German tank vulnerability

[Battlefront.com]

With you to a point Steve, but a Veteran crew would know what the likely deviation would be if you over-estimated range and if in a defensive position, or using crude range finding techniques, know the rough range to target.

The issue is if the system is sensitive enough to respond to a particular adjustment to the degree necessary. As stated above, at some ranges it appears the gunner would not want to take the gamble on this. At close range perhaps he would, but at close range he's got other factors pressing in on him interfering with making a perfect shot.

I cannot remember where I read it but analysis of KO'd tanks in WWII and post-war engagements showed 2/3rds of the rounds struck the turret because gunners were prone to over-estimating the range.

And where did the other 1/3rd of the rounds go? A combination of over the turret, somewhere else on the hull, and short into the ground? I find that very hard to believe. I can believe that 2/3rds of the hits were registered in the turret area because shorts had more of a chance of hitting ground for center mass shots.

Finally found the good link about German tank optics I'd been hunting for:

http://www.weaponsofwwii.com/forum/viewtopic.php?t=2296

A good overview, but I don't think that PDF is really all that interesting as it is mostly a "how to guide" for WWIIOnline. I'm not saying that it's all in error, but guides for games tend to focus on what works within the game instead of what really worked in real life.

Whatever the case is, the evidence that center mass is the better shot in theory and was the better way in real war is pretty strong. Occasional situations that favor more careful aiming are definitely not being disputed. But those situations are, as I have said many times now, very specific and perhaps largely not really applicable to CM as it is played.

Steve

[JoMac]

Steve,

I also remember reading that 2/3 of all hits were to the Turret ( the other 1/3 to the upper & lower Hull...mostly upper ). This was because armor crews overestimating ranges as standard practice.

In regards to the PZIV H/J ( and other armor ), it may have a 50mm Turret face ( about 50% turret front ), but a 80mm curved Turret Mantlet ( other 50% turret front ).

This means you will have a higher survivability chance in CM:BN then you do in CMx1, and especially so if you go hull down.

Joe

[stikkypixie]

Steve,

I also remember reading that 2/3 of all hits were to the Turret ( the other 1/3 to the upper & lower Hull...mostly upper ). This was because armor crews overestimating ranges as standard practice.

In regards to the PZIV H/J ( and other armor ), it may have a 50mm Turret face ( about 50% turret front ), but a 80mm curved Turret Mantlet ( other 50% turret front ).

This means you will have a higher survivability chance in CM:BN then you do in CMx1, and especially so if you go hull down.

Joe

Wouldn't the fact that the turret is the most exposed part of the tank not account for a large part of the 2/3 (assuming that number is correct) more than some sort of overestimating ranges standard practice?

[akd]

Steve,

I also remember reading that 2/3 of all hits were to the Turret ( the other 1/3 to the upper & lower Hull...mostly upper ). This was because armor crews overestimating ranges as standard practice.

In regards to the PZIV H/J ( and other armor ), it may have a 50mm Turret face ( about 50% turret front ), but a 80mm curved Turret Mantlet ( other 50% turret front ).

This means you will have a higher survivability chance in CM:BN then you do in CMx1, and especially so if you go hull down.

Joe

I assume tanks generally seeking hull down positions would contribute to percentage of turret hits.

[JoMac]

Being hull-down may account for some of those 2/3 accounts ( lets say up to 25% of armor at most was in hull-down positions during an engagement at any given time...Not just Normandy, but other engagements during entire war ).

Also, point-blank engagements between 100-250 meters may also account for many of those other 2/3 hits ( turret hits at those ranges tend to occur more often by default...your gun is leveled out with the enemy's gun as a quick shoot out begins ).

So, in general, the above ( hull-down & point-blank ) along with the tendency for over-estimation will usually account for the reason why 2/3 hits are to the turret in all of the war.

[JasonC]

Magpie - you still haven't grokked my point.

It doesn't matter how much the shell has fallen from a laser line out of the barrel, because the gun isn't sighted along the barrel. It is sighted to the target and then the barrel elevated by an amount meant to compensate for the range. How closely that range needs to be estimated to still hit the target is the variable determining shot accuracy. And that varies with the height of the fall from the shell apex, not the total drop from an imaginary point 10 degrees above the horizon or whatever.

When the distance to the target is 1 second of flight time (assume 750 meters per second hereafter), then shell drops from the apex of its flight - at the halfway point of that flight - for half of the flight time. The shell is on the rising half of the parabola before that midpoint, and all the falling part of the parabola from the midpoint to the target. In half a second that fall is 4 feet. In other words, pointing the barrel above horizontal enough to rise 4 feet in half a second of flight, will hit the target, at that distance.

What happens if I think the distance to the target is 600 meters, elevate for 600 meters, but it is really 750 meters? Answer, at 600 meters, 300 meters of the time is falling from apex, the fall of the shot in that time is .4^2 = .16 * 16 feet of 2.56 feet, instead of .5^2 = .25 * 16 feet = 4 feet. I will miss in the vertical by only 1.44 feet, for an error in the range estimate of 150 meters (20% of actual). What if I estimated the range as 900 meters when it was really 750 meters? Now the fall time from apex is 0.6 seconds instead of 0.5 seconds, the height fall I've planned for is 0.36 * 16 = 5.76 feet while the actual is 4 feet - I miss by 1.76 feet in the vertical by being off by 150 meters in my range estimate.

If I'm a sniper aiming an a concealed man, missing in the vertical by 17 to 21 inches might matter, but if I'm shooting a cannon at an entire tank, it won't remotely matter. I can be quite incorrect about the estimated range, and I will still hit the target, because the change in height associated with the change in range from the estimated figure, is tiny at one second total flight time distances.

Now suppose instead the range is 1500 meters. I will of course still hit if I estimate that range correctly and elevate the barrel the correct amount. That correct amount is one that will put the top of the parabola 16 feet above the straight line laser path to the target. But now what happens if I thought the range was 1200 meters? Answer, I aim for 0.8 seconds of fall, which is .64 * 16 or 10.24 feet, and the shot will be low by 5.76 feet. It I estimate the range as 1800 meters, I'll aim for 1.2 seconds of fall, 1.44*16 feet of fall, which is 23.04 feet - and I'll miss in the vertical by 7.04 feet.

20% errors in estimated range lead to misses in the vertical of 1.5 to 2 feet at 750 meters distance, significantly less than half the size of the target. 20% errors in the estimated range lead to 6-7 foot misses in the vertical at 1500 meters distance. That is why I call the first "bowling" - I need to point correctly but the height variations are insignificant compared to the size of the target - and the second "golfing" - typical errors in the estimated range will let the shot sail clear over the target from an center of mass aimpoint, or strike the ground in front of that target. Like trying to get on a green, the distance estimate becomes a separate and indeed the controlling factor.

The increase in the error in the vertical goes as the square of the flight time because that is how the fall of shot is changing. But just as important, it remains well under half the absolute size of a vehicle size target at medium or "combat" ranges" of a single second of flight time, while being equal to or more than half that target size, for typical range estimation errors, at long ranges of 2 seconds of total flight time.

Notice, this is also why the effective "bowling vs. golfing" range for a prone man (or using cover etc) target is more like half a second rather than a full second. A man sized target can't tolerate a miss of 2 feet in the vertical aim point.

Naturally, range estimation also becomes more difficult as the range increases, so the shot difficulty at long range may grow somewhat faster than this. But it is the essential process causing most misses (at least against stationary targets - for moving ones, side to side errors in lead may also become significant, interact with range estimate errors etc).

[JoMac]

I have always wondered how ballistics vs. armor is handled in the real world ( and how it may be portrayed in CMx1 or CMx2 ). I am sure there have been many discussions over the years on these forums regarding this.

You hear reports of T34/76's or M4/75 Sherman's bouncing AP rounds off Tiger 1's frontal armor ( around 110 mm ) at short range.

Lets say for example a M4/75 Sherman w/standard APCBC shoots at the Tiger's upper Hull at 100-250 meters away at 0 degrees, it would normal bounce...right ?

When the AP round hits that Tiger, does it penetrate say 80% of that armor ( full value of its penetration, which is around 80-90 mm ) before breaking up, or does it penetrate far less say 10-25% of that armor before it breaks up ( leaving a 1 or 2" gash or dent in the armor ) ?

Basically, does the degree of armor thickness determine how much a round will penetrate ( including different round types, special armor, or angle of armor ) ?

Will that Sherman penetrate even less if say the armor is 150 mm or 200 mm etc, etc ( basically only denting the armor at best before breaking up ).

Joe

[YankeeDog]

I have always wondered how ballistics vs. armor is handled in the real world ( and how it may be portrayed in CMx1 or CMx2 ). I am sure there have been many discussions over the years on these forums regarding this.

You hear reports of T34/76's or M4/75 Sherman's bouncing AP rounds off Tiger 1's frontal armor ( around 110 mm ) at short range.

Lets say a M4/75 Sherman w/standard APCBC shoots at the Tiger's upper Hull at 100-250 meters away at say 0 degrees, it would normal bounce...right ?

When the AP round hits that Tiger, does it penetrate say 80% of that armor ( full value of its penetration, which is around 80-90 mm ) before breaking up, or does it penetrate far less say 10-25% of that armor before it breaks up ( basically leaving a 1 or 2" gash or dent in the armor ) ?

Does the degree of thickness determine how much a round will penetrate ? Will the above Sherman penetrate even less if say the armor is 150 mm or 200 mm etc, etc ( basically only denting the armor before breaking up ).

Joe

There are entire books written about this. The most accurate, short answer is "it depends."

Specifically, it depends on the exact details of the projectile, the composition of the armor, and the exact velocity and angle of the impact. Here are some of the possibilities. At one end of the spectrum, a projectile can ricochet or break up, and leave little or no gash/dent in the armor. At the other end, the projectile can punch a neat hole right through the armor, and enter the interior of the tank intact.

In between these two ends, there are all sorts of possibilities. The projectile may dig partway through the armor and stick there. In more oblique impacts, the projectile may dig or scrape along the plate, causing some penetration and damage, but then break up or ricochet. If it penetrates most, but not all of the way through the armor, there may be some deformation on the inside surface of the plate, and this may result in spalling, which can cause damage and casualties inside the tank even through penetration isn't technically achieved.

Generally speaking, though, with kinetic penetrators, the more the armor overmatches the projectile, the less the projectile will penetrate at all. So a projectile that can theoretically penetrate 50mm of RHA at a certain range and angle, doesn't necessarily stick 50mm of the way into a 100mm plate. In fact, it will probably penetrate substantially less than this, and may break up or ricochet. But it depends on details so you can't just assume that an armor that is 200% thicker than a projectile's max. penetration will "bounce" the projectile.

Details of the armor type and projectile composition can create different results, too. For example, Some German AFVs such as many PzIVs had face-hardened armor, which was manufactured in a way that created a harder layer of steel on the outer surface of the plate. This type of armor is more likely to shatter the nose of an incoming projectile, which will reduce penetration. But this layer of hardened steel was also more brittle itself, so it did not resist against other types of impacts as well as pure RHA. All steel armor plate is not created equal, either. For example, some Allied tanks had armor made of steel that was somewhat softer than RHA. This armor was less likely to spall or shatter, but was easier for projectiles to push through.

An then you have different shapes/compositions of AP shell/shot; blunt-nosed, sharp-nosed, tungsten-core, which all behave differently on impact.

Yeah; it's complicated. Be glad that BFC has done their research.

[MikeyD]

I recall back in CMBO days the conversation was fixated for awhile on the 'shatter' gap, that magical combination of range and hit angle that would cause US AP shot to shatter instead of penetrate. A little before it or a little after and penetration would occur. I recall Russian rounds had a grove cut around the round to keep cracks from propagating. Ah, the good old days when we had our resident ballistics expert visiting the site.

[JoMac]

Yankee Dog,

Thanx for that great info as it was most helpful indeed.

[Battlefront.com]

From JasonC,

20% errors in estimated range lead to misses in the vertical of 1.5 to 2 feet at 750 meters distance, significantly less than half the size of the target. 20% errors in the estimated range lead to 6-7 foot misses in the vertical at 1500 meters distance.

This is an excellent mathematical reason for aiming center mass. 20% range estimation might sound like a lot for an experienced crew to make, but in the heat of battle and in real combat conditions it is entirely possible. Even if they don't make a range error of that magnitude, when aiming at a target that is perhaps only 3 feet high, 10% error yields a pretty good chance of missing the target completely or striking the center mass. Compare this with a 10% or 20% error when aiming center mass. Results? Either a hit of the lower hull or the turret, guaranteed with all else being equal.

Now, this assumes that the entire frontage of the tank is exposed to fire, of course. If the target is only showing its turret, then that's all that can be aimed for. Therefore, the near sure hit of a center mass shot is now off the menu. Instead the shooter is forced into gambling on a turret front hit because that's all that's available. Which is why hull down positions are a significant advantage.

Which is exactly why the Germans put Panther turrets in the ground for fixed defenses, or dug in tanks like this Panther in Berlin:

Judging from the documentation of what these things did before being overcome, it's pretty clear that the theory of aiming for the turret is harder to implement in real life than might seem. These engagements were often very close quarters.

Steve

[ClarkWGriswold]

I think JasonC did a fantastic job of explaining ballistics in laymans terms and illustrating how much errors in distance estimation can affect a shot. Being an avid shooter, I really love this kind of stuff.

So, now that everyone more or less knows the basics of ballistics and shot trajectory, perhaps Steve can give us an idea of how pixeltruppen in the game spot and judge distance. I'm really curious how the AI handles this, especially if the tank is hull down or not fully visible (in trees, partially behind a building, etc). There was talk (controversy?) earlier about how being hull down is advantageous because there is simply less area to hit (even if the hull is has the thickest armor). In real life the advantages are multiple: Less area to be seen, less accurate range estimation if you are seen, and less area to hit if you are seen and properly ranged.

So, how does the computer "see" targets? Is it done via the number of pixels that are visible to the AI from its position? How does movement affect this? If a unit is seen, how does the AI determine range?

[Magpie_Oz]

I am not sure how to grokk something but when you explain it like you just did we are saying the same thing, just calculating it differently.

[ShakyJake]

One must grokk with deliberation, but never with inordinate haste.

[Wrath of Dagon]

Judging from the documentation of what these things did before being overcome, it's pretty clear that the theory of aiming for the turret is harder to implement in real life than might seem. These engagements were often very close quarters.

Steve

I think we're talking about aiming above the center of mass of the tank rather than aiming at the center of mass of the turret. Results would be quite different.

[Battlefront.com]

I think we're talking about aiming above the center of mass of the tank rather than aiming at the center of mass of the turret. Results would be quite different.

Not really since where you aim establishes your aim point no matter how you aim. Since the aim point is what determines your chances of a hit, it really doesn't matter how you get to that aim point. Look at that Panther in Berlin, for example. A gunner could simply aim for the street in front of the upper part of the hull and bump it up a bit to hit the turret. Or he could aim directly for the turret and bump it down a bit. No matter how you look at this, you aim for the turret you are more than likely to miss the entire tank than if you aim center mass. However you want to think of this, fine, but fundamentally there's no difference based on the method used to aim for the turret.

Steve

[Wrath of Dagon]

I think you misunderstand me. There's a difference between the center of the turret and a point a little above the center of the tank. So you wouldn't be aiming for the turret, you'd be aiming for some point below it, with the expectation that your chance of hitting the turret would be improved, while only reducing your overall chances by a small amount.

[Battlefront.com]

So, now that everyone more or less knows the basics of ballistics and shot trajectory, perhaps Steve can give us an idea of how pixeltruppen in the game spot and judge distance.

The way any computer simulation works, and CM is no different, is to take statistically certain data that is easily quantifiable, known, or otherwise accessible to the sim and use that to approximate a real world effect within the simulation.

The two methods for doing this are "design for effect" (where you determine the result you want then craft the equations to get it) or "engineered" (where you design the equations so they produce effects that you want). The two different approaches are inherently different, but all sims use some form of both. CMx2, however, does almost everything "engineered" and as little as possible "designed for effect". Range estimation is one of those things which is a mix of engineered and designed for effect.

In real life there is a complicated interplay of brain functions and, possibly, equipment involved in estimating range. We do not explicitly simulate any of those things because, well, that would be rather silly What we do is simulate where the eyes of a soldier are scanning, what affect terrain has on degrading that scan, how much soft factors (Morale, Suppression, Experience, etc.) interfere with scanning in general, and other things. The scans then tell the soldier what he can see in the first place, which is probably the most important element in the whole game since EVERYTHING flows from that.

When a soldier sees a potential target inherently the system knows, to within a partial meter, the distance between the two. What the sim has to do is take this perfect piece of information and customize it to suit the individual circumstances of that soldier looking at that one particular target. It uses similar factors as the scan I just described, but with certain basic assumptions made based on what equipment the soldier has available to him. Is he just using his eyes? Does have have binoculars? If there is a gunsight involved, what type? Or for modern, obviously, is there a laser range finder?

From there the system takes the basic quality of LOS from the soldier to the target, modifies it based on the equipment used, and then modifies that based on soft factors. Key amongst these is Experience. In the end the system spits out a value which is then used when calculating the aim point for the weapon. If the value is close to optimal (i.e. 100%), then the aim point is more likely to be correctly placed all else being good (inherent accuracy of the weapon, if either shooter or target moving, etc.) Which means the little pixel warrior might guess the exact range of the target PERFECTLY, but still send the round to the wrong location. Or vice versa, the soldier might put the crosshairs exactly where they should be if the target had been 100m closer, but because he estimated range wrong then he's aiming in the wrong spot. So on and so forth.

Things like Hull Down are a separate thing which the system knows about. So yes, when a shooter is looking to aim at a hull down tank it is aware that the tank is hull down and therefore aims appropriately. Or maybe doesn't even take the shot. The TacAI is programmed to not waste ammo under certain circumstances.

There was talk (controversy?) earlier about how being hull down is advantageous because there is simply less area to hit (even if the hull is has the thickest armor). In real life the advantages are multiple: Less area to be seen, less accurate range estimation if you are seen, and less area to hit if you are seen and properly ranged.

Yup, and all that is simulated. Look at the AAR going on now and see that The_Capt hasn't even seen several of Bil's tanks, despite taking fire from them. Why? Most likely because they are hull down tanks that have a fairly small turret profile (unlike a King Tiger, for example).

Steve

[Battlefront.com]

I think you misunderstand me. There's a difference between the center of the turret and a point a little above the center of the tank. So you wouldn't be aiming for the turret, you'd be aiming for some point below it, with the expectation that your chance of hitting the turret would be improved, while only reducing your overall chances by a small amount.

No, it's still the same thing. When you move the aim point up so that it is effectively pointed at the middle of the turret you are, effectively, aiming at the turret now. The chances of shooting a round over the top of the turret, or into the front hull, are identical to aiming at the turret right from the get go. In fact, I doubt any gunner would move the gun directly from off target to directly at the turret. Most likely all gunners, no matter what, would bring the weapon around so that it was aiming more-or-less center mass first because it can be done quickly, then fine tune adjust from that point. The difference between aiming center mass and aiming for the turret is what happens next.

For a center mass shot the gunner now purposefully optimizes his aim point for the center of the tank. If the gunner wants to try to hit the turret he nudges it towards the turret. No matter how you cut it as soon as the aim point is somewhere above "true" center mass, the chances of striking the turret increase, but so do the chances of missing the tank entirely. In theory a center mass aim point means no extra chance of missing the target.

Steve

[Her Klink]

Steve, will pixeltroopers fire at an enemy target that is obscured by a friendly unit? If so, does the aiming point change if the friendly unit gets out of the line of fine?

[ClarkWGriswold]

Thanks for the explanation on spotting, Steve. I'm curious, when you mention that you modify visual scanning for soft factors, do you take into account different optics and such. In other words, I've read many times that German optics were superior to U.S. optics because they used a more precise range finding method (the "triangle" sight versus the US's more simple crosshatched sight), had a wider field of view, etc. I could see this being simulated by simply saying if vehicle == American then chance to hit = chance to hit * .95 or something similar, but I guess it's also possible you do more in depth calculations.

When AI units are actually spotting, how do you account for things like hull down, cover, etc.? You mention that The_Capt's tanks may not even see Bil's tanks because only the turrets are exposed. How does the computer actually "see"? Is it based upon the actual size of the object that is unobstructed? Is there any factor for movement, color (compared to surrounding objects), etc.? What I'm picturing in my head is an AI unit looking directly at a tank 700m away in clear, open ground. The unit would have X% chance to spot the vehicle based on the size of the vehicle (the number of pixels or square area it occupies) compared to the size of the overall scanned area. Now, if you put a bush in front of the vehicle, chance to spot goes down because less of the object is actually visible. If the vehicle is moving, chance to spot goes up even if exposed area remains the same. If the vehicle is green and the bush is green, chance to spot goes down, but if the vehicle is brown maybe it goes up? If the vehicle fires, chance to spot goes up, but it perhaps goes up more if it doesn't have a flash suppressor or doesn't use smokeless powder. And so on.

Sorry if I'm asking you to explain a lot - and please don't feel obligated to do so - I'm just trying to get a feel for how you model what is essentially a human perception of the world around them. It seems to me to be quite a complex thing with many, many variables to account for. Lots of people can look "right at" a deer standing on the side of the road and not see it, but as soon as it moves "oh! There it is". If the deer were bright orange, most everyone would probably see it. How you simulate this type of thing is what I find interesting.

Hopefully I'm not straying too far from the original topic...

[YankeeDog]

I could see this being simulated by simply saying if vehicle == American then chance to hit = chance to hit * .95 or something similar, but I guess it's also possible you do more in depth calculations.

Yes; the calculations are much more in depth. To reiterate: There is no longer any "chance to hit" or "hit percentage" calculation in CMx2.

What the game does do is calculate the exact direction the gun barrel is pointing, and then ballistic trajectory of the projectile, based on a number of variables. If the resulting ballistic trajectory intersects an AFV, it hits. If not, it doesn't. But the game never makes a "to hit roll" and compares the result to a calculated "hit percentage", which is what CMx1 did.

While I am not speaking with any personal knowledge of the specifics of what is going on "under the hood," assuming that the game does take into account the rangefinding advantages of German gunsights, the way it would do this would be to assume, for example, that a gunner of a certain skill level looking through a German gunsight can estimate the range to a target, to within say, 10% of the actual range, and then assume that the American gunsight in a similar situation allows the gunner to estimate to within say, 20% of the actual range. This would, in turn, affect the exact elevation of the barrel, which would, in turn, affect the chance that the path of the projectile would intersect with the target.

So there is still a "die roll", In a way. In fact, there are probably many -- the "die roll" of how close the gunner's range estimation is to ideal, a "die roll" to determine where within the inherent shot scatter of the actual weapon the shot lies, etc. All these separate die rolls added up create a "to hit %" for any one unique in-game situation. But the game never actually calculates a "chance to hit," as such.

[ClarkWGriswold]

Yes; the calculations are much more in depth. To reiterate: There is no longer any "chance to hit" or "hit percentage" calculation in CMx2.

I'm not talking about a "chance to hit", I'm talking about a "chance to see". The game has some method of determining what an AI unit can and cannot see. That's what I'm trying to determine. How do you simulate eyes scanning terrain?

[Kalibri]

Question, if a tank placed at an angle to it's opponent and not just straight inline at him, in reality it should create higher chances of bouncing off the armor. Does this effect apply to our game?

P.S. I use this idea in Red Orchestra game and the game engine allows to apply angles.

[zukkov]

i can understand wanting to have an idea of how the system works, but you guys asking all the questions know full well you're going to set up tests for all this stuff once you get the game anyway. and that would be true even is steve gave you a complete breakdown of EXACTLY how it all works. :-p