Why a high temperature region is not generating a uphill flow?

I am running several test simulation about the heat convection in a room.

But the plot showed by the result turns out it seems the heat release is not influencing the flow field at all, which should not be correct, can you please enlighten me why?

2 figures are shown below, they are result from 2 different run of simulation, while the first one is set that there is a temperature gradient on the floor and the second one is set that just adiabatic, but the velocity vector as well as the velocity magnitude is exactly same.

I have tried a lot of way to set up the boundary of heat source but it seems not different

This does not makes sense, any how while there is a temperature gradient, there should be an uphill flow and at least the average velocity should be different.

Is it possible that my convergence went wrong?

Hello weixuan_yu,

Can you please share a link to your project here so that we can have a look at it?

Also which analysis type are you using for your project?

Best regards

Dear Block.

Pls find the link which shows the project.


I have run test on both conjugative heat transfer V2.0 and confection heat transfer. The result is the same. No any extra uphill flow is generated.

Best Wishes


The domain size is incredibly small (given the velocity magnitude at the inlet), therefore this problem that you are analyzing is not dominated by natural convection.

It doesn’t mean that buoyancy effects are not playing a role in the simulation, it just means that the advection term is too large in comparison.

PS: both approaches that you ran use the Boussinesq approximation, in case you are interested in reading a little bit more.


Dear Ricardo,

I have adjusted the velocity inlet to 0.004m/s , and it works, thank you.

However, I still don’t fully understand why .

If i use the Gr number to calculate the nature convection, or to be more, I assume the pressure gradient is purely turned into kinetic energy, then the velocity caused by such a 20K temperature even with characteristic length of 0.02m should also be the scale of 10^-2.

Also an experience equation from the following page offers the similar result (but maybe not accurate) ,

Using this equation, if I focus on the height of 0.01 m, the velocity should be 0.16m/s.

in any theory, the scale should be no less than 10^-3, which should show some result in average velocity magnitude while there is not.

Also, I have another project with similar situation where the domain area is 12m*5m, still i cannot see any uphill trend.

link below - simulation- run 20