Sixth ArmyTraining Center: Empirical test of combat effects.

[springer]

Tweber's outstanding sixth army campaign, plus an unusual amount of free time that I have to spend by a computer, have fired up my enthusiasm for AT beyond its normal high levels.

Just for fun, I thought I would do an empirical analysis of Victor's excellent combat engine combined with the sixth army configuration.

This investigation was driven by a question I asked elsewhere about when it is best to just add more units and when it's best to upgrade. I wanted to see if I could discover any effects of experience and upgrades on the interaction of infantry and artillery. The simulations look only at infantry on infantry attacks, which are key elements of the sixth army campaign.

Because infantry survival is vital in the scenarios, I wanted to see what best helps the German infantry to survive. I also wanted to see what helped take out the maximum enemy matériel.

Here's what I would expect (essentially this is AT common sense, so expect nothing new from these assumptions):
Experience is most important:
A) Infantry with high experience will survive better than infantry with low experience
B) Infantry with high experience will destroy more matériel than low experience.
C) Support weapons (infantry guns and mortars) with high experience will destroy more matériel than than support weapons with low material

Upgrades:
A) Infantry II should survive better than infantry I due to the hit point increase.
B) Support weapons II should destroy more matériel than support weapons.

The critical questions I am asking from this test is
A) What are the most efficient combinations of infantry, support weapons, experience and upgrades that maximize infantry survival in the attack?

B)What are the most efficient combinations of infantry, support weapons, experience and upgrades that maximize the destruction of enemy matériel in the attack?

I'll outline the basic situation in the next post.

[springer]

The situation: The Sixth Army has a few more weeks before they have to start the '42 offensive operations. Intelligence has given the Germans fairly precise knowledge of the typical Soviet rifle unit they can expect to encounter. They now need to figure out the tactical configuration that is ideal as they prepare to refit for the campaign.

Setting up their forces in their ad-hoc training ground in Russia, the Germans play a series of wargames simulating a battle in which two German units strike at a group of three inexperienced Russian units. For each force configuration, five identical simulations are run and the results are analysed and their results are statistically combined.

After the five simulations, one variable- for example experience or upgrade level is changed, and five simulations are completed with the new configuration. After all the simulations have been run, the results are tallied.

The picture below shows the basic situation that the German trainers used as their test for all the battles, though experience and upgrade levels can change. (The Russian units also have a horse for the AT gun, that got cut out so the picture could be cropped nicely).

The simulation consists of a single coordinated attack by the German's.

The German HQ has high experience and the German units have low infantry numbers because this is the typical situation for the German '42 army in the campaign, manpower doesn't flow like it used to.

[springer]

The first test of the Sixth Army Training Center (SATC) was not a test of combined arms.  Rather, it addressed the concerns that the German regiments were being reorganized with substantially reduced infantry strength. I wanted to see under what condition starting a regiment with 20 Infantry rather than  40 was reasonable. For my first pass, I tested how 40 INF regiments were compared to 20 INF.

Here are some definitions:
A standard regiment is 20 INF, 1 Infantry Gun, 5 Mortars, and a Horse
An enhanced regiment is 40 INF, 1 Infantry Gun, 5 Mortars, and a Horse
Levels of Upgrade can be I or II
Experience can be low (20 for all units) or high (70 for infantry, 85 for support weapons)

The difference in experience is based on the Sixth Army's '41 campaign. Support weapons often survive long enough to become quite seasoned.  Infantry usually doesn't make it so far, unfortunately.  70 is really optimistic, but by careful combination of seasoned units, it is not infrequent.


Here's the configuration of the first two sets of wargames (five simulations each):
A. One standard regiment (20 INF). Both INF and SW have level I upgrades and low experience
B. One enhanced regiment (40 INF). Both INF and SW have level I upgrades and low experience

The simulations showed a reliable difference in infantry casualties: the enhanced division suffered lost 53% of the attacking infantry and the standard regiment suffered 70% of it's infantry.

This finding concerned the staff officers, but it was pointed out to them that in terms of absolute casualties, the standard regiment faired reliably better than the enhanced regiment (28 INF lost vs. 42 INF lost.)  In other words, the standard regiment lost 16 less infantry per two regiment attack than the enhanced units.

Worse yet, there was not reliable evidence that enhanced units substantially hurt the enemy worse than the standard units.  Yes, there was a difference between the enhanced units and standard units (62% vs. 47%) but because of the high variability in the standard units combat results, these differences were not reliable.

This gave the officers some food for thought.  Larger units may not destroy the enemy that much better than small units, but how many rounds can a unit go on losing 70% of it's infantry?  After two rounds of such losses, the unit will be operating at 10%  of it's original strength (1 or 2 INF).

[springer]

The next step was a first look into the effects of experience: how would 40 inexperienced INF compare to 20 highly experienced INF?

In the simulations, the was no reliable difference in % of infantry lost between the two groups, the 40 inexperienced units suffered 53% and the 20 inexperienced troops suffered 59% losses. In terms of absolute numbers the differences were reliable, the enhanced units (42 INF lost) still took many more casualties than the smaller units (24 INF lost).

It was also interesting that there was no reliable difference in destructive power of the two units: The enhanced, inexperienced unit destroyed 63% of the enemy's matériel and the standard, experienced unit destroyed 65%.

Of course, a 40 INF regiment with high XP would be a lethal weapon.  However, Barbarossa's casualties, particularly those KV-Is and T-34s laying in the way of Kiev had made such units look more like a fantasy of "what ifs". 

The simulation suggests that an enhanced regiment (40 INF)can attack once as an enhanced unit  After the first round, the unit is likely to suffer high infantry casualties and be reduced to being a low experience standard infantry regiment (~20 INF)

[springer]

A further analysis of 40 INF regiments vs. 20 INF regiments was performed. This time, the effect of upgrading the infantry was examined.

So, another set of comparisons were set up looking at the effect of a level II upgrade. 
A. One standard regiment (20 INF).  INF has level II upgrades and support weapons (SW) have level I.  Experience remains low.
B. One enhanced regiment (40 INF). INF has level II upgrades and support weapons (SW) have level I.  Experience remains low.

This simulation showed that the level II upgrade worked to the advantage of the larger units.  In terms of absolute losses, both the enhanced units (26 INF lost) and standard units (26 INF lost) had the same absolute casualty loss.  But percentage-wise, the loss was far less for the enhanced unit than the standard unit ( 33% vs. 65%). This means that an enhanced unit may be good for another such attack before it is reduced to standard unit levels.  In terms of destruction, there was no reliable difference between the enhanced unit and the standard unit in terms of the destruction of Russian power points (64% vs. 51%)

So enhanced (40 INF) highly experienced units may be the way to go, but that could substantially reduce the number of high XP units for the '42 campaign.

However, since small units were more likely to be the norm, the rest of the analysis looked at how the smaller regiments (20 INF) work with different combinations of experience and upgrade for infantry and and support weapons (SW).

Up to this point, here are the lessons learned:
1. Units too heavy in infantry and at level I, risk losing a larger number of these infantry.  This is no problem if there is an endless horde to recruit from, but it is something to think about when available manpower is at a premium.
2. If a unit is infantry heavy, it should have level II Infantry to keep down casualties.
3. When it comes to destruction of the enemy's matériel, the difference between enhanced and standard units (i.e., the number of INF in a unit beyond a minimal, essential amount to cover the SW) is small and highly variable.

[springer]

The main analysis, uses only the standard regiment (20 INF).

In it four aspects of each unit were manipulated with two variations yet.  They are:
Infantry Upgrade (I vs. II)
Infantry Experience (Low [20] vs. High [70])
Support weapon Upgrade (I vs. II)
Support weapon Experience (Low [20] vs. High [70])

All combinations were played (and each combination simulated five times)
Thus, there are 16 combinations all together, making a total of 80 simulations.

The combinations were separately examined in terms of cost to friendly infantry (German INF lost) and Soviet matériel destruction (% of Soviet power points destroyed)

[springer]

The analysis of what reduces German infantry casualties is unexciting. It was expected that infantry experience would play a huge role in infantry survival, and the results confirmed this:

The simulations showed that infantry levels mattered (since Level II gets 25% more hitpoints over Level 1)  : Attacks with level I infantry cost 26 INF in losses (64%), and attacks with level II infantry cost 22 INF (54%). 

Infantry experience was also critical. 20 XP infantry lost 27 INF (69%) and 70 XP lost only 20 (50%).

What was more unexpected by the staff was that artillery XP also had an effect on infantry survival, though it made sense once they thought about it: 20 XP SW lost 25 INF  (61%) and 85 XP SW lost 23 INF(57%).  The effect is small because it is indirect, but it appeared reliable.

By itself, SW upgrades did not save infantry.  Casualty rates were essentially identical for Level 1 and Level 2 SW across the simulations.  However, closer analysis showed that upgrading SW did have an effect, but only for Level II infantry. For level II INF, level I SW lost  21 INF (52%) and level II SW lost 19 INF (46%).  The difference is quite small, and there is no difference for level I INF (Both levels of SW lose about 27 INF). 

These last findings were too variable to be considered reliable. More test have to be done.

Overall, the results fit expectations.  Infantry quality was the primary determinant of infantry survival. 

SW experience can play a small role in saving units.

However, there is no clear effect of SW upgrades on infantry survival, though further testing might show there is be slight one.

[springer]

The last bit of the analysis was looking at what destroyed enemy matériel. In this analysis, it was expected that artillery would be play the major role.

The first results surprised the staff: Infantry upgrade levels do impact the destruction of the enemy units. Overall, Level II Inf destroyed 61% of Soviet power points (SPP) and Level I inf destroyed 58% SPP.  The result is surprising because increasing infantry levels does not directly increase infantry's killing power relative to other infantry. One of the staff members argued that the effect was indirect. The ability of the infantry to withstand damage for extra rounds allowed them to stay in place so that the SW could fire off more shots.  Further analysis supported this idea.

Of course,  infantry experience points (XP) impacted destruction levels: Highly experienced infantry destroyed 64% of SPP while low experienced infantry only destroyed 54% of infantry.

Of course, Support weapon (SW) upgrades made a big difference (since the upgrades increase SWs killing power).  The effects were smaller than for infantry XP.  Level II SW destroyed 63% SPP, whereas level I SW destroyed 56% SPP

By itself,  SW XP did not account for any reliable increase in destruction levels.

There was a  combined effect of Inf XP and Inf level.   Effectively, Infantry with high XP get high kill levels regardless of upgrade level. Both level I and level II with high XP destroyed about 64% SPP.  For low XP infantry, level does make a difference: Level I low XP infantry destroyed 51% SPP and Level II low XP infantry destroyed 58% SPP.   This probably means that Level I low XP infantry don't hang on long enough for the SW to do their work.

There was a combined effect of Inf XP and SW level.  This finding basic confirms the previous idea.  If the infantry has the staying power, the artillery will have it's effect.  Here, High Inf XP and level II SW, High Inf XP and level I SW, and Low Inf XP and level II artillery

There was also an interesting combined effect of Infantry level, SW level and SW experience.  Even when using Level I infantry, SW with both level II capacity and high XP can cause a lot of damage (66% SPP).  However, all Level I infantry with other combinations of SW (level I with low XP; level I with high XP; level II with low XP) do not achieve high kill rates (around 55%).  If the infantry is level II, the kill rates are high for all four SW combinations (around 65%)

[springer]

Here are some additional statistics that are interesting, but peripheral.

If German SW losses are calculated in power points (GPP) here are the losses suffered:

With Level I SW:
Level I Infantry with Low XP: 13 lost GPP per simulation
Level I Infantry with High XP: 0.5 lost GPP per simulation
Level II Infantry with Low XP: 8.5 lost GPP per simulation
Level I Infantry with High XP: 0 lost GPP per simulation

With Level II SW:
Level I Infantry with Low XP: 4 lost GPP per simulation
Level I Infantry with High XP: 0 lost GPP per simulation
Level II Infantry with Low XP: 3.5 lost GPP per simulation
Level I Infantry with High XP: 0 lost GPP per simulation

In 80 of the simulations, there were only 2 that could be considered outright defeats (with PP points equal).

[springer]

To sum up:

The labor-to-results ratio for this exercise may be a bit low as they primarily confirm many AT grognards already knew. (And what we can find out by poking around in Vic's description of the game engine.)

For the sixth army campaign, the results suggest that the most important initial purchase is the Level II infantry upgrade. mainly because it is the best way to saves friendly infantry, which is a rare resource.

Infantry that live longer get the experience that makes the difference: that experience protects SW and allows the SW to do what is it does best, which is remove enemy power points.

Though SW upgrades matter over the long run, SW upgrades seem to have their biggest effect when the crews are experienced enough to take advantage of them. This suggests that SW upgrades are important, but can wait until the crews know what they are doing with the new equipment. In the Sixth Army campaign, this suggests that if a player is strapped for PP, it might be better to take the troops than the SW level II upgrade in the '40 refit.

As the SP crews become veteran the upgrade becomes worthwhile, Veteran SW at level II make even relatively green troops have sharp teeth. The crews might even help reduce infantry losses very, very slightly.

This suggests that it is a reasonable strategy to back up a unit of raw recruits with highly experienced, level II SW. Thus, SWs won't save infantry lives, but it will help the recruits take more enemy PP with them as they go. (In other words, you can watch your Axis Minor allies melt away, but take some solace that the German SWs are helping them stop their fair share of the Soviet Behemoth).

There's one more finding, which may need more testing: it's a mistake to put too many low XP, level I infantry in a unit just to give it more firepower. The main effect, it will have allow more infantry to be destroyed during an attack with very little to show for their effort. It'd be worthwhile to play around and find the ideal infantry to SW ratio for raw recruits.

(Late note: Somehow I misremembered the WaW system. Infantry Guns update as artillery; and mortars as Heavy Inf. This undoes some of my analysis since I'd have to spend PP on heavy infantry and artillery. I guess I'd just plunk my PP on artillery ASAP.)

[springer]

An afterthought:

Playing with the simulation reminds me once again of the many aspects that make AT such a fun game.  In this case, it's the combat engine, which is just a small part of the system. Even though the basic mechanisms are fairly straightforward and feel intuitive, there is still a lot of subtle play in how the elements can be optimally combined.

It also reminds me how much work has gone into making AT. I remember when I bought AT initially, the purchase was intended as a way of thanking Vic for his first effort, people's tactics.  As hard as it is to believe now, I wasn't sure that AT would be better than PT. And I certainly didn't think I would play AT more than PT.  In both cases, I was wrong.

[tweber]

Really interesting stuff Springer.  If you want to play around with this more, the game has a combat sim setting in the options.  If it is toggled on, then the first round of combat is repeated 100 times and the results are reported as averages.  So, you can get more rigorous statistically.
 
I think the point about experience over upgrades is missed in many scenarios where the default (from the ptmaster) for free experience is set to 40%.  If you just wait around, the units gain experience.  In this scenario, there is no free experience so building up a unit with 75%+ experience becomes important. 

[Jeffrey H.]

For those like me that spent some time trying to figure out what "SW" means, it's "support weapons". Beyond that, I'm still reading.

[springer]

@Tweber
Thanks so much for the information on the combat sim.  I think setting up and testing the 80 sims took me nearly 3 hours.  I had started setting up some sims looking at the effects of artillery on softening up the targets first, but felt it would take too much time. 

With the sim setting, I can go back and get results that will be far more reliable. Thanks!

I think your idea of capping free experience to 10% is an excellent one in this scenario.  It makes the player have to focus on the trade-off between using units (to gain experience) and losing units (and losing all experience).

@ Jeffrey H.
I agree that this analysis wound up being pretty hard to follow. Even though it really is just a statistical analysis of a very simple situation that could occur in the Sixth Army '42 campaign, the explanation  becomes complex when untidy because there are so many variables. (Once I realized what I was into myself, I realized I'd lose most readers quickly. However, I wanted to make sure I completed tha analysis. )  I tried to make it potentially interesting by adding "narrative" and a small bit of  humor, but I'm not sure I succeeded.

If I report the results again, I think I need to come up with a better way to synthesize the results: perhaps a table or chart would be better.

[springer]

Using the comments from Jeffrey H. and tweber, I thought I'd try these tests once again, trying to increase both reliability and clarity.

The situation is the same as above. I ran 1000 simulations per situation and rounded the results to the nearest whole number. (I so many simulations to get a statistical sense of the variability, which the simulator- excellent as it is- doesn't report)

Below are graphs of the effects of upgrading (level I or level II) with low XP.

The simple summary:
For attacks using low XP units with small amounts of infantry attacking substantial forces:
1) Upgrades contribute very little to friendly infantry survival.
2) Both infantry and support weapons upgrades contribute to destruction of enemy units.
3) Infantry upgrades contribute 36% more destructive power than support weapon upgrades.

[springer]

Here's my comments on the results:

First, they partially confirm tweber's comments about the importance of XP over upgrade.

At first glance, these results are counterintuitive to what one would assume based on what gains by upgrades:

Upgrading infantry increases hitpoints by 25% but has no impact on infantry firepower vs. support weapons and enemy infantry.  Superficially, I'd expect that infantry upgrades
A) substantially increase survival
B) have little effect on attack damage. 

Neither expectation was correct, though in hindsight I can guess why:

A) Because infantry gains 25% hitpoints, a casual glance would assume that upgrades are translated into a 25% savings in casualties (If 29 INF at level I are destroyed; a 25% absolute reduction would lead only 22 INF at level II being destroyed).  However, actual savings in infantry is only about 7%. Thus, the increased hitpoints translate into increased round survival, not absolute battle survival.

B) Similarly, upgrading infantry doesn't increase infantry firepower vs. support weapons and enemy infantry.  I'd expect that for level I support weapons, there should be no improvement.   However, upgrading just infantry results in greater killing power. (Level II infantry with support destroyed an extra 11% of the Russian Power Points than Level I infantry). I assume that the increased hit points of level II infantry allow them to survive more rounds longer (though they will die), protecting the support weapons longer, and giving them the chance to pound the enemy positions longer.

[springer]

For this simulation of upgrade effects support weapons were given high experience (85 XP), which is quite typical at this point in the campaign. Infantry was left at low experience (20 XP).

The effects of increasing SW XP are much as expected, given that SW is primarily offiensive
The simple summary:
For low infantry XP and high SW XP:
1) Pattern of upgrade effects is almost identical to previous situation when both infantry and SW XP were low. Upgrading infantry increases destructive power but not survivability

2) High SW XP does not substantially change infantry survival. It remains the same as when SW had low experience.

2) High SW XP increases the destructive power of the unit. In all cases high experience SW destroys about 30 more powerpoints (24%-33% more) than low experience SW.

[springer]

For this simulation of upgrade effects, infantry were given high experience (85 XP), which is quite typical at this point in the campaign. Support weapons were left at low experience (20 XP).

Here is a summary of the results, which confirm the first analysis:

1) Infantry experience saves infantry casualties: High XP infantry averages 6 (20%) less casualties than low XP infantry. 
2) Infantry experience increases destructive power as much as artillery experience: High XP level II infantry with low level support weapons has the same destructive power as low XP infantry with high XP SW.
3) For high XP level I infantry, destructive power is even higher with low XP SW than it is with low XP infantry and high XP SW. 

The two corollaries of these results are:
1) As pointed out by tweber, building and preserving unit experience gives bigger gains than upgrades
2) It is worthwhile to give inexperienced infantry experienced SW.  Though infantry losses won't be reduced, the experienced SW will give inexperienced infantry the the destructive strength of high XP infantry

[springer]

Last piece of this analysis:
This has high experience infantry (70 XP) with high experience (85 XP) support weapons.

Here is the summary of results:
1) The pattern of infantry casualties remains essentially unchanged from the 70 XP infantry and low XP SW. This isn't surprising.

2) Both infantry and SW having high experience increases the destructive power of the unit by about 12-21 points (8-16% increase) than having only infantry or SW with high experience.

Possibly this rate could be even higher, but the destruction power of the unit is approaching the limit of the power points in the hex. The three Russian units have 273 power points and a 147-159 destruction rate is about 54%-58% of it's total power points.

[springer]

I thought I would test to make sure that the effects of upgrading from level II to level III were linear with the upgrade benefits of level I to level II. So I simulated some assaults with level III infantry (only 400 simulations per test).

I held (support weapons) SW at level 1 and low experience as the effects of SW seem apparent from previous simulations.

The results are organized somewhat differently, and they are exactly as expected if it is assumed that upgrade effects are linear.

The results reinforce conclusions from previous tests:
1) Upgrades without experience have a marginal effect. (Raw recruits take relatively high casualties no matter what level they are.

2) Experience interacts with upgrades. That is, that bonus on hit points pays off as units gain experience. High experience effectively doubles the number of infantry units that survive.

3) Infantry upgrades continue to increase the destructive power of the support weapons behind them.

4) The effect of infantry upgrades on destructive power is immediate. Upgrades increase the destructive power of a unit even when the infantry is not experienced. Superficially, it looks like the destructive gain diminishes as units gain experience, but this may be an illusion of the size of the attacked units. As mentioned, as power point destruction may level out as total points destroyed approaches 50% of total power points in the defending hex.

________________________

The conclusion that I draw from this is the same as before: the best investment in the Sixth Army scenario is to upgrade the infantry. Even an upgrade to level III may be a good idea.

These simulations particularly generalize to scenarios where: A large portion of the forces have high experience and infantry is hard to come by, and the cost of upgrades is severe (i.e., upgrades take away from replacing scarce units).

It seems to me that for such scenarios, infantry upgrades are essential.

However, heavy infantry upgrade (e.g., mortars, bazookas, and machine-guns) may be a poor investment in the sixth army scenario. Though mortars are a key part of regimental firepower, their overall impact is small and far less than an investment in infantry. (The upgrade also improves bazookas and machine guns, which is a waste of resources. In the sixth army campaign, there are no bazookas. The machine guns are primarily defensive, but the sixth army remains on the offensive throughout the game. Not only that, most machine guns are destroyed before the accumulate any notable experience.) Rather than upgrade the mortars, it's better to buy six new infantry regiments (That's 240 INF, plus all the support weapons).

(The issue of upgrading artillery is more complex in the Sixth Army campaign because stand alone artillery- in addition to the infantry guns- are a core piece of a successful combined arms strategy. For now, I'd favor upgrading based solely on artillery's important role in the campaign.)