[Fraxiel]

Hello everyone, Im currently running a CFD analysis on a conceptual design of our fixed wing UAV, I created the fluid domain (spherical), did the mesh with I would say fairly good quality (Orhtogonal quality is around 0.14 and skewness around 0.74 if I remember correctly), and I ran the first test for the aircraft in cruise at around M=0.73 and at 40000ft and the solution matches up with our hand calculations using excel. However, when I run the same mesh with a lower speed the solution convergence is VERY bad and the solution of the CL and CD are oscillating all over the place. Keep in mind Im fairly new with very minimal applications in CFD however, I know the basics of the flow and how everything should be its just the bugs and certain CFD (known facts for experts) that I lack I guess. I will attach some pictures so I hope you guys could help me.



Length= 30ft

Wingspan= 28ft

Cmean=6.1ft

cruising at M=0.73 at 40000ft (P=18821.73Pa, T=216.661K) good solution and residuals converge.

at takeoff at M=0.18 at sea level conditions (P=101325pa, T=288.16k) bad solution and residuals don't converge.

For the mesh I had local sizing done separately on the wings and tails to have smaller cells there do capture the change in flow properties around 5% of the Cmean (for the wing, Mesh size=0.0985m) and 1% of the Cmean (for the tail, Mesh size=0.0185m)

Surface mesh with the minimum being the smallest mesh size I have= 0.0185m and the maximum as default with the angle set to 12 deg.*

Volume mesh is set to the default values.

and 5 boundary layers.

Density: Ideal-gas

Viscosity: Sutherland

Turbelance model K-omega

solver method Coupled, green-gaused cell based,

Boundary conditions (Pressure far field of the spherical domain) either M=0.73 or M=0.18 depending on the case.

I don't get it when the velocity was higher the solution worked with a reasonable CL and CD value and I checked with varying angle of attack and it keeps matching our hand calculations on top of being reasonable values. However, When I lower the speed to around M=0.18 I get unreasonable values with VERY bad residuals and they never converge. Any help please?

[LuckyTran]

Initial conditions for each? Please don't say I used hybrid initialization... And what means converged? Is this like 10 iterations or 100 000 iterations?

[Fraxiel]

Originally Posted by LuckyTran

Initial conditions for each? Please don't say I used hybrid initialization... And what means converged? Is this like 10 iterations or 100 000 iterations?

Yeah Hybrid.. Like I said Im fairly new so Idk if that's like SUPER BAD or not. the residuals need to converge (decrease) the # of itterations is around (500) because I don't have that much time/computational power.

[LuckyTran]

It's not a sin or anything but you should visually inspect the solution from time to time and check that it makes any sense. If it looks mostly correct except for a few swirlies here and there then just let it crank longer. If you keep cranking and the swirlies don't go away then you can look for ways to stabilize it. 500 iterations is like baby taking their first step. It is premature to say the baby will never learn to walk (never converge) because they fall on the second step. Let it crank.


And don't look at residuals. Look at the plot of the velocity field from time to time.

[Fraxiel]

Originally Posted by LuckyTran

It's not a sin or anything but you should visually inspect the solution from time to time and check that it makes any sense. If it looks mostly correct except for a few swirlies here and there then just let it crank longer. If you keep cranking and the swirlies don't go away then you can look for ways to stabilize it. 500 iterations is like baby taking their first step. It is premature to say the baby will never learn to walk (never converge) because they fall on the second step. Let it crank.


And don't look at residuals. Look at the plot of the velocity field from time to time.

It turns out that I had the back of the aircraft not closed... after re inspecting the mesh, explains why I kept getting the message open faces or something like that. I will crank the solution to around 1500+