Menu
▾
▴
2 Attachments
#2630 yasim lift solutions' sensitivity affects tailGear on ground
Use yasim's command line option to display minimum-IAS-for-lift vs AoA :
yasim pathToSome/model-yasim.xml --min-speed
The console output table shows big changes in lift value for small AoA increments. The lift values are very sensitive to small variations in many of the elemts in yasim configuration, for example, approach speed, AoA, wing stall elements.
By varying elemnts to get as smooth as possible lift curves in the area of takeoff speed it's possible to eliminate what's perceived as 'the ground loop routine' in yasim and have, eg the Beech18 run the entire length of the runway straight line with hands off all controls.
I suspect the rapid fluctuations in lift are causing the gear steer / drag calculations to make for excessive yawing during takeoff roll.
The effect is easily explored using my real-time yasim explorer, ysimi.py at:
[(https://github.com/bitwisetech/yasiplot)]
Where a patched yasim-test.cpp factors the lift so as to appear on the same scale as IAS.
Diff:
With apologies, I am no expert in YASim, or much of aerodynamics. But I can see the L/D curves look rather ... lumpy. :) Assuming your analysis here is generally accepted, what would the next steps be? Do you have some changes to the solvers in mind? And if so, how would we validate / version those? Or is your aim more to start a discussion about possibe changes?
There's been discussion on the forum, 'taildraggers' : someone with real piloting experience of tailgear craft reports that, particularly YASim, models have become impossible to handle during takeoff.
My main aim has been to verify this and to complete a utility that real-time displays all output that YASim has to offer so anyone can interactively adjust YASim elements and alter the 'VERSION' tag and see what effect results on both Lift/Drag curves and IAS curves.
As little as I know of the theory it's stated that, in order to land, wing incidence during roll must be close to stall angle.
YASIM seems to report approx 20% change in lift for a 1 degree change in AoA during roll and that the min-IAS-for-level-flight vs AoA output is particularly sensitive to any changes in YASim configuration file, that looks to me like a bug worth reporting.
I'm hoping that someone with actual experience with the solver can comment on what one's expectations should be for the IAS curves, I'm working on the utility so that it offers instant comparison of YSim output as the 'VERSION' tags are changed: that seems to me a useful endeavor. Thanks.
This turns out to be an artifact resulting from too coarse a step in the data tables. I've run yasim-test with the increments, used to generate the data table outputs, changed form 1.0 to 0.1; the noise in the 'lift at specified speed' becomes much smaller, so no problem.
Last edit: Huntley Palmer 2021-09-01
Huntley, does this imply a change we should make inn the source code itself? I'm not clear where the data tabnles reside, inside Yasim or part of the aircraft? (I kno winside JSBsim that's definitiely part of the aircraft, I thiught for Yasim it's ony paraemters the aircraft supplies)
The way I understand, the source yasim-test.cpp supplies the text report that's used by the main flightgear to log e.g: iterations, CofG, tensors etc .
The source also services YASim command line options that log data tables for plotting. The data table outputs benefit greatly from better resolution, the present AoA step, at one degree, obliterates the fine detail at stall peak. At one tenth degree resolution the output tables are a huge improvement in deatil for less than 5kBy in size
Locally I'm using the tenth step files to display the plots shown above where the imrovement to noise is clear, I included the changed version in case anyone wants to check the improvement, can make a proper patch file if you want one.
Let me think about this and the other ticket. Thinks, the sensible way to do this seems to add extra command options so I'll I will try that out.
Last edit: Huntley Palmer 2021-09-03