DJA 1946 BDAc AI Dogfight Tournament

[Pds314]

21 hours ago, Me1_base said:

Although Al21 is less maneuverable than expected,  shots are really accurate.

That was really surprising to me. I assumed the Alioth81 planes would both turn like a madlad. Also surprising was my L-19 ripping the wings on first turn so much. The only time I saw the L-19 wing rip in testing was testing at the north pole vs dummies. And I figured turning out of a high speed transonic dive against a plane that is effectively stationary in an atmosphere 40% denser than normal was not representative of real combat conditions.

 

Another thing surprising was that when the L-19 had its wing sheared off on impact, it didn't take the dragonfly with it.

 

But yeah, I do most of my aircraft testing in tropical (or they would be tropical if Kerbin had any axial tilt) latitudes. So how it performs at say 45 degrees north or south is a bit of a "who knows?" Apparently the answer is snapping its wings.

Edited July 20, 2019 by Pds314

[Alioth81]

I think at least 2 of the L-19 losses were structural failures.

Regarding agility. It could turn harder but it bleeds too much energy.

Thanks to the 3×30mm it has very weak engines so it cannot afford too crazy maneuvers. 

I would need two steer limiters. One for low and one for high speed

Also structurally the long wings had problens at high G

The strategy here was to have a stable plane with a lot of fire power and redundant systems.

I am a bit worried that it wastes too much ammo. I tested also against very slow planes only.

The other entry is a bit more agile

Edited July 20, 2019 by Alioth81
Some more background

[Pds314]

7 hours ago, Alioth81 said:

I am a bit worried that it wastes too much ammo. I tested also against very slow planes only.

Hmmmm. Well my L-19 is faster than most single engine planes. Like ~308 m/s top speed. It will definitely never reach mach 1 in level flight (although in a dive it can).

Edited July 20, 2019 by Pds314

[SuicidalInsanity]

Round 1, Match 2, between @Alioth81's AL-18 Swift and @Bob_Saget54's Schwalbe Mk VX:

Spoiler

A note about the end of the 3rd bout: The surviving Swift lost its intake in the opening joust, killing the engine and leaving the plane to spend the whole engagement gliding. Bout was called at the point it lost enough airspeed and could no longer maneuver and went into a terminal descent. Given its wingspan, this took a while.

 

[Me1_base]

53 minutes ago, SuicidalInsanity said:

Round 1, Match 2, between @Alioth81's AL-18 Swift and @Bob_Saget54's Schwalbe Mk VX:

  Reveal hidden contents

A note about the end of the 3rd bout: The surviving Swift lost its intake in the opening joust, killing the engine and leaving the plane to spend the whole engagement gliding. Bout was called at the point it lost enough airspeed and could no longer maneuver and went into a terminal descent. Given its wingspan, this took a while.

 

AL-18 Swift has reached expected turning performance, but the use of all moving wings, and no Droop for damper, cause some jitter. 

[Pds314]

It will be interesting vs the Hinden 7-7-30 because:

1: my plane is adjusted not to break its wings, but that decreases turning performance.

2. Turning performance was already not particularly great. It was reasonably efficient but quite slow at changing direction.

3. The 7-7-30 doesn't look like an aerodynamic brick. So I will definitely have a speed advantage but it might be only 40 or 50 m/s. Which means if it overshoots a Hinden at top speed the Hinden might have several seconds of time to fire close enough to it to kill or take it into a turnfight and slower it down.

Edited July 21, 2019 by Pds314

[Me1_base]

On 7/20/2019 at 8:27 PM, Alioth81 said:

structurally the long wings had problens at high G

What kind of problem?

[Alioth81]

7 hours ago, Me1_base said:

What kind of problem?

As I attached wings to other wings and not the main body the inner wing part gets a lot of stress it seems. As the AL-21 has so much wing area I could not reinforce past a certain strength.

If you make the wings stronger but also heavier it means more force on the inner wing part (and as the outermost wings are fully movable they had to had a certain strength)

It was just pure experimental observation that the wings would fall apart as soon as it would fully pitch and try to roll at high speed.

[Me1_base]

11 hours ago, Alioth81 said:

It was just pure experimental observation that the wings would fall apart as soon as it would fully pitch and try to roll at high speed.

different parts of a wing stall at different AoA. stall at high speed means BOOM 

In KSP structural load are less likely to rip the wings apart than aerodynamic stress.

[SuicidalInsanity]

Round 1, Match 3, between @Hinden's H-7-7-30+ and @Pds314's L-24 Broadhead:

Normally these battles go as expected. Sometimes there's a spectacular crash, or a 'how did they survive that!?' moment of craziness. But that first match? 0:45 is going to be difficult to top.
I may have to come up with some sort of merit badge system for craft that, be it act of engineering or act of Kraken, go above and beyond.

[Pds314]

10 hours ago, SuicidalInsanity said:

Round 1, Match 3, between @Hinden's H-7-7-30+ and @Pds314's L-24 Broadhead:

Normally these battles go as expected. Sometimes there's a spectacular crash, or a 'how did they survive that!?' moment of craziness. But that first match? 0:45 is going to be difficult to top.
I may have to come up with some sort of merit badge system for craft that, be it act of engineering or act of Kraken, go above and beyond.

This kinda went exactly how I hoped it wouldn't lol. The H-7-7-30+ has a combination of decent speed, good low/mid-speed acceleration, superb agility, and good energy retention in a turn, and so when it's following my L-24 it can typically provoke it to turn and catch up with it, even from like 800 meters distance or further, at which point my L-24 has no really good options. If it dodges it dies later. If it doesn't dodge it dies now.

0:42: "Target lock acquired. Activate terminal guidance."

Edited July 23, 2019 by Pds314

[SuicidalInsanity]

Round 1, Match 4, between @Me1_base's Zephyrion-Zero and my IA-34:

 

[Pds314]

7 hours ago, SuicidalInsanity said:

Round 1, Match 4, between @Me1_base's Zephyrion-Zero and my IA-34:

 

I'm giving the fascist dorito the "you tried" award. It's an awesome plane even if it didn't win.

[SuicidalInsanity]

Yeah, that was pretty much the rationale for entering a flying wing vs something more optimized for dogfighting - they're cool from both a aesthetic and engineering sense (anyone who's tried to build one without conveniences like thrust vectoring or reaction wheels knows these things aren't easy to get flying, let alone pull off maneuvers); but I had to make a lot of performance sacrifices to get it working.

[Pds314]

11 hours ago, SuicidalInsanity said:

Yeah, that was pretty much the rationale for entering a flying wing vs something more optimized for dogfighting - they're cool from both a aesthetic and engineering sense (anyone who's tried to build one without conveniences like thrust vectoring or reaction wheels knows these things aren't easy to get flying, let alone pull off maneuvers); but I had to make a lot of performance sacrifices to get it working.

Yeah, plus making the AI control them effectively is another challenge. Even with TV since if it dives below minimum altitude it stops thinking about yaw stability.

Such as in this ancient video I made. The flying wing in question had a nasty issue where large sideslip at high speeds caused aerodisintegration. As such it needed constant active yaw stabilization or it would instantly explode. Dive below altitude floor and it is instadead.

Warning: recorded on a potato with unintentional background music.
 

 

Edited July 28, 2019 by Pds314

[OmegaForce]

Creating a flying wing that the AI could pilot was a win of itself.

I attempted a replica of the Horten X but sideslip was problematic. Tried different AOI settings across the span of the wings and moved the COM around along with many other adjustments. Ended up adding vertical stabilizers to the wing tips and it worked but still I  would like to create a true flying wing .Perhaps something similar to the 229 would be a better starting point.

 

[Alioth81]

@SuicidalInsanity

I have some time this evening and even if it takes me past midnight I will rework my flying wing and then challenge yours  (outside the competition)

I thought a flying wing and BDAc AI within this setting will never work and gave up - you inspire me to pick up where I stopped... it already has the looks but it just doesn't like to fly.

I will share it (however good or bad) it will be

[SuicidalInsanity]

Sideslip was fairly easy to counter via proper placement of split ailerons. The more troublesome issue I encountered was outer edge wingtip stall (I'm sure there's some technical aeronautic term for it) where in a banking turn, the outer wingtip would lose lift and cause the whole craft to roll away from the center and/or go into a flatspin, which took some tinkering to solve. But since it's easy to say something is easy, here's some examples to examine/take apart/reverse-engineer:

IA 34 VIII - Tourney entry, originally started out with a tail as a standard aircraft design before slowly evolving into a Ho-229 knockoff (and, as such, ended up incorporating solutions used on the actual Ho-229 prototypes). Single split aileron pair, surprisingly Yaw stable. Solved sideslip/turn instability with 4 degree wing washout and minor wing-tip leading edge dynamic camber adjustment.

IA-39 IV - Second Gen flying wing using lessons from the IA-34. Also a potential entry, but didn't perform quite as well. Higher wing sweep offers more distance between the CoM and the elevators, but costs some lower speed performance. Dual split aileron pairs for attempted smoother yaw control. This time I was attempting better area-ruling and lower drag, so it uses flat wings without washout and instead experiments with wing fences to prevent spanwise airflow stall. Due to lack of washout, wingtip stall during turns instead countered with major dynamic wingtip camber/AoI adjustment. CoM balancing required the inclusion of a pair of Whitcomb bodies to stick fuel in, which had the bonus of also further reducing area-ruling cross section. One of the previous iterations also added small rudders to the rear of the fuel tanks, but I wasn't able to discern any major change in yaw stability.

AI for both is fairly standard, though I did max out steer damping.

[Alioth81]

Here is my pure flying wing design. The AL-23-Boomerang  a Ho-229 look a like. 20x30mm and according to the challenge rules but optimized for design.

I took @SuicidalInsanity recommendations, some online research and a lot of trial and error (also from past failed tries)

Although it is probably not competitive I am extremely proud of the design and it flies much better manually than any flying wing I designed so far  

If you manage to take off (either hit SAS or pull back the stick once over 40 m/s) it flies much better than I expected.

It took some tuning but now the BD AI can handle it as well.

My general advice for flying wings:

Problem 1: Sideslip and yaw stability (less of a Problem once you get the basics)

Basically once you get some yaw it has tendency to worsen or oscillates around your prograde vector

• CoM as far to the front as possible (this way you can get away with moderate sweep). Normally in regular designs I had CoM very close to the center of lift for best maneuverability but this did not work well (they are still fairly close however).
• Some Sweep is required however but less than expected especially once over a certain speed.
• Split ailerons for brakes rudders as far out and/or back as possible
• Cross sectional area should peak behind CoM

 

Problem 2: Adverse yaw (much bigger problem especially when pitching and rolling simultaneously)

2.a Basically when you roll it yaws against the direction you roll (the wing that raises up in the roll creates more drag ->roll right [clockwise] it will yaw to left)

• Solved by giving the ailerons a bit negative incidence also solved by giving the outer wing (where the ailerons are) some wash out.
• Result is that the ailerons on the downward turning wing (which are deflecting upwards) have more relative deflection against the airflow and thus create more drag to compensate the higher drag on the upward moving wing (where the ailerons deflect downwards and create less drag relative to the airflow).

2.b. Very very bad adverse yaw when you roll and pitch at the same time. This will multiply the problem 2.a. by a lot (I think this is the banking turn @SuicidalInsanity mentioned). Roll right, pitch up --> yaw to left

• Solved by ensuring all roll components also have a pitch part mixed in (elevons)
  • theory is same as in adverse yaw but the additional pitch component makes sure that the side that is rolling downwards (and deflects upwards) deflects even more upwards thanks to the pitch component and creates more drag to compensate the hard adverse yaw (on the other side the downward deflection of the ailerons for the roll is compensated by a upwards deflection due to pitch input to create less drag).
  • you basically have to tune the mix of pitch and yaw in the elevons to get an ideal point where the drag difference in a "banking turn" is compensated (together it should add up to 100%)
  • this took quiet some trial and error and is still not perfect but now manageable for the AI.
  • In fact I believe this is one of the biggest problems why the AI could not handle flying wings as it likes to fully deflect pitch and roll together

3. Once you optimized the plane for those problems:

• Fine-tune the innermost elevons or elevators so that you get a level flight without to much pitch control Input.
• Typically you have to deflect them downwards to create a pitch down moment because the outer ailerons are probably deflected upwards because of wash out or incidence you set to compensate adverse yaw).

edit added a short video of the Boomerang in action

 

 

Edited July 30, 2019 by Alioth81
spelling corrections more background and formatting

[Sidestrafe2462]

GAH! I MISSED It AGAIN

[SuicidalInsanity]

Round 1, Match 5, between @dundun92's IV-5.Mk IVa and @Me1_base's Agravain FCS Unlocked:

 

[SuicidalInsanity]

Round 1, Match 6, between @OmegaForce's A.F.C-12 Cosmic Star Blazer and @Noir's Blizzard MkV:

Spoiler

Alternate video title: Fly closer! I want to hit them with my plane! 
Two rammings, at least two very near misses... Blizzard pilots, 'Melee Combat' is not part of the air-to-air paradigm.

 

[Noir]

Return of the phoenix.

 

Spoiler

5:00 min mark: This is fine...

 

This is not fine!

Spoiler

Obligatory meme

In Non-Gif form

 

Edited August 1, 2019 by Noir

[Mr. Peabody]

16 hours ago, SuicidalInsanity said:

Round 1, Match 6, between @OmegaForce's A.F.C-12 Cosmic Star Blazer and @Noir's Blizzard MkV:

I like how the lame duck still took an aircraft down with it in stride.

[Pds314]

17 hours ago, SuicidalInsanity said:

Round 1, Match 6, between @OmegaForce's A.F.C-12 Cosmic Star Blazer and @Noir's Blizzard MkV:

  Hide contents

Alternate video title: Fly closer! I want to hit them with my plane! 
Two rammings, at least two very near misses... Blizzard pilots, 'Melee Combat' is not part of the air-to-air paradigm.

 

Wow. They should have made their planes out of some crazy high part count multi-layered assemblage of wings to try to survive mid air collisions