Natural convection simulation divergence issue

[Ali Fauze]

Dear Openlb community,

I am currently working on a natural convection simulation based on the RayleighBernard3D example. The simulation involves aspirating an air flow within a rectangulare chanel, with one heated face and the other 3 faces cold, and there are 2 faces defined as the inlet and the outlet.

But my simulation diverges. The y+ value become high, and the same for the velocity (as seen in the output below).

If anyone has insights into why this might be happening or could offer assistance in resolving this issue i would greatly appreciate it.

[Timer] step=1872; percent=52.5843; passedTime=1854.35; remTime=1672.09; MLUPs=3.64843
[LatticeStatistics] step=1872; t=7.88688; uMax=0.253201; avEnergy=0.000308495; avRho=1.29214
[LatticeStatistics] step=1872; t=7.88688; uMax=0.253201; avEnergy=0.000279843; avRho=0.561948
[getResults] yPlusMax5=1.46316
[getResults] yPlusMax6=1.17847
[getResults] yPlusMax7=1.0356
[getResults] yPlusMax8=2.33729
[Timer] step=1896; percent=53.2584; passedTime=1876.9; remTime=1647.24; MLUPs=3.75894
[LatticeStatistics] step=1896; t=7.98799; uMax=1.25186; avEnergy=0.000321607; avRho=1.29219
[LatticeStatistics] step=1896; t=7.98799; uMax=1.25186; avEnergy=0.000291737; avRho=0.562247
[getResults] yPlusMax5=2.35587
[getResults] yPlusMax6=2.30773
[getResults] yPlusMax7=0.902924
[getResults] yPlusMax8=2.22089
[Timer] step=1920; percent=53.9326; passedTime=1919.53; remTime=1639.6; MLUPs=1.98829
[LatticeStatistics] step=1920; t=8.0891; uMax=0; avEnergy=nan; avRho=nan
[LatticeStatistics] step=1920; t=8.0891; uMax=0; avEnergy=-nan; avRho=-nan
[getResults] yPlusMax5=inf
[getResults] yPlusMax6=1293.36
[getResults] yPlusMax7=inf
[getResults] yPlusMax8=inf
[Timer] step=1944; percent=54.6067; passedTime=1942.18; remTime=1614.48; MLUPs=3.74251
[LatticeStatistics] step=1944; t=8.19022; uMax=0; avEnergy=nan; avRho=nan
[LatticeStatistics] step=1944; t=8.19022; uMax=0; avEnergy=-nan; avRho=-nan
[getResults] yPlusMax5=5.6955
[getResults] yPlusMax6=inf
[getResults] yPlusMax7=inf
[getResults] yPlusMax8=inf
[Timer] step=1968; percent=55.2809; passedTime=1964.64; remTime=1589.28; MLUPs=3.77266
[LatticeStatistics] step=1968; t=8.29133; uMax=0; avEnergy=nan; avRho=nan
[LatticeStatistics] step=1968; t=8.29133; uMax=0; avEnergy=-nan; avRho=-nan
[getResults] yPlusMax5=5.74115
[getResults] yPlusMax6=inf
[getResults] yPlusMax7=9.03935
[getResults] yPlusMax8=inf

PS: yPlusMax5 is the maximum y+ value on the face number 5 (material number 5).

Best Regards
Ali

[mathias]

There can be a lot of sources for unstability (bulk, boundary, initial condoitions, ..). One criteria for bulk instabilits is considered in Krüger’s book and well be studies also at our next spring school.

[Ali Fauze]

Hello mathias,

Thank you for your response. Does an ‘opening’ boundary exist in OpenLB to allow the flow to enter or exit freely?

BR,
Ali

[mathias]

Yes, there are serval options for free outflow conditions they are refferred to as “concection” in OpenLB, it is du/dn = 0 as well as d\rho/dn = 0.

Best
Mathias

[Ali Fauze]

Dear Dr. Mathias,

I appreciate your prompt response. After running the simulation with the Rayleigh-Bénard 3D example parameters, adjusting Ra and Pr values, I am pleased to report that the simulation did not diverge. However, the results obtained were unexpected. I observed the flow temperature and density both increased, contrary to the anticipated decrease in density. I attempted to fixe this issue by modifying the ThermalUnitConverter, and changing the physics model. Unfortunately, the simulation now diverges at the first time step.

Given that my ∆T is approximately 205°C, I am uncertain about the appropriate characteristics to assign to the ThermalUnitConverter at this temperature. Could you provide guidance on which temperature the characteristics shoud be assigned (Tmax or T0 …)?

Additionally, I would like to inquire if there is a recommended method for tracking convergence, like the transport equations used in RANS simulations.

Best regards,
Ali

[mathias]

These question are beyond the scope of this forum. Some of them we answer at the spring school where we have a week to go into details. The next school is in March 2024.

[Author]

Posts