[Santiago]

Originally Posted by cyln

@alexeym

I varied relTol between 0.1 and 0.01. I am getting the same error.
In CFX, I assigned mass flow rate at the outflow boundary. It is so robust, converge no matter what. I wanna assign the same BC in OF.

@Santiago

I am a CFX user, however, switching to OF. I need to capture the velocity gradient at the outlet as accurate as possible.

You need a precursor simulation in order to start with this one. Again, you are solving Incompressible flows and mass must be conserved at all times. Usually, inflow/outflow BC's what they do is to integrate mass on domain and then somehow assing the mass defect (outflow = (sum u dvol )/ Vol - inflow ) to the outflow(s); obviously such defect cannot be big because it will violate the integral relation of mass conservation (inflow=outflow) rendering the Cauchy condition on the Poisson equation invalid. A good start could be a potential flow, either generated using potentialFoam or generated manually. Another option is to set a fixedValue velocity BC for the inlet and zeroGradient/outflowInflow velocity BC in the outlet

Look it up in this way, you start your simulation (a channel, one outlet one inlet normal to x) from a zero field with your inflow BC. The predictor equation will update U* only by diffusion, and since the outflow velocity is equal to zero, you'll get a linear horizontal profile of the intermediate velocity. The poisson equation will throw a linear function in x for the pressure gradient which will generate a velocity profile that monotonically decreases along x, reaching zero at the boundary since this boundary is usually updated after. The inflow/outflow BC will force the mass flux on the exit to be equal to the integral over the volume of the said velocity profiles minus the inflow flux. As you may have seen, two problems arise: (1) a -possibly- large discontinuity is set on near the outlet, (2) the integral conservation of mass is not guaranteed by this method, since what it is imposed on the outflow BC are mass flow INCREMENTS.

[alexeym]

Originally Posted by cyln

I varied relTol between 0.1 and 0.01. I am getting the same error.

And why not 1e-3? 1e-5? 1e-8? Since adjustPhi function is trying to adjust mass flux calculated as H/A, reducing solution error can help. Since all your velocity BCs fix value, there is no way to adjust mass flux. Btw zero gradient for velocity for one of the outlets is not an option?

In CFX, I assigned mass flow rate at the outflow boundary. It is so robust, converge no matter what. I wanna assign the same BC in OF.

So, in CFX you in fact set a) fixed pressure at the outlets, b) set mass flow rate outflow BCs? I.e. you use "Mass Flow Rate: Scale Mass Flows" in terms of CFX 11.0 theory guide.

[veeturi]

I'm not sure if this helps anyone but here's what I did:

As explained a lot of people in this thread, If you use flow rates at all the outlets, that's ill posed. Instead, impose flow rate at inlet, calculate percentage flow at one of the outlets, and leave the other outlet as 0 pressure. If >2 then do the same for the others i.e. specify flow rate at all the outles except the farthest one and specify that as 0 pressure.

My U file is as follows

boundaryField
{
INLET
{
type flowRateInletVelocity;
volumetricFlowRate 0.0005;
value uniform (0 0 0);
}

OUTLET1
{
type flowRateOutletVelocity;
volumetricFlowRate 0.000205;
value uniform (0 0 0);
}

OUTLET2
{
type zeroGradient;
}

WALL
{
type noSlip;
}
}

and P file was

boundaryField
{
INLET
{
type zeroGradient;
}

OUTLET1
{
type zeroGradient;
}

OUTLET2
{
type fixedValue;
value uniform 0;
}

WALL
{
type zeroGradient;
}
}

Hope this helps