WW2 BAD-T 2 - BDA AI Dogfight Tournament [SEMIFINALS]

[DoctorDavinci]

28 minutes ago, Bob_Saget54 said:

Wow, the senior Aerospace Engineer going up against the high school junior who wants to become an Aerospace Engineer. Talk about coincidence!

I don't really have high hopes for this version, but the next iteration of my plane works a lot better.

Also, @DoctorDavinci, care to share some experience when you get the chance?

Lol ... I'm not an Aerospace Engineer and not a senior (although I'm getting there) ... I've just worked for companies that built things for the aerospace industry as well as military hardware and avionics.

One thing I would suggest, in regards to your education, is to not specialize in any one discipline of engineering and to gather a broader understanding of as many aspects of what interests you. Specialization, if done too early, breeds in weakness and can create a narrow minded approach to engineering problems. It is best to gain knowledge of many aspects of a certain discipline ... By doing this you will become a more valued employee as being able to do many things well is better than being able to do one thing really good

Your career after finishing school will direct you to where you should specialize, don't make that decision until you have a better understanding ... Making a decision without understanding or having all the pertinent knowledge involved can end up stopping you in your tracks

9 minutes ago, tetryds said:

Hey, do you think that it would be interested if I commented on aspects of the airplane that probably made it lose?

Some things are only noticeable when I pilot them, or check on the curves and derivatives, or even on the aftermatches, which would go unnoticed by the creators, mainly when fighting against yourself or piloting in a certain manner.

There was even a case with an airplane where I only really noticed a big issue with it when checking out how it flies with MAF, and explained a lot what happened on the match.

So, if you want, I can do this after I post a battle, it wouldn't take much time for me and it may help everyone to improve their designs.

I would only comment about the airplane which loses, but in the end there will be only one airplane which will pass so yeah

Yes ... Me thinks that this is a good idea

It will definitely help the defeated combatant improve on their designs for the next BAD-T Tournament ... Constructive criticism is always a good thing as long as the person being criticized understands it isn't to beat them down but actually to help prop them up

Edited April 12, 2016 by DoctorDavinci

[Doke]

19 minutes ago, tetryds said:

Hey, do you think that it would be interested if I commented on aspects of the airplane that probably made it lose?

Some things are only noticeable when I pilot them, or check on the curves and derivatives, or even on the aftermatches, which would go unnoticed by the creators, mainly when fighting against yourself or piloting in a certain manner.

There was even a case with an airplane where I only really noticed a big issue with it when checking out how it flies with MAF, and explained a lot what happened on the match.

So, if you want, I can do this after I post a battle, it wouldn't take much time for me and it may help everyone to improve their designs.

I would only comment about the airplane which loses, but in the end there will be only one airplane which will pass so yeah

I think this is a great idea.

[Bob_Saget54]

3 hours ago, tetryds said:

Hey, do you think that it would be interested if I commented on aspects of the airplane that probably made it lose?

Some things are only noticeable when I pilot them, or check on the curves and derivatives, or even on the aftermatches, which would go unnoticed by the creators, mainly when fighting against yourself or piloting in a certain manner.

There was even a case with an airplane where I only really noticed a big issue with it when checking out how it flies with MAF, and explained a lot what happened on the match.

So, if you want, I can do this after I post a battle, it wouldn't take much time for me and it may help everyone to improve their designs.

I would only comment about the airplane which loses, but in the end there will be only one airplane which will pass so yeah

I think it's a very good idea as well. There's nothing wrong with constructive criticism.

[Van Disaster]

Thirded on broad skills - aside from maths & physics I took literary subjects in my last two years of school, and then started mech eng at uni; didn't get on too well because of health issues, and when I came back I switched to software eng because it was rapidly expanding and suited me a little better. Never regretted the couple of years of mech eng or the literature though, I've used all of it in many seemingly unrelated areas.

I always enjoy comments & discussion on aircraft design, especially when it's focussed.

[gridghost]

11 hours ago, tetryds said:

Hey, do you think that it would be interested if I commented on aspects of the airplane that probably made it lose?

Some things are only noticeable when I pilot them, or check on the curves and derivatives, or even on the aftermatches, which would go unnoticed by the creators, mainly when fighting against yourself or piloting in a certain manner.

There was even a case with an airplane where I only really noticed a big issue with it when checking out how it flies with MAF, and explained a lot what happened on the match.

So, if you want, I can do this after I post a battle, it wouldn't take much time for me and it may help everyone to improve their designs.

I would only comment about the airplane which loses, but in the end there will be only one airplane which will pass so yeah

This is an excellent idea. It would help in making even better planes in the future iterations of BAD-T

[MarvinCZ]

I say "yes, please" on the plane design feedback.

[tetryds]

Another battle coming out today!

Due to using two screens, shadowplay glitched and actually recorded at 1440x900 resolution, this will greatly impact the video quality, but it was not intentional.

I could run the battles again but I would rather not do that, I would be forced to run it over and over again until the outcome is the same, and/or have to edit it later to match the results of the actual battle.

So I will stick to a lower video quality this time, and watch out for this problem on the future.

 

I think that the main problem with giving feedback to the losers is that it's not about pointing out the pros and cons, but greatly highlighting the cons, that can be underwhelming for someone who just lost a battle, even though this is just a game.

[gridghost]

And it's up

[Van Disaster]

3 hours ago, tetryds said:

I think that the main problem with giving feedback to the losers is that it's not about pointing out the pros and cons, but greatly highlighting the cons, that can be underwhelming for someone who just lost a battle, even though this is just a game.

Or it could be just the thing to make them feel better when they know *why*. Also, it *is* about pointing out pros - "This was good, in fact better than the opposition - however this other area needs improvement, and here's how and why". I think people enter these things expecting to lose and hoping to win ( at least people I've asked, anyway ).

Edited April 14, 2016 by Van Disaster

[crashsolo36]

Well Not Bad for a little craft that came over on a boat from Jamaica... oh and i didnt go to school of aviation, engineering or any thing like that.. but i went to YOUTUBE - SCOTT MANLEY High and KSP Realism Overhaul 101..BK-(BuzzKill).

plus CON -  i hate how it wobbles when it acquires a target and the guns sights are slightly off..

[alp3r]

@crashsolo36 The fact that your guns somehow managed to rip off your enemies wing really amazed me. How many of those 7.7s did you put on it? The plane was wobbling all over the place but somehow it managed to keep the guns on target as well.

[crashsolo36]

I have like 6. 2 onsight the other 4 are offsight.. I think that's why it wobbles when it gets a target.. But this is just round 1 let's see wat happens round 2.

[Van Disaster]

The wobble is because the AI pilot turns the sensitivity up when it shoots - currently that's a compromise between shooting straight or snappy maneuverablility you have to make. You could make the craft naturally resistant to pitch change - aerodynamically, bigger fixed horizontal stabs would help, or you could put more weight at the ends of the craft ( which increases moment of inertia - the bigger they are the more resistance to rotation - assuming your CoM isn't at the end of the craft(!)) which would damp it down somewhat, but then it also doesn't want to change direction when you want it to. The other way is balancing aero very tightly and using small control surfaces, but the downside of that is you've got a narrower range of conditions for best maneuverability - not least because your CoL moves around with speed.

Apparently F2 gives you gun convergence tools, I only discovered the other day...

Horizontal tail waggling is easily solved by smaller rudder & less rudder movement, which works fine until you try and build something assymetric.

Spoiler

 

Edited April 14, 2016 by Van Disaster

[tetryds]

The wobble is causes by the noise generared by the quantization of the mearusements on the simulation, which adds noise to the response of the system and decreases the phase margin.

Basically, you are limited on your gain and damping parameters when dealing with digital data, if you increase the deflection your max gain and damping values decrease.

This is a very complex problem, there are many papers on it and I don't expect it to be solved on BDArmory, it's simply too far out of the scope of the mod. I am attempting a few solutions on MAF, it's very hard to get it to wobble already but I want to eliminate it for good. Then I man give Baha directions on how to minimize it.

What you can do now is simply add more natural damping to the system and balance the elevator deflection , so, bigger tailplanes. As a penalty you lose a bit of turning hability, but you gain on energy conservation. Usually, simply decreasing the deflection already helps, but bigger tailplanes allow for more deflection.

Tightly balancing aero works very well, you can tweak it so that your curves remain about constant, and remembering that these airplanes rarely go past mach 0.6, if ever.

Maelstorm, from the first BAD-T, was tweaked like that, unfortunately it died on the final because of the AI

 

Now to the feedback:

The Yellow Moth is a good airplane, in this case it simply lost because it was an inferior design, not because there were a major flaw that killed its performance like on a few other airplanes, it was just suboptimal design choices.

A few changes that would improve its behavior would be a lighter engine, which would also move the CoM back, a standard tailplane instead of all-moving (which shouldn't be allowed, but np), slightly longer wings and maybe a longer fuselage.

The major problem of the design was the engine, it's simply too heavy for such a light craft, and being light it should be focused on turnfighting, so a two-stage supercharger would work better (D-series engines). To put it simple, it's a turnfighter with a B&Zner engine, that is like putting the engine of a race car on a monster truck.

You sould try changing that and rebalancing the airplane, with less weight you gain lift, CoM back, you gain maneuverability with less stress, also gain climb rate and top speed, the design has potential.

Also try to focus on the lowest possible weight, at a point your wings are so light that making them heavier won't make them survive for longer, so go for the lowest possible that can handle the aero stresses.

[DoctorDavinci]

what he said ^^^^^

Although I do have to say that my Bad Company regularly hits Mach 0.7 and I've seen her go as fast as Mach 0.74

Tip for the happy tail conundrum - Set your steer factor and your damping to max and design around that if your unstable in flight and flapping around while firing. You may need to reduce the steer factor and damping a little bit but for the most part if you have a good air frame design your damping should be at 7 or higher and your steer factor shouldn't be below 18 ... You do this by tweaking your control surfaces' deflection angle and their control percentage

Also, don't forget to pay attention to your minimum altitude as it is easy to crash into the ground if you can't pull out of a dive

 

Edited April 14, 2016 by DoctorDavinci

[MarvinCZ]

24 minutes ago, tetryds said:

The major problem of the design was the engine, it's simply too heavy for such a light craft, and being light it should be focused on turnfighting, so a two-stage supercharger would work better (D-series engines). To put it simple, it's a turnfighter with a B&Zner engine, that is like putting the engine of a race car on a monster truck.

I never understood the engine differences beyond weight & simple power. What are the effects of the two-stage supercharger on the aircraft?

I also remembered how I made my craft heavy, I gave it the heaviest engine We may read about that after the next fight.

[Van Disaster]

Point for those who understand them: Steer factor & Steering damping are labels on coefficients in a PD controller - factor is the proportional coeff & damping is the derivative coeff. Currently there's no integral which is what is usually used to handle noise ( it may appear soon ) - but setting up PID for an aircraft is a monumental pain in the rear anyway if it doesn't self adjust ( this one *does* seem to self adjust, but maxes itself when it shoots ). Also there's no facility to adjust how the pilot handles changes in dynamic pressure, but that's less important for WW2 dogfighter speeds - although it does make a difference at the faster end.

If you want to get a feel for aircraft PID controller use, go and play with Pilot Assistant. That has considerably more controllers than BDA's pilot has but it's a useful learning experiment. Aircraft seem to like more derivative coefficient than you'd use on a less dynamic installation which if you think about what it's doing, is not that surprising.

I bashed out a simple spreadsheet for engine choices which included craft mass and wing area - there's two obvious leaders depending on installation requirements. I'm curious if the supercharger stages are actually implemented in the engine curves, they all seem to conk out at somewhere around 7500-8000m. [Edit: they are, it's just not *that* much difference]

Edited April 14, 2016 by Van Disaster

[BahamutoD]

1 hour ago, DoctorDavinci said:

what he said ^^^^^

Although I do have to say that my Bad Company regularly hits Mach 0.7 and I've seen her go as fast as Mach 0.74

Tip for the happy tail conundrum - Set your steer factor and your damping to max and design around that if your unstable in flight and flapping around while firing. You may need to reduce the steer factor and damping a little bit but for the most part if you have a good air frame design your damping should be at 7 or higher and your steer factor shouldn't be below 18 ... You do this by tweaking your control surfaces' deflection angle and their control percentage

Also, don't forget to pay attention to your minimum altitude as it is easy to crash into the ground if you can't pull out of a dive

 

I have actually had better results from keeping steer factor and damping low.  Usually 4 and 2 or 6 and 3.  The planes don't need a whole lot of inputs when just flying around, and since the gains get boosted during aiming, setting them lower will minimize the wobbling during aiming.

[tetryds]

1 hour ago, Van Disaster said:

I'm curious if the supercharger stages are actually implemented in the engine curves, they all seem to conk out at somewhere around 7500-8000m.

There are differences, but I would rather if @ferram4 explained them.

Despite seeming small, they are not trivial, but you are right, weight and static thrust are very important.

@MarvinCZ not really, every engine is useful, you only need to find the best one for your airplane.

So maybe it was the best choice, maybe it wasn't, but was not too bad, or maybe the enemy has a big problem or is just not that good, hard to know, that's why I run the battles and on a best out of three manner 

[ferram4]

For the supercharger settings there are only two curves, one which is a single-stage supercharger that has a crit alt around ~4.5km, and one which is a two-stage supercharger with crit alts around 1.5km and 6km and a crossover near 2km (I think).  The main difference is that the single-stage will reach a lower altitude before its power starts to drop off, while the two-stage can reach a higher altitude before it finally completely runs out of breath, but it pays for it with a lower-altitude loss in power.

And as for the plane designs and AI settings, you people are insane with your high settings.  Seriously, your designs are all so much better than they're performing because they can't keep their guns on target.  Which I suppose gives realistic hit numbers but not for the right reasons.

[Van Disaster]

Tetrydis: Yeah, I was curious if I'd missed something so I reflew pairs of engines - the B37-D is still sputtering to a halt at 8700m, which is a bit low for a high altitude engine; I know the US radials seem to generally use turbos, but there were Merlins with superchargers ( two speed as well as two stage ) working up to ~40k ft so I was hoping for something similar. You can see the performance difference - I stuck a B38 on the other wing & watched the climb - but I guess I've not had any test fights at altitudes where the difference would be noticeable.

I have AI settings of around 5.5/2 or 7/3 at most for these. High settings worked well in BDA 9, but not anymore. I tried Steer factor all the way down to 2-ish, but there was a marked reluctance to turn sharply while still wanting to turn fight, and kD was really awkward to set up.

5.5/2 demo, still wobbles slightly when shooting at speed but I'm not sure you can get rid of all of it and still be as snappy as I like craft to be. Less steer factor would probably be in order if anything, it's rather hard to tune a controller that's only used at max values for a tiny percentage of flight time.

Spoiler

 

Later edit: for comparison this is at 4/1.5 - as you can see there's less overshoot, but there's still long period oscillation which means the controller is underdamped; at this point raising the damping should help, but I suspect it'll result in short-period oscillation instead. Going below 4 ( I tried the same craft at 2.5 ) meant the pilot wasn't using the full range of the controls, so this is near my happy lower limit for this particular craft. I'd spend some time carefully finding the actual lower limit I think; pulling all the ammo out of every craft gives you room to tweak if you use one of the extended time deceleration mods.

Spoiler

https://www.youtube.com/watch?v=cc0L_u-Motw

 

Edited April 14, 2016 by Van Disaster

[SuicidalInsanity]

I've been told there have been a few requests for some of the BAD-T arenas. Since all of the BAD-T arenas have now been seen, I've made them publicly available for all who want them:

 

 

 

 

[DoctorDavinci]

I have the next two craft files for the second tier and will be setting them up for battle shortly ... @Pds314's Yellow Moth vs the Hurles - Bot Richard's

Good luck to both of you

Edited April 15, 2016 by DoctorDavinci

[tetryds]

It's not mine, it's Bot Richard's

I may not be able to run a battle today because I have a test tomorrow, but I will on the weekend.

There is also that I am getting used to the flow of running them, so before I would take a lot of time checking and re-checking everything so that nothing wrong happens, now I already know what to check or not and that saves me a lot of time.

[DoctorDavinci]

17 minutes ago, tetryds said:

It's not mine, it's Bot Richard's

I may not be able to run a battle today because I have a test tomorrow, but I will on the weekend.

There is also that I am getting used to the flow of running them, so before I would take a lot of time checking and re-checking everything so that nothing wrong happens, now I already know what to check or not and that saves me a lot of time.

I see said the blind man