[cm_jubayer]

I have quick question. If I want to simulate forced convection heat transfer over a horizontal flat plate which solver would be the best choice?

Jubayer

[deji]

Hello. I am simulating free and mixed convection turbulent boundary layer flow with a low Mach number solver that I came up with based on fireFoam. So the question for you is what regime of forced convection are you simulating, is it incompressible or compressible?

Cheers,
Deji

[fabian_roesler]

Hi

You should go for buoyantPimpleFoam or buoyantSimpleFoam depending on your problem, whether it is steady state or not. If you have incompressible flow you can go for the boussinesq solvers and if fluid density is constant in addition you can set the volume expansion coefficient to zero.

Regards

Fabian

[cm_jubayer]

Thanks deji and Fabian. I am dealing with incompressible flow and natural convection is negligible compared to forced convection. At first I thought buoyant solvers solve U equation based on gradient of density only, so I added temperature to simpleFoam solver. After running my case with the temperature added simpleFoam solver, I am getting huge continuity error. But, now I see buoyantSimpleFoam/buoyantSimpleFoam has pressure term as well in the U equation. I will follow Fabian's advice which is to set alpha=0 to treat the flow as incompressible. Thanks.

[cm_jubayer]

I tried to mean buoyantBoussinesqSimpleFoam/buoyantBoussinesqPimpleFoam instead of buoyantSimpleFoam/buoyantPimpleFoam.

[fabian_roesler]

Hi

You should better go for buoyantBossinesqSimpleFoam. This solver is for incompressible flow with Boussinesq approximation for natural convection. There you can set the volume expansion coefficient to zero (no natural convection anymore).

Fabian

---

Well, didn't read your last post. So you're on the right track.

Fabian

[cm_jubayer]

Really need your help guys. As I said earlier in this thread that I wanted to simulate forced convection over a flat plate and compare the Nusselt number values with the Nuseelt number correlation for the turbulent boundary layer over flat plate [Nu = 0.037*(Re^0.8)*Pr^(1/3)]. I need this to see how OpenFOAM performs in case of forced convection heat transfer and also to educate myself so that I can use the knowledge for my research with much complicated geometry.

I am using low-Re SST komega model with very low turbulence (~0.01%) at the inlet. I have a uniform velocity (20 m/s) at the inlet. My domain is just a long box with bottom of the box as the plate (uniform fixed temperature). Sides of the boxes are empty (2D). I am using buoyantBoussinesqSimpleFoam with beta=0 and g=0 (no natural convection). After running the simulation, I am getting very low heat flux and thus very low Nusselt number compared to the turbulent boundary layer correlation . I ran the same geometry with same boundary condition in FLUENT and got a good match. Then I dug deep and found that both FLUENT and OpenFOAM uses gradT to measure heat flux. And there I found that the value of gradT at the wall (with near cell) is really low in OpenFOAM compared to FLUENT which is giving me low heat flux values. Can anyone suggest why my temperature value at the near wall cell is so different in OpenFOAM than FLUENT? Thanks.


Jubayer

[Lodda]

Originally Posted by cm_jubayer

Really need your help guys. As I said earlier in this thread that I wanted to simulate forced convection over a flat plate and compare the Nusselt number values with the Nuseelt number correlation for the turbulent boundary layer over flat plate [Nu = 0.037*(Re^0.8)*Pr^(1/3)]. I need this to see how OpenFOAM performs in case of forced convection heat transfer and also to educate myself so that I can use the knowledge for my research with much complicated geometry.

I am using low-Re SST komega model with very low turbulence (~0.01%) at the inlet. I have a uniform velocity (20 m/s) at the inlet. My domain is just a long box with bottom of the box as the plate (uniform fixed temperature). Sides of the boxes are empty (2D). I am using buoyantBoussinesqSimpleFoam with beta=0 and g=0 (no natural convection). After running the simulation, I am getting very low heat flux and thus very low Nusselt number compared to the turbulent boundary layer correlation . I ran the same geometry with same boundary condition in FLUENT and got a good match. Then I dug deep and found that both FLUENT and OpenFOAM uses gradT to measure heat flux. And there I found that the value of gradT at the wall (with near cell) is really low in OpenFOAM compared to FLUENT which is giving me low heat flux values. Can anyone suggest why my temperature value at the near wall cell is so different in OpenFOAM than FLUENT? Thanks.


Jubayer

Hello Jubayer,

have you meanwhile found a solution for your problem? Im working on a similar case and my heatflux is also to low.

Best regards

Lodda

[cm_jubayer]

Hi Lodda,

you can try 'nutUSpaldingWallFunction' at the wall for nut, 'omegaWallFunction' for omega. These are continuous wall function that gives profile up to y+ =0.

Jubayer