[foamF]

Hi Formers,

I am running a case to simulate the inlet of a pumping station. The water is flowing via three inlet openings to three wet wells which are inter-connected with some openings. Each of the wet well is equipped with 2 pumps. The scenario under simulation is a low-flow condition - 3 inlets with 0.67m3/s for each (total 2.01m3/s) and 1 pump-on ONLY (others are off). I use interFOAM as the solver and kOmegaSST as the turbulence model. The inlet/outlet condition is simply velocity inlet and pressure outlet. There are also patches as atmosphere and wall. (Note: I used similar setup for other many cases working fine.)

During the simulation, I monitor the residuals, pressure at the atmosphere patch and flow rate at the outlet patch, as usual, so as to determine whether the steady state reaches or not. I also looked at the log file, indicating that the time-step continuity errors are small enough, which are in the order of 10^-7 (local), 10^-8 (global) and 10^-5 (cumulative). I used Paraview to visualize the results. I tried to extract the outlet patch, and compute the flow rate at outlet patch (which is nearly the same as that using OF flowRatePatch. HOWEVER, the flow rate at the outlet is only about 1.86m3/s, which is only about 92.5% of the total inlet flow.

MY QUESTION is why the total flow apparently looks not conservative but other results look reasonable?

When I have a look at velocity variation at the outlet patch, it looks the velocity variation is significant (due to the swirling behavior). My guess is that the mesh at outlet patch is not fine enough to capture the velocity variation, so there would be "error" when applying integration to compute the flow rate. Using a finer mesh could result in a outlet flow closing to the inlet flow. At the moment, I am running another case with a finer outlet mesh, but it really need some time for the simulation (near 2M grid is hard for my workstation).

Any expert here can comment on my observation and give some advice to improve the situation? (Note: To make the simulation faster, I used Gauss upwind for the turbulence as the divergence scheme and max Courant No. up to 6. But, I didn't think it is the reason leading the apparently "not conservative" flow for steady state, right?)

[akidess]

You are violating the stability criterion with your Courant number.

[foamF]

do you mean that the situation I encountered is due to high Co no.? may i have further elaboration from you?

based on my understanding, implicit scheme is used in OF. stability shouldn't be affected too much by Co no., unless it is too large resulting in high fluctuating results.

Instead, Co no. affects hidden some temporal data. Provided that only the steady state is of concern, Co no. should not be an issue (unless steady state can be achieved).

[akidess]

interFoam is explicit.

[Ford Prefect]

Originally Posted by akidess

interFoam is explicit.

So I guess it is important to specify which Courant number we are talking about.


This works just fine for many cases:

maxCo    2;
maxAlphaCo    0.5;