Question about how accuracy is calculated

[roachbeef]

I've read in this forum that shells are not simulated individually, and that whether or not they hit is determined at the moment of firing.

When ship A is pursuing ship B in a stern chase and they have a closure rate of less than 0.05 mile per minute (or 88 yards per minute), it seems that the dispersion pattern looks something like this:

 

Assuming that Ship B is 200 yards in length, the shells would fall anywhere between 50 to 600 yards short or long. Range was 22-ish miles. I would say that the shell dispersion is up to 300 yards short or long of the target. Given that the two ships are basically at the same distance, and that Ship A has been firing 400 shells from a 16" Mk5 gun with maxed-out rangefinder, RDF, and radar, would that degree of continuous missing be something to be expected? Especially given that horizontal dispersion (perpendicular to the line of fire) seems to be extremely low. You would think that if wind and distance are constant and seas are calm that the gunners would eventually stop firing short or long, yet accuracy seems to be independent of the amount of shells fired.

 

Is the vertical dispersion supposed to be that much larger than the horizontal dispersion? I know it would be a bit larger but not to the extent that there's practically no horizontal dispersion and only vertical dispersion.

Edited January 31, 2020 by roachbeef
change image to increase clarity

[arkhangelsk]

You mean you fired 400 rounds (over 40 tries given a typical layout of 9 guns) with maxed out 16 inch guns and didn't get a single hit? This I'll have to see. One reason why everyone thinks the secondaries are pretty useless is that the modern guns can kill destroyers (get one hit) at long range with maybe 5 salvoes.

Are you sure you did the basics like keeping your ship balanced? What did the computer predict the hit rate was?

[disc]

I think that the odd dispersion pattern is a product of the modified shell trajectories. The computer has (probably) decided whether each shot will hit or miss from the moment of firing, so the visual shell is literally turned midair to hit or miss the target. This is why radical maneuvers are pretty ineffective even at long range. Possibly the strange pattern of overs and unders is an unintended consequence of this system.

Edited January 31, 2020 by disc

[roachbeef]

1 hour ago, arkhangelsk said:

You mean you fired 400 rounds (over 40 tries given a typical layout of 9 guns) with maxed out 16 inch guns and didn't get a single hit? This I'll have to see. One reason why everyone thinks the secondaries are pretty useless is that the modern guns can kill destroyers (get one hit) at long range with maybe 5 salvoes.

Are you sure you did the basics like keeping your ship balanced? What did the computer predict the hit rate was?

Sorry, the orange marks in the image as well as between the red vertical lines on the graph both represent hits. 

It was a 1940 Modern BC (the one with a clean deck); overall hit rate was a bit less than 10% per gun, and I removed secondaries and balanced the ship to increase stability.

Edited January 31, 2020 by roachbeef
additional comment

[roachbeef]

I conducted a test in custom battle. 2 ranging shots + 17 salvos from 2x3 16" guns from a 1940 Modern BC. Despite zooming out as much as possible, sometimes shots landed outside the field of view.

https://imgur.com/a/Z3ylyfl

Range was constant at 16.7 km (54790 ft). Using the tool at (https://www.omnicalculator.com/physics/trajectory-projectile-motion) and Wolfram Alpha you eventually figure out that initial angle of firing is roughly 0.171 radians or 9.86 degrees. Of course, this is assuming no air resistance but that was too much work.

From the above image, if we assume a roughly 150m ship we get about 1.8 km of vertical dispersion. To go short that would require 19 mils (milliradians) of vertical deviation at the barrel (assuming no horizontal deviation); too go long that would require... also about 19 mils of vertical deviation. And that's without considering the shots that fell outside the screen. Wouldn't normal warships get 2~5 mils of dispersion? Seems a bit off for 1940s tech. I'm not complaining about the accuracy itself; more about how it's being presented graphically.

Edited January 31, 2020 by roachbeef
calculations

[akd]

Remember that visual dispersion is used to represent both precision errors (pure dispersion) and accuracy errors.  You will never see a tight pattern land away from a target due to a bad target solution.  This makes it impossible to judge whether patterns are normal or not.

[roachbeef]

18 minutes ago, akd said:

You will never see a tight pattern land away from a target due to a bad target solution.

Given the constant range, bearing, and relative speed, low accuracy and high precision still seems possible: see below image and the close pattern of the four shots on the right. What you're stated might still be true, but I still believe that the way precision and accuracy are graphically represented seems a little skewed towards exaggerating vertical dispersion. Even if it was in the case in real life, the game seems to be exaggerating 200-600 yard dispersion to be up to 20 times worse (when you consider the shots that fell out of the 2500-yard field of view).

 

 

[akd]

Some sub-set of shots landing closer together can still be part of a poor precision pattern.  Again, there is no separate treatment of precision and accuracy in game, there is just “accuracy” with some accuracy factors tied to factors that would affect precision in real life (e.g. multiple gun in turret accuracy malus).  There is only greater or lesser accuracy, and this is visually represented by increasing or decreasing dispersion patterns.  This is also why stacking accuracy factors can lead to impossibly tight dispersion patterns.  The dispersion is fake, not an actual ballistic calculation.

Edited January 31, 2020 by akd