Page 1 of 2

Artillery algorithms need a tweak

Posted: Tue Jan 03, 2023 1:51 am
by GloriousRuse
Ran into what seems like some extreme formula interactions, deeply in my favor, while playing End of the Road as NATO vs AI.

In short, German artillery is killing over one vehicle per round of basic HE.

In total, the nine gun M109 battery has fired 60x rounds in two "neutralization" missions.

Image

During that time they have killed 68x vehicles, mostly BMPs. To be fair, they did catch a large traffic jam in what I assume was "move hasty" road column on the roads, observed by scouts, but this seems like a bug or modelling flaw.

Image

The most devastating fire mission was the first one, which killed 1.5 vehicles per round fired.

Image

The second fire mission killed .76 vehicles per round fired, putting it on par with direct fire tank main gun at favorable ranges in FCSS.

I think that you may have various multipliers or "shots on target checks" stacking up in a manner that is causing artillery shells to go beyond precision guidance and into the super human.

Update: the mission finished with 98x vehicles killed for 180 rounds of HE, for a final floating average of .54 kills per round - including 18x more tanks. The single largest tank kill collection was another traffic jam, where a single 30 round mission killed most of that armor; this is reinforcing the idea that there is something off in how target density is affecting losses.

Re: Artillery algorithms need a tweak

Posted: Tue Jan 03, 2023 9:50 am
by WildCatNL
Do you have a save game (from prior the barrages) so I can try to reproduce the behavior?

Background: the game models each individual shells (and sub-munitions), and their interaction with the target. The fire mission is fired as several volleys over time, each volley being an sheaf with shell overlap. The sheaf is aimed at the largest concentration of targets (if spotted) with an error which again is reduced by spotting. Shells within the sheaf have their errors. Targets systems are either spread out over the hex or mimicking road formations (Move Hasty) with intervals.
A well placed 155mm HE shell is likely to impact multiple targets with its blast effect when targets are closely packed. As a player, you don't want to create traffic jams while being observed...

Re: Artillery algorithms need a tweak

Posted: Tue Jan 03, 2023 12:11 pm
by Searry
Killing a traffic jam with artillery is a classic in these games and I think it's pretty accurate, no need to change that.

Re: Artillery algorithms need a tweak

Posted: Tue Jan 03, 2023 8:59 pm
by byzantine1990
In my experience artillery does too much physical damage and not enough to readiness.

A company of tanks will be massacred in minutes but the last tank will have almost full readiness after half an hour of shelling.

Re: Artillery algorithms need a tweak

Posted: Tue Jan 03, 2023 11:55 pm
by Ichili
It seems to me that movement is never really simultaneous in this game. Like in a sort of chess play, at any given time only one unit is moving only one hex, while the other units remain momentarily in their places. By contrast, in the "real world" three units advancing in tandem along the same road and at the same speed would never "step" on each other. But in the game, as one of those units moves forward, it gets stacked with the one in front of it. At that precise moment, an artillery mission falling on them would do twice as much damage.
Another example: lets imagine that my WP opponent has three unobserved units behind the edge of a forest and waiting to move forward one hex in a coordinated fashion, exactly at the same minute. I have three NATO units defending. In the real world, the three units would emerge in front of my units simultaneously and receive my defensive fire at the same time. My units would have to divide their fire among the three targets. But in the context of the game, only one unit moves forward one hex, gets in sight of my units and immediately receives the full effects of defensive fire from my three units. As a result, the enemy unit is decimated. Now, the second unit moves forward, and once again, it also receives concentrated fire from my three units and gets decimated. The third unit moves forward in turn and the same outcome happens again.
That is what i meant when I said that movement is never really simultaneous in this game.

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 3:57 am
by kevinkins
I believe your observation describes the WEGO system in Flashpoint:

Flashpoint Campaigns: Southern Storm has an innovative WEGO asynchronous turn resolution system
---------
In general, asynchronous is an adjective describing objects or events that are not coordinated in time.
--------
The WEGO system in a tactical game like Combat Mission has the fire and movement of units occur across the battlefield at the same time. We would need input from the programmers as to why one is used rather than the other. I think the WEGO in Stalingrad is also asynchronous. It's not a tactical level game either. While it can be tedious, the CM player can avoid pile-ups with planning. Most fighting in CM occurs dismounted where only huge scenarios have long lines of transport. But you still get punished for piling up at choke points etc..

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 4:41 am
by GloriousRuse
For some odd reason it won't let me attach .Sav to PMs. If you have an email I can send it; the rest of the fire mission data should be in there well enough to create a rough sheaf replication in the worst case.

Should be a nine gun lazy W with the sheath laid in at ~39 degrees to the target road column (a BN of BMPs), I think CEP will be ~95x45 at the 14km or so, but your weaponeers will know that better than me.

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 5:36 am
by kevinkins
Try zipping the sav file and then attaching archive to the PM.

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 6:55 am
by TarkError
kevinkins wrote: Wed Jan 04, 2023 3:57 am I believe your observation describes the WEGO system in Flashpoint:

Flashpoint Campaigns: Southern Storm has an innovative WEGO asynchronous turn resolution system
---------
In general, asynchronous is an adjective describing objects or events that are not coordinated in time.
--------
The "async" is that the opposing sides don't have the same time allowances (e.g. side A gets to issue orders every 15 minutes vs. side B who can do it every 12 minutes--that's why it's called asynchronous); it does not correspond to Ichili's observation of sequential movement which is another topic at hand

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 1:27 pm
by Searry
What the WP player has to do and in some scenarios the NATO player as well is to set the delays properly so that traffic jams are minimized and when assaulting they are synced as best as possible so all the different units can support each other from as many hexes as possible. I don't know if the AI manages this since I only play multiplayer.

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 3:33 pm
by kevinkins
[/quote]
The "async" is that the opposing sides don't have the same time allowances (e.g. side A gets to issue orders every 15 minutes vs. side B who can do it every 12 minutes--that's why it's called asynchronous); it does not correspond to Ichili's observation of sequential movement which is another topic at hand
[/quote]

Thanks. There is more than one difference between the two WEGO systems - FP vs CM. Never really thought about it until now.

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 4:02 pm
by GloriousRuse
Another one, I was playing NATO in Steel Rain vs the AI. 6x gun M109 mission; I saw so many little explosions after one ten round "pulse" of a mission that I let the full thirty play out (no more visible explosions) ended the game and hopped to the Soviet TOC:

Image

Two companies had basically been evaporated by, to all appearances, the first ten rounds of artillery to fall on them. While I realize it is possible for the theoretical blast radius of a 155mm to cover multiple vehicles in a dense formation, and that some of these losses represent mobility and firepower kills, I am incredulous that virtually every round fired not only manages to hit multiple vehicles, but that the fragmentation effects on each are sufficient not just to serve as readiness hit, but inherently deadlining damage.

Re: Artillery algorithms need a tweak

Posted: Wed Jan 04, 2023 8:13 pm
by StasSche
I notices that the counter-battery fire is way overpowered: I played PBEM++ "A brief moment in time" and I lost 9 out of 12 and my opponent 16 out of 18 SPAs within just 5 fire missions.

Re: Artillery algorithms need a tweak

Posted: Thu Jan 05, 2023 2:43 pm
by Tcao
wait, so Counter Battery works?

for some reason I have never saw CB fires. All I can do is manually CB , reading the intelligence panel, find out the latest enemy artillery position estimation, then assign the neutralized fire mission into that hex

Re: Artillery algorithms need a tweak

Posted: Thu Jan 05, 2023 3:04 pm
by GloriousRuse
It does. And when it lands on SP Arty, it is almost always ruinously effective. Not so much on mortars for some reason, and towed guns are a mixed bag, but it's not unusual to see an entire battery crippled in one "pulse", let alone one mission.

If they really are laying sheaths of CEPs over the hex, I suspect that the battery positioning (300+m wide, 100+m deep), + CEP (35+m wide, 75+m deep) and theoretical bursting radius of 155/152 (50m circular) is coating most of the hex in theoretical "damage could happen here" zones...which is, indeed, the reason sheaths exist...and then treating each and every vehicle in it as "potentially dead, apply damage check", against which the thin skin armored vehicles just get fire hosed.

Re: Artillery algorithms need a tweak

Posted: Thu Jan 05, 2023 3:44 pm
by 22sec
I would also say compare the number of fallen out to dead units after the barrages. The inclusion of fallen out is unique to FC, and many think strictly in terms
If units dying or not. This is amazing for campaigns, or even two scenarios linked as a campaign (working on that at the moment).

Just a side note to the conversation.

Re: Artillery algorithms need a tweak

Posted: Fri Jan 06, 2023 12:23 am
by GloriousRuse
It's about three to one, but even going by K-kills this seems way high.

I believe those 7 kills happened in the opening 10 rounds, but even if it was 30...4.5 rounds per catastrophic kill is extraordinarily lethal.

Then we get into what constitutes fallen out/too damaged to continue fighting in the old soviet modeling. It certainly is possible for a near miss to track a BMP, or even send non catastrophic shrapnel into the turret at certain ranges and angles, but by and large something like losing a rangefinder or even the electric turret drive are unlikely to render one out of combat.

To your point, It may br a case that 30m out effects are being treated as mission kills overly generously.

Re: Artillery algorithms need a tweak

Posted: Fri Jan 06, 2023 1:52 pm
by cristianwj
StasSche wrote: Wed Jan 04, 2023 8:13 pm I notices that the counter-battery fire is way overpowered: I played PBEM++ "A brief moment in time" and I lost 9 out of 12 and my opponent 16 out of 18 SPAs within just 5 fire missions.
actually, my SPARTY vanished in 2 minutes, Lol
But it's because my SPARTY have conduct three missions and haven't relocated position, so I think the result could be acceptable.

Re: Artillery algorithms need a tweak

Posted: Fri Jan 06, 2023 6:29 pm
by byzantine1990
GloriousRuse wrote: Fri Jan 06, 2023 12:23 am It's about three to one, but even going by K-kills this seems way high.

I believe those 7 kills happened in the opening 10 rounds, but even if it was 30...4.5 rounds per catastrophic kill is extraordinarily lethal.

Then we get into what constitutes fallen out/too damaged to continue fighting in the old soviet modeling. It certainly is possible for a near miss to track a BMP, or even send non catastrophic shrapnel into the turret at certain ranges and angles, but by and large something like losing a rangefinder or even the electric turret drive are unlikely to render one out of combat.

To your point, It may br a case that 30m out effects are being treated as mission kills overly generously.
The argument that we are only tracking a BMP is kind of rediculous when the entire infantry squad is rendered combat ineffective regardless.

Again, I would be much happier if artillery was less lethal but much more damaging to readiness. Keep the slowing effects as well.

Re: Artillery algorithms need a tweak

Posted: Fri Jan 06, 2023 6:33 pm
by byzantine1990
GloriousRuse wrote: Fri Jan 06, 2023 12:23 am It's about three to one, but even going by K-kills this seems way high.

I believe those 7 kills happened in the opening 10 rounds, but even if it was 30...4.5 rounds per catastrophic kill is extraordinarily lethal.

Then we get into what constitutes fallen out/too damaged to continue fighting in the old soviet modeling. It certainly is possible for a near miss to track a BMP, or even send non catastrophic shrapnel into the turret at certain ranges and angles, but by and large something like losing a rangefinder or even the electric turret drive are unlikely to render one out of combat.

To your point, It may br a case that 30m out effects are being treated as mission kills overly generously.
The argument that we are only tracking a BMP is kind of rediculous when the entire infantry squad is rendered combat ineffective regardless.

Again, I would be much happier if artillery was less lethal but much more damaging to readiness. Keep the slowing effects as well.