Differences between methods used in AVL and Flow5

[antoinelmb]

Hi Team,

I used to work with a custom-made tool combining Xfoil and AVL (viscous drag added to avl potential results) . From what I understood, Flow5 provides different types of potential flow solver, namely LLT/NLLT, VLM and 3D panels while AVL only uses the VLM and a flat wake model (where flow 5 also implements VPW).
The methods of resolution incorporated in flow5 then seem to encompass everything developped in AVL, is that correct ? If not, what does AVL do additionally and/or differently ?
Also I'm not sure what is meant by CdAVl in the extra drag option. It is defined in AVL as profile drag but then why adding another term of viscous drag ?

Thanking you in advance for your answers.

[Flying-Geek]

Hi,
I haven't worked much with AVL, but to my knowledge there are no features in AVL for which there isn't at least an equivalent one in flow5.

The equation shows all possible drag components in an analysis. The Cd_ind and Cd_visc are what are normally calculated by flow5 unless you disable the viscous analysis, so the Cd_visc is only the viscous drag coefficient calculated by flow5s viscous analysis model. The "Extra drag" tab then gives you the option of defining constant drag coefficients + reference area (the last term of the equation) and enable the AVL type parabolic drag, which is what the Cd_AVL represents to separate it from the other components.

Cheers,
Stefan

[antoinelmb]

Hi Stefan,
Thanks for your answer.

I understand the usage of optionnaly adding custom drag components by defining the coefficient and reference area.
But I don't understand why one would chose to add both Cd_AVL and Cd_visc as they are both reprenting the same quantity in my understanding (profile/viscous drag).

Best,
Antoine

[Flying-Geek]

Viscous drag is very difficult to predict, so the results of the viscous analysis can be useful if done right, but are often far from the truth. The same applies to the parabolic drag model. There are certain situations where it can be helpful to combine the two to tweak the results in a certain direction. It is also useful to mimic additional parasitic drag effects created by other components or interactions, that are not purely wetted area dependent.

Cheers,
Stefan

[antoinelmb]

@techwinder, as the main developper of the tool, what is your point of view on the differences with AVL ? Do you agree with @Flying-Geek that there are no features in AVL for which there isn't at least an equivalent one in flow5 ?

[techwinder]

Yes Stefan is right, both applications are VLMs which gives similar results. flow5 uses exclusively the direct approach, i.e. define the operating point and calculate the results. AVL can also calculate in the inverse mode: define a target e.g. CL or Cm and trim the parameters to achieve the target.
Other than that AVL builds a fixed mesh on the geometry and changes only the B.C. corresponding to wind direction and surface rotations. flow5 actually rotates the flaps and builds the mesh on the modified geometry. This may lead to different predictions because it implies for instance that AVL ignores the gap effect at the edge of the flaps. Whether this is significant or not I don't know. We had some extensive discussions with Stefan on the matter some time ago but we could not reach a firm conclusion.
Lastly AVL determines the stability modes by diagonalizing the full 8x8 state matrix, whereas flow5 computes independently longitudinal and lateral modes by diagonalizing two 4x4 matrices.
There may be other differences, but then I don't know because I only use the basic features of AVL for benchmark purposes.

[antoinelmb]

Alright, thank you for your detailed answer.