LES: cyclic U and non-cyclic scalar transport

pippo2013 · February 16, 2017, 05:58

Dear all,

I am using OF 2.3 in order to simulate a periodic urban canyon.
I managed to correctly perform the simulation using cyclic boundary conditions.

However, now I want to add a scalar source from the bottom of the channel, hence cyclic conditions do not work because the scalar field is not cyclic at all! I would like to impose periodic conditions just for U, and fixed value (=0 ) at the inlet and open boundary condition at the outlet for scalar concentration.

I have tried with direct mapped (recompiling it since it is not included in 2.3) without succeeding...

Could you please help me??

Thanks in advance!!

agustinvo · February 17, 2017, 11:32

If I am not worng, the problem is that you don't want to have recirculation of your scalar, right? otherwise it's always changing and so on. For periodic channel flows, what people use to do is to put a heat sink, taking into account the heat you introduce through the walls. Maybe you can try to do this for your scalar.

Or, what you can do, is to recirculate your velocity field before your scalar source, and you have this "cyclic" condition before. Only if the velocity is not affected by the scalar, I think you could do this.

Luiz · February 10, 2020, 14:11

Does anyone has an idea to this problem?
I tried the above solutions but applying a damping function to the scalar didn't work, and simulating the flow and then the scalar isn't an option due to the time required to simulate.

Different to this approach, I tried the mapped BC, although I need to allow backflow since it is part of my physical problem. In order to allow backflow, I tried mapping the pressure from a totalPressure outlet, but this leads to divergence. So either the flow works and the tracer is cyclic or the tracer doesn't go periodic but the simulation diverts.
Does anyone have ideas for this case?
Also, working on OpenFOAM v1912.

February 16, 2017, 05:58	LES: cyclic U and non-cyclic scalar transport	#1
pippo2013 New Member Join Date: Apr 2013 Posts: 26 Rep Power: 13	Dear all, I am using OF 2.3 in order to simulate a periodic urban canyon. I managed to correctly perform the simulation using cyclic boundary conditions. However, now I want to add a scalar source from the bottom of the channel, hence cyclic conditions do not work because the scalar field is not cyclic at all! I would like to impose periodic conditions just for U, and fixed value (=0 ) at the inlet and open boundary condition at the outlet for scalar concentration. I have tried with direct mapped (recompiling it since it is not included in 2.3) without succeeding... Could you please help me?? Thanks in advance!! wes1204 and hhu_lulu like this.

February 17, 2017, 11:32		#2
agustinvo Senior Member Agustín Villa Join Date: Apr 2013 Location: Alcorcón Posts: 313 Rep Power: 15	If I am not worng, the problem is that you don't want to have recirculation of your scalar, right? otherwise it's always changing and so on. For periodic channel flows, what people use to do is to put a heat sink, taking into account the heat you introduce through the walls. Maybe you can try to do this for your scalar. Or, what you can do, is to recirculate your velocity field before your scalar source, and you have this "cyclic" condition before. Only if the velocity is not affected by the scalar, I think you could do this.

February 10, 2020, 14:11		#3
Luiz New Member Luiz Oliveira Join Date: Aug 2018 Location: Napoli, Italy Posts: 24 Rep Power: 7	Does anyone has an idea to this problem? I tried the above solutions but applying a damping function to the scalar didn't work, and simulating the flow and then the scalar isn't an option due to the time required to simulate. Different to this approach, I tried the mapped BC, although I need to allow backflow since it is part of my physical problem. In order to allow backflow, I tried mapping the pressure from a totalPressure outlet, but this leads to divergence. So either the flow works and the tracer is cyclic or the tracer doesn't go periodic but the simulation diverts. Does anyone have ideas for this case? Also, working on OpenFOAM v1912.