function.

Terms 5 and 7 were computed in Tecplot using volume integration, excluding the 930 mm inlet region.

Despite these considerations, the equation does not balance.

I am wondering whether the MKE equation is expected to hold in Fluent simulations or if discrepancies are common. I suspect the issue might stem from the scale of Term 1, which is approximately 100 to 1000 times larger than the other terms.

I would appreciate any insights or recommendations regarding this issue.

Thank you in advance!

[sbaffini]

Originally Posted by hjhjrh

Hello,

I am conducting a CFD simulation of blood flow in the aorta and have encountered an issue where the MKE (Mean Kinetic Energy) equation does not hold.

The aortic model includes a 930 mm inlet extension to ensure a fully developed velocity profile, as this was also done in the experiment. I have verified that the simulated velocity profile matches the experimental data.

The GCI value is 0.1%, and the timestep size is set to 0.5 ms, determined based on the CFL condition.

Initially, I used the SST k-ω model, but since it could not resolve Reynolds stresses, I switched to the Reynolds Stress Model (RSM).

Since the velocity profile is uniform, I simplified the MKE equation (https://www.docsity.com/en/docs/mke-...-notes/409345/) by neglecting terms 3 and 6. The individual terms were computed as follows:

Term 1 (Material derivative of MKE) was computed by evaluating MKE at different time steps, using spline interpolation, and then differentiating.

Terms 2 and 4 were computed in CFD-Post, using manually defined expressions and the areaAVE function.

Terms 5 and 7 were computed in Tecplot using volume integration, excluding the 930 mm inlet region.

Despite these considerations, the equation does not balance.

I am wondering whether the MKE equation is expected to hold in Fluent simulations or if discrepancies are common. I suspect the issue might stem from the scale of Term 1, which is approximately 100 to 1000 times larger than the other terms.

I would appreciate any insights or recommendations regarding this issue.

Thank you in advance!

That's not how balances actually work. You either:

1) Solve full DNS equations and from them directly compute every term in the balance equation you are interested in (as per your slide) or...

2) Stick to the balance equations you are actually solving. In particular, for a RANS approach, you have a mean kinetic energy equation deriving from the single equations you are solving. Which, in turn, involves the one resulting from the mean momentum conservation equations AND how turbulent kinetic energy appears in it (which in turn depends also on specific details like if k was absorbed in the pressure or not). If you don't have access to the code you used for the computation I see it as a very difficult route.

Besides this, of course, there might be other specific issues related to your results which nobody can spot at this level