[hansenka]

Hei,
Sort of new to this world, especially natural convection. I have the same problem as many other people in this forum, but can't find the correct boundary conditions.
I am simulating a 20m high building with a heat source inside, 1 outlet and 4 inlets, two below and two at same level as heat source. I am struggeling to find boundary cinditions that are usabel for this domain. The flow is circulating like shown in the picture with short outlet. Struggeling with back-flow, and in this simulating, I have sort of cheated, using the inletOutlet for temperature, where the return flow is hoter than inside the domain.
On other option is by cheating and setting fixed values for inlet velocity, but I am trying to avoid that.
I may have to move the outlet further away, but I im interested in other ideas. Also playing with y_plus around heat source to improve the heat transfer.
I am naturally struggeling for convergence.

The bounday condditions:
P_rgh
boundaryField
{
inlet-basement
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
inlets-above-floor
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
outlet
{
type prghTotalPressure;
p0 $internalField;
}
wall-ground
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-basement
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-pot-room-walls
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-pot-room-roof
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-monitor
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}

wall-floor
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-pot-abovefloor
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-pot-belowfloor
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
wall-pot-bottom
{
type fixedFluxPressure;
gradient uniform 0;
value $internalField;
}
frontAndBackPlanes
{
type empty;
}

T


inlet-basement
{
type fixedValue;
value uniform 288;
}
inlets-above-floor
{
type fixedValue;
value uniform 288;
}
outlet
{
// type zeroGradient;
type inletOutlet;
inletValue uniform 360;
value $internalField;
}
wall-ground
{
type zeroGradient;
}
wall-basement
{
type zeroGradient;
}
wall-pot-room-walls
{
type zeroGradient;
}
wall-pot-room-roof
{
type zeroGradient;
}
wall-monitor
{
type zeroGradient;
}

wall-floor
{
type zeroGradient;
}
wall-pot-abovefloor
{
type externalWallHeatFluxTemperature;
mode flux;
q uniform 3000; //Heat flux [W/m2]
kappaMethod fluidThermo;
value uniform 300; //pot_room temp [K]
}
wall-pot-belowfloor
{
type externalWallHeatFluxTemperature;
mode flux;
q uniform 1000; //Heat flux [W/m2]
kappaMethod fluidThermo;
value uniform 300; //pot_room temp [K]
}
wall-pot-bottom
{
type externalWallHeatFluxTemperature;
mode flux;
q uniform 1000; //Heat flux [W/m2]
kappaMethod fluidThermo;
value uniform 300; //pot_room temp [K]
}
frontAndBackPlanes
{
type empty;
}
}

U, There is serveral combinations here, to have a insight at what I have been trying. In my head, outletInlet for inlets and pressureInletOutletVelocity are the one to choose, since the flex between fixed value and zeroGradient, depending on the flux, but this results in 0 flow at inlets and the simulation crashes.

boundaryField
{
inlet-basement
{
/* type outletInlet;
phi phi;
outletValue $internalField;
value $internalField;*/
// type zeroGradient;
type freestreamVelocity;
freestreamValue $internalField;
// type noSlip;
// type fixedValue;
// value uniform (0 0 0);
}
inlets-above-floor
{
/* type outletInlet;
phi phi;
outletValue $internalField;
value $internalField;*/
// type zeroGradient;
type freestreamVelocity;
freestreamValue $internalField;
// type noSlip;
// type fixedValue;
// value uniform (0 0 0);
}
outlet
{
// type zeroGradient;
type pressureInletOutletVelocity;
value uniform (0 0 0);
}
wall-ground
{
type noSlip;
}
wall-basement
{
type noSlip;
}
wall-pot-room-walls
{
type noSlip;
}
wall-pot-room-roof
{
type noSlip;
}
wall-monitor
{
type noSlip;
}

wall-floor
{
type noSlip;
}
wall-pot-abovefloor
{
type noSlip;
}
wall-pot-belowfloor
{
type noSlip;
}
wall-pot-bottom
{
type noSlip;
}
frontAndBackPlanes
{
type empty;
}

}

Epsilon
inlet-basement
{
type turbulentMixingLengthDissipationRateInlet;
mixingLength 0.0168;
value $internalField;
}
inlets-above-floor
{
type turbulentMixingLengthDissipationRateInlet;
mixingLength 0.0168;
value $internalField;
}
outlet
{
type inletOutlet;
inletValue $internalField;
}
wall-ground
{
type epsilonWallFunction;
value $internalField;
}

alphaT
inlet-basement
{
type calculated;
value $internalField;
}
inlets-above-floor
{
type calculated;
value $internalField;
}
outlet
{
type calculated;
value $internalField;
}
wall-ground
{
type compressible::alphatJayatillekeWallFunction;
Prt 0.71;
value $internalField;

For k (wall=kLowReWallFunction), Similar to epsilon, and nut equal to alphat for inlets and outlet.

I have gone back to a simpler case, to implement inflation layers in the mesh around heated walls, but still not fixed the problem.


I am intersted in your thoughts, especially the boundary conditions for U.

[piu58]

I don't understand your sketch fully.

For U, you used noslip b.c. which is adequate. I recommend to have at least one outlet. May be you have, at the upper part of your geometry. For this outlet, I recommend zeroGradient for U.

If you have a density driven flow, you have expanding of the fluid of course. It is easier simulating a free outflow (or inflow) in comparision to pressure changes.

[hansenka]

If you study one of the pictures, all the inlets and outlets are described. Outlet at top, as you mentioned.

I have tried zeroGradient for U_outlet, without luck. I'm trying one more time to see if it works, just in case.

[hansenka]

as i thought, zeroGradient was not successfull.