Mass balance in the FDS

[manjunath reddy]

Hello Fire modeller,
 
I'm working on compartment fire simulation. 
The room ( of size 0.4x0.4x0.4 m3) is attached to the adjacent air block (of size 0.4x0.4x0.6 m3) (Please find the Attached file - Geometry.png for more detail). Both the domains are connected at the bottom and at the top, and the size of connected region is 0.03x0.4 m2 (each). The fire source defined is at the centre of the floor inside the room. Open boundary conditions are defined over the Side 1, side 2, side 3, side 5, side 6 and a wall is defined for side 4.
 
I ran the simulations in FDS as well as in fireFoam solver(solver in OpenFOAM for fire simulations). From the results of FDS, I studied the pattern of the mass flow rates across the open boundary conditions which shows that flow is coming in from the side 3 and side 5  whereas flow is going out from the side 1, side 2 and side 6. More detail of the result can be found in the attached Graph “massFlowrate.png”. 
 
As one can observe that the domain is symmetrical about the centre XY plane, from that one can deduce that Side 1 and Side 3 should have same kind of flow behavior - either the flow will go out from each face or will come in from each face. But it is not happening in the present scenario. The noise introduced by the FDS will make the flow asymmetric (which can introduce some difference in mass flow rate) but the doubt here is that, will this asymmetricity alter the flow so much that the net flowrate from one face is entering the domain and leaving from other?
 
And I added mass flow rates of all sides at every time it should be 0 but it is not zero not even close to zero (You can see in the graph massFlowrate.png with legend total mass) it means that mass is not conserving in the FDS? Please some one help me why the flow is not conserving?
 
Thanks
Manjunath Reddy

[dr_jfloyd]

The sign of the mass flow device indicates the flow direction and not inflow or outflow. You have flow in through side 1 and in through side 3.  The signs on the two faces differ since inflow on one side is a different direction than inflow on the other side. You need to correct the signs to reflect inflow and outflow before doing your sums.  At 50 s I see roughly an outflow of 0.04 (side 6) and inflows of ~0.15 (2), ~0.05 (1), ~0.05 (5), and ~0.01 (3).  The inflows sum to ~0.04.

The net instantaneous mass flow through the open boundaries will not necessarily be zero, the total mass in the domain will change over time as the average density in the domain changes due to the fire. While the compartment is heating up there will be a net mass flow out of the domain. Once steady-state is reached, then the net mass flow should be close to the  fuel mass loss rate (it will still fluctuate due with average density fluctuations).

[manjunath reddy]

Floyd,
Thank you very much for your reply. Now it makes sense, I corrected the signs and added the mass flow rates and plotted(Attached massFlowrate_new.png), net flow rate is close to zero as expected.

One more question,

Is there a way to switch off the turbulent fluctuations completely to make the simulation symmetric?
I kept NOISE=.FALSE. but still results are not symmetric, I think burner vent also introduces some kind of random fluctuations(Same as of TurbulentInlet boundary condition) which breaks the symmetry, If that is the case, is there a way to change it?

Thanks
Manjunath

[dr_jfloyd]

If you want the simulation to be symmetric (which is not advisable since for an LES code that will impact the computed entrainment), use a MIRROR boundary condition.  You might be able to maintain symmetry longer by updating all radiation angles at once rather than in 5 increments. It will add to your computational cost, but will avoid adding an asymmetry due to the radiation solver. Without a mirror boundary condition, it is difficult to maintain symmetry forever. Any slight roundoff error at the limits of machine precision will eventually result in a loss of symmetry.