Learning Forum

[dzaki8598]

Hi,

I am trying to simulate a steady-state 2D flow with natural/free convection. As an example, this is the schematic of the simulation. It is a 3x3 room with a solar chimney. The pressure inlet is on the right hand side of the picture. The heat flux comes from the left side of the picture (below is a velocity contour, the heat comes from the high velocity surface side in the picture)

[dzaki8598]

(EDIT)
I have managed to reach the steady solution. Apparently, in some cases it is recommended to not initialize the natural convection simulation using forced convection simulation first. Now, I have reached stable residuals and the monitors (volume average velocity and temperature) are almost very steady, also the contours are the same as the results with the forced convection initialization.

[Rob]

Good to hear.
As an aside, natural convection flows tend to exhibit transient behaviour so the solver may struggle to get a single steady result. However, the variation in the flow with time is small so we often use the steady solver with monitors to confirm the flow field is converged: residuals tend to get stuck before reaching convergence.

[dzaki8598]

Thank you for your answer, Rob.
Also, I want to add that the continuity residual of my simulation may not have reached convergence (although it's going stable), but from the monitors the volume average velocity has steadied. As for the volume average temperature, the monitor hasn't been as steady as the velocity monitor, although the difference of value between each iteration is very small (but the graph is still increasing at a relatively low rate.
Although that may be the case, I have validated my simulation using a reference paper, and surprisingly it shows a very agreeable results with the paper!
So, is my simulation good enough now in order for me to analyze the physics?
Thank you
Regards Dzaki

[DrAmine]

If the monitors plotted over the iterations are not changing anymore towards the end of time step then you probably reached steady state. I always recommend doing mesh and time step sensitivity analysis to verify the methods but I will keep that decision to you!

[DrAmine]

And natural convection are inherent unsteady: even if you are lucky with steady solver you should see that as a best practice!

[DrAmine]

You residual plot are bad, though. Can you add screenshots of your mass and heat flux reports. Thanks!