[EGY]

Hi everyone.

I'am doing simulation of free convection in a adiabatic fluid domain which is consist of air. I have steel cylinder middle of the domain same length with my fluid domain and my cylinder have heat flux from bottom to top. In addition my cylinder's mean temperature reach 350 Celsius. In Initilazing the situation i set these;

Air Meterial : Ideal gas
Referance Pressure : 1 atm
Bouyancy Model : Bouyant
Bouyancy Referance Density : 0.54 (Density of air for 625 Kelvin)

In solver control state:

Timescale Control : Physical time scale (set as 0.3 sec)

In Advection Scheme Convergence Control setting;

Continuty, Enerrgy, Momentum i set all of it physical time scale 0.01 sec


But when fuild domain temperature reach 400 Kelvin my momentum and velocity residual values start increasing and oscilating. My velocity vortex occurs unphysicaly and temperature have more way to converge.

I know it was a vert long post but it is about my senior design project and I stucked this point.

[ghorrocks]

Is this a steady state simulation? If so, I have 2 comments:

* If the domain is adiabatic and you have a heat source in it then how can it have a steady state? For steady state the heat in needs to equal the heat out. So how does the heat leave the domain? This would explain the increasing temperature.

* Free convection simulations often cause large scale transient flow structures. These structures cannot be captured by steady state simulations and require transient simulations to converge.

[EGY]

Yes it is steady-state simulation.

I do not want to heat out from my system, I want to observe in high temperature of free convection,conduction and radiation in this system

In image 2 you can see our geometry the outer side of fluid domain is adiabatic and just bottom boundry of cylinder have heat flux only and top boundry cooled to 300 Kelvin. Conduction occurs in cylinder and there will be free convection to fluid domain from cylinder. So I think this case is steady_state.

Every run I made system converged and temperature values I get are correct. On the other hand my velocity vortex occurs unphysically. You can see it in image 1 I send.

Thanks for helping me.

[ghorrocks]

I see - so the heat path is from the heat source to the 300K top boundary. This is OK.

Why do you say the vortex is unphysical? This is what commonly happens in free convection flows and this is why you cannot model these as steady state, they require transient simulations.

[EGY]

I get it but velocity vectors should be follow this path ;

1) Go top from bottom near the cylinder.

2) Then creating vortex.

I have done another steady state solution in very common systems and we get this vortex profile in image now I am sending.

So I get this correct solution in steady - state. After it for complate my design just my geometry and little radiation initilazing changed these are not suppose to effect velocity. But it is more sense to solve it in transient if we can not get right solutions like you said

[ghorrocks]

OK, so it looks like the downwards flow is wrong. This is probably because the flow was not converging and this was caused by either the numerical stability issues listed here (http://www.cfd-online.com/Wiki/Ansys...do_about_it.3F) or because the flow was transient and therefore a steady state run is never going to converge.

[EGY]

Really thanks for sharing your knowlage and time for me.

Wish you best with your works.