Braced vs cantilever design - engineers needed

[J2_Trupobaw]

That's a lot of hit boxes flying around online, though. And,once career mode comes, lots of hit boxes hogging single machine. The game is dropping FPS when 21 players simply decide to go in one formation already :).

I wold leave probability of hitting the wire itself and used presence of wires as safety net reinforcing the spars. MAYBE add all the wires as third invisible spar located in box between wings, hits calculated the same way existing spars are, and degrading one "wire" per bullet hit. That would solve the problem while using existing code and DM engine.,

Edited September 8, 2021 by J2_Trupobaw

[1PL-Husar-1Esk]

2 hours ago, J2_Trupobaw said:

That's a lot of hit boxes flying around online, though. And,once career mode comes, lots of hit boxes hogging single machine. The game is dropping FPS when 21 players simply decide to go in one formation already :).

Not really,  21 vectors to track by modern CPU is easy,  net code transfering that information is what cose some hiccups and even a disco. 21 AI planes would be something totally different on older CPU. 21 object in the sky with relative low polygons and active lod is easy for GPU to render.  Those collision detection are easy for CPU and even more to GPU to calculate,  matrix math. 

[ZachariasX]

9 hours ago, unreasonable said:

The only reason I bothered to post is because some people here appear to believe that hits to the flying wires are almost impossible,

 

TL;DR: It takes more than one shot on average to sever a cable. There are many cables. Flying and landing wires are more like steel rods, not cables and take considerable more punishment that "cables". They are also absurdly overdimensioned in comparison to their actual load, because they have to qualify for more than just pull load.

 

 

Maybe we should add that both litz wire and solid steel wire are rather resistant to bullets. They are both elastic, especially the litz wire that can even cope with partial cuts. (Hello "control rod" failures, I am talking to you!) Also, flying wires are sometimes profiled steel rods, see here:

 

 

Hence, one bullet hit that merely touches the wire simply doesn't equal always a cut wire. Cables rarely snap in other places but attachment points. If they snap by themselves then it is by improper rigging or material failure.

 

Anectote:

Spoiler

That happened for instance to Ernst Udet when suddely in flight when a flying wire of his Aviatik B somehow just broke. The aircraft went out of control due to the sudden wing twist. The wings stayed on though and he spiraled downwards. As he had no way to override the wingtwist with the ailerons, he added and reduced throttle to push his circles toward his lines. He could see the upper wing loose and the flying cable fluttering in the wind. The clevis connecting the wire broke. Then his navigator (and superior) Justinus also realized that they would never make it back to their lines and stood up in the cockpit, walked out on the wing. he sat down near the center strut of the wing holding on to that, feet dangling in front to the leading edge. Unfortunately, that didn't help enough and Udet made him return to his seat. Either way, Udet knew he couldn't hold the wheel to keep them on course. having reached the cockpit again, Justinus shattered the bulkhead between him and Udet and helped twisting the wheel. Still using selective throttle usage to keep the on course, they managed to cross back over their lines. After some initial hooray, they found a blacksmith that made them a new clevis to attach the wire. While that was done they annonced their return by phoning their aerodrome. But just before they fixed their aircraft themselves some officers buzzed in congratulated them for their feat, then asking for the defective part. "Repaired" was exactly what they didn't want to hear, but the blackmith could produce the broken part again from his trash. Later they heard that just before another Aviatik went down, both dead, probably due to the same failure.

 

There's lots of wires on these crates and snapping one makes the load distribute on all remaining wires. Structural flexiblitly permits that, as long as deformantion doesn't exceed the structures deformation limits.

 

Thus, shooting wires affects primarily the planes trim first. (Flying wires are four of many wires, so statistacally speking in terms of "whole body probabilities", another cable is expected to go before the flying wires.) It doesn't make the wings just go in all cases. You primarily need the front flying wire to go. For all the rest you have certain redundncy, plus the imposed trim, especially under g load would make you reconsider ANY coarse maneuvers. I don't know how many rounds it takes to snap a finger thick steel cable or 2 or 3 millimeter thick steel litz wire. But to snap it with one shot, I figure you'd have to be awfully lucky. Looking at the attachment points, I'd be way more optimistic for terminal damage with a single shot (that has to hit a two inch size target).

 

I'd reduce the DM area for sudden, catastrophic wing failure to 4 x 25 cm2, representing both attachment points of the flying wires. As for the wires, I just don't think flying wires give in that quickly. If I say the whole target area in the "full body probability" is 8 x 2 meters, then we have 25 / 1.6E5 = 0.0002 (to be generous) for a single hit probability.

 

This gives me:

 

#shots               p 1 1.00 10 1.00 100 0.98 1000 0.82 2000 0.67 3000 0.55 3500 0.50

 

some 3'500 rounds to fire to make a wire snap just by hosing the whole aircraft. You gotta be very lucky to sever a flying wire just by hosing the aircraft.

 

If I take the diameter (I know @Holtzauge can do that much better than me) of flying wires as 50 mm2, then with ~1'200 N/mm2 I get 60 kN strenght in pull, some 6 tons*). E.g. on an SE5a, flying wires go up in a ~30° angle, this means that the outer strut can, per flying wire, carry 3 tons, after which it would exceed the pull resistance of the flying wire. 4 * 3 = 12 [tons], the weight they can carryif the plane is rested on the outer struts. 12 / 0.6 = 20 [g]. Thus, that rigging is such that it could support the plane past 20 g.

 

Adding that failure probability to the probability of severing attachment points by chance, then we are still in the area of hosing your entire ammo supply on your mark to make for a snapped cable if all you do is hosing the entire structure instead of landing your shots where it hurt. Btw, the pilot is half a square meter of a target, and would you go for palm sized, low visibility targes instead? Anyway, if my fag packet math is wrong, please correct me.

 

 

Of course, these are all arbitary assumptions, but you can see that every way we turn it, the rigging strenght is always far, far in excess of the structural strenght required. Redundancy in wiring further amplifies that. Failure points are specific and local. And it doesn't matter if you use 2 or 3 mm liz or whatever streght, as long as you still get the desired function of that cable. You use as much material as possible, as it makes the whole spiel less elastic. And this is key.

 

But in a nutshell, it is obvious that wires and flying wires in particular are absurdly overdimensioned in terms of pull force they ever have to deal with. This for the simple and practical reason that they have to be structurally rigid enough for being free of harmonics in the slipstream as well as being free of any elasticity under any reasonable load whatsoever. That is ANY load until the breakup point of the airframe, as that would result in changing the rigging of the wing.

 

I think it is a bad idea to factor in snapped cables in a model that assumes any touch of a bullet with a terminal damage, unless you make an arbitarly, very low probability to suit nothing but your gameplay. Else, you're prone to have what we have now. It is hard shooting steel cables if you are not shooting (shrapnel) blades at them. Shrapnel will cut wire with ease and it is commonly used for that purpose in barrages to cut barb wire. But bullets are highly inefficient to cut wire.

 

----------------------

*) Yesyesyes, tons is not force, but it try to keep things simple. Yes, too simple.

 

[No.23_Triggers]

7 hours ago, 1PL-Husar-1Esk said:

I would left probability to more discrete math and make all those wires hit boxes,  then when hit , roll a  dice for that wire to snap.  Same to spars and other important construction elements, 3 or more close enough hits would break it in significant G load and collapse the wing. Bullets just passing the linen if they not meet wood construction,  objects or wires. Fabric tearing I would make only at high speed dives adding  bullets concentration as factor. The plane shaking as it is now  I would remove entirely,  add engine shaking if severe damaged and lost of lift  if fabric is torn. Do our PC can make all that calculation with small latency?  It should as I don't see environment to demanding,  only ground AI is killing  performance,  but we don't use  much of it in ww1 scenarios any way. No big or any trench warfare sadly. In ww2 there are front lines  where AI is fighting each other , and lots objects are moving, AI planes fighting, those hog performance a lot. 

Not that I really know, but I'd imagine (supported by some old DM convos with AnP) that more hitboxes = more processing power and other complications, which makes things overly complex and not viable (as much as I'd like to see how the DM felt with intricate hitbox placement!) 

For the SPAD as an example, you'd end up with something roughly like this (assuming you'd model spar hitboxes too): 


EDIT 2: ^^^ That's actually not right. There are a lot more wires per wing on the XIII. 



That's 23 hitboxes for the upper wing alone - plus 4 for the wires, plus, say, 20 for the lower wings? Then you have to do the rest of the plane, etc, etc - and that's all before you add the "modifier" for if a hit breaks a wire / spar / etc or not.  I can imagine that it's all just a little too complicated to be viable. 

EDIT: Now imagine how many hitboxes would be in the centre of a 30-plane furball!


Agree with the crazy shaking when you get shot in the tail. I really don't get that. 

 

Edited September 8, 2021 by US93_Larner

[AndyJWest]

There are hidden wires in a conventional wire-braced biplane - the wings have diagonal bracing wires internally. Lose one of those and it will have significant effects on rigidity. Likewise, the fuselage has a lot of bracing wires, all of which matter.

[unreasonable]

1 hour ago, ZachariasX said:

 

TL;DR: It takes more than one shot on average to sever a cable. There are many cables.

 

 

I am not going to go through this point by point but just to clarify: I was responding to the proposal made earlier that a broken flying wire would probably make the wing fold up.  So I calculated the probability of hitting the flying wires on a Camel wires compared to a hit to the pilot, ignoring other cables. 

 

Nobody is suggesting that you should aim at the wires, this is pure straw man stuff. The issue is, given the known inaccuracy of most shooting in WW1, how likely are you to hit one of those wires vs hitting the pilot given that you are getting shots into the general area? 

 

I am happy with my area estimates for wires and pilot: the average surface area of a human male is (now) apparently 1.9m^2. It might have been a little lower for a WW1 pilot but let's take that. As a cube, that would be 1.9/6 = 0.32 m^2 Obviously the area depends on the angle of view, but I think 0.5m^2 is the very upper limit if the pilot has no protection. a800394 used 3 sqft as the area of the pilot in the P47 test at an angle at which the pilot had no protection = 0.28m^2 

 

So what happens if a flying wire (or pilot) is hit: does it lead to aircraft loss?  MvR observed wing folding on a few occasions: something was happening (to these braced designs) and you cannot just assume that it was always because the target pilot panicked and pulled 10gs.  If you have a source that shows that bullet hits to flying wires hardly ever break them please share.

 

OT - Bullet hits to barbed wire are inefficient at breaking it, but that is because barbed wire is not under much, if any, tension.  BTW shrapnel is not much better in that case, you need HE, as the British found at the Somme. 

 

edit: we need one of our gun-toting members to suspend a very heavy weight from a beam using a 2-3mm aviation wire and shoot a rifle at the wire: observe and report. I find the idea that a bullet travelling at 800m/s would just push the wire to one side implausible: but I am willing to change my mind if someone has the empirical evidence!

 

Edited September 8, 2021 by unreasonable

[J2_Trupobaw]

1 hour ago, US93_Larner said:

For the SPAD as an example, you'd end up with something roughly like this (assuming you'd model spar hitboxes too): 


EDIT 2: ^^^ That's actually not right. There are a lot more wires per wing on the XIII. 





 

You can remove spar hitboxes by extanding hitbox ahead and behind the spar to include the spar. Spar is hit when both areas that intersect at spar register a hit :).

 

But if I were to speculate I'd rather expect that the wing is one box, and the DM looks at angle of hit, projects the wing on orthogonal surface, checks what percentage of projection is made by projection of spar 1 / spar 2 / intersection of spar projections, then rolls a dice. As long as we keep to principle of bullets hiting whole area evenly and independently (as in unreasonables post) the result is the same.

[ZachariasX]

1 hour ago, unreasonable said:

we need one of our gun-toting members to suspend a very heavy weight from a beam using a 2-3mm aviation wire and shoot a rifle at the wire: observe and report.

That would be interesting, yes.

 

I do see your point. But the flying wires should for the most part really be seen as finger thick solid steel rods, and not 2 or 3 mm thick „wires“.

[SeaW0lf]

Searching about bracing wires in a few books on Kindle, I found accounts that a cut bracing wire could be either fatal or just damage depending on the circumstances. I wish I had a few dozen books on Kindle to make a more complete search, but from what I read, some of them had more than one wire failure in their tours, turning moot the argument that they were just the ones that survived. I’ve seen accounts enough to have an idea that bracing wires dangling in the wind was not that extraordinary or meaning certain death.

 

As usual, and this applies to the damage model as well, I think these failures should lean more towards freak accidents or partial damage than just a critical / deadly hitbox. I think I mentioned here before, but McCudden had one of his elevators control wires cut once but he continue his patrol ‘because he had another elevator functional’, which is surprising. Gould Lee has also similar accounts regarding bad scraps that did not impede him to finish his patrol, including three (3) instances when his plane had bracing wires cut in combat or by flak. Two of them I think he only found out on the ground.

 

Here are two accounts that somehow contradict each other, although in the second he was diving.

    

1 - Once, we got into a fight with an Albatros, the first we’d seen. It attacked us on our side of the lines. We had quite a scrap with it, we were in a BE2 and it wasn’t a very well-known aircraft from the point of view of fighting. There were so many wires on it, I often shot away my own wires in a fight ‘cause they got in the way. Joslyn, Mauriel; Malinovska, Anna. Voices in Flight.

 

2 - I was diving at such a pace that my speed-indicator needle went off the dial, and stayed there (...) Meanwhile, they were taking it in turns to dive and pot at me (...) The wires screamed above the roar of the engine. The whole machine shuddered, as though about to break up. One shot on a bracing wire would have done it. It’s a bloodsome feeling, waiting for a bullet in the back or for your wings to fold up. Lee, Arthur Gould. No Parachute.

[unreasonable]

I have no doubt that bracing wires could be cut without the plane being destroyed: a lot of bracing wires, if I understand correctly, add to the general strength and rigidity of the structure, but the concern here is specifically what happens if you cut or damage a flying wire. These are the ones, typically two on each side, that hold the upper wing down against the lift force that is trying to fold them up.  It sounds like Arthur Lee may have been thinking about flying wires in the second case, although if you are really pushing your machine perhaps breaking other wires could be very dangerous too.

 

Given that the area in which a bullet could hit a flying wire is much larger than seems intuitively obvious, the chances of getting a hit on a wire, compared to on the pilot, is surprisingly high. What we still do not know is how often a hit on a flying wire would break or damage it significantly. My hunch is fairly often, Zacharias' is almost never. But your first example show that bullets could indeed break bracing wires in general.

[ZachariasX]

I think these are good points. I don‘t say that bullets wouldn‘t touch wires. Actually I am very much with @unreasonable about such a probability. My point is rather that the flying wires are the toughest ones by far and the least ready to give in unless you hit them at very specific points. I am counting 44 wires on that SE5a. Adding them and using unreasonables math, that shows a very, very high likelyhood of a wire being touched by a bullet. Far in excess of any anectotes or reasonable prospect of surviving a salvo. I don‘t really think a wire cut four each time a bullet tuoches a wire would have been the prospect of the pilots back then.

Edited September 9, 2021 by ZachariasX

[No.23_Triggers]

5 hours ago, SeaW0lf said:

Searching about bracing wires in a few books on Kindle, I found accounts that a cut bracing wire could be either fatal or just damage depending on the circumstances.

Adding to the anecdotal pilot accounts, Charles J. Biddle had a great quote about a Spad VII diving so hard after getting surprised by five Albatroses that all its flying wires were de-tensioned and the wings warped:



Also happened across this picture of Everett T. Buckley's SPAD after being shot down. Check out the state of the wires (and struts) here! 



EDIT: Another interesting relevant find - this time a letter from a B.E.2 pilot who had the displeasure of running into Oswald Boelcke's Fokker E.IV. A line towards the bottom reads '"We had most of our controls shot through & had to land & crashed very badly". Apparently Boelcke fired 'hundreds' of rounds at the B.E. 



 

Edited September 9, 2021 by US93_Larner

[ZachariasX]

Some more fag packet math. If I made that wrong, please correct me.

 

For this, I just make assumptions, to see how things would play out.

 

I start with a given likelyhood of having a cable (flying wire) severed by a single shot. Then taking the following variables to make up for a monthly operation of a made up squadron:

 

p "flying wire hit with given salvo"			0.0002	0.006
average # missions flown per day			2	2
average # shots received per fight			30	30
% enemy contact			20%	20%
% of losses due to wire failure			10%	10%
# planes per sqadron			16	16

 

The shots received are the number of shots that pass through the "DM Box" between the wings. From this, I derive what happens to those 16 planes over the course of 30 days, assuming they are replaced. The important thing I added is that I took the percentage of planes doomed by flying wire failure as a fixed percentage. Planes get shot down, but not always have that cable failure been the cause. It happened in some cases, and I use that as a statistical constant to get an idea of what that would make total losses.

 

What I get is this:

 

Survivability per mission			0.9988	0.9645
Survivability per month			0.93	0.11
monthly losses due to flying wire hits			1	14
monthly total losses			11	142

 

Thus, if we take a plausible "bullet touches wire" probability as a catastrophic hit probability, then p=0.006 is grotesquely high. In order to get the net monthy loss rate sustainable, almost all aircraft had to be doomed by failing wires.

 

This hit probability is the absolute killer criterion that overrides anything else. You get that wrong, no remotely sane other assumption can mend the issue. If the catastrophic hit probability is beyond 0.0002, you're just dead. A probability of 0.0001 gives me 6 losses per month which is about borderline. p=0.0002 gives me Bloody April, and beyond that you are just dead and nothing will save you except not flying.

 

 

Edited September 9, 2021 by ZachariasX

[Holtzauge]

Seeing the discussion on the probabilities of cutting a wire with machine gun fire I think @ZachariasX made good points about the number of wires in a typical scout and the difficulties of severing a wire and if it was as easy IRL as shooting 100 rounds to sewer a lift wire then with the same reasoning all manner of support and control wires would be popping much as we have in game today for control wires. So on that point I’m leaning more towards @ZachariasX estimates.

 

About the risk for structural failure maybe I was unclear in my first post: IMHO the “insta-fail” wires are the lift wires on the forward spars (one on each side). Why? Well I think the wing will simply diverge: The center of lift will be forward of the flexural center of what now remains of the wings structural system so the wings lift will bend the forward spar which will increase the aoa which will increase the load and turn into a vicious cycle breaking the wing. If a rear lift wire is severed the situation becomes more complex: Now the flexural center is behind the center of lift so no divergence. So now everything is hunky dory right? Well maybe not: This turns the aileron on that side (assuming it’s on the upper wing) into a gigantic Flettner tab so the wing may twist so you get a form of so-called aileron reversal. In the other case when the aileron is on the lower wing the upper wing would most likely whether wane to some extent meaning it would not provide the same lift derivate but you could probably nurse it home to safe landing. The so-called landing wires (those going from the upper part of the cabane strut down to the outer part of the lower wing) could probably be shot off and flap in the wind with no larger consequences unless you fly very fast or intend pulling negative g’s.

 

However, I have no strong opinion on this support wire modeling other than “modeling” the forward lift wire failure would be nice. Modeling the other support wires in the DM is maybe setting the bar to high. My point in the first post was simply that cantilever and braced wings should not be miles apart in resistance to spar material whittled away but both more leaning towards being equal if anything. What I suggested was that as icing on the cake, the braced designs should roll the dice for a catastrophical wing failure which OTOH should be a very rare occurrence indeed and I deeply apologize for using such hyperbole as winning the lottery and a one in a million shot.

 

To make everyone happy, I suggest implementing the hits needed per hitbox to sever lift wires as a slider that can be set between 1-10000 hits per hit box both for QM and MP servers!

[unreasonable]

Just now, ZachariasX said:

 

 

This hit probability is the absolute killer criterion that overrides anything else. You get that wrong, no remotely sane other assumption can mend the issue. If the catastrophic hit probability is beyond 0.0002, you're just dead. A probability of 0.0001 gives me 6 losses per month which is about borderline. p=0.0002 gives me Bloody April, and beyond that you are just dead and nothing will save you except not flying.

 

 

 

What overrides anything else is the hit probability to the pilot. Not only is the pilot the largest single vulnerable area, but we know that he (she/it/they) is highly vulnerable to MG bullets!

 

If you want to calculate overall vulnerability, you have to take into account all of the ways a hit can down a plane: I would suggest something like the method in the US Ballistic Report I mentioned earlier as per my P-47 analysis long, long ago. You need to create a table something like this (numbers are purely illustrative and you would need at least another decimal place at the end. I will estimate something for the Camel later if I feel the urge....).

 

 

 

 

 

You could add a row for the "spar" if you want, or even "hits" to some area that do not actually hit the plane, if you also want to factor in shots fired and accuracy.

 

Even if "bullet touches flying wire" = kill, which was not my initial estimate, the majority of losses would be down to pilot hits, just due to the area comparison, at least in  tractor scouts, pre-incendiaries. (Pushers more engine damage, after mid 1917 more flamers)  My contention is simply that you did in fact get some small but appreciable proportion of wing folding losses and that this is supported by the wire area analysis and the anecdotes and MvR reports. The only way this could not happen is if the flying wires were somehow immune to damage, moving out of the way like a guitar string plucked by a plectrum. (Actually, they break sometimes too...  )   

 

The longevity vs hits is then a function of the "all hits" p of loss and the number of hits taken.  I suspect that while your number of hits taken per sortie includes shots "in the box" it is still massively too high. Even in Bloody April the majority of RFC missions saw no action, and even in those that did the majority of planes would have taken no hits at all, I suspect. The real problem for longevity is that some appreciable proportion of hits to a plane, especially a small tractor scout, will hit the pilot, and this does tend to terminate the sortie. 

 

 

[ZachariasX]

24 minutes ago, unreasonable said:

What overrides anything else is the hit probability to the pilot.

Totally agree. He‘s also in the center of where one aims.

 

My musing was more directed at the general likelyhood of the critical barcing to fail. And that part (as it turned ou) can on only be at risk with a probability around 0.0001. Regardless of how likely it is for the pilot to die. I factored this in in the „total losses“. I even think that 10% is a high guess for fatalities cased directly by the loss of a critical bracing. It is for sure a gross oversimplification, but it gives a ballpark.

 

If you do follow the urge and make a more refined spreadsheet, then I should  be really interested to see the results, as academic of an exercise it might be. 

[unreasonable]

I need much more coffee if I am going to try to make that table more real: but going back to your exercise, you make an incorrect assumption.  "The important thing I added is that I took the percentage of planes doomed by flying wire failure as a fixed percentage".  If the p of hitting/breaking the wires is changed, there is no reason to assume that the p of hitting the pilot (or fuel or engine) should also be changed. Similarly you could armour the pilot, but that would have no effect on hits to the wires.

 

Because the loss rate from wire hits is a small proportion of the total to start with, (which I agree), doubling it or dividing by whatever makes only a small change to the total loss rate. I could not exactly duplicate your numbers, but I hope this table makes the point clearly. If we are uncertain about the effect of hits to areas A, B, C etc, we should start with the most certain, which is the product of the area of the pilot and the p of a hit being kill, then add on from there keeping this constant.

 

 

 

[ZachariasX]

1 hour ago, unreasonable said:

doubling it or dividing by whatever makes only a small change to the total loss rate.

That was my working hypothesis:

 

I look at combat reports in the evening and I distribute the reports of the goners on two stacks. One stack is for the one killed by bracing failure, the other stack for all other causes. The assumption is that all other causes are 9 times the bracing goners. (Variable can be changed.) This means that of 100 planes, there are 10 with shoot bracing, and 90 everyother  lethal failure. All I need then to calculate the bracing goners to find out the expected total goners. As the proportions are (arbitarily) fixed, a multiple of the bracing goners cause the same multiple in all other goners. It affects total number of goners only.

 

I definitely have to look over my little table again, but you are definitely better at statistics than me. So if you find the nerve to go for a complete spreadsheet, I'd be interssted seeing it.

 

 

 

[unreasonable]

I agree that not all shots that contact wire would make them fail - but I would really like to know what is a reasonable proportion.  I could find a number less than 10% plausible, given that the hits - including grazes - must have been in the same order of magnitude as hits to the pilot, but PKs were much more common. 

 

But let me put my objection to your quantification in another way. 

 

If the pilot (my lower estimate) is 0.3m^2 in a Camel "box" of ~11.7m^2 (pixel counting from drawing) then the pilot alone is 2.56% of the box. So the probability of a shot in the box hitting the pilot is 0.0256 quite independently of what happens to wire hits. (Or sortie numbers, or total shots fired etc).

 

Let's optimistically (for the pilot) say only 75% of those hits are terminal. Then the probability of a shot in the box killing the pilot = 0.0192

 

The probability of a shot in the box being a kill from all causes cannot be lower than that unless the pilot is protected in some way. If you think that wire kills are say 10% of the total, ignoring flamers and engine kills completely, their probability would be 0.0192/9 =  0.0021, lower than 0.006 but ten times higher than your low case that you think is consistent with your assumptions about sortie loss rates in general.

 

Add in some flamers and engine kills and you are looking at maybe 0.025- 0.030

 

In reality, your high survivability per mission of 0.9645  - or p to die per mission 0.0355 - is quite close to that for a single shot in the "box". If that is inconsistent with your assumptions about sortie and combat rates, hits taken etc then your assumptions are wrong. 

 

Qualitatively I would say that taking accurate fire (ie in the box in this case) was much more dangerous than people think, but happened much less often. If an enemy scout gets behind you and is close enough and a good enough shot to get hits into your "box" then your chance of survival is worse than evens unless you can escape the fire immediately. This does not fit well with the long-dogfight ethos, unfortunately. 

[ZachariasX]

Actually, the survivability of a mission must be above 0.95. If you are below it, you lose your war and your airforce is made short work of. 5% loss rate is the utmost limit you can go, 2% being very high losses making it near impossible for a pilot to survive his tour.

 

In my very coarse attempt, I tried to simulate the fate of a squadron in relation to the hit probability. It is not very helpful to have a survivabilty of 0.93 as human beings are extremely poor in making sense of actual probabilities. This is why casinos exist in the first place. What I wanted is a "simulator" that shows the direct effect of your hit probabilities on actual loss numbers on the roster. As we do know how such a roster looks like, we can easily see if our guessing chances actually makes sense in a way of reproducing actual losses on said assumed squadron. The difference in survivability of 0.97 and 0.94 makes all the difference in the world, whereas it is irrelevant if you get 0.76 or 0.65, because in both of the latter cases it means equal instant termination of your sqadron, while in the 0.97 and 0.94 may be the difference of winning an losing a war.

 

When one simulates such things, you always have to start out with assumptions that one "knows" and some variables that we want to play with to see how they affect the whole system. My assumption was essenttially a wrong one, as I took the primitive "one hit = death" approach, but coarsly reduced the hit area where I could imagine such a thing to occur. Using the one hit assumption just kept my math small enough to fit on the fag packet. I think your assumption of 10 bullets on a cable per terminal event is a sound one and something that could be worked with.

 

Thus having a spreadsheet where I can define:

 

Respective sizes of hit areas

• total target area
• respective hitbox area
• hits required per hitbox for terminal event
• # of aircraft
• # of sorties flown per aircraft per day/month
• # of shots that are fired upon one aircraft in a sorty with enemy action (N)
• % of sorties with enemy action; no action = 100% survival, action = cumulative survivability after N shots fired

 

Which should produce:

• # aircraft losses per month
• # losses due to hits on respective hitbox

 

This way I could see if mainly aircraft perish due to PK, wing failure, engine failure (or whatever hitboxes one created). I would liketo see if one could reproduce a plausible loss register. If so, then the assumed respective vulnerabilities may be reasonable.

 

Right now in the game, we see that the outcomes of fights are not what we expect them to be. This has many reasons, one is that the game is not the same environment as the real world for pilots back then, different flying style, shooting abilities etc. But in the spreadsheet one should be able to vary these parameters to match or case and see what happens. For instance, if we are great shots, then we reduce the number of hits required per hitbox respectively.

 

[1PL-Husar-1Esk]

During duels vs my squad mates,  I aim at pilot and engine, never the wings but frequent I inflict jamming of control surfaces. I have good efficiency do to very low dispersion.

 

When I fight vs ACE AI , very frequently damage which I take in first place is jammed controls surfaces.  Probability of jamming surfaces is  calculated on whole fuselage so the way you guys are trying to calculate semi real probability of loosing controls is very interesting. 

Edited September 10, 2021 by 1PL-Husar-1Esk

[unreasonable]

1 hour ago, ZachariasX said:

Actually, the survivability of a mission must be above 0.95. If you are below it, you lose your war and your airforce is made short work of. 5% loss rate is the utmost limit you can go, 2% being very high losses making it near impossible for a pilot to survive his tour.

 

 

 

"The overall loss rate for Bomber Command operations was 2.2 percent, but loss rates over Germany were significantly higher... 5%" yeah Google, but easy enough to check in depth. Overall crew losses about 60% of the total enrolled. Not sure of the USAAF numbers - similar for European operations at least until 1945? The truth is out there. So 2% may well be a pretty average overall loss rate for an air force - on the winning side. Nobody said young men were smart. 

 

It is impossible to infer p of a hit to kill from overall loss rates: there are too many unknown and probably unknowable variables, in particular the average hits/sortie. (Ie the last two on your list).   In reverse, is possible to fit a 0.02 p of a hit in the box to kill with a rate of losses per sortie of 2%. It just means that the average number of hits taken in the box per sortie was about one.  In practice a power law applied: the vast majority of sorties taking no hits at all, the unlucky few getting a number. 

 

Even if you could do it, I am unconvinced that it adds to the understanding of the DM mechanics. For instance, it does not translate well to any specific plane: eg 2-seaters usually took quite a few more hits to bring down, as did pushers. So I will not be spending time on this.

 

You cannot recreate historical outcomes, even if your mechanical simulation is absolutely perfect (ours of course is not...), because you will have both un-historic behaviour and mechanical aids that real pilots did not have. I see no way around that. Get a complete novice, disallow fixed gun-sight views, turn turbulence on in MP, ban team speak/chat etc... he is still in a much more favourable position than a real WW1 pilot, but imagine the squealing in the forums if you tried to impose that in the game.... 

 

Really all I wanted to do was investigate how easy it would be to 1) hit and 2) break a flying wire, compared to hitting and killing the pilot: not simulate the whole WW1 air war! 

[ZachariasX]

38 minutes ago, unreasonable said:

It is impossible to infer p of a hit to kill from overall loss rates: there are too many unknown and probably unknowable variables, in particular the average hits/sortie.

We do this all the time. With the DM in our game. The problem is that our in game bottom line does not match the expectations of some. Question is what would you have to change to make a bottom line to your liking? What are the factors?

 

s=1/2*a*t^2 makes a great lot of assumptions as well and only works in the little playpen of our world. Let's be strong an make us a meaningful playpen that helps us grasp at least the dimensons of the involved factors.

 

If we cannot know, that closes a great many doors and we are left with just arbitary speculation.

 

Btw, 5% losses means of a force of 600 aircrtaft, 300 Lancasters gone after 10 missions, requiring 900 aircraft buit every month to sustain the numbers if you were to fly once each night. Three years of war requires ~30'000 built, but they made ~7800 pieces. Average loss rate is usually very, very low if you stay in business. Or, you just don't fly often. That is the alternative. Imagine it's war, but you don't fly. (Story of the F-22)

[unreasonable]

We are going to disagree and leave it at that.  The DM is about damage per hit - not about how many sorties anyone flies or how often they get into fights.  Our playpen is not a simulation of an air war, playing as the high command.  At best it is a pilot's eye view simulation of a specific mission, or in an SP career a series of missions with the boring ones missed out.  If the developers did try to get sortie outcomes on a par with historical averages nobody would play, because of the lack of action.

 

In summary back on the OP's topic:

 

In MvR's reports, in 12 cases out of 80, he used words that indicate that wings were shed during the fight, not as the victim was spinning or diving OOC. (another 5).  That is 15% of his kills, all braced designs.  Assuming all these away as flukes or being due to some other factor, is pure confirmation bias.  

 

But some were 2 seaters, others scouts, he was not always using the same ammo, we do not know how much the target pilot actions contributed. Then there is the random element. Given that, I would find the 10-20% range to be a plausible proportion of wing kills/total kills for braced targets, at least for the pre-incendiary/HE era, although interestingly the wings/kills ratio then was only slightly lower, but the sample is getting small.  The only other piece of direct evidence that we have is the effective area of the flying wires.     

 

Unless anyone has any new data or analysis on the wires, I have no more to add. 

[No.23_Triggers]

27 minutes ago, unreasonable said:

we do not know how much the target pilot actions contributed.

McCudden's recounting of aircraft he shot down might be a decent source to shed a little bit of light on wingoffs vs pilot actions - he tended to include a good deal of detail. Examples: 

"I picked out the Albatros who was East of this formation and, opening fire at 200 yards, released my triggers about 50 yards short of the Albatros, who's left wings at once fell off, and then the whole machine fell to pieces at about 9,000 feet ". 

"I very quickly secured a good firing position, and after firing a good burst from both guns the Rumpler went down in a vertical dive and all its wings fell off at 16,000 feet [from 19,000]"

 

[unreasonable]

Which I would interpret as, in the first case the wing failure was a bullet damage effect, and in the second you might assume that the Rumpler exceeded it's VNE, probably after a PK, maybe with no or minimal bullet effect on the wing. Similar to the MvR examples you have to take an educated guess, it is never completely clear cut.

 

So if those were MvR reports I would have added the first to my 12, but not the second which would be with the other 5 where structural failure was observed but probably not the cause of the crash.  

[No.23_Triggers]

29 minutes ago, unreasonable said:

Which I would interpret as, in the first case the wing failure was a bullet damage effect, and in the second you might assume that the Rumpler exceeded it's VNE, probably after a PK, maybe with no or minimal bullet effect on the wing. Similar to the MvR examples you have to take an educated guess, it is never completely clear cut.

 

So if those were MvR reports I would have added the first to my 12, but not the second which would be with the other 5 where structural failure was observed but probably not the cause of the crash.  

Yeah, I drew the same conclusion - #1 being catastrophic damage of some kind and #2 being an Out-Of-Control overspeed or pilot error (knowing McCudden's gunnery prowess, probably the former). There are a few more mentioned, which all read with a similar level of detail and suggest different outcomes (one interesting mention of McCudden assuming that fire damage cause the wings to shear away) 

[=IRFC=Gascan]

I like the idea of taking test shots at some wire of the appropriate material. I don't think it is reasonable to put the wires under the sort of tension they might be at 4 or 5 g's, but we can see what sort of damage is inflicted on the wires. A bullet impact will either sever it completely or, more likely, leave a dent, crack, or flaw. I'm sure there are plenty of research papers on how wires fail if they have a kink, crack, or dent in them, so we can probably determine how much the wire loses strength by looking at what a bullet does to the wire.

 

A bit off topic, but it is also possible to build a model of, say, a Sopwith Camel's controls. You can then cut the control cable and see how the controls respond depending on where the cable is cut.

[Guest deleted@83466]

I’m not sure what the point of that exercise would be, other than somebodies academic curiosity. Are you saying you want the damage model to not only accurately represent structural consequences of individual snapped wires, but also frayed or weakened wires as well?  That seems to be getting way beyond the scope of a “survey sim” game where you get 10 planes for 70 bucks, doesn’t it?

[J2_Trupobaw]

They are trying to understand how AI D.IIIa can blow wings off the Spad with single bullet from 500m  .

[ST_Catchov]

19 minutes ago, J2_Trupobaw said:

They are trying to understand how AI D.IIIa can blow wings off the Spad with single bullet from 500m  .

 

I think this is the point isn't it? Probabilities and mathematical theorems are one thing but a few bullets causing wing failure too often is another.

[unreasonable]

2 hours ago, ST_Catchov said:

 

I think this is the point isn't it? Probabilities and mathematical theorems are one thing but a few bullets causing wing failure too often is another.

 

How can you decide that a "few" bullets cause anything "too often", unless you have a mathematical model of their probability?  You obviously do, you just have not quantified it.

 

 

[US41_Winslow]

9 minutes ago, unreasonable said:

How can you decide that a "few" bullets cause anything "too often", unless you have a mathematical model of their probability?  You obviously do, you just have not quantified it.

You look at combat reports and loss records and read their memoirs and notice how (1). Airplanes didn’t flop to pieces and (2). See that they regularly returned with more than 30 bullet holes.  In the game itself, you can also see how the non-braced airplanes behave realistically and how the wingshedding issue is only a problem with braced airplanes.  Now to be clear, I’m not saying airplanes shouldn’t fall apart after a very small number of rounds, just it should be a very very rare occurrence.

[unreasonable]

8 hours ago, Miners said:

You look at combat reports and loss records and read their memoirs and notice how (1). Airplanes didn’t flop to pieces and (2). See that they regularly returned with more than 30 bullet holes.  In the game itself, you can also see how the non-braced airplanes behave realistically and how the wingshedding issue is only a problem with braced airplanes.  Now to be clear, I’m not saying airplanes shouldn’t fall apart after a very small number of rounds, just it should be a very very rare occurrence.

 

Quantify "very very rare".

 

1) If you actually read MvR's combat reports, you will find that he is very explicit that ~15% of his victims did indeed flop to pieces while he was engaging them. (All braced).

 

He does not always give rounds fired, but for all of the reports where he does, the average number used was 251 - which includes mostly 2-seaters, many of which were pushers which could absorb a large number of hits to the engine and crash slowly. For his later kills, mostly scouts, his average reported usage was less than 100 shots. We can reasonably assume that a minority of these shots hit anything at all. 

 

2) Of course they did. That is entirely consistent with what I have said. Here is some more fag packet maths.

 

If you look at the rear area of a Camel from a roughly 30 degree offset side and above - a typical shooting angle, you will see that the area of the pilot is about 2.4% of the total area of the plane, ignoring struts and wires. So a shot from that angle hitting a plane at random (the actual plane, not "the box") has a p of hitting the pilot of 0.024

 

Leaving aside other things that could be hit, that means that the pilot has the p not to be hit after 30 hits of (1-0.024)^30 = 0.48

So nearly half the Camels shot at from that angle would not have had a pilot hit after 30 rounds. 

 

Generalising, if the total probability of being downed per hit from all causes was 0.035 (which I suspect is close for a Camel, at that angle, possibly a little higher) the p not to be downed after 30 hits would be =  0.34       Still over a third of the targets, which could, if they succeed in disengaging, RTB. 

(Planes with a larger area compared to the area of vulnerable systems would have better ratio).

 

But in this scenario 1 out of every 29 Camels is downed by the very first hit,  of which some would be due to wing failure after a wire hit. (~15% based on the MvR sample). ie 1 out of  193

Edited September 11, 2021 by unreasonable

[Holtzauge]

 

To begin with, if I take the crosshatched area on the left in the figure below for the Albatros scout and calculate the 50% probability of losing one of the two forward lift wires I get around a 100 rounds needed as well. But who in heavens name shoots a 100 rounds into that rectangular box evenly distributed without hitting anything in the central parts of the scout? How is that even remotely possible? I my opinion you can't use the aircraft's area in any projection and assume an even distribution of bullets into that: Well you can on paper of course, but it has no bearing on reality and fag packet maths seems an appropriate term for it.

 

Here is an alternative calculation method connected to the attached picture below:

 

Assume pilot hit area 0.2 sqrm (0.45x0.45 m)

 

With this and the normal distribution assuming 40% of the rounds hit within small 0.5 std dev circle the pilot has a very slim chance of surviving more than 2-3 rounds.

 

Leaving the high probability of hitting the pilot aside and using the normal distribution of fire depicted under the Albatros D.V scout then we conclude that the area where the lift wires are located will only receive 60% of the bullets fired. Let’s keep that in mind for now.

 

OK, to cut the wire I think at the very least 1 mm of the bullet should be outside and using the sketch in the upper right of the attached figure this means that a bullet will cover about 10.5 mm to be sure to cut it. The lift wires are roughly 3 m long. Calculating the area we get circa 0.063 sqrm.

 

Aiming central in the scout we then have an area with a diameter of circa 9 m where the bullets will be normally distributed and calculating probabilities assuming the wire area above I get roughly 650 bullets needed in that area but remembering that only 60% of the bullets fired hit this area so almost 1100 bullets need to be fire to sewer one of the two forward lift wires with 50% probability.

 

But a scout has 1000 bullets so given this it could happen right? Well not in my opinion: Using the normal distribution the area where the pilot is sitting has now received a wopping 400-450 bullets and the pilot would be a sieve. So statistically speaking its far more likely that something vital in the central parts will be hit long before there is even the remotest chance of severing any lift wire.

 

Granted, I'm no statistician so I may be wrong but I think this would be a more appropriate way to calculate the wire hit probability.

 

 

[J2_Trupobaw]

11 hours ago, Miners said:

You look at combat reports and loss records and read their memoirs and notice how (1). Airplanes didn’t flop to pieces and (2). See that they regularly returned with more than 30 bullet holes.  In the game itself, you can also see how the non-braced airplanes behave realistically and how the wingshedding issue is only a problem with braced airplanes.  Now to be clear, I’m not saying airplanes shouldn’t fall apart after a very small number of rounds, just it should be a very very rare occurrence.

There is survivor bias in combat reports - people whose planes floped to pieces tended not to write them.

[unreasonable]

4 hours ago, Holtzauge said:

 

To begin with, if I take the crosshatched area on the left in the figure below for the Albatros scout and calculate the 50% probability of losing one of the two forward lift wires I get around a 100 rounds needed as well. But who in heavens name shoots a 100 rounds into that rectangular box evenly distributed without hitting anything in the central parts of the scout? How is that even remotely possible? I my opinion you can't use the aircraft's area in any projection and assume an even distribution of bullets into that: Well you can on paper of course, but it has no bearing on reality and fag packet maths seems an appropriate term for it.

 

Here is an alternative calculation method connected to the attached picture below:

 

Assume pilot hit area 0.2 sqrm (0.45x0.45 m)

 

With this and the normal distribution assuming 40% of the rounds hit within small 0.5 std dev circle the pilot has a very slim chance of surviving more than 2-3 rounds.

 

Leaving the high probability of hitting the pilot aside and using the normal distribution of fire depicted under the Albatros D.V scout then we conclude that the area where the lift wires are located will only receive 60% of the bullets fired. Let’s keep that in mind for now.

 

OK, to cut the wire I think at the very least 1 mm of the bullet should be outside and using the sketch in the upper right of the attached figure this means that a bullet will cover about 10.5 mm to be sure to cut it. The lift wires are roughly 3 m long. Calculating the area we get circa 0.063 sqrm.

 

Aiming central in the scout we then have an area with a diameter of circa 9 m where the bullets will be normally distributed and calculating probabilities assuming the wire area above I get roughly 650 bullets needed in that area but remembering that only 60% of the bullets fired hit this area so almost 1100 bullets need to be fire to sewer one of the two forward lift wires with 50% probability.

 

But a scout has 1000 bullets so given this it could happen right? Well not in my opinion: Using the normal distribution the area where the pilot is sitting has now received a wopping 400-450 bullets and the pilot would be a sieve. So statistically speaking its far more likely that something vital in the central parts will be hit long before there is even the remotest chance of severing any lift wire.

 

Granted, I'm no statistician so I may be wrong but I think this would be a more appropriate way to calculate the wire hit probability.

 

 

 

No, it really is not a better way, or I am sure the actual OR scientists would have used it. Why not make the group even smaller: then if the bullets hit the pilot, of course they do not hit the wires -  objective achieved!

Obviously, for any specific attack, the point of aim and group size affects the results: but in air combat, for every case of hits where a small group is centred on the pilot, there is another where the group barely overlaps with the plane at all, or the firer is moving relative to the target and spreading his group over a large area.  It is not so much that the hits in a given group are randomly distributed over the whole target - but that over a series of engagements the groups are distributed randomly enough on the target area that the whole population of hits is. 

 

That is the difference between analysing one specific attack with a known point of aim and group size - which is what you are doing - as opposed to analysing the results of the whole population of attacks.  That is also why the reference area has to include the whole target: we are interested in all of the groups that hit, not just the optimum group for PKs.

 

The inconvenient fact for all of you clinging to the idea that braced aircraft are hardly ever shot down due to wing failure, is that the best documented sample we have strongly suggests that ~15% of them were.  Explain that away.

 

[Holtzauge]

3 hours ago, unreasonable said:

 

No, it really is not a better way, or I am sure the actual OR scientists would have used it. Why not make the group even smaller: then if the bullets hit the pilot, of course they do not hit the wires -  objective achieved!

Obviously, for any specific attack, the point of aim and group size affects the results: but in air combat, for every case of hits where a small group is centred on the pilot, there is another where the group barely overlaps with the plane at all, or the firer is moving relative to the target and spreading his group over a large area.  It is not so much that the hits in a given group are randomly distributed over the whole target - but that over a series of engagements the groups are distributed randomly enough on the target area that the whole population of hits is. 

 

That is the difference between analysing one specific attack with a known point of aim and group size - which is what you are doing - as opposed to analysing the results of the whole population of attacks.  That is also why the reference area has to include the whole target: we are interested in all of the groups that hit, not just the optimum group for PKs.

 

The inconvenient fact for all of you clinging to the idea that braced aircraft are hardly ever shot down due to wing failure, is that the best documented sample we have strongly suggests that ~15% of them were.  Explain that away.

 

 

Where does the 15% of the total number of kills come from? Two questions here: It seems you claim 15% were “wing failures”? First of all who collected the statistics? You or someone else? 15% would equate to circa 12 cases of 80 so first of all what in MvR’s phrasing of the kill made you determine it was a wing failure? Second: IF we indeed determine it WAS a case of 12 “wing failures” how many of these were due to wires being shot off and how many were simply due to overloading of the airframe (much more likely IMHO) due to a panicking pilot? Please do elaborate.

 

As to who is clinging to what we have different opinions: Calculating a hit probability using an even distribution of hits in a 30 degree projection of the pilot area versus the area of the airplane is IMHO a totally paper constructed scenario with no bearing to IRL. What about the area in between the wing tips, nose and the rudder? Are there magically no hits there? To use your own words: Explain that away.

 

In comparison, using a normal distribution with a circular hit probability area centered around the point of aim (say the  pilot) is a much better estimate but if you want to “cling” to an even distribution over an airplane projection discounting hit probabilities in between to make your point then be my guest, just don’t expect me to agree. I guess we will just have to agree to disagree and leave it at that.

 

Edited September 11, 2021 by Holtzauge

[No.23_Starling]

3 hours ago, unreasonable said:

 

No, it really is not a better way, or I am sure the actual OR scientists would have used it. Why not make the group even smaller: then if the bullets hit the pilot, of course they do not hit the wires -  objective achieved!

Obviously, for any specific attack, the point of aim and group size affects the results: but in air combat, for every case of hits where a small group is centred on the pilot, there is another where the group barely overlaps with the plane at all, or the firer is moving relative to the target and spreading his group over a large area.  It is not so much that the hits in a given group are randomly distributed over the whole target - but that over a series of engagements the groups are distributed randomly enough on the target area that the whole population of hits is. 

 

That is the difference between analysing one specific attack with a known point of aim and group size - which is what you are doing - as opposed to analysing the results of the whole population of attacks.  That is also why the reference area has to include the whole target: we are interested in all of the groups that hit, not just the optimum group for PKs.

 

The inconvenient fact for all of you clinging to the idea that braced aircraft are hardly ever shot down due to wing failure, is that the best documented sample we have strongly suggests that ~15% of them were.  Explain that away.

 

That 15% number, again memoirs can be tricky to interpret. MvR often writes claims like “I fired 400 rounds and he then went out of control and fell to pieces”. This could be wing failure from having surfaces and bracing damaged, but could also be a dead pilot falling on his controls and over stressing the airframe in a power dive.

 

McCudden’s memoirs are a bit more helpful as he specifically mentions when he saws of the wing of the odd Rumpler. I think @US93_Larner produced a tally from Mac’s book.

 

You can’t take “I shot him to pieces” or “I shot him then he fell to pieces” to mean the same as a wing being shot off. If you take that logic then we have memoirs of US pilots talking about shooting Dvii’s “to pieces” counting as cantilever wings being shot off. More likely the US pilots riddled them and the pilots were mortally wounded.

 

The MvR quote which still comes to mind is “meat or metal”. Squishy pilots make larger targets than the odd tiny crucial wire.

[unreasonable]

Extracted from MvR reports as per translation in Franks et al "Under the wings of the Red Baron"

 

Rep 23 "After the 400th shot, plane lost a wing while making a curve"

Rep 24 "The wings of the plane I attacked came off"

Rep 28 "After a short fight my opponent lost both wings and fell"

Rep 39 "After a very few shots the plane broke to pieces and fell near Vimy"

Rep 47 "...shot at him from the closest range until his left wing came off."

Rep 49 "The plane I had singled out broke to pieces while curving."

Rep 51 "After a short fight my adversary's plane lost it's wings".

Rep 55 "I managed to break one of the reconnaissance planes with my fire. The fuselage fell with the inmates..."

Rep 56 "We spotted an enemy artillery flyer whose wings broke off in my machine gun fire"

Rep 66 " I shot down the last opponent, a Bristol fighter. He lost his wings, and Lt Gussmann brought him down".

Rep 67 "Under my machine gun fire, both wings broke away from the aeroplane in the air".

Rep 77 "I shot at an enemy plane some 200m away. After I had fired 100 shots. the enemy plane broke apart".

 

That is 12 cases. I am counting 80 as the total, as per Franks' assessment. 15%

 

This excludes a number of others in which he uses words indicating  that the wings came off while the plane was diving OOC. In most of the quotes above it is clear that the wing loss or breakage occurred while he was attacking, not after he had stopped because he knew the enemy was finished.

 

You can quibble about any specific claim - MvR had to interpret what he thought we saw, and we have to interpret what he wrote - but MvR's claims have been analysed more than anyone else and stand up very well. 8 of these 12 explicitly say "wings": that is still 10% 

 

Analysis of his claims by historians show MvR to be a remarkably accurate observer, given the circumstances, with very few claims in serious doubt. In my view it is highly improbable that when MvR said "broke to pieces" he actually meant "did not break to pieces", although I agree that may be an appropriate reaction to a completely unverified claim by an excitable no-name pilot.

 

This historical record, taken as a whole, clearly refutes the "win the lottery theory". Losing wings under fire was a material risk for these planes. 

 

 

 

Edited September 12, 2021 by unreasonable