PDA

View Full Version : M1A2 gunnery



kbluck
29 Apr 03, 16:32
I'm becoming more and more curious about how the game resolves fire. You mentioned the game is using MILES standards, but based on my results they sure do seem pessimistic.

The final straw prompting me to write was just now. I had a tank platoon in ambush of a gaggle of marching antitank units. The tanks were lurking behind a ridge line, and pulled forward about 100m up to the crest to attack. 4 stopped M1s with side and rear shots on 6 BRDMs and 3 MTLBs, clear red LOS to all targets (I checked using both Info and LOS fan), range about 2000m. All four tanks were firing.

I would expect those poor enemy saps to be toast in about 20 seconds. Literally. In 20 seconds, an M1 should get off about 3 well-aimed shots. (Shot#1 at 0 seconds.) 12 shots, 9 targets. Surely a 75% kill rate is not too optimistic in this situation.

Nope. The engagement went on over 90 seconds, and by that time two BRDMs managed to pull out of range still alive. Even so, a platoon of America's most modern tanks only managed to score seven kills on a line of sitting ducks after at least 40 shots? That flies in the face of both training gunnery tables and real world experience.

Part of the problem is overkill, i.e. all four tanks would fire at the same target. However, at least a couple of times, I saw all the tanks fire with no kills to show for it.

It just doesn't seem right to me.

--- Kevin

Pat Proctor
29 Apr 03, 19:08
Were the targets suppressed? If they were not, then you did not hit them. We are refining the pH algorithm (percent hit, which is seperate from the pK algorithm, percent Kill). pK is based on MILES II, but the pH is based on a host of factors, including movement and posture of units, illumination, suppression, and virtually every other factor in the game.

I can not really comment on what may have prevented you from getting a kill on every shot. I am answering this posting away from a development platform, so I don't know the pK for a flank shot on a BRDM or MTLB. You can check this yourself by using "Vehicle Specs" from the Help menu.

I would make a few comments.

1) The pH algorithm is being "tweaked", though mostly to correct issues with small arms and dismounts, and aircraft vulnerability. The tank fire pH is the baseline for the entire algorithm, so we are relatively confident in it. It has "served with distinction since the first "BCT: Brigade Combat Team" shipped (on 3.5" floppies!, 6 years ago).

2) I can not make an assessment of what was a factor in your particular incident without a saved game file (I can give you an e-mail, if you have the situation saved as a save game state and can provide a short blurb about what you observe where).

3) 20 seconds for three shots is not a realistic standard in a combat environment for the average M1A2 crew. On a static range or even a table 8, this may be achievable, but when a tank platoon is firing and maneuvering, executing berm drills, and, oh yeah, being shot back at, we stand by the 15-20 second per shot figure in the database. Some tank crews can exceed this standard, some can not. We settle on this as an average.

4) There is a certain level of abstraction in any simulation of real life, whether it is an engineering program or a computer wargame. In ATF, you will be able to see example of unrealistic results, such as that dismount team that just won't die, or the tank that just can't be killed with artillery. These are as much functions of the random number generator in your computer as functions of the program. But the RESULTS at the macro level of your tactics against the tactics of the enemy AI are, in our opinion, rock solid, and the simulation, when examined at the hierarchy level, is as far as we have been able to ascertain, as or more realistic than any defense department simulation on which real life and death decisions are made.

Deltapooh
29 Apr 03, 20:10
I would like to see some improvements in this department. I once brought an entire company to bear on just two T-72S tanks, and took heavy casualties. (I believe they fired missiles though I surpressed them, or thought I did.)

On another note, I have to commend you kbluck. Your analysis seems to be very detailed. You're noting things I ignore because I just too busy dying and all. :D

kbluck
29 Apr 03, 20:15
so I don't know the pK for a flank shot on a BRDM or MTLB. You can check this yourself by using "Vehicle Specs" from the Help menu.

100% from any angle using 120mm. This indicates to me that the targets were not being hit at all.



we stand by the 15-20 second per shot figure in the database.

OK, you're right, I meant 30 seconds. However, just to nitpick, I don't think reload time should be simply added, since the gunner can continue to aim while the gun is being loaded. On the other hand, certain Soviet tanks with autoloaders do restrict the gunner from aiming during loading. I guess this comes back to the "What is the FDC doing while missions are firing" question. If its all an abstraction, fine, but why arbitrarily split up the fire time factors in that case? Just have a single time to fire.


The tank fire pH is the baseline for the entire algorithm, so we are relatively confident in it. It has "served with distinction since the first "BCT: Brigade Combat Team" shipped (on 3.5" floppies!, 6 years ago).

I still find it hard to believe that a reasonably proficient M1A2 platoon, firing at plainly visible targets, in daylight, broadside, from a static position, taking no return or indirect fire, with no smoke or other concealment, would fail to destroy the entire formation in short order.


I can not make an assessment of what was a factor in your particular incident without a saved game file

OK, I respun a similar situation using Death Valley Defense. The tanks in question are just north of PL Martin and should open fire on the AT column in front of them as soon as you unpause. If you check LOS you'll find all 4 tanks have an excellent view of the engagement area. (Ignore all the other goofy action... I neglected all of it while fast-forwarding to the interesting part.)

One thing that is confusing the issue is that they are firing with .50 cal as well as main gun. On a typical run just now, depsite every tank starting with a clear view of every target, it took 30 seconds for them to reach out with 120mm and actually kill something. Another is the overkill --- they're definitely ganging up on single targets. That certainly reduces their effectiveness quite a bit. I guess the PL was absent the day they taught the class on "depth of fire" in AOBC. ;) But seriously, I think the algorithm of "fire any weapon capable of damaging the target when in range" is a bit unrealistic. The TC should be spotting targets and controlling fire in this situation, not fooling with the .50 cal.

At any rate, targets consistently escape the ambush when I run it. :mad:

Thanks,

--- Kevin

kbluck
29 Apr 03, 20:25
On a side note, in the same game, can you give me an opinion on why my fire teams are so ineffective? On their sweep through the OPFOR dismounts in the center there, they failed to kill anything, while the OPFOR killed several of them. I tried putting them to ground when receiving fire, but even so OPFOR kept on getting them while not even being suppressed by my guys, even when I had 5 or 6 fire teams hosing down a single target. I also tried using a "support" force on "Suppress" without any better effect. (All the red Xs you see are from artillery fires.)

Thanks,

--- Kevin

Pat Proctor
29 Apr 03, 22:51
First of all, sorry if I came off as overly defensive in my last post. I re-read what I wrote and it came off pretty assholish. Sorry. I just had a jump today that rang my bell pretty hard. I'm a bit crotchety.

I will personally debug the algorithm again. There may be some error or exageration of the moving target speed-flank calculation. I will check it out.

For your dismounts... As I said, we ARE adjusting the algorithm for dismounts AND small arms. Small arms are too hard to hit with and dismounts are too easy to see, too hard to kill. This has been fixed and will be posted with the 1.02 update (its coming, its coming:eek: )

But there are some things that are not going to change, update or not. Dismounts, in the open, walking up on enemy dismounts, dug in with overhead cover are going to die. Unless you have overwelming odds (ala D-Day) in your favor, you are in trouble. Ask Pickett.

If you are going to use dismounted infantry to attack a dug in position, you are going to have to:

1) use suppression, either direct or indirect fire.
2) gain fire superiority
3) and not assault from the front (view range decreases as you move farther from the facing of the unit).

This is just as true in real infantry operations as it is in ATF.

kbluck
30 Apr 03, 00:46
I will personally debug the algorithm again. There may be some error or exageration of the moving target speed-flank calculation. I will check it out.

The more I replay it, the more convinced I become that the problem isn't with the pH or pK algorithms as such. When the 120s do fire, they seem to kill pretty consistently.

I think the problem is that the 120s aren't firing optimally.

I reviewed the spot logs carefully, and noticed a recurring pattern. The main guns only fired about every 30 seconds. This seems slow, since by my understanding of the Vehicle specs the load and fire time for the 120mm adds up to 15 sec. It doesn't seem to be a spotting problem, since the .50 cals keep up a rather steady stream of fire.

So, if the tank crews have plenty of targets and can fire the main gun every 15 seconds, the question is: why don't they?

The other big issue is overkill. A major drag on unit fire effectiveness. When all four units in a platoon fire at the same, almost-guaranteed kill, that's two or three wasted shots.

The obvious solution is to have some sort of coordination happening between vehicles to minimize overkill. Now, I don't know what is easy or hard for you in your system, but perhaps it would be easier, instead of messing with the target selection algorithm, to instead somehow stagger the fire times for each subunit in a platoon. I presume they already don't target destroyed vehicles, so having them fire at different times would have much the same effect; if a targeted vehicle goes tits up before the unit fires, I presume it would automatically pick another. Perhaps something as simple as a random modifier to the ready-to-fire time, i.e. 15 seconds +- 2 seconds, for example?

Thanks,

--- Kevin

Pat Proctor
30 Apr 03, 01:18
While it appears simultaneous, the tanks are taking turns firing at targets. So, if it appears that two vehicles both kill one target, what in fact has happened is that the first shooter shot and did not kill the vehicle and the second vehicle got it.

The tanks are firing 50 cal while they wait to load the next round. Sometimes sounds step on each other and override eachother, as more than one type of weapon fires. What may be happening is two tanks are waiting to load, and firing 50 cal in the interim. The pK is low and/or they are either not hitting (problem with pH algorithm and small arms that I described before). Then a tank loads a main gun round and fires the tank round and kills the vehicle. Since this is the last sound requested, it stomps out the MG fire. The rendering of all three shots happens simulatneously. The result: It looks like all three tanks fired main gun rounds and all three killed the target.

kbluck
30 Apr 03, 12:49
While it appears simultaneous, the tanks are taking turns firing at targets. So, if it appears that two vehicles both kill one target, what in fact has happened is that the first shooter shot and did not kill the vehicle and the second vehicle got it.

Well, that's good to hear. Appearances can be deceiving.

I was looking to the spot report log to try and figure out exactly what was going on. It seemed awfully sparse on the firing. I noted that 3 tank didn't seem to be firing at all, according to the log. I ran an info dialog on it during the entire engagement, and noted that it had clear LOS. Then I noted the ammo counters were ticking down. So, it appears that the spot log doesn't log everything. As an example, the spot log only shows a total of 8 main gun shots over the entire engagement period of 2.5 minutes, of which 5 were hits. However, at least two more targets were destroyed and not logged, and presumably a number of other missed shots went out and weren't logged. Given a ROF of 15 seconds --- oh, heck, lets say 20 seconds --- I'd expect a tank platoon to pump out as many as 32 main gun shots during that time period.

Also, I noticed on my other computer (the one that won't show the options dialog) that sometimes it doesn't log anything at all.
And, of course, even at best its really hard to read because its so narrow that you have to move the horizontal scrollbar back and forth constantly to read the whole line, because you can't (ahem) resize the window. :D

So, in addition to the spot log discrepancies, now I'm back to the question of why these dang M1 gunners are missing so much.

Thanks,

--- Kevin

Pat Proctor
30 Apr 03, 20:13
I am re-evaluating the ROF. The Mortar investigation has covered a problem when alot is happening at once. Lagging does not "take up the slack" and execute multiple events when a long time (which in ATF is a couple of seconds) between turns (times the unit is cycled through in the game loop.

This will probably fix a lot of the anomolous ROF things you have seen.

The event log ommissions are intentional. If every shot fired was logged, the log would cycle through max entries and events would drop off every minute. The log only records the most important thing that happens every turn (turns are transparent to the user, but are, at a minimum, one second long, but vary in length through out the game, due to variable processing speeds.

kbluck
01 May 03, 03:47
This will probably fix a lot of the anomolous ROF things you have seen.

What's the ROF on your boot squishing bugs? :D



If every shot fired was logged, the log would cycle through max entries and events would drop off every minute.

Can we get a "log to file" option that would include everything?

Ideally, the spot log would have some sort of filter for display purposes. It would be nice to have "high interest" type events jump out, like low/out of ammo or aborting mission. Right now, they'd be lost in a sea of "yeah, yeah, you're shooting, yay" events.

Thanks,

--- Kevin