Which DM do you like more - before 3.008 or after?

[Panthera]

So now at least we know how the Germans arrived at their probability estimation, which was gun camera footage + testing, and thus it can stand on its own. I think I also located the later German tests, and they are essentially identical to the British ones but also describes/demonstrates how the Ha41 explosive filler dramatically improved upon the incendiary effect of the M shells. 

 

In regards to the number of hits that the Germans determined it would take to down a B17/B24 (4), it is worth remembering that this conclusion was largely based on observing aircraft hit in flight, and not solely on static ground targets not subject to the forces associated with flight. This does lend the German analysis some extra value/credibility on the subject over the rest.

Edited December 17, 2018 by Panthera

[sevenless]

20 minutes ago, Panthera said:

So now at least we know how the Germans arrived at their probability estimation, which was gun camera footage + testing, and thus it can stand on its own. I think I also located the later German tests, and they are essentially identical to the British ones but also describes/demonstrates how the Ha41 explosive filler dramatically improved upon the incendiary effect of the M shells. 

 

In regards to the number of hits that the Germans determined it would take to down a B17/B22 (4), it is worth remembering that this conclusion was based on observing aircraft hit in flight, and not solely on static ground targets not subject to the forces associated with flight.

 

I think that settles it then. We only need the game to reflect the destructive power in a sensible way and to normalize the in game weapon effects and we have the best possible sim on the market. They especially need to think about how to simulate the blast effect of the 30mm and the 20mm Mine.

 

As for the other german tests you mentioned, are those: ?

 

 

If not, do you mind uploading them?

Edited December 17, 2018 by sevenless

[6./ZG26_Gielow]

We all have have read those numbers all around the place, nice to see some documentation that originated those numbers in books and websites. 

[Panthera]

23 minutes ago, sevenless said:

 

I think that settles it then. We only need the game to reflect the destructive power in a sensible way and to normalize the in game weapon effects and we have the best possible sim on the market. They especially need to think about how to simulate the blast effect of the 30mm and the 20mm Mine.

 

Yes, the local damage dealt (blast effect) definitely needs to be increased. IMHO however shrapnel damage, at least to powerplant systems, is too severe as I wouldn't expect hits to the outboard parts of a wing to damage the engine.  On other hand pilot vulnerability to shrapnel seems too low, as even hits directly next to the cockpit (e.g. wing root) never seems to hurt the pilot. So it's all abit odd atm.

23 minutes ago, sevenless said:

As for the other german tests you mentioned, are those: ?

 

 

If not, do you mind uploading them?

 

Yeah those should be the ones.

 

 

 

 

 

 

 

 

______________________________________________________________

 

 

 

 

 

 

Oh guys do I have a treat for you! Got some good stuff coming up next!

 

 

 

Edited December 17, 2018 by Panthera

[sevenless]

3 hours ago, Panthera said:

Oh guys do I have a treat for you! Got some good stuff coming up next!

 

It´s x-mas soon isn´t it`?

[Panthera]

Alright guys, here's the early christmas present. Only took me the better part of the afternoon to translate back to proper english, sort out the pictures/films and upload them. ? (Good thing my holidays started yesterday)

 

 

British MK108 trials against heavy bombers

Targets: Halifax wing center panels and cantilever sections with protected fuel tanks and controls

Conditions of tests: The wings were fired upon directly from behind

 

The tests were mainly to test the ability of the shells to set fire to the tanks, so there were only 2 HE shots and 1 HE / T shot fired. 2 of them would've probably been lethal due to aerodynamic damage. 

 

The table with the results of the shelling:

Spoiler

 

The numbering is in accordance with the table.

 

26. (HE / T) Results: "Possibly lethal aerodynamically"

Spoiler

 

27. (HE) Results: "Not lethal"

Spoiler

 

28. (HE) Results: "Probably lethal aerodynamically"

Spoiler

 

Of the 4 incendiary shells fired, 1 caused serious structural damage so that in addition to the good igniting properties (all 4 shots set fire to the tanks), the shells also seem to have a significant high-explosive effect.

 

Spoiler

 

 

Later on, 3 more HE shells were tested against the Halifax's fuselage, the damage of each was assessed seperately:

Spoiler

 

Photos of shelling, numbered & lettered in accordance with description above:

 

Fig A

Spoiler

 

Fig B

Spoiler

 

Fig C & C(I)

Spoiler

 

Conclusions:

 

 

 

Firings vs Stirling fuselage

 

2 HE and 2 HE / T did not cause serious damage.

 

The table with the results of the shelling:

Spoiler

 

Spoiler

 

 

 

ADEN & MK108 trials: Firings vs Lancaster wings

 

Shooting conditions: Attack from 20 ° below ahead

 

2 HE shells were tested. Damage is lethal aerodynamically, but not structurally. Thus, the entire front edge of the wing collapsed, covering the top and bottom surfaces in front of the front side member in a section of ~ 2.5 m . A hole of 25x15 cm formed in the front wall of the side member. Many fragments pierced the tanks. The fuel line, wiring to the engine control circuit and power cables in front of the front side member were either completely or partially broken.

Spoiler

 

The table with the results of the shelling:

Spoiler

Spoiler

 

Conclusions:

HE and HE / T are ineffective at inflicting lethal damage with single hits when firing at the fuselage of four-engine heavy bombers such as Stirling and Halifax, but can produce serious aerodynamic damage to the wings of the Halifax and Lancaster. Judging by the damage inflicted, a lethal attack would result from 2-3 HE (HE / T) hits. Similarly, for lethal structural damage to the wings will need about 3 hits with incendiary shells (not taking into account the fuel tanks).

 

 

Testing the spread of splinters:

Target: box folded from 2 mm (14 swg) mild steel sheets

 

Here is what the device looks like:

Spoiler

 

The shelling of the target showed that the fragments from the incendiary projectiles continued to move in the forward direction, even after punching several sheets in their path. At the same time, the fragments from HE and HE / T scattered mainly after breaking through the first sheet:

Spoiler

 

The penetration & fragmentation pattern through the sheets for different types of shells:

Spoiler

Spoiler

Spoiler

 

Incendiary effect vs heavy bomber fuel tanks

 

For the initial stages of testing, mock-ups of the Lancaster wing were built. The first two types of targets simulated a wing attacked directly from behind.

Target I: lining the upper surface + rear side of the spar. Used Blenheim tanks, filled on 7/8 + made a protection similar to Lancaster tanks

 

Target II: the same + sheet durali between them, representing the structural elements.

 

Target III: simulates a tail attack from below with a pitch angle of 20 °
 

Targets illustration:

Spoiler

 

The following is statistics on fires for three targets, depending on the type of projectile:

 

As you can see, HE and HE / T have a slight incendiary effect due to the fact that they explode on the surface. A small number of fragments reaching the tanks cannot cause serious damage, and fire does not occur. The incendiary shells on the hand functioned very well, so much so that all targets except II were ignited by each of the projectiles. In regard to Targert II, only 2 out of 5 shots caused fires, which is explained by a more complex trajectory for the projectile.

 

Analysis of shells:

 

Target II Inc. shell I (Fire)

Spoiler

 

Film of shooting (Projectile direction → ?

Spoiler

 

 

 

Objective II Inc. Projectile II (No Fire): In this case, the projectile exploded prematurely and carried a sheet of duralumin

Film of shooting (Projectile direction ← )

Spoiler

 

 

 

Objective II Inc. Shell III (Fire)

Spoiler

 

Film of shooting: (Projectile direction ← )

Spoiler

 

 

Objective III Inc. Shell II (Fire):

Spoiler

 

Film of shooting: 

Spoiler

 

Objective III Inc. Shell III (Fire)

Spoiler

 

Film of shooting:

Spoiler

 

Objective III Inc. Shell IV (Fire):

Spoiler

 

Film of shooting:

Spoiler

 

Target III HE Projectile I (Fire)

Spoiler

 

Film of shooting:

Spoiler

 

Target III HE Projectile II (Fire):

Spoiler

 

Film of shooting:

Spoiler

 

Target III HE / T Shell I (Fire):

Spoiler

 

Film of shooting:

Spoiler

 

 

Tests vs Halifax wings

 

Targets: Halifax wings with tanks.

Shelling Conditions: The wings were attacked directly from behind. They shot at the upper surface, sometimes flap or aileron.

As part of the test, tanks that were not attacked directly were filled with water. If they were left empty, they could have exploded, which actually happened after the shelling with the first projectile (after which it was decided to fill the rest).

 

Spoiler

 

Of the 4 shells each one lit up the tanks. One of them even had to go through the entire aileron, the aileron control and the rear side of the spar before breaking through the tank and causing a fire:

Spoiler

 

Examples of shelling: 
 

Projectile I:

Spoiler

 

Film of shooting:

Spoiler

 

Projectile IV:

Spoiler

 

Film of shooting:

Spoiler

 

 

Final Conclusion:

 

HE and HE / T shells are unsuitable for the ignition of fuel tanks due to their premature detonation before reaching the tanks, and the nature of the spread of fragments except for attacks from below with a pitch angle of ~ 20 °. At the same time the Incendiary shells worked perfectly for damaging and setting fire to the protected tanks in the wings of heavy bombers during attacks from the rear hemisphere both directly at the tail level and from the bottom.

 

 

Sources :

• Trials of German 30mm Ammunition, Gordon, HWB and Macdonald, JA, Orfordness reseach station FT343, May 1945
• 30mm Ammunition, Gordon, HWB, Orfordness reseach station FT359, August 1946
• Trials of Aden 30mm, HWB and Smith, AE, Technical Note No. Arm.440 Orfordness reseach station FT377 , July 1950 ( given comparisons with German 30 mm )

 

Edited December 18, 2018 by Panthera

[sevenless]

1 hour ago, Panthera said:

 

Sources :

• Trials of German 30mm Ammunition, Gordon, HWB and Macdonald, JA, Orfordness reseach station FT343, May 1945
• 30mm Ammunition, Gordon, HWB, Orfordness reseach station FT359, August 1946
• Trials of Aden 30mm, HWB and Smith, AE, Technical Note No. Arm.440 Orfordness reseach station FT377 , July 1950 ( given comparisons with German 30 mm )

 

WOW! When the brits test stuff, they test it thoroughly. Thanks for sharing. Very interesting read.

[unreasonable]

22 hours ago, Panthera said:

So now at least we know how the Germans arrived at their probability estimation, which was gun camera footage + testing, and thus it can stand on its own. I think I also located the later German tests, and they are essentially identical to the British ones but also describes/demonstrates how the Ha41 explosive filler dramatically improved upon the incendiary effect of the M shells. 

 

In regards to the number of hits that the Germans determined it would take to down a B17/B24 (4), it is worth remembering that this conclusion was largely based on observing aircraft hit in flight, and not solely on static ground targets not subject to the forces associated with flight. This does lend the German analysis some extra value/credibility on the subject over the rest.

 

A question for the German speakers - is the document clear that the number of hits given - 4 for the 3cm - is an average number required for a B-17 kill, or the maximum required? I read "mittelwerte" as average, but is that right? 

 

If it is an average that suggests that the number required could be from 1-8 or more.  4 suggests a single shot kill probability of about 15%,  (1-0.15)^4 = 0.52, at the top end of the range for the US estimates on the B-25 (twin engine) but not wildly inconsistent.  Some further thoughts:

 

I have to assume that the people doing the testing and assessing the likelihood of loss were sensible enough to realize that planes shot by cannons are normally in flight.  As for camera estimates being more reliable than tests I disagree. The problem is that many - perhaps most - US bombers shot down by fighters had fallen out of formation due to damage, usually from flak.  It would be very difficult to disentangle the earlier effects from that of the gun attack. 

 

What I can well believe is that in a head on attack 4 hits would be about the average required: that is because eyeballing the frontal plan of a B-17 I estimate the area of the nose and cockpit to be about 15% of the total. So if you get 4 hits, about half the time at least one will hit the nose and cockpit with disastrous effects for the pilots. Structural damage would be irrelevant.

 

Well done for finding this document.

 

 

 

 

 

 

 

[sevenless]

29 minutes ago, unreasonable said:

A question for the German speakers - is the document clear that the number of hits given - 4 for the 3cm - is an average number required for a B-17 kill, or the maximum required? I read "mittelwerte" as average, but is that right? 

 

Average is correct. I translate the core part:

 

Für die einzelnen Kaliber ist folgende Anzahl von Treffern erforderlich, um den Abschuss eines viermotorigen Bombers zu erzielen:

Kaliber 2 cm, Minengeschoss, Sprengladung 18g
18 Treffer
Kaliber 3 cm, Minengeschoss, Sprengladung 72 g
4 Treffer
Kaliber 5 cm bzw. 5,5 cm. Minengeschoss, Sprengladung 350g bzw. 420g
1 Treffer

Diese Werte sind als Mittelwerte aus Frontfilmen und Erprobung ermittelt worden

For each caliber, the following number of hits is required to achieve the kill of a four-engine bomber:

Caliber 2 cm, mine shell, explosive charge 18g
18 hits
Caliber 3 cm, mine shell, explosive charge 72 g
4 hits
Caliber 5 cm or 5.5 cm. mine shell, explosive charge 350g or 420g
1 hit

These values have been determined as averages from combat footage and testing

 

29 minutes ago, unreasonable said:

 

If it is an average that suggests that the number required could be from 1-8 or more. 

 

 

I would expect something roughly like this:

 

 

Edited December 18, 2018 by sevenless

[Panthera]

It should be noted that based on the tests posted above the British also arrived at a 3-4 hit average to down a four engined heavy bomber with the Mk108 (3 hits to the wings was considered likely to cause structural failure in flight, so a P-47 surviving 4 wing hits seems quite outlandish). Also of note is that the British tested the effect of impacts from pretty much every conceivable attack angle. By comparison the US tests were clearly less thurough and decidedly lacking in detail. The German estimates I consider as probably the most accurate as it's based on actual recorded results in combat in conjunction with long running ground testing. 

 

In the end the British & German estimates are in very good agreement, with the only possible outlier being the US probability analysis, depending on how you choose to interpret said analysis ofcourse. Thus I believe we can say with pretty good certainty that the ingame damage model for the Mk108 HE(M) shells is in need of adjustment, mostly in terms of local blast damage dealt, as the evidence for this is pretty overwhelming at this point. 

 

1 hour ago, sevenless said:

 

WOW! When the brits test stuff, they test it thoroughly. Thanks for sharing. Very interesting read.

 

Indeed, but they were also about to decide on a new main armament for their own fighters, so they were going into pretty unusual detail with this.

 

Edited December 18, 2018 by Panthera

[Operatsiya_Ivy]

3 hours ago, unreasonable said:

US bombers shot down by fighters had fallen out of formation due to damage, usually from flak.  It would be very difficult to disentangle the earlier effects from that of the gun attack. 

 

I would consider this speculation. It is certainly true that they were easy prey and easily shot down but  from our stand point it is very difficult at best to get an idea on this.

However I fully expect the experts during that time to differentiate flak damage from 30mm hits. Especially in earlier years when fighters still retreated during flak to avoid friendly fire.

Edited December 18, 2018 by Operation_Ivy

[unreasonable]

5 hours ago, Operation_Ivy said:

 

I would consider this speculation. It is certainly true that they were easy prey and easily shot down but  from our stand point it is very difficult at best to get an idea on this.

However I fully expect the experts during that time to differentiate flak damage from 30mm hits. Especially in earlier years when fighters still retreated during flak to avoid friendly fire.

 

It would be easy for the US to tell the difference between 30mm hits and flak damage of planes that managed to RTB, I agree with that. Also easy enough when you test fire. It would not be easy to see what damage caused a crash from wrecks in most circumstances, unless the plane was successfully force landed on it's wheels or belly. It would not be easy to see how a plane was damaged from a few seconds of grainy gun camera film.

 

Gun cameras were not standard issue in the GAF,  one wonders about the sample. Is it in the report? I do not see any detail about the derivation of the numbers of shells to down a plane at all apart from the sentence mentioning the cameras and tests, only a long discussion about the ammunition and firing time required to achieve the desired number; but as I said I may be missing something.

 

8 hours ago, sevenless said:

 

Average is correct. I translate the core part:

<snip>

These values have been determined as averages from combat footage and testing

 

 

I would expect something roughly like this:

 

Thanks for translation.   

 

In terms of the graph, the question is what are you trying to show.  If you are showing number of aircraft shot down after n hits, I very much doubt that it would look like that at all.  This will not be a normal distribution. In fact, the number shot down by 1 hit should be the largest number, since the number of targets available to be shot down by the second shot is lower than that available to be shot down by one. That is if hit effects are independent. If not, there may be a slight hump, but the distribution will still be very flat.

 

See spoiler for how that works in detail, also why it matters to say exactly what "average" means. The mean number of shots required to shoot down a bomber in a sample will not usually be the same as the number at which half the sample is shot down in this kind of distribution.

 

Spoiler

 

 

 

Edited December 19, 2018 by unreasonable

[sevenless]

3 hours ago, unreasonable said:

 

In terms of the graph, the question is what are you trying to show.  If you are showing number of aircraft shot down after n hits, I very much doubt that it would look like that at all.  This will not be a normal distribution.

 

Well, we have to keep in mind that all this is and will be speculation forever, however I don´t see any convincing reason why there wont be a normal distribution with the most planes (mean) shot down with four hits and lesser planes with 3 or 5 hits and so on as shown by the standard normal distribution graph.

[unreasonable]

19 minutes ago, sevenless said:

 

Well, we have to keep in mind that all this is and will be speculation forever, however I don´t see any convincing reason why there wont be a normal distribution with the most planes (mean) shot down with four hits and lesser planes with 3 or 5 hits and so on as shown by the standard normal distribution graph.

 

Then you really need to study some more statistics. We are not selecting a random replaceable sample from a population with non-discrete measurements, like say height for people. There are loads of possible distributions in nature: although I suppose to someone with standard, everything looks like a deviation . 

 

Joking aside, a normal distribution is impossible here: when you compound probabilities you are always going to get a long tail, asymmetric distribution, to which Std Dev analysis is inapplicable. Take rolling a fair dice for example, until you get a 6. On 1/6 of runs you will get it on the first throw. There is a slim possibility that you could roll one hundred times and never get a 6. If you graph number of throws until get a a 6 against number of runs you will get the downwards sloping distribution similar to my first table. Run that with the p to kill of 1/6 and that is exactly what you will get.

 

The plane damage case is more complex since later hits could have compound effects. My second table shows a scenario for this. This still leaves a tail, but it is shorter and the distribution is flatter at the LH end. This is a reasonable model for the data given, but that is all.  If you see something wrong in my workings please point it out.
  

I agree that there is considerable speculation going on here: for instance, we still do not know if the "4" given for the 30mm is the mean: it could be the median, or the mode. That matters. They are not the same thing, but all are "averages".   We do not know how it is rounded, except that it clearly is. We do not know exactly how the figure was generated: so far all we appear to have is a German report that uses that figure for calculation of firing times and ammunition expenditure and says it is from tests and camera analysis, but no detail of how. And yet this is supposed to be conclusive?

 

I have no problem with having this conversation; what I have a problem with is people jumping to conclusions.

 

 

 

 

[sevenless]

2 hours ago, unreasonable said:

 

I agree that there is considerable speculation going on here: for instance, we still do not know if the "4" given for the 30mm is the mean: it could be the median, or the mode. That

 

 

Can´t comment on your other points, but on this one the german report is cristal clear: It is the mean. If in german the word "Mittelwert" is used without further explanation it is the "arithmetisches Mittel". Direct translation from german to english would be "Arithmetic mean". If they would have meant Median, or geometrisches Mittel or quadratisches Mittel they would have written that.

No room left for interpretation. So we either take it as written by them back then or we put out our individual crystal balls and start speculating.

 

Just to explain: In german the words "Durchschnitt" (engl. average), "Mittelwert" (engl. mean) both are synonymous for "Arithmetisches Mittel" (engl. arithmetic mean)

Edited December 19, 2018 by sevenless

[=EXPEND=13SchwarzeHand]

3 hours ago, unreasonable said:

  If you see something wrong in my workings please point it out.

 

Yeah, the multiplier of 1.1 that you use, is a number that you basically pulled out [edited] ( to match a survival rate calculation that is wrong at the basic level because you assume independent events)

 

[edited]

Edited December 19, 2018 by SYN_Haashashin
Respect other members. Last warning

[II./JG77_Manu*]

3 minutes ago, sevenless said:

 

Can´t comment on your other points, but on this one the german report is cristal clear: It is the mean. No room left for interpretation. So we either take it as written by them back then or we put out our individual crystal balls and start speculating.

 

Exactly. I don't get all this crystal ball speculation while there is hard data available.

If we start speculating here, we could also start speculating about many many FMs , that are modeled after "hard data" sheets - those being taken from much-cared prototyps for representative reasons and sometimes more then 10% deviating from actual flight tests of combat proven aircraft (I surely won't say the models to not start another war). 

 

The credo of this game has always been to take any "hard data" as fact and model the game after it, without any further speculation. So this should also be applied to gun damage. Nonsense like 

15 hours ago, unreasonable said:

US bombers shot down by fighters had fallen out of formation due to damage, usually from flak.

should definitely not be taken into account, especially when they are not true like in this case. 

 

 

[unreasonable]

1 hour ago, =EXPEND=SchwarzeDreizehn said:

 

Yeah, the multiplier of 1.1 that you use, is a number that you basically pulled out [edited] ( to match a survival rate calculation that is wrong at the basic level because you assume independent events)

 

[edited]

 

The multipliers I chose show what the distribution would look like if the first single shot kill probability was 0.15 and there was some compounding.  Even if you take a mean of 4 as gospel,  to actually calculate the compound effects of hits, you have to know the number of planes that were hit n times that were killed but also how many survived. We do not know this, so we have to  assume some combination of first shot kill probability and compounding effect to generate a distribution.

 

None of these distributions is remotely close to normal: that is the point I was addressing.  A single shot kill is much more likely than under a normal distribution.

 

 There are an awful lot of ways you can distribute numbers and get a mean of 4.  No one is stopping you proposing an alternative distribution.  Reproduce the analysis and type in whatever numbers you want. 

 

1 hour ago, II./JG77_Manu* said:

 

Exactly. I don't get all this crystal ball speculation while there is hard data available.

If we start speculating here, we could also start speculating about many many FMs , that are modeled after "hard data" sheets - those being taken from much-cared prototyps for representative reasons and sometimes more then 10% deviating from actual flight tests of combat proven aircraft (I surely won't say the models to not start another war). 

 

The credo of this game has always been to take any "hard data" as fact and model the game after it, without any further speculation. So this should also be applied to gun damage. Nonsense like 

should definitely not be taken into account, especially when they are not true like in this case. 

 

 

The problem is that it is not very hard data. We have a single number; a mean apparently (thanks for clarification @sevenless; rounded by +/-  12.5%, for the estimated hits to kill on an aircraft that is not even in the game! We do not know the data from which that was extracted, so we have no idea how reliable the number is, to what kind of attack it relates, or what they counted as a kill.

 

The game has to be modeled on hard data from the ground up, that means breaking down the individual components, not just somehow plugging in an outcome that you like. 

By my tests the HE rounds (not including the mineshells which I have not tested) are actually rather more destructive than the US OR report proposes. People who have tested the 30mm are in some disagreement over it's effects.  Current results are reasonable IMHO even if the graphics and details of damage mechanisms could be more detailed and accurate.

 

 

 

[II./JG77_Manu*]

8 minutes ago, unreasonable said:

The game has to be modeled on hard data from the ground up, that means breaking down the individual components, not just somehow plugging in an outcome that you like. 

By my tests the HE rounds (not including the mineshells which I have not tested) are actually rather more destructive than the US OR report proposes. People who have tested the 30mm are in some disagreement over it's effects.  Current results are reasonable IMHO even if the graphics and details of damage mechanisms could be more detailed and accurate.

 

But isn't this (the part I bolded in your post) exactly the problem with the current DM? Let's assume you are right with your assessment that 20mm and 30mm are about right in the aspect of the number of hits that are needed to down an aircraft. Then we have a big problem - the reason why the aircraft get shot down.

In real life they were shot down by several reasons like structural failure, pilot death or engine failure, all depending on where you hit the enemy aircraft. While in game the aircraft will in most cases die because of engine failure. You can hit the enemy aircraft at the wing several times, but it will still go down because of engine failure. So assuming you are right with the numbers of hits needed, it's exactly like it shouldn't be - plugged in a total outcome (aircraft kill) that is dependent of numbers of hit - but not for the right (physical) reasons. 

Edited December 19, 2018 by II./JG77_Manu*

[unreasonable]

@II./JG77_Manu*  I have much more sympathy with that formulation of the issue: I had also noticed how fragile engines can be to HE shell splinters from the 37mm flak.

 

But that is not the same as plugging in a number as an outcome: just observing that the game's ratio of hits/kills is broadly in line with what we think it should be, even if the causes are perhaps not.  From a pure realism POV I agree that a DM and graphical approach that more closely produced expected results at a component level would be desirable. I have never said that I am satisfied with the current DM, although I do think that 3.08 overall is an improvement. 

 

The US OR report is quite clear that the 3cm shell tends to create kills in a different way to a conventional 37mm,  this has never been in dispute, I would be happy to see this more closely reflected in the game.

 

I get the feeling, however, based on certain people's efforts to dismiss that report, that you could change the composition of the total kills a fair bit, but that would not satisfy some people who mainly want a more lethal hit/kill ratio for large HE shells. 

[=EXPEND=13SchwarzeHand]

37 minutes ago, unreasonable said:

None of these distributions is remotely close to normal: that is the point I was addressing.  A single shot kill is much more likely than under a normal distribution.

 

 There are an awful lot of ways you can distribute numbers and get a mean of 4.  No one is stopping you proposing an alternative distribution.  Reproduce the analysis and type in whatever numbers you want. 

 

Of course the distribution will not be normal. Of course I’m not going to post alternative distributions because all of them will be just as unfounded as yours and offer no additional value.

[216th_Jordan]

If you real 1337 you know only pre 3.008 can keep up with you Skillzz!

 

PS. Please don't take this too seriously, just couldn't let a bad 1337 joke slip !

 

Edited December 19, 2018 by 216th_Jordan

[Panthera]

46 minutes ago, II./JG77_Manu* said:

But isn't this (the part I bolded in your post) exactly the problem with the current DM? Let's assume you are right with your assessment that 20mm and 30mm are about right in the aspect of the number of hits that are needed to down an aircraft. Then we have a big problem - the reason why the aircraft get shot down.

In real life they were shot down by several reasons like structural failure, pilot death or engine failure, all depending on where you hit the enemy aircraft. While in game the aircraft will in most cases die because of engine failure. You can hit the enemy aircraft at the wing several times, but it will still go down because of engine failure. So assuming you are right with the numbers of hits needed, it's exactly like it shouldn't be - plugged in a total outcome (aircraft kill) that is dependent of numbers of hit - but not for the right (physical) reasons. 

 

Yes, the discussion has never centered around wether a weapon is effective or not, but instead about wether the type of damage it inflicts is realistic.

 

As I've said myself many times now the Mk108 is certainly still a very effective weapon ingame, mainly due to the fact that a hit with it almost guaranteed to damage the targeted aircrafts' powerplant system(s).  The big problem is that this not how the MK108 mainly brought down aircraft in real life, which by design was by causing as much structural damage pr. hit as possible to the point where the target simply came apart. 

 

Now with the help of studying the British tests we have a very good idea of what type of damage to expect when hitting certain parts of an airframe with 3cm HE(M) or Inc. shells, and we can definitely say for certain that these results are not reflected ingame.

 

PS: Does the ingame MK108 even have incendiary shells available? If not it's definitely worth advocating for.

 

24 minutes ago, unreasonable said:

@II./JG77_Manu*  I have much more sympathy with that formulation of the issue: I had also noticed how fragile engines can be to HE shell splinters from the 37mm flak.

 

What Manu wrote has been the formulation from the beginning, but you've ignored it at every corner.

 

 

 

Edited December 19, 2018 by Panthera

[unreasonable]

2 minutes ago, Panthera said:

 

That has been the formulation from the beginning.

 

I do not think it has not just been that: or are you now saying that you accepted the single shot kill probabilities for 3cm vs B-25 and P-47 in the US report all along?

 

I got the distinct impression that you thought they were implausibly low.  

[Panthera]

30 minutes ago, unreasonable said:

 

I do not think it has not just been that: or are you now saying that you accepted the single shot kill probabilities for 3cm vs B-25 and P-47 in the US report all along?

 

I got the distinct impression that you thought they were implausibly low.  

 

No, like I've said from the beginning the US analysis makes sense based on what it's trying to figure out, i.e. the probability of a kill from a single randomly placed hit by 3cm HE. But that's as far as it goes, it makes no attempts at figuring out how many hits on average it would take to shoot down either aircraft, no cumulative effects were considered and neither was lethality vs hit placement addressed. Hence it is of little use when trying to assess what type of damage a hit to a specific location is likely to produce, which has been what the discussion was about from the beginning, first starting with the wings of the ingame P-47 taking up to 4 hits to sever which just shouldn't be possible. 

 

 

 

Edited December 19, 2018 by Panthera

[JtD]

5 minutes ago, Panthera said:

... But that's as far as it goes, it makes no attempts at figuring out how many hits on average it would take to shoot down either aircraft, no cumulative effects were considered...

 

Actually it does that.

---

Regarding the basis for the German figures, the US evaluated German evaluations of ~200 gun cam films, which the Germans used for their studies. Due to the mix of weapons installed at all the fighter aircraft investigated and other reasons, it's hard to get a specific number of weapon specific hits per kill from that. However, it can be assumed that the basis for the German evaluation is not smaller than these 200 films. What I also remember, for the decision about bomber killed / not killed, the number of visible hits was less relevant than the rank of pilot making the claim. :) Never ever take anything at face value.

---

What I keep noticing now is the general lack of fires in game, I don't recall having scored a single kill based on fuel tank fire. Not just for the MK108, but for all calibres I've tried.

[unreasonable]

23 minutes ago, Panthera said:

 

No, like I've said from the beginning the US analysis makes sense based on what it's trying to figure out, i.e. the probability of a kill from a single randomly placed hit by 3cm HE. But that's as far as it goes, it makes no attempts at figuring out how many hits on average it would take to shoot down either aircraft (no cumulative effects considered), and neither does it deal with lethality vs hit placement. Hence it is of little use when trying to assess what type of damage a hit to a specific location is likely to produce, which has been what the discussion was about from the beginning, first starting with the wings of certain aircraft. 

 

 

 

 

 

Well again you are just not getting it right about what the report actually says. It very definitely deals with lethality vs hit placement, how on earth do you think they arrived at the probabilities for kills vs structure, pilot, fuel cells or engines? They fired at various parts and looked at what happened: just as in the UK trials but far more often and with more weapons/shells.  That it why it is so useful.

 

As for compound damage: none of the reports really deals with this in any depth, for the simple reason that it would be absurdly difficult to test. Compounding the probabilities independently gives the upper bound of hits required, and for most of the calibers considered will be very close to the true result.   The OR report may not give the compounded damage effect of hits, but neither does the German report with a mean kills to hit: it does not give first hit kill probabilities either!   Arguably you need them both to get a full picture, and you cannot calculate one from the other, you can only find a combination of mean, compound effect and first shot kill p that look reasonable on all variables.  

6 minutes ago, JtD said:

<snip>

---

What I keep noticing now is the general lack of fires in game, I don't recall having scored a single kill based on fuel tank fire. Not just for the MK108, but for all calibres I've tried.

 

On the fires: try the Sopwith Camel: that bursts into flame when you give it a hot look. My tests of P-47 vs 3.7cm flak also involved a few fires. Maybe you need a lot of damage to defeat the self sealing on the WW2 planes?

Edited December 19, 2018 by unreasonable

[sevenless]

7 minutes ago, JtD said:

Regarding the basis for the German figures, the US evaluated German evaluations of ~200 gun cam films, which the Germans used for their studies. Due to the mix of weapons installed at all the fighter aircraft investigated and other reasons, it's hard to get a specific number of weapon specific hits per kill from that. However, it can be assumed that the basis for the German evaluation is not smaller than these 200 films.

 

Any source on that?

[JtD]

4 minutes ago, unreasonable said:

Maybe you need a lot of damage to defeat the self sealing on the WW2 planes?

 

0.50/13mm were deemed sufficient, 0.30/7-8mm were not. A hit with a 20mm or even larger calibre weapon into a fuel tank would give a high probability for fire, provided it's not just a pure AP round. Fires should be (one of) the main reason(s) for losing an aircraft.

[II./JG77_Manu*]

6 minutes ago, JtD said:

 

0.50/13mm were deemed sufficient, 0.30/7-8mm were not. A hit with a 20mm or even larger calibre weapon into a fuel tank would give a high probability for fire, provided it's not just a pure AP round. Fires should be (one of) the main reason(s) for losing an aircraft.

 

Shame that this is not modeled in the game. I have never ever seen a fuel tank fire in this game so far, not before and not after 3.008

[unreasonable]

4 minutes ago, JtD said:

 

0.50/13mm were deemed sufficient, 0.30/7-8mm were not. A hit with a 20mm or even larger calibre weapon into a fuel tank would give a high probability for fire, provided it's not just a pure AP round. Fires should be (one of) the main reason(s) for losing an aircraft.

 

I know, I am just wondering how the DM deals with it.  Fire is certainly in there somewhere. I certainly used to see them occasionally on Migs in the last 3.07 career I flew in F-4 and F-2s. 2 out of the 23 of my current runs of 37mm flak vs P-47 were flamers, quite spectacular ones at that.  

[II./JG77_Manu*]

9 minutes ago, unreasonable said:

 

I know, I am just wondering how the DM deals with it.  Fire is certainly in there somewhere. I certainly used to see them occasionally on Migs in the last 3.07 career I flew in F-4 and F-2s. 2 out of the 23 of my current runs of 37mm flak vs P-47 were flamers, quite spectacular ones at that.  

 

I see engine fires on the Mig occasionally, but I have never seen a fuel tank fire...

[unreasonable]

2 minutes ago, II./JG77_Manu* said:

 

I see engine fires on the Mig occasionally, but I have never seen a fuel tank fire...

 

How do you tell the difference?

[II./JG77_Manu*]

5 minutes ago, unreasonable said:

 

How do you tell the difference?

 

Shouldn't a fuel tank fire surmise from the wings? I didn't check if the Mig-3 has wing tanks tbh. But I know there are aircraft with wing tanks in BoX and I have never ever seen a wing burning

Edited December 19, 2018 by II./JG77_Manu*

[Solmyr]

14 minutes ago, II./JG77_Manu* said:

 

Shouldn't a fuel tank fire surmise from the wings? I didn't check if the Mig-3 has wing tanks tbh. But I know there are aircraft with wing tanks in BoX and I have never ever seen a wing burning

 

As far as I remember, neither here, always front fuselage (engine bay) fires. Maybe one better informed could tell us more on that modelisation ?

 

About the MiG-3, I mostly fly her in my career and sometimes I had to check for minor damages on one of my wings. Here are some fuel tank refill plugs, so I guess yes there are fuel tanks in the wings.

Edited December 19, 2018 by Solmyr

[=EXPEND=13SchwarzeHand]

1 hour ago, JtD said:

Actually it does that

 

It makes an attempt. That is all though. The method of using single shot probabilities and extrapolating multi shot kill rates by the use of a survival rate method is exactly what has been criticized here, and that rightfully so.

The assumption is that the probability of a kill stays the same across „n“ hits.

It does not take into account that the probability is dependent on the previouse state. A second hit should be more likely to produce a kill due to cumulative effects in reality. IMO this simple method has only been included in the test due to the lack of any other measuring method.

That does not mean that cumulative damage is taken into account in any sensible way, but only under the most basic assumptions. This is definitely not the strong point of the study.

The average hits needed to destroy a plane taken from actual observations should be a better indicator than the theoretical values produced by the test.

Edited December 19, 2018 by =EXPEND=SchwarzeDreizehn

[unreasonable]

32 minutes ago, II./JG77_Manu* said:

 

Shouldn't a fuel tank fire surmise from the wings? I didn't check if the Mig-3 has wing tanks tbh. But I know there are aircraft with wing tanks in BoX and I have never ever seen a wing burning

 

Not sure I have come to think of it: but plenty of burning fuselage/engine areas.

 

8 minutes ago, =EXPEND=SchwarzeDreizehn said:

 

It makes an attempt. That is all though. The method of using single shot probabilities and extrapolating multi shot kill rates by the use of a survival rate method is exactly what has been criticized here, and that rightfully so.

The assumption is that the probability of a kill stays the same across „n“ hits.

It does not take into account that the probability is dependent on the previouse state. A second hit should be more likely to produce a kill due to cumulative effects in reality. IMO this simple method has only been included in the test due to the lack of any other measuring method.

That does not mean that cumulative damage is taken into account in any sensible way, but only under the most basic assumptions. This is definitely not the strong point of the study.

The average hits needed to destroy a plane taken from actual observations should be a better indicator than the theoretical values produced by the test.

 

It is also included because a) it establishes upper limit of the number of shots, and b) it is very close estimate for the MGs and even the 20mm cannon: it is only with the larger shells that it is likely to be an issue.  How much of an issue? None of us know for sure.

 

Unfortunately all we have as a mean is an obviously rounded "4" for the B-17, which  was also at least partly estimated from tests, not purely from operational observations, which as JtD points out have problems of their own, on an aircraft that we cannot test in game.  Not much to go on there. 

 

If we had the distribution we would have a better idea: the problem is that there are an infinite number of ways you can combine first shot kill probability and the increase in that probability with further hits, to generate your mean of 4.  

 

We have yet to see any estimates from anyone here of what the initial shot kill probability should be for the planes in the game, and how it increases with later hits. What is stopping you?

 

 

[=EXPEND=13SchwarzeHand]

2 hours ago, unreasonable said:

If we had the distribution we would have a better idea: the problem is that there are an infinite number of ways you can combine first shot kill probability and the increase in that probability with further hits, to generate your mean of 4.  

 

This is the point you don’t seem to understand.

There is an infinite amount of distributions, period.

This distribution does not have to be dependent on the first shot kill

propablity or as your assume a subsequent modification of that first probability within a survival rate calculation.

That is why I am not keen on producing any alternative model to yours, which will be -just like yours- based on unfounded assumptions.

In this case the actual measurement is the best indiacator. Anything else is speculation.

 

As an addendum, I believe firmly that the pictures showing the damage done by a 30mm shell in this case are more suitable evidence for implementing a realistic approach to the MK108 in the game than the rather weakly supported „models“ that have been suggested here.

Edited December 19, 2018 by =EXPEND=SchwarzeDreizehn

[6./ZG26_Custard]

If you take the example of a 262 firing all four cannons at once and hitting more or less the same area with 4 30 mm rounds, I would think damage would be absolutely catastrophic in most cases. Looking at the testing results that have been posted recently, a single cannon firing on a heavy bomber would seem to create several different results based on the averages ( and the tests). I hope at  some point down the line developers will look further into the damage model particularly for HE rounds.

[JtD]

1 hour ago, =EXPEND=SchwarzeDreizehn said:

As an addendum, I believe firmly that the pictures showing the damage done by a 30mm shell in this case are more suitable evidence for implementing a realistic approach to the MK108 in the game than the rather weakly supported „models“ that have been suggested here.

 

I agree, but...

 

In case of the nicely picture documented test results we must not forget that these were effective hits - for instance, all these hits occurred in positions and at angles that allowed the projectile to penetrate and explode inside of the structure. This will not be the case for every hit in combat and that should not be forgotten when shaping expectations. You'll need one effective hit to kill a Spitfire, not just any one hit.