Cm graph different T1 methods

Robert S.Robert S. asked 3 months ago

Hello.
T-1(20m/s) STL Tri-Uniform Thick analysis was conducted.

Depending on the mesh number, there are not many changes in CL and CD, so it can be understood,

However, the Cm change is very different.

Mesh 3,986 / 6,222 / 22,948 as shown in the picture.

At this time, we analyzed Tri-linear Thick for mesh 9,986 and found that the Cm graph is linear, but the Tri-Uniform Thick graph is Curved.

What data is similar to and applicable to the real world?

Robert S.Robert S. replied 3 months ago

I`m sorry. can`t insert imgae files
I wonder, why Cm difference each other.
mesh is weight?

André Deperroistechwinder Staff replied 3 months ago

The evaluation of the pitching moment is tricky in panel methods:
(1) When using linear analyses, e.g. T1, the wake panels remain aligned with the x-axis and are not oriented in the wind direction when the aoa changes.
(2) The intersections of the wake panels with the body panels create parasite numerical interactions which hinder the evaluation of pressure coefficients and therefore of the moments. On the other hand, it does not perturbate the calculation of lift and drag which are evaluated in the far-field plane instead.

The solution is to use T6 polars with the VPW.
T6 polars are non-linear, i.e. the wake is aligned with the wind direction for each aoa which solves issue (1)
The VPW reduces the interactions which corrects problem (2)

If you have a look at the Jibe2 experiment in the online doc (https://flow5.tech/docs/flow5_doc/Validation/Jibe2.html), the “model with the fuselage” exhibits the same “curved” polar for the Cm=f(aoa) curve as you describe in the case of T1 polars, but no longer in the case of the T6/1 + VPW.

Hope this helps,
André

Robert S.Robert S. replied 3 months ago

Thank you for Andre

If then,
Depending on the mesh number, Polar is slightly different.
How many numbers are good data?
Of course, the larger the mesh, the greater PC load,
I wonder how much would be appropriate.

The initial numbers will change a lot polar,
But There’s not much change polar between a lot of numbers…

1000?
10000?

André Deperroistechwinder Staff replied 2 months ago

I’m afraid that I don’t have a good answer. The way I do it:
– use tri-uniform always
– start with small mesh sizes so that the calculations run fast
– set higher panel densities in area of locally high pressure gradients
– once the design gets more mature, progressively increase the mesh size until the results stabilize; usually there is no need to increase the density too much: small mesh sizes are good enough and consistent with the precision that can be expected of a panel method
– finally run lengthy T6+VPW analyses to get reliable Cm values

Robert S.Robert S. replied 2 months ago

Thank you Andre
it is best answer for me.

have good day!