.50 cal damage, or lack there of

October 16, 2020

So after that ball of numbers can we definetly erase from the equation the factor "bad aiming ???

October 16, 2020

Depends on your criterion. 1029 shots to get 28.8 hits per kill, even allowing for a proportion of these being fired after the target was already effectively dead, is not exactly stellar shooting. Clearly there were some who were much better and others much worse.

It is good enough for an average shot to shoot down one or two targets per sortie, depending on your load out. If you think you should be getting multiple kills per sortie - then you would have to shoot much better than that and/or find some targets that do not see you at all.

What is more important though, for understanding the DM issues, is what happens when hits are made. But it is a problem if people are assuming they are hitting many times in cases where they actually are not.

October 16, 2020

27 minutes ago, Tatata_Time said:

So after that ball of numbers can we definetly erase from the equation the factor "bad aiming ???

Right?

?

It is all part of what was called in Master Gunner School as, "The Error Budget!"

October 16, 2020

I'm rarely flying P-51 or P-47 in pure fighter vs. fighter sorties (for obvious reasons *cough* *cough*) but going back through our server stats, I found one sortie where actually I didn't shoot ground targets but only bombed them, so all bullets were against single engine fighters, and I got a total of 5.

1x 109K-4, 1x109F-4, 1x109E-7, 2x Fw-190A-5.

Number of bullets hit the target: 983

Number of PK kills: 4

Number of Plane damage kills: 1 - only the very last 190 went down due to damage, and then it was just two minutes after I let her go.

Not that I'd insist on this sortie being representative for anything in any way, just pointing out that odds can be much worse than needing a hundred hits for a kill.

http://www.sas1946.rocks:8000/en/sortie/19978/?tour=1

Mike

October 16, 2020

5 hours ago, [DBS]Browning said:

I spent this morning trawling through the combat box stats again.

It is very tedious and time consuming and I've completely lost the will to do any more after collecting stats from 100 P-51 sorties. If anyone wants a comparison to other planes, they must do it themselves (or, ideally, write a script to automate the process).

I only gathered stats from sorties in which the P51 only hit a single target and in which the target was only hit by the single P51.

The average time from first hit to the death or bailout of the enemy was 96.4 seconds. The Longest time was 821 seconds and the shortest time was under one second (pk in 4 hits)

Average accuracy was 2.8% with a high of 22.1% and a low of 0.2%

Average rounds per kill was 28.8 with a high of 82 and a low of 4.

Edit:

This thread is about comparing the experiences in game to the historical reality. Understanding what is currently happening in the game is vital to that.

Thank you for doing the digging. I'm not sure that these numbers tell us anything conclusive. The number of hits to kill seems to line up with the tests and historical average, but the average time until the target went down of 96 seconds is troubling. That could be time where the enemy was still killing people. 96 secs is a long time in a dogfight. It might be good to know the avg. time from first hit to kill for a plane with HE ammo. I wish the stats could give us more info.

1 hour ago, SAS_Storebror said:

I'm rarely flying P-51 or P-47 in pure fighter vs. fighter sorties (for obvious reasons *cough* *cough*) but going back through our server stats, I found one sortie where actually I didn't shoot ground targets but only bombed them, so all bullets were against single engine fighters, and I got a total of 5.

1x 109K-4, 1x109F-4, 1x109E-7, 2x Fw-190A-5.

Number of bullets hit the target: 983

Number of PK kills: 4

Number of Plane damage kills: 1 - only the very last 190 went down due to damage, and then it was just two minutes after I let her go.

Not that I'd insist on this sortie being representative for anything in any way, just pointing out that odds can be much worse than needing a hundred hits for a kill.

http://www.sas1946.rocks:8000/en/sortie/19978/?tour=1

Mike

These stats are very unusual. The gunnery accuracy is about 20x better than normal and the number of hits is off the charts. Are you sure there isn't a bug somewhere in your stat gathering.?

Edited October 16, 2020 by BCI-Nazgul

October 16, 2020

15 hours ago, =DMD=Honza said:

Now i wonder why germans switched to AP-API ammo only, in 44 for 13mms instead of HE ?

Becuase now only ammo 13mm in game uses is HE, i tryed also for test to replace german 13mm HE with german 13mm AP that is in game so 109s shoot only AP then, and no longer was i able to down 10+ 51s or 47 AIs with 2x13mm only like its posible now with bugged full HE belt.

16 minutes ago, BCI-Nazgul said:

...

These stats are very unusual. The gunnery accuracy is about 20x better than normal and the number of hits is off the charts. Are you sure there isn't a bug somewhere in your stat gathering.?

Its more of how you would have acc% vs AI in SP as this was also vs AI just in MP server, when you know your attacking AI and no other humans to fear of stealing your kill or getting you , you can wait for better opotunities to shoot stedy and have high % like you would have in SP missions as AI bhaves predictable and you dont have pressure like vs humans or with humans, but still you can see awradge is ~190 bullets per airplane to down it, is that suposed to be normal for shoting down enemy airplanes with .50s ?

October 16, 2020

4 hours ago, [DBS]Browning said:

The damage percentages are absolutely useless. I only used the number of rounds that hit.

I see, sorry, I misinterpreted what you said.

October 16, 2020

4 hours ago, [DBS]Browning said:

Only single engine axis aircraft. Edit: Also, I remember there is one 262 kill in the stats used.

The previous figure of 31.5 hits per kill from a smaller sample size may have included 110s as I was not aware that they listed as "Fighter" in the stats. They are not included in the new figures.

Edit2:

It should also be remembered that there will be a certain amount of over-kill for the majority of kills. The stats appear to stop counting hits after a PK or bailout, but for other kill types, we should expect that the attacker often continues to fire and score hits, even after the target is damaged beyond the point it can fly.

Due to reduced maneuvering of damaged planes, it may sometimes be the case that the majority of hits occur after the plane is too damaged to stay airborne.

The problem I see of taking into account only kill samples (sorties that end with a kill) and single engament kill (combat with only one attacker) is that is biased towards to the effective ones. I understand it was chosen that for the sake of simplicity and consistency of the samples (to eliminate other variables) but at the same time you are only picking the cases in which the attack hit something "important in the game" and is thus effective.

If you take into account those attacks in which several allies gives pass after pass to down a plane or those single attacker cases in which after several passes the target "merrely" flew away you would get different results. Obviously, I have no idea how to do it without throwing into the calculation many other variables and inaccuracies (for instance,when you have multiple attacker is almost imposible to know how many hits the target sustained. Or that is almost impossible to know for a target flying away when/why the attacker stop the attack).

Edited October 16, 2020 by HR_Zunzun

October 16, 2020

47 minutes ago, CountZero said:

Becuase now only ammo 13mm in game uses is HE, i tryed also for test to replace german 13mm HE with german 13mm AP that is in game so 109s shoot only AP then, and no longer was i able to down 10+ 51s or 47 AIs with 2x13mm only like its posible now with bugged full HE belt.

Its more of how you would have acc% vs AI in SP as this was also vs AI just in MP server, when you know your attacking AI and no other humans to fear of stealing your kill or getting you , you can wait for better opotunities to shoot stedy and have high % like you would have in SP missions as AI bhaves predictable and you dont have pressure like vs humans or with humans, but still you can see awradge is ~190 bullets per airplane to down it, is that suposed to be normal for shoting down enemy airplanes with .50s ?

According to @[DBS]Browning's analysis (of 100 sorties?) in competitive MP, the average number of hits for a P-51 to get a kill was 28.8 So no, ~190 hits does not look normal. Just accepting that both numbers are real and not some artefact of the stats collection, there could be a rational explanation for this huge difference.

If you were firing from mostly a six o'clock position, where the pilot is relatively well protected, and your hits were fairly well distributed over the aircraft, I can see that you might take that number of hits to down a109/190 in BoX. (Whether it should is another matter). The apparent size of the pilot's exposed parts is small, dispersion of hits means no wing breakages, and then there is the difficulty of setting fire to fuel tanks.

In a fight against a manoeuvring human target you are more likely to get high deflection shots with a reasonable view of the pilot. Much harder to get a high hits/shot ratio than when tucked in behind an AI, but a much better kill/hit ratio as PK is - quite correctly - the main vulnerability of fighter types, at least in terms of quick kills.

October 16, 2020

The issue with this statistical sample is that it only shows a part of the picture. The issue with the AP rounds isn't so much that they can't kill planes. They can, obviously. The issue is if you hit a plane with AP rounds but don't kill it outright there is little to no effect on the plane. There is little speed loss and in so far as I can tell no loss to maneuverability. I don't know if there is a way to capture this phenomenon in a statistical analysis of damage events. Perhaps looking at planes that were damage by .50 AP only firing planes who then go on to win fights? Something closer to a bench test would be a better way to plumb the depths of this occurrence.

October 16, 2020

If you are only choosing sorties where kills were scored, you’re eliminating a lot of data that should be counted. Also, how are you accounting for cases where multiple pilots engaged the same target? The hits are only accounted on the shooters end, you have to dig through the targets log to see if there were multiple attackers, and the you’d have to look at each attacker and hope they’d only shot at one target.

An example: https://combatbox.net/en/sortie/log/909591/?tour=26

If you just looked at the log of the player who scored the kill, you say it took 78 hits to kill that FW190, but if you keep digging there were 121 hits from two of the attackers, plus some percentage of my 98 hits (I did some strafing afterwards).

Another example: https://combatbox.net/en/sortie/857926/?tour=25

This looks great on the surface - 19 hits for a kill. But once you dig in there you see he was also damaged by a P-51 (who attacked multiple targets, so we can’t tell how many hits he scored on my kill) and a Tempest (who did some strafing so the same problem). And what the stats don’t really make clear is that he appeared to just crash while trying to avoid the Tempest, and likely hadn’t actually sustained serious damage yet.

October 16, 2020

1 hour ago, KW_1979 said:

If you just looked at the log of the player who scored the kill, you say it took 78 hits to kill that FW190, but if you keep digging there were 121 hits from two of the attackers, plus some percentage of my 98 hits (I did some strafing afterwards).

Yeah, I checked both. That's largely what made it such a lengthy and tedious process.

October 16, 2020

6 hours ago, CountZero said:

Becuase now only ammo 13mm in game uses is HE, i tryed also for test to replace german 13mm HE with german 13mm AP that is in game so 109s shoot only AP then, and no longer was i able to down 10+ 51s or 47 AIs with 2x13mm only like its posible now with bugged full HE belt.

Its more of how you would have acc% vs AI in SP as this was also vs AI just in MP server, when you know your attacking AI and no other humans to fear of stealing your kill or getting you , you can wait for better opotunities to shoot stedy and have high % like you would have in SP missions as AI bhaves predictable and you dont have pressure like vs humans or with humans, but still you can see awradge is ~190 bullets per airplane to down it, is that suposed to be normal for shoting down enemy airplanes with .50s ?

Hey CountZero you are right. UB data file is MG-131.Check that data file out in your MOD. UB has been MG-131 entire time. M2 has been nerfed from the start...

People showing amazing silly game data just stop.

October 16, 2020

15 minutes ago, CrazyhorseB34 said:

Hey CountZero you are right. UB data file is MG-131.Check that data file out in your MOD. UB has been MG-131 entire time. M2 has been nerfed from the start...

People showing amazing silly game data just stop.

They are differant, it has some same parts but UB is UB, 131 is 131, they are not same, here are extracted files so you can see:

12.7mm HE (one i used for mod for .50)

Spoiler

Class_name = "CBatchBallistics"
object_name = "BULLET_RUS_12-7x108_HE"

//12.7мм патрон к пулемету УБ разрывной (ОФ)

////// PhysicsBody properties
VisualImage=0,"graphics/ammo/ammoRmgSmk.mgm",false
visualradius=4
SoundScript=""
ImageAttr=21 //IA_NOMINPIXELS | IA_NOINTERPOLATION

//CollisionBody="graphics/ammo/ammo.col"

NoDirecion=true
NoCollision=true

// дальность, скорость на этой дальности (на 0м - указать на 5км/ч меньше расчетной), пары броня-урон за броней. Броню больше, чем в первой паре, не пробьет
// за базовый взят урон на дальности 100м при нулевой броне, на макс. броне урон равне 0.25 от базового, на 0.8 макс. брони урон равен 0.5 от базового
// урон при нулевой броне уменьшается с дальностью по соотношению квадрата скоростей
// скорости по дальностям взяты из лога утилиты расчета .bin файлов пули (расчет таблиц наведения), на 10км/ч меньше ресчетной
// т.к. если скорость окажется больше расчетной на данной дальности, то дамага не будет
// базовый урон расчитан по соотношению кинетических энергий на дальности 100м между данной пулей и референсной пулей SHELL_GER_20x82_AP (=1000)

Armor=1,890,   9,160,   7,400,   0,641
Armor=100,786,   8,125,   6,312,   0,499
Armor=500,499,   4,50,   3,126,   0,201 // опорная бронепробиваемость пересчитана по формуле де Марра из референсной бронепробиваемости SHELL_RUS_20x99_HE на близкой скорости
Armor=2000,195,   2,0,   1,8,   0,31

// аэродинамические коэфициенты подобраны исходя из совпадения графика скорости по дальности для эталонной пули (посчитано на балл. калькуляторе)

Gage=12.7               //калибр
DefaultStartSpeed=890   //дефолтная начальная скорость при стрельбе из простых пушек (наземка)
MaxDistance=2000       //предельная дальность, дальше которой объект удалится
Mass=0.039               //масса пули
Mkr=0.99               //значение числа Маха, после которого начинается повышение Cx до момента M=1, зависит от формы тела но не от размера
Mmax=2.0               //значение числа Маха, до которого завершается понижение Cx, начавшееся при M=1, зависит от формы тела но не от размера
Cx_0=0.23               //значение к-та сопротивления при (M <= Mkr), зависит от формы тела но не от размера
Cx_max=0.6               //значение к-та сопротивления при (M = 1), зависит от формы тела но не от размера
Cx_1=0.55               //значение к-та сопротивления при (M >= Mmax), зависит от формы тела но не от размера
R=0.00635               //радиус пули
LiveTime=7               //время жизни объекта (большее из двух: время полета на MaxDistance или время горения трассера + 3.5с)
TimeToDestroy=0           //время жизни объекта после попадания в скрытом состоянии, что бы трейл трассера не пропадал пока не расствориться
TracertShowtime = 2.12   //время горения трассера
MaxRedirections = 0       //максимальное кол-во рикошетов

debug=false

DefaultBulletSpeed = 890.0 //дефолтная начальная скорость при стрельбе из комплексной (авиационной) пушки неопределенной модели
BulletSpeed = "UB", 890.0 //начальная скорость при стрельбе из комплексного (авиационного) пулмета УБ

HitDummy="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_object.txt"
HitMisc="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_object.txt"
HitWater="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_water.txt"
HitGround="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_ground.txt"
HitArmor="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_object.txt"
HitBuilding="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_building.txt"
HitMetal="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_object.txt"
HitForest="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_39g_2g_object.txt"

shellType = 6
// SHELL_TYPE_CV = 1,// Бронебойный - CV - Armour Piercing with Cavity Bursting Charge
// SHELL_TYPE_AP = 2,// Бронебойный сплошной - AP - Armour Piercing Solid
// SHELL_TYPE_HV = 3,// Подкалиберный - HV - Armour Piercing Composite Rigid (High Velocity)
// SHELL_TYPE_HT = 4,// Кумулятивный - HT - Hight Explosive Anti Tank (HEAT)
// SHELL_TYPE_SH = 5,// Шрапнельный - SH - Shrapnel
// SHELL_TYPE_HE = 6,// Осколочный - HE - High Explosive and Fragmental
// SHELL_TYPE_SpHV = 7,// Подкалиберный катушечный - SpHV - Armour Piercing Spool-Shaped (Spool High Velocity)

131 HE

Spoiler

Class_name = "CBatchBallistics"
object_name = "BULLET_GER_13x64_HE"

//13.0мм патрон к пулемету MG131 разрывной (ОФ)

////// PhysicsBody properties
VisualImage=0,"graphics/ammo/ammoYmgSmk.mgm",false
visualradius=4
SoundScript=""
ImageAttr=21 //IA_NOMINPIXELS | IA_NOINTERPOLATION

//CollisionBody="graphics/ammo/ammo.col"

NoDirecion=true
NoCollision=true

// дальность, скорость на этой дальности (на 0м - указать на 5км/ч меньше расчетной), пары броня-урон за броней. Броню больше, чем в первой паре, не пробьет
// за базовый взят урон на дальности 100м при нулевой броне, на макс. броне урон равне 0.25 от базового, на 0.8 макс. брони урон равен 0.5 от базового
// урон при нулевой броне уменьшается с дальностью по соотношению квадрата скоростей
// скорости по дальностям взяты из лога утилиты расчета .bin файлов пули (расчет таблиц наведения), на 10км/ч меньше ресчетной
// т.к. если скорость окажется больше расчетной на данной дальности, то дамага не будет
// базовый урон расчитан по соотношению кинетических энергий на дальности 100м между данной пулей и референсной пулей SHELL_GER_20x82_AP (=1000)

Armor=1,750,   7,112,   6,281,   0,449
Armor=100,654,   6,85,   5,214,   0,342 // опорная бронепробиваемость пересчитана по формуле де Марра из референсной бронепробиваемости SHELL_RUS_20x99_HE на близкой скорости
Armor=500,382,   3,29,   2,73,   0,117
Armor=2000,183,           1,7,   0,27

// аэродинамические коэфициенты подобраны исходя из совпадения графика скорости по дальности для эталонной пули (посчитано на балл. калькуляторе)

Gage=13.0               //калибр
DefaultStartSpeed=750   //дефолтная начальная скорость при стрельбе из простых пушек (наземка)
MaxDistance=2000       //предельная дальность, дальше которой объект удалится
Mass=0.034               //масса пули
Mkr=0.89               //значение числа Маха, после которого начинается повышение Cx до момента M=1, зависит от формы тела но не от размера
Mmax=1.3               //значение числа Маха, до которого завершается понижение Cx, начавшееся при M=1, зависит от формы тела но не от размера
Cx_0=0.17               //значение к-та сопротивления при (M <= Mkr), зависит от формы тела но не от размера
Cx_max=0.57               //значение к-та сопротивления при (M = 1), зависит от формы тела но не от размера
Cx_1=0.55               //значение к-та сопротивления при (M >= Mmax), зависит от формы тела но не от размера
R=0.0065               //радиус пули
LiveTime=7.3           //время жизни объекта (большее из двух: время полета на MaxDistance или время горения трассера + 3.5с)
TimeToDestroy=0           //время жизни объекта после попадания в скрытом состоянии, что бы трейл трассера не пропадал пока не расствориться
TracertShowtime = 1.75   //время горения трассера
MaxRedirections = 0       //максимальное кол-во рикошетов

debug=false

DefaultBulletSpeed = 750.0 //дефолтная начальная скорость при стрельбе из комплексной (авиационной) пушки неопределенной модели
BulletSpeed = "MG131", 750.0 //начальная скорость при стрельбе из комплексного (авиационного) пулмета MG131

HitDummy="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_object.txt"
HitMisc="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_object.txt"
HitWater="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_water.txt"
HitGround="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_ground.txt"
HitArmor="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_object.txt"
HitBuilding="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_building.txt"
HitMetal="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_object.txt"
HitForest="LuaScripts/WorldObjects/Explosions/HEprojectiles/HE_013mm_38g_1g_object.txt"

shellType = 6
// SHELL_TYPE_CV = 1,// Бронебойный - CV - Armour Piercing with Cavity Bursting Charge
// SHELL_TYPE_AP = 2,// Бронебойный сплошной - AP - Armour Piercing Solid
// SHELL_TYPE_HV = 3,// Подкалиберный - HV - Armour Piercing Composite Rigid (High Velocity)
// SHELL_TYPE_HT = 4,// Кумулятивный - HT - Hight Explosive Anti Tank (HEAT)
// SHELL_TYPE_SH = 5,// Шрапнельный - SH - Shrapnel
// SHELL_TYPE_HE = 6,// Осколочный - HE - High Explosive and Fragmental
// SHELL_TYPE_SpHV = 7,// Подкалиберный катушечный - SpHV - Armour Piercing Spool-Shaped (Spool High Velocity)

bug with 131 guns in game is that they show they should be using mix of AP and HE 13mm ammo, but they use only HE ammo and no AP becuase of bug. So you can notice if you shoot only with 13mm german guns they fire only HE even though picture when selecting ammo says its mix.

Edited October 16, 2020 by CountZero

October 17, 2020

The top MG131 in the He111 H-16 certainly uses a mixed belt. If you are careful you can fire one shot at a time, each one on a different hit box on a static target. The surface damage graphics clearly alternate: one at the minimum level - presumably the AP shot, then the next hitting, a different area, at the second level.

Edited October 17, 2020 by unreasonable

October 17, 2020

9 hours ago, KW_1979 said:

If you are only choosing sorties where kills were scored, you’re eliminating a lot of data that should be counted.

Considering your first point here: that depends to some extent on what you want to know. Part of the problem is that the attacks and hits are not coming at random angles, even if we treat them as hitting randomly from a given angle.

It seems to me that there are two types of .50 cal attacks on 109s/190s:

1) Dead six - with an extremely low p of kill per hit. (< 0.005)

2) Angle off 10 degrees plus - with a much higher average. (~0.02)

I would not be surprised if @[DBS]Browning 's sample included a high proportion of Type 2. Adding in lots of Type 1 attacks without kills will obviously increase the aggregate average hits/kill but it does not add much information except the ratio of the attack types that MP pilots are using.

I agree it would be better if all of the attacks of a given type could be tracked, kill or not, to determine the mean average total hits/killed target. With a big enough sample you actually do not need this: you can deduce the p to kill of the incremental shots purely from the shape of the distribution of kills. But we are unlikely to have a sample large enough to do that convincingly.

It is, however, a property of these distributions that whatever the likely average p of kill, the actual numbers of hits required to kill will have a tail well out to the right. For instance, if you take 100,000 targets and subject them to one hit at a time with a p to kill of 0.02, the average hits/kill will be 50, at each stage, but half the sample will be shot down with 35 hits or fewer. You will still have ten percent of the sample left after 113 hits, and five percent after 148 hits.

Edited October 17, 2020 by unreasonable

October 17, 2020

1 hour ago, unreasonable said:

Considering your first point here: that depends to some extent on what you want to know. Part of the problem is that the attacks and hits are not coming at random angles, even if we treat them as hitting randomly from a given angle.

It seems to me that there are two types of .50 cal attacks on 109s/190s:

1) Dead six - with an extremely low p of kill per hit. (< 0.005)

2) Angle off 10 degrees plus - with a much higher average. (~0.02)

I would not be surprised if @[DBS]Browning 's sample included a high proportion of Type 2. Adding in lots of Type 1 attacks without kills will obviously increase the aggregate average hits/kill but it does not add much information except the ratio of the attack types that MP pilots are using.

I agree it would be better if all of the attacks of a given type could be tracked, kill or not, to determine the mean average total hits/killed target. With a big enough sample you actually do not need this: you can deduce the p to kill of the incremental shots purely from the shape of the distribution of kills. But we are unlikely to have a sample large enough to do that convincingly.

It is, however, a property of these distributions that whatever the likely average p of kill, the actual numbers of hits required to kill will have a tail well out to the right. For instance, if you take 100,000 targets and subject them to one hit at a time with a p to kill of 0.02, the average hits/kill will be 50, at each stage, but half the sample will be shot down with 35 hits or fewer. You will still have ten percent of the sample left after 113 hits, and five percent after 148 hits.

The problem with the way the samples are chosen is that you select only the cases that are suscesful. For the question we are trying to answer is giving us an incorrect answer. 34 hits per kill is not the average of hits you need to get for shooting down a plane. It is the average you use when you do shot down a plane. And I wouldn't be surprised if in those ones PKs are over represented.

In order to better answer our question one the way I think the samples should be selected is all the sortie in which one plane attack only one target and this is being attacked by this plane only and then note the hits and the outcome (shot down or not). With all that data then make groups of different numbers of hits (0-10,10-20-20-30 etc.) and the outcome (shot down or not) in each group. Although this method have its own sort of problem and biases, I think it will give you a better representation of the efficacy of the gun.

October 17, 2020

I would much rather people who want more or different data gather it them selves, rather than just complaining that it isn't there.

I checked 140 sorties in which single engine Axis fighters landed at base.

Of those sorties, 6 planes received damage from 50cals. Of those 6, 4 of the attackers damaged no other aircraft.

Here are the sorties:

Attacker 1 - Target 1

Attacker 2 - Target 2

Attacker 3 - Target 3

Attacker 4 - Target 4

The number of hits without a kill where 10, 4, 74 and 5 respectively.

It took me over an hour to get just these four results. A sample of 4 is obviously of little use.

It is, of course, possible to be hit by a million rounds and survive, providing they all hit the same point on the wing tip. I don't think this data is useful compared to the number of rounds needed to score a kill.

Edited October 17, 2020 by [DBS]Browning

October 17, 2020

5 hours ago, HR_Zunzun said:

The problem with the way the samples are chosen is that you select only the cases that are suscesful. For the question we are trying to answer is giving us an incorrect answer. 34 hits per kill is not the average of hits you need to get for shooting down a plane. It is the average you use when you do shot down a plane. And I wouldn't be surprised if in those ones PKs are over represented.

In order to better answer our question one the way I think the samples should be selected is all the sortie in which one plane attack only one target and this is being attacked by this plane only and then note the hits and the outcome (shot down or not). With all that data then make groups of different numbers of hits (0-10,10-20-20-30 etc.) and the outcome (shot down or not) in each group. Although this method have its own sort of problem and biases, I think it will give you a better representation of the efficacy of the gun.

You are right up to a point, but these two numbers - average hits on planes shot down, and total hits/ planes shot down - will have mathematical relationship when we have a reasonably large sample, which 100 planes shot down is. So from the average hits on killed planes we can deduce the kills/total hits with good confidence, unless, as I pointed out earlier, there are two types of attacks with radically different characteristics, allowing for a large number of high hit, non-shoot downs, in which the key variable is the prevalence of each attack.

For instance: assume a p to kill of 0.02, making a hit at a time on the survivors of each round. For a 100 plane sample:

After the first hit, we have 100 hits, 2 planes shot down*, average hits per kill = 50, average hits on killed planes = 1.0

After 30 hits we have 2,273 hits, 45 planes shot down, average hits per kill = 50, average hits on killed planes = 14.0

After 60 hits we have 3,512 hits, 70 aircraft shot down, average hits per kill = 50, average hits on killed planes = 24.6

After 100 hits we have 4,337 hits, 87 aircraft shot down, average hits per kill = 50, average hits on killed planes = 34.7

Note you are still have 13 aircraft that are not shot down after 100 hits.

*On any given run you might have more or less but do this enough times and the average here is 2.

The average hits on killed planes will gradually converge on 50 (1/0.02) but very, very slowly. Which is why it is not a particularly useful number for this sort of problem. The number of hits required to shoot down half the sample is much more meaningful - in this example 34 hits.

This model - p per hit of about 0.02 to 0.03 looks a reasonable first estimate for @[DBS]Browning's analysis. It is consistent with the historical data such as it is, also allows for a considerable proportion of kills that require over 100 hits.

Edited October 17, 2020 by unreasonable

October 17, 2020

4 hours ago, [DBS]Browning said:

I would much rather people who want more or different data gather it them selves, rather than just complaining that it isn't there.

If you refer to me, I certanly wasn´t complaining neither demanding to anyone to do anything. Just pointed out what I considered it was a incorrect data gathering method for the question at hand. Just my opinion given in a reasoned way. I am sorry if it sounded to you otherway.

1 hour ago, unreasonable said:

You are right up to a point, but these two numbers - average hits on planes shot down, and total hits/ planes shot down - will have mathematical relationship when we have a reasonably large sample, which 100 planes shot down is. So from the average hits on killed planes we can deduce the kills/total hits with good confidence, unless, as I pointed out earlier, there are two types of attacks with radically different characteristics, allowing for a large number of high hit, non-shoot downs, in which the key variable is the prevalence of each attack.

For instance: assume a p to kill of 0.02, making a hit at a time on the survivors of each round. For a 100 plane sample:

After the first hit, we have 100 hits, 2 planes shot down*, average hits per kill = 50, average hits on killed planes = 1.0

After 30 hits we have 2,273 hits, 45 planes shot down, average hits per kill = 50, average hits on killed planes = 14.0

After 60 hits we have 3,512 hits, 70 aircraft shot down, average hits per kill = 50, average hits on killed planes = 24.6

After 100 hits we have 4,337 hits, 87 aircraft shot down, average hits per kill = 50, average hits on killed planes = 34.7

Note you are still have 13 aircraft that are not shot down after 100 hits.

*On any given run you might have more or less but do this enough times and the average here is 2.

The average hits on killed planes will gradually converge on 50 (1/0.02) but very, very slowly. Which is why it is not a particularly useful number for this sort of problem. The number of hits required to shoot down half the sample is much more meaningful - in this example 34 hits.

This model - p per hit of about 0.02 to 0.03 looks a reasonable first estimate for @[DBS]Browning's analysis. It is consistent with the historical data such as it is, also allows for a considerable proportion of kills that require over 100 hits.

Again, the problem I see with that method is that by using susccesful sorties only you asume that you include, in your kills, all type of damage when that is not neccessarily true.

Depending on the flaws of the current DM, with regard to AP damage, then your average on only suscessful sorties (what Browning has collected) will represent most probable causes of a kill that the current DM allows.

Let me try to explain with an example. If in our case, following your method, our PK chance is 0.02 (Just invented for the example) and it was in line with real life but severing control cables to cause a depart from controlled fly is 0.004 but in real life was 0.017 (both invented too for the shake of the example), then is easy to see that the later is going to dissapear from you averages because the average amount of hits on target needed will exceed what the nature of air combat will allow you.

The method I explained is far from perfect (as it has its own caveats) but if you get the probability of kill for the different amount of hits, you will get a better picture of how many hits , on average, you need for downing a plane. And most important, it will let us see whether or not the number of occasions in which you put an excesive or ridiculous amount of hits on a plane and he manage to go back to base is overepresented (as is the feeling of some of us). Going back to base with 5-10-15 impacts is reasonable. Going back to base with 70 hits should be very unlikely.

Obviously, as Browning has correctly pointed out, that will require an enormous amount of data collection time and effort to do it. I do not know if anyone will have the time and-or the will.

October 17, 2020

Knowing the kill probability of a hit (here after "PKill") also allows its to directly infer the survival probability of a hit. Further more, neither the PKill, nor the survival probability can be calculated from hits that don't result in a kill. This is why is is not useful to have data from hits that don't result in kills.

Even if it was the case that it is common to survive 500 hits, so long as the shooter is on the dead six, this will still decrease the PKill, even if only kills are looked at.

Now there could be some other mechanic going on. For example, perhaps, due to a bug, a gun does no damage at all of it hits the rudder. That would be a problem that looking at the stats would not reveal, but it would still result in a lower PKill.

Likewise, if a gun caused a plane to explode every time the red navigation lights where hit, that's not a problem the stats will reveal, but it will still result in a higher PKill.

Such stats don't reveal bugs with the damage model, but they do show us how likely the average hit surviveable as much as they show is how likely each hit will result in a kill.

From this, you can infer, without any other information, how likely it is that a plane can RTB after a given number of hits.

Edited October 17, 2020 by [DBS]Browning

October 17, 2020

Browning's type of analysis does "get the probability of kill for the different amount of hits". As I have tried to explain, you do not actually need the total number of hits on the undead planes to work this out, you only need the number on the dead planes. The shape of the distribution of kills over hits tells you the probability of each incremental hit causing a kill.

edit: See Browning just made same point.

13 minutes ago, CountZero said:

So if they have tool that do thousounds and thousonds tests quickly then they can easy see if their game is inline to ww2 data.

This topic can easy be debunked then if you let tool do planty of tests and show all is ok and we are just biased to one outcom.

Although if you have only 2 types of ammo to represent all ammo types and not historical belts,

Good point on their tool - they could actually replicate the US Ballistics lab reports...... as I tried to do for 20mm HE and P-47 using LAA: took weeks.

The lack of a specific incendiary effect is something I have been going on about for years: FC is a 1918 game and certainly they were in routine use. Although the FC crowd (if there is one left) would go nuts if their planes were made any more fragile, it would actually be a leveller, since the entente planes are now mostly structurally much weaker. Certainly setting fire to those leaking 109/190 fuel tanks somewhat faster would help with the .50 cal complaints.

Edited October 17, 2020 by unreasonable

October 17, 2020

29 minutes ago, unreasonable said:

Browning's type of analysis does "get the probability of kill for the different amount of hits". As I have tried to explain, you do not actually need the total number of hits on the undead planes to work this out, you only need the number on the dead planes. The shape of the distribution of kills over hits tells you the probability of each incremental hit causing a kill.

But this is only if, as you mentioned in one of your previous posts, the attacks are all similar (and thus the damaged caused). And that is why I think that method, if we are right, won´t show the cases in wich the ammo is unrealistically ineffective (basically attacks from 6 oclock that is the most common attack). And this is in line with what my method, I think, it would let you find out. Because it will showed the results of those attack method results that Browning sampling would be just ignoring.

Also, take into account that neither methods is going to deal with the other main complaint that is the ability of 109/190 to maneouvir unexpectedly well after being hit a big amount of times.

By the way, I used Countzero mod with the UB instead of the M2 and use is in a few rounds of QMB against g14. In that very limited test, the difference was abysmal and jaw opening. Almost any 2 second bust was a guranteed kill. Even considering the lack of explosive/incendiary material in the AP and the noticiable, but not big, advantage in kinetic energy of the UB vs the M2 in real life, the difference is not what I would expect.

As has been pointed out there seem to be a huge inbalance in between what AP vs HE/I can do in the game. If that was a true representation of real life performance then I bet everybody would have changed to I/He-only ammo belts.

Edited October 17, 2020 by HR_Zunzun

October 17, 2020

I think the DM does not handle the effects of AP correctly either through lack of complexity and/or how it handles penetration of multiple parts of the plane. We know for sure the aero effects are way below par.

Edited October 17, 2020 by BCI-Nazgul

October 17, 2020

22 minutes ago, HR_Zunzun said:

But this is only if, as you mentioned in one of your previous posts, the attacks are all similar (and thus the damaged caused). And that is why I think that method, if we are right, won´t show the cases in wich the ammo is unrealistically ineffective (basically attacks from 6 oclock that is the most common attack).

All you have to do to show this type of attack is ineffective or not is static testing where you can place shots, if you have a reasonable RL basis for comparison: really no need to select some set of sorties from MP.

9 minutes ago, BCI-Nazgul said:

I think the DM does not handle the effects of AP correctly either through lack of complexity and/or how it handles penetration of multiple parts of the plane. We know for sure the aero effects are way below par.

Maybe on the first point - I have yet to see any evidence to support the second. Photos of 30mm Incendiary exit holes certainly do not qualify. More like the 13mm HE is way OP.

October 17, 2020

6 minutes ago, unreasonable said:

All you have to do to show this type of attack is ineffective or not is static testing where you can place shots, if you have a reasonable RL basis for comparison: really no need to select some set of sorties from MP.

Of course you do not need it and a test would be representative. My point was to point out the possible mistakes you could incurr if using the MP sorties (and giving an reasoned alternative).

Regarding the tests, a static one will only give you part of the picture. PK, fires or Structural damage. Control system damage, engine loss of performance and aerodynamic penalties would be left out of the test.

6 minutes ago, unreasonable said:

Maybe on the first point - I have yet to see any evidence to support the second. Photos of 30mm Incendiary exit holes certainly do not qualify. More like the 13mm HE is way OP.

I gave some information regarding the effect of AP rounds in real life as perceived by the pilots and reported on the AAR. That is, that seeing pieces flying off was something at least, relatively common with AP (so far aprox 200 mentions out of 700 reports and one pilot commented that was "customary" so, what one can expect). Pieces flying off from the wing are no doubt going to cause aerodynamic penalty.

The problem with this is like with any other damage effect in real life, the data is limited and subjected to error and innacuracies (whether is coming from a test or a personal report). So the frequency and the magnitude is always going to be subjected to personal interpretation. What we can at least try to do is eliminate the extremes. And this is what all this complain is all about.

October 17, 2020

On 10/16/2020 at 5:48 PM, BCI-Nazgul said:

These stats are very unusual. The gunnery accuracy is about 20x better than normal and the number of hits is off the charts. Are you sure there isn't a bug somewhere in your stat gathering.?

There's a simple reason why the accuracy is so much better than "normal":
I'm doing this CFS stuff for way too long now and I'm generally shooting at point blank range if the target performs evasive manoeuvres and only use distance (200-300 yds) deflection shots (which I dare to say I'm pretty good at) if the victim is completely unaware of my presence.
If you look at the results of my other sorties, you'll have a hard time to find any hit rate in the single digit range.

44 minutes ago, HR_Zunzun said:

I gave some information regarding the effect of AP rounds in real life as perceived by the pilots and reported on the AAR. That is, that seeing pieces flying off was something at least, relatively common with AP (so far aprox 200 mentions out of 700 reports and one pilot commented that was "customary" so, what one can expect). Pieces flying off from the wing are no doubt going to cause aerodynamic penalty.

You gave sufficient information but it gets largely ignored.

Instead we now see the thread getting turned into some high sophisticated bullet counting battle, based on so many assumptions that by any reasonable scientific standards the efforts are useless at best.

@sniperton nailed it already:

On 10/15/2020 at 10:19 PM, sniperton said:

- no need to be proven here that a great many people flying aircraft using API ammo, represented as using AP ammo in-game, have the feeling that those aircraft underperform - they do feel.

Yet there's still a few - but very loquacious - people trying to persuade us that what we all see and feel is just "normal".
They'll have their reasons, some call it "agenda".

1 hour ago, unreasonable said:

Photos of 30mm Incendiary exit holes certainly do not qualify.

Because... you say so?

I have yet to see evidence to support your take on it.

Guys we keep discussing things all reasonable people already agreed upon ten pages ago.

Let's stop the trolling.

If the Devs would have wanted to let us know their opinion about it and/or whether or not they're willing to change any of this, they would have had plenty of opportunity already.

No need to beat this dead horse any further.

Mike

Edited October 17, 2020 by SAS_Storebror

October 17, 2020

2 hours ago, unreasonable said:

Maybe on the first point - I have yet to see any evidence to support the second. Photos of 30mm Incendiary exit holes certainly do not qualify. More like the 13mm HE is way OP.

OK, as far as I'm concerned the horse isn't dead until the devs say it's dead. ?

But, here are some things we found about the aero damage that don't seem right...

Yes, we both agree that the 13mm HE is OP. But, here's my take on the AP aero stuff. From your tests and ours we both found that it takes MANY hits in the same area (you call them hit boxes) with AP to cause a significant speed loss and/or loss of handling. I think you said "you have to get to 'level two' damage before aero effects show up." In reality ANY hole in a wing is going to cause a slight speed loss through lower lift and drag. Bullet holes in metal are indented on the side that is struck and jagged on exit side. Both are not good for smooth air flow over the surface. The fact that the aero damage is quantized by the game is problematic. I also do not believe that aero damage is additive across the whole airframe. In other words, let's say I get 10 hits on the right wing, 3 on the left and 7 on the fuselage. From what it appears since none of the individual "hit boxes" has reached "level two" the plane continues to fly normally even though it's been hit by a total of 20 rounds just not all in the same area. In real life this would add a lot of drag to the plane, but in IL2 it does nothing really. In WWII it was found that just polishing a plane could add 5 or 10 mph to it's top speed, just imagine what a bunch of jagged .5" or even .303 holes would take away from it's top speed. The other thing we found is that the aero damage appears to have an upper limit for speed loss. One 20mm or 13mm hit to a wing causes a lot of speed loss, but more after that do not further reduce the speed. This all or nothing aero damage is also not right. The plane should simply become slower and slower until it simply can't stay airborne. Again this more of a penalty for AP ammo because numerous smaller hits aren't going to count for much past a certain point (even if the aero damage worked correctly.) Granted the plane will probably die from something else first, but with AP, barring a fire or PK, it's all the hope you've got in the current situation. I really don't know how the .303 armed Hurricane (which should absolutely riddle a plane with small holes) is going to be able to win any fights under the current DM and maybe that's our best hope for the devs to fix this situation.

Edited October 17, 2020 by BCI-Nazgul

October 18, 2020

On 10/16/2020 at 3:07 AM, [DBS]Browning said:

I spent this morning trawling through the combat box stats again.

It is very tedious and time consuming and I've completely lost the will to do any more after collecting stats from 100 P-51 sorties. If anyone wants a comparison to other planes, they must do it themselves (or, ideally, write a script to automate the process).

I only gathered stats from sorties in which the P51 only hit a single target and in which the target was only hit by the single P51.

The average time from first hit to the death or bailout of the enemy was 96.4 seconds. The Longest time was 821 seconds and the shortest time was under one second (pk in 4 hits)

Average accuracy was 2.8% with a high of 22.1% and a low of 0.2%

Average rounds per kill was 28.8 with a high of 82 and a low of 4.

Circling back to the small set of data we're arguing over - this doesn't pass the sniff test to me (I'm not trying to suggest any dishonesty or anything like that - just that's its not a representative sample). As I already pointed out, filtering out unsuccessful sorties is already artificially selecting best case outcomes. For another, in my extensive DM testing in 4.006 using static targets, the only way to get kills in that low of a round count was either to hit the pilot or to put all the shots into the engine. That was from a static gun shooting at a static target - not representative of the kind of hit spread you'd seeing in actual flying.

So I dug into my own stats going back a few weeks. Only counting P-51 and P-47 sorties, and discounting any where I did some strafing (and thus the hit number is uncertain), and counting in numbers from the other pilots who either assisted my kills or who I assisted I got these results:

33 blue aircraft shot down, by 2947 hits for an average of 89.3 hits per kill. If I exclude any multiengine aircraft (there were a fair number of 110s, 262s, and Ju88s in that set) it was 20 kills from 1407 hits for an average of 70.35 hits per kill.

But that's a small sample size and mostly coming from one pilot. So I decided to pull Krupinskii's stats for this month - one of the top pilots in the game, lots of dogfighting - should be a really optimistic setup to get low round per kill totals. I did the same thing where I traced all the airplanes he shot, as well as anyone assisting him, and I just ignored any extra rounds coming in from Spits, Tempests, P-38s, flak and in one case some friendly fire from a 190. So this is a really optimistic set of data that includes some 20mm and heavy flak which I can't account for exactly. 3693 hits for 56 kills - 65.9 rounds per kill.

So if we split the difference between my data and [DBS]Browning, the average hits per kill is about 50 - and if the average pilot is hitting at 5% (which may be optimistic based on his numbers) - then it takes 12.8 seconds of gunfire on average to score a kill with 6 x .50s in game right now. And I think that's a really optimistic number.

October 18, 2020

8 hours ago, BCI-Nazgul said:

OK, as far as I'm concerned the horse isn't dead until the devs say it's dead. ?

But, here are some things we found about the aero damage that don't seem right...

Yes, we both agree that the 13mm HE is OP. But, here's my take on the AP aero stuff. From your tests and ours we both found that it takes MANY hits in the same area (you call them hit boxes) with AP to cause a significant speed loss and/or loss of handling. I think you said "you have to get to 'level two' damage before aero effects show up." In reality ANY hole in a wing is going to cause a slight speed loss through lower lift and drag. Bullet holes in metal are indented on the side that is struck and jagged on exit side. Both are not good for smooth air flow over the surface. The fact that the aero damage is quantized by the game is problematic. I also do not believe that aero damage is additive across the whole airframe. In other words, let's say I get 10 hits on the right wing, 3 on the left and 7 on the fuselage. From what it appears since none of the individual "hit boxes" has reached "level two" the plane continues to fly normally even though it's been hit by a total of 20 rounds just not all in the same area. In real life this would add a lot of drag to the plane, but in IL2 it does nothing really. In WWII it was found that just polishing a plane could add 5 or 10 mph to it's top speed, just imagine what a bunch of jagged .5" or even .303 holes would take away from it's top speed. The other thing we found is that the aero damage appears to have an upper limit for speed loss. One 20mm or 13mm hit to a wing causes a lot of speed loss, but more after that do not further reduce the speed. This all or nothing aero damage is also not right. The plane should simply become slower and slower until it simply can't stay airborne. Again this more of a penalty for AP ammo because numerous smaller hits aren't going to count for much past a certain point (even if the aero damage worked correctly.) Granted the plane will probably die from something else first, but with AP, barring a fire or PK, it's all the hope you've got in the current situation. I really don't know how the .303 armed Hurricane (which should absolutely riddle a plane with small holes) is going to be able to win any fights under the current DM and maybe that's our best hope for the devs to fix this situation.

I do not really dispute much of this: especially that we can discuss what we want! If you get a penalty from one hit box at stage two I would expect a further penalty from the same level of damage from a second hit box. But I have not tested that. With my He111 control test I would need three targets - one undamaged control, one to be damaged only on one area, the third to be damaged on two. I will have a go at this soon - but my flying and shooting skills may not be up to it. Which is where the developers' tool CountZero mentioned would be useful.

As for the levels of penalty associated with each level of damage: obviously the game currently simplifies the damage graphics to make it work, rather than showing individual hits, but it must be counting individual hits and/or the damage from in order to determine which graphic level to show. Presumably, therefore, it could add lift/drag penalties at a more fine grained level. I assume that the developers do not consider this a worthwhile use of resources currently. The overall problem with this whole issue, is that we just do not have very good evidence of the real world effects although we all agree that there are some; if the developers have based their estimates of lift/drag effects on some data source I do not know what it is; perhaps they are just guessing; but at the moment that is all we are doing.

Just going back to the Hurricane and the 303s (and Emil's 7.7mm) for a moment: it only takes one of these bullets to puncture a radiator or kill a pilot. At close range against a lightly armoured target, with eight 303s you have a better chance of hitting the pilot or radiator than with six .50 cals. Where they will suffer most will be against heavily armoured planes with radial engines.

I know MP get fed up with planes streaming fuel and coolant fighting on, but in the Battle of Britain any plane with a punctured radiator (or in the Emils' cases possibly both), or leaking fuel tank, was getting not back to base, unless it was an RAF plane directly over it's base. A very large proportion - how large I do not know - of GAF losses were B kills in US terms: after damage they were in control but they could not fly the 100-150 miles they needed to RTB. (And for most of that distance a bailed out or force landed crew would drown or be captured).

October 18, 2020

2 hours ago, KW_1979 said:

Circling back to the small set of data we're arguing over - this doesn't pass the sniff test to me (I'm not trying to suggest any dishonesty or anything like that - just that's its not a representative sample). As I already pointed out, filtering out unsuccessful sorties is already artificially selecting best case outcomes. For another, in my extensive DM testing in 4.006 using static targets, the only way to get kills in that low of a round count was either to hit the pilot or to put all the shots into the engine. That was from a static gun shooting at a static target - not representative of the kind of hit spread you'd seeing in actual flying.

So I dug into my own stats going back a few weeks. Only counting P-51 and P-47 sorties, and discounting any where I did some strafing (and thus the hit number is uncertain), and counting in numbers from the other pilots who either assisted my kills or who I assisted I got these results:

33 blue aircraft shot down, by 2947 hits for an average of 89.3 hits per kill. If I exclude any multiengine aircraft (there were a fair number of 110s, 262s, and Ju88s in that set) it was 20 kills from 1407 hits for an average of 70.35 hits per kill.

But that's a small sample size and mostly coming from one pilot. So I decided to pull Krupinskii's stats for this month - one of the top pilots in the game, lots of dogfighting - should be a really optimistic setup to get low round per kill totals. I did the same thing where I traced all the airplanes he shot, as well as anyone assisting him, and I just ignored any extra rounds coming in from Spits, Tempests, P-38s, flak and in one case some friendly fire from a 190. So this is a really optimistic set of data that includes some 20mm and heavy flak which I can't account for exactly. 3693 hits for 56 kills - 65.9 rounds per kill.

So if we split the difference between my data and [DBS]Browning, the average hits per kill is about 50 - and if the average pilot is hitting at 5% (which may be optimistic based on his numbers) - then it takes 12.8 seconds of gunfire on average to score a kill with 6 x .50s in game right now. And I think that's a really optimistic number.

It would be great if you and @[DBS]Browning could share and/or merge your data sets, at least to see if the methodology is consistent.

I would point out, though, that it is entirely natural that you will get some very low numbers in any large set: precisely because hitting the pilot is possible from most angles. In fact, if hit effects are independent (which was the assumption used by 8th AF Operations Research in recommending the use of all API ammunition) the most common number of hits to down an enemy fighter would have been 1. Ie more planes downed by 1 hit than by 2 hits, by 2 than 3 etc.

My caveat is that in practice, because you are firing bursts from several guns you do not score only one hit at a time, but even allowing for this you should expect to see kills at numbers throughout the whole range from a fairly low level. Firing in groups of hits will also distort the average number needed to actually down the aircraft, by roughly half the group size. Ie if you always get 10 hits at a time, and get a kill after 70 hits in 10 groups, your best guess of the actual hits to kill is not 70, but ~65 If you can see exactly which bullet caused the kill then this is not an issue.

The average number is not the most common number - far from it. It is also likely to be considerably higher than the number required to shoot down half the targets.

If the average hits/kill is 50, the p of A kill on 109s/190s is 0.020 - which is a bit higher than the US report A kill for .50 cal hits to P-47s at 0.017 (ie , 59 average) which would have been mostly PKs. I would see that as a plausible outcome.

If 70, p to A kill 0.014 I agree that looks too low.

Ideally someone can retest the P-47 as a reference point, since it is the only one we have.

Edited October 18, 2020 by unreasonable

October 18, 2020

1 hour ago, unreasonable said:

I would point out, though, that it is entirely natural that you will get some very low numbers in any large set: precisely because hitting the pilot is possible from most angles.

Ideally, in an attempt to evaluate DM issues, any sortie with a pilot kill won't count.

Mike

October 18, 2020

1 hour ago, unreasonable said:

Ideally someone can retest the P-47 as a reference point, since it is the only one we have.

Shooting into the nose from low 12 o'clock (which I believe was the configuration they were using for that one) I averaged 24.3 rds of .50 over 10 passes to either knock out the P-47's engine or set it on fire - no other kill mechanism was apparent with this shot, which makes sense. Data set was: 41, 24, 8, 30, 12, 9, 23, 23, 30, 43. This was far and away the highest average of the BoBP fighters I tested this way, most being around 15. I'm pretty sure that the US vulnerability study was just looking at the pK of a single round from this angle though, and not considering compound damage from multiple rounds. In that light, the chance of a single round knocking down an undamaged P-47 from that angle has to be real close to zero.

The other portions of that study that I mimicked were P-38 & B-25 fuel tanks, (32.4 and 39.3 rds on average to ignite, which seemed way low compared to their compound damage numbers, but of course theres the AP vs API issue) and Me262 jet engines. I figured the 262 was close enough to a Bell jet engine to at least make an interesting comparison - 9.7 rounds on average from the from and 6 from the rear for a fire or engine kill - once again higher they what they got, but its a different engine ultimately, and the AP vs API issue is likely a factor again.

October 18, 2020

3 hours ago, SAS_Storebror said:

Ideally, in an attempt to evaluate DM issues, any sortie with a pilot kill won't count.

Mike

For examining issues with individual components of the DM - lift/drag penalties, setting fire to fuel leaks etc - I am convinced that damage specific tests are more likely to provide convincing data. In my P-47 vs 20mm cases I was able to identify the specific locations of hits and the p of the plane crashing as a result. (Much too high!) But that required running the same highly controlled test 100+ times. I am not sure if you can extract that level of detail from actual sortie logs, MP or SP.

The Pilot's armour and aspect vulnerability is part of the overall DM, and PK/PWs were and should be a high proportion of single seat fighter A kills. So if you exclude PKs the overall probability of an A kill would be much lower.

3 hours ago, KW_1979 said:

Shooting into the nose from low 12 o'clock (which I believe was the configuration they were using for that one) I averaged 24.3 rds of .50 over 10 passes to either knock out the P-47's engine or set it on fire - no other kill mechanism was apparent with this shot, which makes sense. Data set was: 41, 24, 8, 30, 12, 9, 23, 23, 30, 43. This was far and away the highest average of the BoBP fighters I tested this way, most being around 15. I'm pretty sure that the US vulnerability study was just looking at the pK of a single round from this angle though, and not considering compound damage from multiple rounds. In that light, the chance of a single round knocking down an undamaged P-47 from that angle has to be real close to zero.

The other portions of that study that I mimicked were P-38 & B-25 fuel tanks, (32.4 and 39.3 rds on average to ignite, which seemed way low compared to their compound damage numbers, but of course theres the AP vs API issue) and Me262 jet engines. I figured the 262 was close enough to a Bell jet engine to at least make an interesting comparison - 9.7 rounds on average from the from and 6 from the rear for a fire or engine kill - once again higher they what they got, but its a different engine ultimately, and the AP vs API issue is likely a factor again.

500m, 20 degrees below and 20 degrees off - close to 12 low, but PKs were certainly possible - actually the single most likely cause of an A kill. If you are dead on 12 low then you would not be able to get at the pilot. The A kill probability from their angle was 0.017 - but a lot of that was the pilot.

The US study considered both single hits and compound hits - but it said, essentially, that compound hits were only relevant, in the case of relatively small calibres like .50 cals, for fuel tank hits, and even there the detailed studies of the effect on further hits on previously holed fuel tanks suggested that .50 cal API hits were likely to cause fires, but not sufficient after one fire to cause loss of the aircraft. In short - it is complicated, but we have to start somewhere. They did not consider a more qualitative loss of combat effectiveness from damage - I assume because they had no way of quantifying that.

My take on the report is that the writers thought that the overall effect of compound hits was sufficiently small, that it would not effect any practical recommendations on calibre etc, which is with ultimate purpose of the report, so they were happy to leave it out of most sections.

The figures in the main table in the report are first derived from firings in which specific components were hit and assessed: then they derived the overall probabilities, by taking the probability of a kill from a hit on a component, divided by the proportion of the shots from the given angle that would hit the component, based on the the aspect. So we could in principle reconstruct a probability table for any angle given the aspect of the components (known) and the effect of hits on each component (modified from the base case where necessary).

So the table gives an A kill probability due to a hit to the engine of 0.001 (some more decimals would help!) which is the product of a the relative projected area of the engine from the specified angle and the p of an A kill if the engine is hit. From their diagram and my own playing around with the viewer I estimated the relative projected engine area at 19% for that angle, including props since I was originally looking at HE including splinter damage): it would be much more the closer to 12 o'clock you get. So the probability of a hit to the engine producing an A kill would be 0.001/0.19 = 0.0053 (given qualification about rounding).

From your data the game's .50 call ball is trashing P-47 engines much faster than the USBR thought API-T M20 would. The USBR B kill for the same shot was rated at 0.011 (includes A kill of 0.001) This would give you overall kills much closer to your figures, but they would be fail to RTB kills, not immediately crashing with engine failure or in a satisfying fireball.

I have to say this does not surprise me - what I found when I tested HE 20mm against P-47 in a long dead DM version is that it was about twice as effective in creating A kills as the USBR thought it was. (For which I got roundly abused by one of the worst headbangers of the mineshell lobby).

Generally in this game, I suspect weapons are not ineffective measured against historical reference points, they are too effective, especially in the short term - even before including issues like jams and duds which are not modelled at all. I can understand why this should be so: people want to have fun blowing things up. Seeing a plane limp off leaking various fluids and smokes is no more satisfying to us, even if in fact it never RTBs, than it was to USAF pilots who wanted confirmed Ace status, so that they could score a lot of chicks in Piccadilly before they went home to Dullville, Idaho.

Edited October 18, 2020 by unreasonable

October 18, 2020

23 minutes ago, unreasonable said:

For examining issues with individual components of the DM - lift/drag penalties, setting fire to fuel leaks etc - I am convinced that damage specific tests are more likely to provide convincing data. In my P-47 vs 20mm cases I was able to identify the specific locations of hits and the p of the plane crashing as a result. (Much too high!) But that required running the same highly controlled test 100+ times. I am not sure if you can extract that level of detail from actual sortie logs, MP or SP.

The Pilot's armour and aspect vulnerability is part of the overall DM, and PK/PWs were and should be a high proportion of single seat fighter A kills. So if you exclude PKs the overall probability of an A kill would be much lower.

I hear what you say and I see your point.

The reason why I was suggesting to take out pilot kill sorties is that in case of the damage caused by cal .50s, my bums feeling is that they do way too little damage to planes, yet they are quite capable in killing pilots as long as you hit the cockpit section from the "right" angle, read: Lightly armoured parts inbetween only, no armour plate like the dreaded six-o-clock attack agains 109s.

Wading through the server logs I see that when I fly a P-47 and engage enemy fighters, almost 90% of the kills I score are pilot kills against the enemy aircraft, and that seems odd to me.

Either it's too easy to kill the pilot (doesn't feel like) or too hard to cause any other fatal damage to other planes (that's what it feels like) with cal 50s.

The latter you won't see in your numbers if you include the pilot kills I'm afraid.

Mike

October 18, 2020

I also take your point - I will have to think about it for a while. Given the potential issues with fuel tanks and our propensity to keep firing until the target is obviously dead, you may well be right.

Need a lie down to ponder this. Edit probably just confusing myself at the moment so edited out until a little clearer....

Edited October 18, 2020 by unreasonable

October 18, 2020

Here is the HE ammo in a p47. It's rather good at starting fires. A bit too good IMO.

October 18, 2020

Is that the 13mm He?

October 18, 2020

20 minutes ago, HR_Zunzun said:

Is that the 13mm He?

I think it's the 12.7 HE

The 13mm HE has similar results

What's interesting, is that in my 3 flights with the HE ammo in 50 cals, the kills were mostly from fire. It's almost as if they've stumbled upon a good way of simulating incendiary rounds

Edited October 18, 2020 by 71st_AH_Barnacles

October 19, 2020

On 10/9/2020 at 2:34 AM, Cpt_Siddy said:

There is sooo many people in here who have never fired a gun, let alone 50 cal, in this thread....

There are also a lot of military vets and gun nuts in here. Those of us across the pond generally have a lot more experience with firearms than anyone in Europe. We have a gun culture that Europeans in general do not.

On topic:

It appears that most of us agree that HE damage is overblown in both the 20mm and 13mm MG's. The .50s need a tweak and most of all the damage model needs more depth.

/thread

Edited October 19, 2020 by ACG_Smokejumper

.50 cal damage, or lack there of

Recommended Posts

Tempus

unreasonable

CrazyhorseB34

SAS_Storebror

BCI-Nazgul

CountZero

RedKestrel

HR_Zunzun

unreasonable

-SF-Disarray

357th_KW

[DBS]Browning

CrazyhorseB34

CountZero

unreasonable

unreasonable

HR_Zunzun

[DBS]Browning

unreasonable

HR_Zunzun

[DBS]Browning

unreasonable

HR_Zunzun

BCI-Nazgul

unreasonable

HR_Zunzun

SAS_Storebror

BCI-Nazgul

357th_KW

unreasonable

unreasonable

SAS_Storebror

357th_KW

unreasonable

SAS_Storebror

unreasonable

Barnacles

HR_Zunzun

Barnacles

ACG_Smokejumper

Browse All

Activity