Re: [Flightgear-devel] YASIM Comment (was New aircraft: SF-25)
Hi Ron, 2011/8/16 Ron Jensen : > > I took a look at the YASim solver today, first comparing the Inertial tensor > with the inertias coming out of aeromatic. Not too far apart, but the 2 yasim > aircraft I looked at were both 50% higher than the aeromatic numbers. > > Then I took a look at the -g switch and was rather shocked at the curves yasim > generates. It generates CL, CD and L/D. My plot software computes force as > (lift^2+drag^2)^0.5. Lift and drag are perpendicular by definition so this > gives the total force in the lift/drag plane. > > I made 3 plots with Tat's A6M2 and Helijah's Katana and Gloster-Meteor models: > http://www.jentronics.com/fgfs/temp/yasim/yasim-A6M2.png > http://www.jentronics.com/fgfs/temp/yasim/yasim-katana.png > http://www.jentronics.com/fgfs/temp/yasim/yasim-glostermeteor.png > > And then I loaded some data I had from a NACA 4 digit 0015 airfoil into the > same plot definition: > http://www.jentronics.com/fgfs/temp/yasim/naca0015.png > > I was pretty shocked to see the YASim charts drag numbers all seem to fall off > at stall. This is counterintuitive to me, I expect drag to increase at stall, > that is what keeps the speed low during the stall. Also, the lift/drag (L/D) > ratio is actually the tangent of the angle the aerodynamic force is acting > in. On the real airfoil it rapidly approaches 0 after the stall because the > force is acting nearly parallel to the free stream velocity and drag > dominates the ratio. This seems not to be the case with the YASim models. That is because you are comparing apples and oranges. What YASim outputs as a drag is actually the total drag including the induced drag. I guess that for the NACA 0015 you used the classical CD curve which is basically the friction drag only. Hence the differences. However YASim uses the well known formula CDi = k * CL^2 to assess the induced drag (line 224 of src/FDM/YASim/Surface.cpp). The problem is that this formula is obtained with Prandtl lifting line theory and is therefore only applicable in the linear domain (small AoAs) while YASim uses it for any AoA including those close to and beyond stall. The drop you see in the drag is the direct consequence of that : induced drag drops after the stall because the lift itself drops. Bertrand. -- Get a FREE DOWNLOAD! and learn more about uberSVN rich system, user administration capabilities and model configuration. Take the hassle out of deploying and managing Subversion and the tools developers use with it. http://p.sf.net/sfu/wandisco-d2d-2 ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASIM Comment (was New aircraft: SF-25)
Hi Adrian, On 08/16/2011 12:46 PM, Adrian Musceac wrote: > Viktor, > During the long hours which have been spent by Emilian and I tweaking the > IAR80 FDM, I have found out that three tag parameters have a very heavy > influence on the solver: [approach aoa="" glide-angle=""] and [cruise glide- > angle=""] > Basically if you get these two right your L/D ratio will come inside a usable > range. After that you could tweak the other wing parameters like stall etc. > I suggest taking a look at the IAR80 for some other ideas. Thanks, tweaking the touchdown AoA was definitely useful. My problem is that I don't have any solid figures for the AoA, so anything I put in there would be a guess. The best clue I could find was that the Falke touches down tail wheel first at around 65-70 km/h. That would require an AoA of about 10-20 degs wouldn't it? But that's not really an approach speed but a good approximation of the stall speed and critical AoA. Would that help the solver? Cheers, Vik -- Get a FREE DOWNLOAD! and learn more about uberSVN rich system, user administration capabilities and model configuration. Take the hassle out of deploying and managing Subversion and the tools developers use with it. http://p.sf.net/sfu/wandisco-d2d-2 ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASIM Comment (was New aircraft: SF-25)
On Tuesday, August 16, 2011 06:23:06 Ron Jensen wrote: > I took a look at the YASim solver today, first comparing the Inertial > tensor with the inertias coming out of aeromatic. Not too far apart, but > the 2 yasim aircraft I looked at were both 50% higher than the aeromatic > numbers. > > Then I took a look at the -g switch and was rather shocked at the curves > yasim generates. It generates CL, CD and L/D. My plot software computes > force as (lift^2+drag^2)^0.5. Lift and drag are perpendicular by > definition so this gives the total force in the lift/drag plane. > > I made 3 plots with Tat's A6M2 and Helijah's Katana and Gloster-Meteor > models: http://www.jentronics.com/fgfs/temp/yasim/yasim-A6M2.png > http://www.jentronics.com/fgfs/temp/yasim/yasim-katana.png > http://www.jentronics.com/fgfs/temp/yasim/yasim-glostermeteor.png > > And then I loaded some data I had from a NACA 4 digit 0015 airfoil into the > same plot definition: > http://www.jentronics.com/fgfs/temp/yasim/naca0015.png > > I was pretty shocked to see the YASim charts drag numbers all seem to fall > off at stall. This is counterintuitive to me, I expect drag to increase at > stall, that is what keeps the speed low during the stall. Also, the > lift/drag (L/D) ratio is actually the tangent of the angle the aerodynamic > force is acting in. On the real airfoil it rapidly approaches 0 after the > stall because the force is acting nearly parallel to the free stream > velocity and drag dominates the ratio. This seems not to be the case with > the YASim models. > > Thanks, > Ron > Hi Ron, Yasim does a very good job at simulating normal flight conditions, if given the right values in the config file. However, once the plane has departed from normal flight, all bets are off, partly due to some quirks in the L/D calculations, partly because simulation limitations. If you compare the usual NACA graphs of CL/aoa to the output of the CL calculation function inside Yasim, you will see why the results look somewhat odd: the Yasim function is based on a bicubic interpolation between CLmax and the lowest after-stall value which is 1. This interpolation draws most of it's parameters from the config xml, notably stall *width* and *peak* among others. Apart from the fact that very little data is available even for established wing profiles inside stall conditions, the config values (stall width etc.) are also used in other ecuations throughout Yasim. You can modify them to obtain a believable behaviour after departure from normal flight, but other stuff will break in normal flight :-) I have made some changes to the L/D calculations in my own Yasim branch, however they are not yet usable. I will attach a KmPlot file which contains the ecuations and graph of delta CL vs delta aoa as lifted from the code. The final ecuation, h(x), represents CL/aoa. Viktor, During the long hours which have been spent by Emilian and I tweaking the IAR80 FDM, I have found out that three tag parameters have a very heavy influence on the solver: [approach aoa="" glide-angle=""] and [cruise glide- angle=""] Basically if you get these two right your L/D ratio will come inside a usable range. After that you could tweak the other wing parameters like stall etc. I suggest taking a look at the IAR80 for some other ideas. 1 1 1 -5.288780438988257 8.324450098829855 -7.39529900057323 7.963450289942801 1 1 1 1 1 f(x) = (x−0.21)/0.42 −π π g(x) = f(x)∙f(x)∙(3−2∙f(x)) − π h(x) =((0.5∙1.5)/0.21)∙(1−g(x))+g(x) −π π DejaVu Sans DejaVu Sans DejaVu Sans -- uberSVN's rich system and user administration capabilities and model configuration take the hassle out of deploying and managing Subversion and the tools developers use with it. Learn more about uberSVN and get a free download at: http://p.sf.net/sfu/wandisco-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] YASIM Comment (was New aircraft: SF-25)
On Monday 15 August 2011 11:18:53 Gary Neely wrote: > Looks like a great start. The first thing I would do before anything > else is make sure your CG is positioned reasonably. In your SF-25, the > CG is much too far back; given the forward-swept wings, it looks to be > about a meter behind MAC: > > E:\FlightGear projects\sf25b>yasim sf25b-yasim.xml > Solution results: Iterations: 1292 > Drag Coefficient: 10.955851 >Lift Ratio: 291.677826 >Cruise AoA: 1.469686 >Tail Incidence: 2.793443 > Approach Elevator: -0.014301 >CG: x:-0.900, y:-0.000, z:0.284 > > Inertia tensor : 1831.357, -0.000, 78.171 > [kg*m^2] -0.000, 2075.542, 0.000 > Origo at CG 78.171, 0.000, 3856.738 > > The command-line YASim solver is showing CG at x=-0.9, well behind the > wing's root chord position at x="-0.371". I took a look at the YASim solver today, first comparing the Inertial tensor with the inertias coming out of aeromatic. Not too far apart, but the 2 yasim aircraft I looked at were both 50% higher than the aeromatic numbers. Then I took a look at the -g switch and was rather shocked at the curves yasim generates. It generates CL, CD and L/D. My plot software computes force as (lift^2+drag^2)^0.5. Lift and drag are perpendicular by definition so this gives the total force in the lift/drag plane. I made 3 plots with Tat's A6M2 and Helijah's Katana and Gloster-Meteor models: http://www.jentronics.com/fgfs/temp/yasim/yasim-A6M2.png http://www.jentronics.com/fgfs/temp/yasim/yasim-katana.png http://www.jentronics.com/fgfs/temp/yasim/yasim-glostermeteor.png And then I loaded some data I had from a NACA 4 digit 0015 airfoil into the same plot definition: http://www.jentronics.com/fgfs/temp/yasim/naca0015.png I was pretty shocked to see the YASim charts drag numbers all seem to fall off at stall. This is counterintuitive to me, I expect drag to increase at stall, that is what keeps the speed low during the stall. Also, the lift/drag (L/D) ratio is actually the tangent of the angle the aerodynamic force is acting in. On the real airfoil it rapidly approaches 0 after the stall because the force is acting nearly parallel to the free stream velocity and drag dominates the ratio. This seems not to be the case with the YASim models. Thanks, Ron -- uberSVN's rich system and user administration capabilities and model configuration take the hassle out of deploying and managing Subversion and the tools developers use with it. Learn more about uberSVN and get a free download at: http://p.sf.net/sfu/wandisco-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel