Learning Forum

[marius1.gratzl]

Hey, I am doing a simulation of a mechanical heart valve. Since my overset approach failed and my dynamic mesh approach also failed, I took a step back and only simulate certain leaflet angles without any movement.

While simulating laminar flow, everything works fine but as soon as I switch to turbulent (SST-k-omega) I run into the same error at any attempt.

Some information on the geometry: Entire fluid domain 150mm in length, a minimum diameter of 17mm, and maximum width of 30mm.

Mesh ≈ 5.000.000 cells with inflation layers on all wall boundaries and two refinement regions in the areas of interest (See Screenshots).

Orthogonal quality > 0.01

Simulation setup: SST-k-omega (default settings) – Methods tried: Simplec, Coupled, and Piso

Time step: 0.002s and 0.001s

Iterations/TS: 80-120 (multiple simulations with different IT/TS)

Time simulated: 0.8s (one cardiac cycle)

Inlet: Mass flow Inlet with the boundary condition below (watch out for the values! Image below shows the mass flow in l/min, while in Fluent kg/s is used!)

Outlet: Pressure Outlet

No matter which settings or mesh alterations, I always get the same Floating Point exception error at about the same simulated time (between 0.29s to 0.31s, see screenshot), at the most negative mass flow.

Console:

 ....
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 10901 cells 

 12808  8.2199e+07  9.2068e-03  1.7513e-02  1.1618e-02  4.2917e-01  4.2450e+00  0:01:31   32

 turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 15871 cells 

 12809  8.5765e+07  7.2338e-03  1.4483e-02  8.1843e-03  2.6342e-01  6.8199e-01  0:01:23   31

 turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 21356 cells 

 12810  1.0002e+08  6.8773e-03  1.3285e-02  7.0888e-03  2.0145e-01  2.1047e-01  0:01:22   30

 turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 29060 cells 

 12811  1.2231e+08  6.1226e-03  1.2139e-02  6.2705e-03  1.7502e-01  1.9840e-01  0:01:21   29

        Stabilizing k to enhance linear solver robustness.

        Stabilizing k using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: k

 turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 26956 cells 

Divergence detected in AMG solver: k

Divergence detected in AMG solver: k

Divergence detected in AMG solver: k

Divergence detected in AMG solver: k

Divergence detected in AMG solver: k

Divergence detected in AMG solver: k

Divergence detected in AMG solver: k

Error at host: floating point exception

Error: floating point exception

Error Object: #f

Y+ values throughout the simulation stay between 0-2, although looking at the y+ values when the error occurs, they increase to almost 300.000.

I do have reversed flow on 100% of the outlet faces, but since I also have a negative inlet mass flow at that time, this shouldn´t bother me, should it?

I am also confused by the initial mass flow value at time step 0. When plotting my mass flow in Matlab, the initial value is 0 l/min (equal to 0 kg/s) but when using the same function in Fluent, the initial value is 0.03 kg/s, why?

If someone could assist me in eliminating that error or tell me why this even occurs, I would be very thankful.

Regards,

Marius

[DrAmine]

Is the flow turbulent or laminar? Can you do a rough estimate of the Reynolds Number? 
You wrote you OQ is more than 0.01: okay but that is a warranty for smooth run. I recommend to have a OQ not smaller than 0.1 or 0.2 and Cell Volume Change should not exceed 1.2. Aspect Ration should be smaller than 100. 

I do not understand your "mass flow" BC and how you are flowing in to or flowing out of the domain: is your BC a sort of pump sucking in and out? You can if the density is constant use instead of velocity inlet with dedicated fow direction but changing the direction of the inflow (outwards now)

[marius1.gratzl]

My flow is turbulent with Re>30.000

My inlet BC is set to Massflow inlet, with the massflow rate you can see in the image above. And yes, think of it as a pump pumping fluid in and out of the domain. Negative massflow = sucking fluid out and positive massflow = pumping fluid in