FAR Hypersonic Lift calculation (WaveRiders)

[Kerbinator-Fras]

As far as I know,FAR L/D is very awful at hypersonic speed (even if I use skylon aerodynamic shape which almost satisfied area ruling, it is still only 2 L/D max at mach 9)
I'm now looking for high performance lift-assisted Earth SSTOs
Yes FAR is accurate in subsonic and transonic region, but what will FAR do at hypersonic speeds?
As for waveriders, their hypersonic lift is mainly compression lift, which is because shockwave are attached onthe front line, and shock wave can get sealed and have pressure difference between upper and lower sides. As a result, it gives a good compression lift that supports a high L/D up to 10 at Mach 6 (in the waverider module that its shape optimised for M6). Also, for a M6-rated waverider its L/D is higher-then-lower at AoA>0 (peaks at about AoA 10deg), and much lower at AoA<0 because of damage to aerodynamic flow stability.
As for latitual performance I didn't get any data related datas tho :-(
I also get the information that lift of waverider that can fly at low speed mainly comes from low pressure of eddy stream on both sides of upper surface's 2 concaved surface lines between middle raised line and strake edges. Before 25deg AoA as eddy atream get stronger lower the pressure of upper side, making lift even higher. After 25d AoA the stucture break and result in loss of lift.

Possible solution
Also I noticed a fact that FAR has FARWingModule, So may it be done through a FARwaverider Module that may include a series of constants for a CPU-efficient but accurate solution (3D CFD is way too performance-costly......?)
Note 1: may include meshworks to support that because it is nearly impossible to get waverider shape from stock parts
Note 2:Parameters to be included in the module: for example, Optimal_Mach, voxel pressure data (as curvature of mach) or..........
Note 3: interaction needed

Note: Problem is that whether FAR can actually see the shockwave?
As as mentioned in #8 ,it seems that FAR is just applying forces directly on voxel, and then add it to parts?

[dkavolis]

I just don't know a good way to support waveriders. The issue is their performance greatly depends on having the shock wave fully contained on the underside of the body. Otherwise, high-pressure leaks to the upper surface and greatly reduces lift and increases drag. I don't know a way on how to simulate that, let alone for arbitrary shapes that KSP needs to handle. You could have a surrogate model but it would have to interpolate over Mach, temperature, angle of attack and angle of slip. Add in also interactions with any other attached parts and I just don't think it's feasible.

Note: Problem is that whether FAR can actually see the shockwave?

No

As as mentioned in #8 ,it seems that FAR is just applying forces directly on voxel, and then add it to parts?

No, forces are added to parts. Voxels are used for calculating exposed areas, determining shielding from the airstream and calculating cross-sections.

[Kerbinator-Fras]

ferram4 2021/01/24
The difficulty shows up in handling all the little effects in real-time for that sort of thing
Ideal wave rider at or above its ideal Mach number? Should be easy enough, the calculations for a conical shockwave are a small numeric mess, but the results can be put in a lookup table to make it easy
Ideal wave rider below its ideal Mach number? Now the leakage calcs get nasty. Ideal wave rider in sideslip? Okay, now you've got weird unbalanced leakage to deal with
Real wave rider that doesn't just have a shock at the bow, but a bunch elsewhere? Now you're handling shock-shock interactions and I'm not 100% certain where to even start with that
Lots of that could be tabulated out and stored, but you'd have huge lookup tables to reference for all the Mach numbers, pitch angles, sideslip angles, etc. in flight. Not cheap to precalculate and not cheap in memory usage
Bright side is that it is potentially more well-specified than the fuzzy logic "is this a wing" sort of geometry stuff I was trying to do to update the wing-specific calculations to something less terrible, so some form of a wave glider sim could probably be implemented and debugged faster if it proves to be possible to implement in real time with minimal impact on overhead
Keeping overhead down is a big part of FAR's design requirements. It can't slow down things to do things in slower-than-real-time
in KSP-RO discord, principia channel
https://discord.com/channels/319857228905447436/480397772248580098/802814853135335425

[Kerbinator-Fras]

well, my main concern is towards something like stock multiplanetary SSTO, and RSS SSTO with FAR, and hypersonic L/D is the essense. So FAR's hypersonic L/D should go close to reality, instead of going far lower than reality (even for normally winged aircraft). I think the best thing for FAR is that it exterminate the node effect of magical stock aero (for example, by using something like handreds of side intakes and fairings to build up an oddly shaped (imitating real craft) will encounter tremendous drag in stock aero, but normal drag with FAR.
For stock aero, by using fairing clippiboi magic (fairing clipping for fueltanks can be done in FAR, however true essence of clippiboi's magical stock aero performance is because stock node occlusion to the parts exposed to free airstream.For stock aero, part nodes left free in airstream have higher drag, and drag is nearly ignorable once all free nodes are occupied and the vessel is in supersonic (to simulate supersonic airflow?), and . By using this tactic, you may build traditional winged stock SSTOs up to ~4.5 which is already enough for highly efficient RAPIER-Nerv SSTOs that is able to pack 8000dv in LKO, or orbit Earth with stock RAPIER-Nerv SSTO. (heatshield/flag magic can even yield >100ish L/D because of either compensation lift utilization or broken drag cubes)
So a thing is that, L/D=4 is the bare minimum for stock/RSS SSTOs to be efficient. For realism data, well-designed winged aircraft(no utilization of compression lift) may reach 3; and waverider at or over optimum Mach is 6~10 (will be better with a small positive AoA).
Main concern should be here, capability is the problem, and accuracy is just a secondary factor.
And to make things easier, it's good that, in supersonic aero, any effect can only travel backwards, so no need to worry about it. Making a waverider mesh and forbid any surface attach on it (except rear node) may be a good choie.

[DAG299]

@dkavolis I apologize if I should've opened a new issue instead of continuing this one, as this one's quite old now, but I have a suggestion that might make it possible to simulate compression lift without being too computationally expensive.

Basically my suggestion is this, instead of making the shock cone a 3D surface, wouldn't it be easier to model it as a 2D surface(such as a triangle) superimposed below the vessel, whose angle with the axis(of the vessel) varies as a function of the Mach number. (To make these calculations even simpler, only do this below the vessel)

Then, to calculate compression lift at a particular Mach number, use the percentage area of the shock triangle that the waverider covers, apply a suitable function, and apply this compression lift effect.

The shock triangle's shape can also be made to vary with Mach number, and as the craft will only superimpose the triangle completely above a specific speed, that speed will have the max compression lift, with diminishing effects below this speed.

As the incidence angle perpendicular to the triangle's plane reduces with higher Mach numbers(essentially the shock-cone gets closer or more tightly-wound around the craft), there would also be some velocity at which the shock triangle is closest to the the craft(without intersecting it), and this would yield the absolute maximum amount of compression lift. A corollary to this is that if the vessel surface actually intersects or pokes out of the shock triangle, it would lead to a huge increase in drag(as is true in real life.)

Because triangles are relatively simple shapes, I think it'd also be possible to vary them as the shape generating the compression shock changes(blunt bodies, sharp edged etc.) However, I suppose the effects of this on the actual shock incidence angle will be minuscule.

Like @Kerbinator-CN said, these triangles can also be pre-computed, to further reduce computation time. Simply pre-calculate the triangle shapes before hand for several Mach numbers, then extrapolate using several float curves and use these values in-game.

I actually don't know a lot about coding, but if you want, I can probably look up some values of the shock incidence angles, and shock-cone shapes, that could be used to model these 'shock-triangles'. I think this would actually be a reasonable way of modelling compression lift, without being too hacky(As Ferram intended FAR to be.)