mav-jp

i know that all Falcon4 versions use HFFM , that is for sure... the question is did FF5 change the Falcon FM core ? answer is no therefore Falcon4 FM remains shit :)

ho yeah i have already investigated that possibility, i.e creating a CL/CD curve for each surface and compute locally the speed, alpha, beta to get the force from each surface and then integrating the whole thing It works..., but : 1) is extremly difficult to create the data (you need to know the CL/CD of elevator, rudder , etc etc.. which is difficult to have). 2) you dont simulate interaction of the surfaces with the others, so you need aditionnal modeling for interaction. all in all..it works but it's difficult to get it work right ...whereas globalmodeling can be much more accurate (all effects are in already)

reading the NASA implentation is a good start :)

correct , and after 40 years of testing, some people have concluded that for aerodynamic effects, most of the dependances are of the second, third of more orders which allows us to simplify / reduce the problem. Why the hell do you want to re do the work already done ?

you have never seen NS modeling ? you need to add some turbulence equations modeling, (like kolmogorov) plus boundary limits equations blah blah blah ... what about FF5.5 ???

what you are saying ha no meaning to me: *Whether you have an interpolated linear model and you know perfectly what are the number of parameters you want to use and the order of importance of it, in that case, i think the NASA model is quite correct and as explained you dont need 8D data... * whether you have a non linear model and the last known is Navier stokes... I dont deny you could propose a linear 8D model but it can NOT be unknown , you need to know it to measure it - or compute it with non linear model.

the fact that you would chose 8 parameter is already a model ;)... you can not say 'i have no model', even NS is a model ... and by the way, I dont have "My" model, NASA had a model ;)

Anyway , to get back on the original subject. FF5.5 has the same falcon4 FM ? ;)

In that case the 8D assumption is already incorrect, because it comes from a simplified model. you then need a more complex model rather than an interpolated model. but i'm saying complex model is NS and is not usable for real time computation. 8D has no interest because you use dat or order of importance too low. so there is no "in between" like 8D table there is 1) NS 2) Physical interpolation which needs ~ 2 Mb to get accurate modeling, adding data would be useless i dont believe in the 3) simplified NS , which is what X plane does... :)

this is what i call math masturbation and has absolutly no interest for engineering purpose ;) ;)

Well what people tend to forget is that EVERYTHING IS A MODEL, even the Navier Stokes Equations are a model ; the Turbulence equations are a model....so saying you can find the analytic solution of the problem is pure Math mind masturbation. The reality : * Navier Stokes equations + Turbu + boundary limit equations are a model and can not be used at the moment in real time. * For Real time simulation you can use * Simplified real time computation models * Physics model with tabular data (pre calculated by NS or testing) I'm saying when you are able to get sufficient data, tabular model is far far better than simplified real time computation. however, getting those tabular data is very time consuming and can be costly. NOW this is only for aero forces, of course after that you need to use a real EOM solver to compute the trajectory Falcon4 has very poor Physics modeling and tabular data Falcon4 have no EOM solver. End of story cheers

This assumption is very incorrect, saying that the function is unknown is saying that we have absolutly no knowledge of what aerodynamics is. OF course we know what kind of function is interpolating the best the reality (see equation i wrote above) and THANK GOD people have worked during hte 20th Century to understand the order of importance of PArameters !!!! You way of thinking remind me myself while i was studying purely math at University (before studying enginnering ). Think ENGINEERING, think PHYSICS :)

Your assumption of equal dimension is incorrect: As the parameters have diffferent order of importance, the dimensionnal table to decribe their effect is not identical. stop thinking Math or Code, think Physics, you need to know the order of importance. To give you an exemple : if you take Cm (alpha,Beta, deltah, lef, sb,q ,ds) Cm can be expressed (read page 38 in Appendix B): Cm (alpha,Beta, deltah, lef, sb,q ,ds) = Cm(alpha,Beta,deltah) * Nu,deltah(deltah) + DELTACm,Lef(1-deltalef/25) + DELTACm,sb(alpha) *(deltasb/60) + Cbar*q/2V*[ CM,q(alpha) + DELTA Cm q, lef (alpha) *( 1 - deltalef/25)] + DELTACm(alpha) + DELTA Cm,ds(alpha,deltah) * Cm(alpha, Beta,deltah) has 20 alpha Bkpts, 19 beta Bkpts, 5 deltah Bkpts. this means a table of 1900 float values * Cm,lef(alpha, Beta) has 20 alpha Bkpts, 19 beta Bkpts, this means a table of 380 float values * DELTACm,sb(alpha) has 20 alpha Bkpts, this means a table of 20 float values *CM,q(alpha) has 20 alpha Bkpts, this means a table of 20 float values * deltah(deltah) has 5 deltah Bkpts, this means a table of 5 float values * DELTA Cm q, lef (alpha) has 14 alpha Bkpts, this means a table of 14 float values * DELTACm(alpha) has 20 alpha Bkpts, this means a table of 20 float values Finally to accuratly describe Cm , you need here 2359 float values. Size in the RAM : 2359 * 4 = 9436 bytes = 9,436 Kb !! Assumin you need more or less the same size for the 5 other force and moment adimensionnal coefficient (which is wrong because you dont need as much for Cn or Cl for instance)... you would get : 9,436 Kb * 6 = 56,616 Kb. Now , in the extreme case, you need all those coefficients for lets say 20 mach values (reynolds), Overall 1,132 Mb is needed.... NOT A BIG DEAL for the RAM, Believe me, the problem is not the storage in the RAM of those coefficient, the problem is the calculation of those coefficients and to WRITE THEM in a file LOL ;)

Your assumption is incorrect , In m'y previous post i was stating the dependancies but i should add the order of it becuse in order to correctly and accuratly compute a variable dependant of a second order parAmeter you dont need N but most likely N/50 variables. Read the Nasa document i have provided the simulation used in there has been used to develop thé fLCS of the real. As you CAN see there are indeed thousands of variable here but nothing close to 1 Gb :) and Bélieve me this was accurate enough to test and develop the deep stall situations which means a lot ( extrême enveloppe ) cheers

Aditionnaly some parameters have a importance of a second order , for example Cl vs deltar , therefore a simple table like 20 coefficients is only required to handle the coupling... second order parameters dont need as much values as first order :) i can GARANTEE you that our computers today have FAR enough RAM to handle the necessary tabular data to go donw to second order influence which is by far too much already of course, getting those coefficients is quite tricky and time/cost consuming. For the F16, hopefully, Wind data tunnel exist and of course experiment has been done for a reynolds that covers most of the flight regime... for Cl,Cm,Cr (mach 0.6) And HFFM covers other Reynolds for CL/CD because it is retro engineered :) the time when real time aero parameters computation is better than tabular is not here yet, maybe in a dozens of years ?