Ground Combat Theory Academy

Hello, I have been wanting to make a topic that really analyses Aurora C# Ground Combat, as in what interesting effects emerge despite its simple foundations (as it always happens), and what possible conclusions could be drawn from it.
I had been very active in a similar thread in the old forum some years ago when I coincidentally had just been playing as GDI and came out of it finding that even in otherwise complete defeat encounters with moderate 1.5TL disadvantages on near equal tonnage deployments, a single ground unit class just kept standing and winning all the battles. (more to that later) Since I learned some new things while writing and thinking about it back in the forum, I fear the ultimate point of it all went a bit under in some confusion.
That is why, -this time-, I wanted to place a neutral, large calibre thesis alpha-strike that really aligns all the mathematics right from the start, so that all that remains doing is uncovering implications thereafter.

Before start, just also a motivational disclaimer:

Expand

My own gameplay philosophy in Aurora is actually 95+% RP based, which means in ground combat too I will simply build all kinds of thematic designs, APCs, guard towers, “medics” infantry, -you name it-, no matter their objective efficiency.
However, I am also a person that likes to understand systems. I can throw the conclusions away for artistic purpose, but I am still curious about the truth.
This was merely an exercise to figure out how everything really works itself out to be.

Formulas

I am starting with the basics, but derive from there to two full design efficiency formulas, - one for duels, one for battlefields.
The most foundational Aurora equation is this kill chance for a shot that had already hit:

(1) Kill_wpnX_pShot

Kill_wpnX_pShot
(“T” always stands for the stats of the Target/Enemy)
Right away I want to foreshadow a result by drawing attention to that AP and DMG, as well as Armor and HP originate from the same two source TL-numbers, especially in cases of even stat units and weapons. Seeing that, the formula could be effectually simplified to either close to or even literally:
Kill_wpnX_pShot_approxConclusion

Next we include the hit chance from Aurora basics in completion:

(2) Hit+Kill_wpnX_pShot

Hit+Kill_wpnX_pShot
However, note that 0.2 is merely a stretch factor to scale encounter length that applies to both sides, and thus cancels out. Then, since this is about comparing design efficiency, design independent factors like morale don’t matter, and environmental factors are unnecessary complexity, -mostly applying to both sides again anyway-, that can be left outside the scope of this argument. The Hit+Kill formula in effect therefore simplifies to:
Hit+Kill_wpnX_pShot_simple

Next we need to include all the shots a weapon can fire. For this it is necessary to realize there are two possible situations to compare. One is the battlefield with a target rich environment, in which case the kill performance can simply be multiplied by the shot amount:

Kills_wpnX

Kills_wpnX
Should this result in more than 100%, it simply means the weapon kills more than one of the enemy design ‘per turn’ on average.
However, on a direct unit vs unit comparison, only one “kill” can ever be had, which makes the equation slightly more involved if you want to understand the designs in this “duel” environment:

Kill_wpnX_Duel

Kill_wpnX_singleT

Then you do this nearly identically in a second layer when including all weapons X a design has, finalizing the two ultimate kill chance equations vs their opponent:

(3A) Kills

Kills
And for duels:

(3B) Kill_Duel

Kill_singleT

Now, to get the actual design efficiency vs the opponent, you can primitively ask the question “how many of them do I kill before they kill me?”:

(4A) Efficiency

Efficiency_pTarget_primitive

And though that gives some very interesting results, it also isn’t exactly fair, because an Oliphant doesn’t actually count as one. In most cases the real efficiency comes rather from the principle to inflict as much cost as possible on as low a budget as possible. Cases like this one may arise:


The yellow numbers below designate the Efficiency per unit first, -looking terrible for the Infantry against the tank-, but after factoring in build cost in the second yellow line, suddenly the infantry is on top as it is so vastly cheaper.
To take into account cost, there is an asterix for later (here:*), but on the simpler end and by principle, you can append the equation with this:

(4B) Efficiency (size or cost)


Interestingly in Aurora, only mass and Armor thickness determine cost of ground units, not the tech level, so you can easily convert the numbers into each other using the official rule:

(4C) Cost-Size Conversion

Cost
Sometimes you might even want to draw up a mixed metric where you optimize the cost-damage inflicted for a certain tonnage of ground troops you can drop at once, so cost/m would also be a very legal consideration. (Pulverizing a mountain by the lowest bidder in m/cost… may be not that useful however :P)

Finally it would be great to judge a whole design’s performance against multiple types of threats to get an overall assessment of validity. This is a topic about analyzing unit types against all others after all, so there must be some objective numerical condense to make that visible.
Naively it is just the average efficiency against all enemy types:

(5A) Summary Efficiency

Efficiency_general
Of course, that doesn’t take into account the fact that some unit types like infantry or lighter vehicles would likely make up a much larger share of an actual battle force, so you could add some incidence factor to it:

(5B)

Efficiency_specific
The problem with that however is that you never really have enough battlefield intelligence to know these incidence numbers, and thus can only make a close guess. I ended up using the general formula and simply adding more infantry types to the battlefield, so it should at least roughly be representative.

At this point we are capable of writing out a complete efficiency formula for duels:

(6) Duel/Unit Efficiency Complete

Efficiency_specific_LongForm
(the arrow notation on the right is half-bogus, but the Latex editor I was using had no other passable alternative)
It looks impressive through size, but is really simple as relations go, so this goes in all the Excel work for getting Unit vs Unit Efficiencies.

Formulas Round 2 - True Battlefields

Naturally, nobody will be shocked to find out that the huge Behemoth Obliteratrix Cannon ‘shell-shock, the new rock’ heavy tank will have an easy time rolling over a normal infantry man, - even if cost adjusted. By applying the 3A “Kills” formula, I got comparisons like these…:


…Where the tank has a maximum hitmod(0.6)*100% kill chance per shot, so 4x60% = 2.4 “kills per turn”, while the AMR infantry with its mere 1/45 chance can’t even get to to 1 once you factor in it is 43.5 times cheaper, thus ending with roughly a 2.5 cost efficient encounter for the ‘not even anti-infantry’ tank vs. ‘supposed anti-tank’ infantry.
However, to arrive at a real battlefield there is a part that is easily overlooked, -including by me at first in the old forum and also this revision once more-, which is that you need to set up a cost equivalent encounter beforehand.
The domination of the tank above is achieved because he can keep shooting at more targets after dispatching the first, but these additional targets somehow get no chance to fire back at him, which would have shifted the balance.

So in order to have an actual fair fight, you need to set up every unit against a cost equivalent number of enemies of the target’s design (seen in the screenshot through the “# Encountered” box on the right), which is done by satisfying this simple formula:

(7) Fair Force Balance

(With U_0 just being 1. T as the numerical count of Targets. Thus, T can be fractional and even smaller than 1 if more expensive, but that is fine since all formulas are continuous and apply just the same.)
What follows now is the derivation that takes this and leads to the full battlefield efficiency equations. Since they are long and not really useful by themselves, I just package them into a box for those interested:

To whom it may concern

In any encounter the winner will always be efficiency >1 and the loser <1.(unless exactly matched) Since we already normalized the two sides for cost, we should be able to figure out both sides’ efficiencies by simply looking at the percentage of forces the winner lost at the time the losers’ force reached 0.
For that end we start with the force change rates:

Force Change Rates per Turn


Note that we are using the “Kills”, not “Kill%” equations here. Similarly to the shot multiplication, we can just multiply by force count as we assume there are always targets left. Just think of it as an infinite line of tanks against an infinite line of infantry. Though they are infinite, we still know they both spent the same amount of resources, so there are cost/cost_T infantry for every tank. They will always be able to keep shooting and contributing if by chance they kill it early.

Now, having both number counts related to each other makes for easy setup for differential equations, so here is one side (you only need one):
ForceChangePerTurn_Interrelation
Substitute:
Substitution_Kills
Substitution Result:
Substitution_ForcesResult
This is a famous formula. Known equations that satisfy being themselves again after two derivations are hyperbolic functions, so a good general solution is:

Force Count - General Solution

ForceCount_generalSolutions
At t=0, cosh will be 1 and sinh be 0, so we have trivial constants:
ForceCount_TrivialConstants
To get the non-trivial constants, one route can be to combine this known solution for t=0:
ForceCount_InitialChangeRate
With what our general formula produces (also at t=0, but not yet inserted here):
ForceCount_SolutionChangeRate
Again with cosh(0)=1 and sinh(0)=0, you now get the non-trivial constants as:


And thus a non-nebulous, usable force count equation:

Force Count - Actual Solution


Interesting to analyze about this is that the cosh-Term basically designates the initial population count of each side, while the sinh-Term represents attrition stemming from the enemy.
The Kills geometric average appearing as a multiplier to time is also very interesting, as it simply signifies the encounter’s end state approaching faster the better either -or both- sides are at killing. It does counter-intuitively however not mean that more killing or attrition will be done to yourself if for example only your own unit gets better at killing. Because in such a case, the end time t_e itself would be smaller too, counter-balancing and even over-compensating the effect. (not to mention attrition would be reduced by inverse relation to your killing)

So from this we can now check which side reaches 0 first by setting U and T 0. Remember that tanh = sinh/cosh. Here is the example for U=0:
ZeroPopulationTime_Intermediate

Zero Population Times

The tanh function is only defined between -1 and 1 and so an expression inside the tanh could resolve in error if it is bigger than 1. Most interestingly that happens exactly when that side is the winner, as once the other side is destroyed first, your own forces never reach “Zero Population”.
(There is also a rare edge case where both sides are exactly equally matched, thus both having ‘inner expression’=1. It means both sides only asymptotically approach 0, never actually reaching it.)

Now, that is interesting. We now have an isolated term that doesn’t even have to do with time anymore and seems to predict the outcome of a battle in itself. Squaring it for convenience, you can now try at defining an invariant, a numerical constant predictor of the matchup that doesn’t change over time:

Invariant Definition

Invariant_Definition

(8) Invariant

Invariant
(If you derivate this after t and plug in the U’ and T’ from the start of the bonus section, you can see these indeed cancel out.)
The right side of this equation is especially interesting as it holds values accessible without calculation at the start of a matchup. From this we can actually predict the winner of the battle before it even starts, and without need for any cumbersome tanh-calculations.
The rules stand as - "1) If [ Kills * U²_0 > Kills_T * T²_0 ], T gets destroyed. 2) Both are equal at equal (duh). 3) U gets destroyed if smaller.

Since the invariant term has now given us much more handy force count equations compared to the bonus section, we can now move to directly determining the remaining forces factor that we set out to get, as it directly represents final efficiency. Plugging in these times, depending on who we now know will win beforehand:
ZeroTimes
We get:

(9) Remaining Forces Factor

The final efficiency for the winner is thus determined by inserting the term of the winner into this:

(10B2) Efficiency - Battlefield (cost)

Efficiency_pTarget_cost_Battle
The inverse of this will then be the efficiency of the loser.
We had the cost necessarily baked into the formula, but it can now be extracted back out to arrive at primitive unit count elimination efficiency and from there mass efficiency (or any desired combination):

(10A) Efficiency - Battlefield

Efficiency_pTarget_primitive_Battle

(10B1) Efficiency - Battlefield (size)

Efficiency_pTarget_size_Battle

Lastly, finally, you can write the full battlefield efficiency formula out the same way it was done for duels in (6), in a despite the troubles getting there, surprisingly easier to compute form:

(11) Battlefield Efficiency Complete

The cost and unit efficiencies just determined in 10A and 10B1 can be re-applied in the same way to the inner part of of the sum as they had been in (6).
Another interesting thing to note here btw. is already a conclusion that I want to frame, stemming from the mere condition of winning here again:

C1: “An increase in cost of a design is only justified if it leads to at least a quadratic raise in killing efficiency against its enemies.”

Because, as you can see, cost itself has quadratic impact on performance due to the cheaper side not only being less costly when individually destroyed, but also appearing with more “individuals” to begin with at the same time. In effect, this is increasing defense on a secret route not caught by the classic defensive stats alone.(more “lives”, while a tank only ever has 1)

Thus, if a bigger weapon comes with *1.1 cost, it better at least provide *1.21 better killing chances.

That concludes the theoretical base. Now that all the tools exist to really look into what does what and how well it fares, I put it all into an Excel file to make it quicker to calculate. I will show you some findings in the next section, but everyone can play around with it if they want here: Aurora GC Analysis v1.1.xlsx

Analysis

To start out with, I found it would be interesting to maybe settle a question everybody could quickly stumble across and see if that reveals something in the process already.

“What is the best anti-infantry tank?”

Using this model


Against softest targets of same TL it dispatches 68 light PWI in duels and 48 in battlefields.
Against many types of infantry, with the exception of any LAV, we find a very efficient design:

However, increasing armor from light to medium reveals something profoundly interesting about Aurora GC rules:

All the anti-infantry performance stats remained exactly the same! Our tank model got 2 times as expensive from doubling its armor, but since even LAV could just barely penetrate the light armor, weapon efficiency against it went down by exactly (1/2)^2, as dictated by (1). Since this combines with C1 to cost(*2)^2 = efficiency(*2)^2, we arrive at exact parity.
However, notice that against the group of 3 vehicles to the right the extra armor actually decreases efficiency against them? That is because some of their weapons (the MAVs and MACs) actually overshot the armor in the beginning, so adding more actually put the anti-infantry tank more into their target focus, thus not being able to “harvest” the extra efficiency from more armor to offset the quadratic inefficiency loss from cost.

This reveals a tragic truth about Aurora GC mechanics (no criticism though), by applying the C1 to:

C2: “An increase in armor thickness is only ever barely justified if the original armor wasn’t (or was barely) penetrated to begin with.”

Or a more compact, final conclusion version:
C2B: “An increase in armor thickness is never worth it.”

Now if that sound sacrilegious to you, welcome to the show and the why I opened this topic.
But think about it. Increasing the armor through thickness by a factor f can at best only decrease enemy killing chance by 1/f². At the same time, cost is directly linearly related to armor thickness (cost~f), thus our win efficiency determination (Kills_T*cost^2) becomes 1/f² * f² and cancels out exactly.
At the same time however, this cancellation does not occur at the same 1/f² yield for any weapon that overshot the previous armor, while still demanding f² battlefield efficiency on the negative side.
So we have 0 gains and sometimes loss. The only conclusion is to never do it.

Uncomfortable truths need to be spoken, but as my disclaimer reads, I will also ignore them, - just know them.

Just to make it clear, this rule only pertains to ‘armor thickness’. As by the cost formula (4C), any increase in base armor from TL comes free of charge and remains a huge boon.

Also, this is a cost consideration only. If you are looking for the highest performance for a certain drop size (mass limited deployment, not cost limited), then armored units are of course at least neutral if not better performing. A boarding shuttle crew in heavy armor will always be the better pick.
…But armor thickness is useless in an open battlefield where both sides can throw whatever is available.

Heavier in other ways

Okay, so making the thing heavier from skin is not a good idea. What if we also increase the motor and make it heavier as a HVeh anti-inf tank instead?



Yields in all regards are once again slightly lower than with the light MVeh. We increased size (thus cost) from 50 to 68 here. A 36% cost increase vs a 50% hp gain should at least in those cases where dmg wasn’t high enough yield a slight increase, or not? Ah, but the HVeh is also 33% easier to hit, thus the inefficiency rise is actually cost^2*Hitmod_increase, which comes out to virtual cost factor of 57%, - slightly worse than what we get from hp.
Exactly what shows up in the graph, and I checked, but this exact slight 7% inefficiency shows up throughout all tech level differences too. I will spare the armor thickness experiments as those are now foregone conclusions.

So you can’t make a better anti-infantry tank by going heavier, but can it be better by being lighter? Yes.

Performance increase in all regards, including LAV infantry! Its only counter in fact seems to be the lightest possible LAV with on the other side even anti-vehicle MAV tanks performing at a loss, what is going on?
The combination of square increase from saving about a third in cost and another third from better hit mod (~27/8 gain) has more than offset the loss of only half the kill shots from one weapon (50% inefficiency).

Now we have a design that performs well not just in an anti-infantry role, but shapes up as potential anti-all superior design. Can we improve by going even smaller and not using a vehicle at all? Well, not entirely as well. We have already seen HCAP infantry being outmatched by real anti-infantry tanks. However, CAP infantry still reveals something interesting:


Is this dramatic anti-vehicle efficiency from a supposed anti-infantry trooper? The only weakness shown here being the HCAP vehicle from just before?

Well, as infantry come with 0 base size, the weapons now follow the same direct application of C1 as the armor before, which is that lighter is always better except if efficiency increases by at least square that saving. HCAP is more than 50% more expensive for only potential 50% AP gain, and thus really inefficient in every single situation. I looked at the all TL comparisons, and HCAP is literally never an improvement over CAP for infantry against any unit at any TL, so a micro conclusion:

C3: “Infantry should never use HCAP over CAP.”

Again, this is for cost consideration only. Density of firepower is a perspective where HCAP is better unless your TL exceeds enemy TL so far that your CAPs can even shred tanks with guarantee.(but then you wouln’t care about efficiency anymore, so HCAP>CAP for mass, always)

You might also want to conclude the same for PWI vs PW, but here the 20% cost vs potential 25% offensive gain lead to more ambiguous outcomes.

Despite that small outlier, I think it is time to celebrate the previous findings by formulating a super summary conclusion so far:

C-Super: “Light designs are cost efficient. Heavy designs are size efficient.”

It might sound obvious if stated like that, but is by no means guaranteed from just some rules and seemingly random numbers for vehicle and weapon weights as they are. However, the balancing is so well done that Aurora C# simply blooms into this great position where both paths of design have their own clear benefits, thus leaving none obsolete, and even satisfying the RP-perspective.

Battlefield Hierarchy

With that foundation settled, there still remains the question and the lead from before that indicated possible existence of a ‘best unit’ (by cost) semi-allrounder with few specialized weaknesses.
We had an extremely well performing light vehicle, and an anti-infantry infantry that somehow was immensely efficient against vehicles as well. Hell, CAP infantry is actually so good, it performs favorably even one TL upwards against a somewhat balanced 55% infantry battlefield (own unit is TL3):


(Great average 1TL up mostly due to stellar performance against LAV infantry, which with 15% is most likely overrepresented here.)
To spare all the detailed individual investigations, it would be better to just take this entire theatre…:

…and see how it performs against itself in one go:

I marked our champions light CAP infantry as well as the two light vehicles, because they clearly stand out.

So things to note:

  • All light infantry, save PWL -the greatest failure of all-, is actually performing stellar well, being only ever beaten by dedicated anti-infantry but literally no other vehicle class, and even then never to a truly disastrous degree like <0.1.
    It appears the “light is most cost efficient” rule can be appended to include the unit classes themselves after already conquering armor and weapons.
  • In fact, the greatest winner here (and in other tests) is the LAV light infantry, though it also has glaring weaknesses that can be much more easily exploited compared to the all round fit other infantry types. Also, the numbers convey undue advantage from 45% vehicle content still being a rather too large share of true battlefield composition.
  • You can clearly see the previously reported “armor thickness makes things worse, always” statement in action.
  • Heavy vehicles really have no place in cost considerations.
    The semi anti-infantry light variant (18) is barely performing against anti-infantry tanks, and otherwise infantry too armored to be in the field anyway.
    The heavy HAV (20) at least works as HVeh and anti-infantry tank hunter, but those positions are also performed better by other designs.
    Finally, my favorite, my original GDI Mammoth Tank template (MAV+MAC) HVehH, is the greatest loser of all, having no strengths other than anti-infantry tanks, against which he is also just 3rd place.
  • Tank destroyer MVehs with HAV, despite overshooting other MVeh, are actually -true to design- very effective against all medium or heavier vehicles, thus becoming the only non anti-infantry tank design to be viable.(though still outclassed by light LAV infantry)
  • Finally the two most popular light vehicle designs, where the MAV version is the most efficient tank destroyer in the game (even above LAV infantry) but with a series of glaring weaknesses (also against said vindictive LAV infantry), while the HCAP version is shinning through and through as the king of the battlefield, only being beaten by both these grumpy envious louts.

For comparison to the last, its CAP version too:
TheatreVsSelfMap_LvehAdjust
As you can see, while average performance increases by notable further 25% and the LAV-inf is left beaten, it also comes with a reinstated allergy to the common main battle tank and other heavy vehicles. In my opinion the avoidance of straight counters is in summary more important to be crowned a battlefield king (otherwise a 1million one-unit specialist could claim it), but that might be taste.

Since lots of unit types seem to be made obsolete by others that are just simply better than they themselves in all matters and do more, a sort of core unit composition and hierarchy has now emerged, which I see mostly as this:


(Single LAV LVeh {0.15} or double LAV MVeh {0.21} are actually the most efficienct anti HCAP LVeh designs, but who really builds these?)

So all tanks are useless, except maybe the anti-infantry tanks, who can however all without exception be better represented in that role by the HCAP LVeh.
Seeing that and thus taking the tanks entirely out of the picture, we arrive at a classic paper-rock-scissor situation, where:

  • AT infantry and AT walkers/skimmers --beat–> regular walkers/skimmers
  • Walkers/Skimmers --beat-> all other infantry
  • All other infantry --beat-> AT infantry and AT walkers/skimmers
  • Everybody: Eats tanks for breakfast. (infantry sometimes scared)

Those are the conclusive rules for Aurora GC cost efficiency. Lay thine eyes upon them and despair. (or or you may like it, idk..)

Visuals Hinting at Model Weakness

The last section deals with perhaps a hook for further discussion, because I could myself not solve it in any definite way. Remember that I said in the very beginning that in my GDI ground battles there was only one unit that stood against all tides and even defeated enemies of higher tech levels?
That unit was no M60 armed Humvee LVeh. Those were in fact the first to go.

The winner was the Mammoth Tank. A unit the math says is the worst of the bunch. When all other units of the division had been eradicated to single digits or close to, the original 24 strong contingent of HVeh multi-purpose tanks ended the battle with still 18 of them left, mowing down hundreds of enemy PWI and even LAV.
The fourth power (Attack/Defense)^4 law had their armor deny even technologically advanced enemies, and their bigger cannons actually came into focus when classical AP weapons could not touch the enemy anymore.
“Shooting with cannons after sparrows?” Yes, but what if those sparrows suddenly developed shield generators? Then the cannons are the only calibre left still hurting, so you gotta do it.

Looking at it more naively, I actually found that a multi-purpose HVehH was usually only matched by any weight HAV designs and more anti-tank specialized medium HVehs:

However, that is not because getting weapons into focus was the entire story either. What really turned it around is the uncaptured effect of cover.
[Edit: A misconception and error had me thinking all units were target size=1. Corrected below, but I leave the original text bit.]

Wrong Original

In Aurora, units aren’t targeted by size, but by unit, so everyone is essentially just “size 1”. That curiously means that large and expensive units effectively apply a shrink factor when becoming part of a force group. The larger, the more they get hidden/overshadowed by friendly crowds in matters of cost calculations, and thus the more they benefit from cover.

A calculation example (also in the Excel file):

A tank of cost 12 might get targeted by 100x0.12 cost enemy infantry at once, being easily destroyed. However, if he appears with a cost equivalent contingent of 100 infantry himself, the enemy also receiving another 100, we have two equal cost forces again, ours with 101 units, the enemy with 200.
The chance that the tank gets targeted suddenly shrinks to 2/101 of what it was before, thus representing about 98% cover. Naively he now has 50 times as much time to deal with the enemy infantry that is only twice as large (though, of course, cover will deplete over time). Given that the tank could in some situations (like tech advanced aliens) be much better at guaranteeing kills than the infantries around him, this force composition suddenly is better than any of both units on their own.

In Aurora, units aren’t targeted by cost, but by size. Since larger vehicles expand in cost faster than in size, it curiously means that large and expensive units effectively apply a shrink factor when becoming part of a force group. The larger, the more they get hidden/overshadowed by friendly crowds in matters of cost calculations, and thus the more they benefit from cover.

A calculation example (also in the Excel file):

A tank of cost 12 might get targeted by 100x0.12 cost enemy infantry at once, being easily destroyed. However, if he appears with a cost equivalent contingent of 100 infantry himself, the enemy also receiving another 100, we have two equal cost forces again.
Though an infantry man comes at 1/100th the cost, they still weight 1/10th of a tanks weight. Thus the 100x infantry force of equal cost to the tank will get targeted 10 times more often due to their sprawling size.
The chance that the tank gets targeted suddenly shrinks to 2/11th of what it was before, thus representing about 82% cover. Naively he now has ~5 times as much time to deal with the enemy infantry that is only twice as large, - more than doubling his personal performance. (though, of course, cover will deplete over time) Given that the tank could in some situations (like tech advanced aliens) be much better at guaranteeing kills than the infantries around him, this force composition suddenly is better than any of both units on their own.
[/Edit]

Somewhere in the formulas section I wrote that cheaper units shine because their cheapness leads to quadratic increase in efficiency from also granting them a second line of defense upgrade from getting ‘extra lives’.
Well, it turns out other units can borrow that part, as simply by standing next to those ‘extra lives’, they kind of “rub off”. The effect essentially allows to merge large attack power of heavy units with large extralegal defense stemming from ‘many lives’ the tank doesn’t otherwise have access to.
Yes, it won’t work for too long, but longer the more you add, and every bit of time is more valuable to the ‘100% kill chance but too few shots’-tank as it is to the ‘low kill chance but many shots’-crowd.(rather, the later doesn’t even get as much cover due to ..being a crowd)
Cover benefits the large, singular units most. [Edit] Just 1 infantry man means a huge tank receives half damage for a time.
Just a few infantry men can mean a huge tank receives half damage for a time. [/Edit]

So while the Mammoth Tanks in that historic encounter might have nearly lost all cover, even the bit that was still left meant their mere 18 remaining units were so unlikely to be targeted that they could continue their devastating artillerist handiwork until no enemies were standing.

…In have failed to capture this effect in straight mathematics. It depends too much on real battleforce composition on both sides. I don’t think that outside of complete simulations of battles there can be a straightforward formulaic predictor as those demonstrated above. It really “depends”.
For now you can simply approach the reality of being integral to a working battle force by tweaking the cover option I added in the sheet, though remember that each unit type would need a different value. (maybe some standard for size-to-cover gain could actually be worked out? By looking at PW infantry as standard perhaps? Recommendations welcome)

Though it isn’t really any final conclusion this time, and even draws into question previous conclusions made, I think it is still time to celebrate this final revelation just emerging out of the Aurora basic rules setup:

C4: “Force composition actually matters”

Great for RP, great for keeping all game elements relevant. Not so great for calculation exactness. :sweat_smile:

So that might be potentially disappointing that all that could be read above in the end is thrown into disarray. Perhaps see it as a buddhist zen voyage, where you learn about all the complexities, only to after years realize it is all simple again, ..but now you know why just a bit better than before. :slight_smile:

So until further GC enlightening might graze us, I will close the post as I did back in the old forum on similar topic:
“All hail the Mammoth Tank!”

///Edit 28.10.26: One of the final battlefield efficiency formulas was inaccurately calculating the side of the targets. I changed it to show only “our” side. The calculations had used the correct one already, so just a visual issue that is updated in the new excel file too.

17 Likes

Good analysis and nice write-up. If I understand you correctly - I’ve had a couple of Friday night sweeteners - then light is the way to go when cost is factor but tonnage isn’t and heavy is the way to go when cost is not a factor but tonnage is. The CAP+INF thing is a little surprising and really good to know.

As far as my English knowledge allows me to understand, you are analyzing the effectiveness of troops mainly in relation to their Cost. Cost-effective units is a good colony defense and an addition to the police infantry.

But this cannot be applied to offensive forces, where efficiency per tonnage is much more important (In the usual case - however, you can just build 200+ troop transports if you want). Besides, it’s obvious that cost-effective INF and LVH will never be able to handle UHV because they don’t have enough armor-piercing weapons.

In my opinion, Ground Combat are a clash of our own ground doctrine with the opponent’s doctrine in various conditions (attack/defense, tonnage/cost limitation). The main question when creating Ground Forces is “What kind of enemy are we creating these units against?” When creating the doctrine of our Ground Forces, we must choose between versatility and specialization in countering different types of units.

We usually don’t know the enemy’s doctrine until we encounter them. However, the NPR doctrines differ very little, so an experienced player can immediately create troops that effectively counteract them. Just build a UHV with CAP, LAV and MAV… I would like to see more heavy units among the NPR troops, but I’ve never even seen an HVH (or there were so few that I didn’t remember it).

First of all, this is tremendous.
Extremely well conceived and executed.

One point that makes me scratch my head:

But from the base ground combat rules:

Hostile formations are checked for their weighted size. This is based on actual size for front line size, 25% for support and 5% for rear-echelon. Each hostile formation is given a range for potential selection, based on its weighted size.

Am I overlooking something else about the target selection process?

Edit to add:
And after formation target selection, the element target selection is:

Once a front line formation (or a light bombardment element in the Support position) has been matched against a hostile formation, each friendly individual unit (a soldier or vehicle) in that formation engages a random element in the hostile formation, with the randomisation based on the relative size of the hostile formation elements.

1 Like

I don’t find this to be true in cases of large-scale invasions.
As you say, you can always build more transports.
When building an army to invade major hostile colonies, I am always constrained by how quickly I can train the troops, not by how quickly I can build the transports.

Targeting is based on relative size, not number of units.

image

3 Likes

Thank you for the writeup! I like the analysis. I know in theory there is a way to create a linear algebra matrix that will resolve NvN matchups of hetrogenous forces but thats definitely way beyond my skillset. Even just having the math and spreadsheets for 1v1’s is a good tool. And I’m also in the RP’r category where i’ll include wasted tonnage of support and noncombat units.

Maybe we’ll get some big additions to ground combat in 3.1 or 3.2. There are a few overall effects I think would be worth exploring and possibly adding since I do love ground combat.

1st: Frontage -

As the number of units on one side skews a growing % of the larger formation should be forced to skip their attack rolls.

2nd: Screening -

In good combined arms, infantry screen vehicles, and vehicles guard infantry. We do already have the target weighting by relative size, but I would like to see that extended by medium+ sized vehicles being able to essentially give infantry a saving roll against lighter weapons.

While as a counterpoint even the sturdiest unit will get mission killed by light infantry if it is on its own, regardless of armor protection. That loops back to the first point, where if you have some absurd frontage value, like say 1000 infantry vs 1 tank, the larger force starts to get asymmetric attack rolls as their conventional ones falloff due to the frontage.

3rd: Suppression -

A good mix of artillery and heavy weapons should be able to suppress infantry to some extent. There would need to be an asymptotic limit because you don’t want to be able to stunlock by minmaxing. Doing so would make Light Bombardment relevant. This is also where things like air support, orbital support, and backline forces would really start to shine.

2 Likes

I understood this as a targeting multiplier for only the position in the battlefield, but I should have questioned it more. Steve has now confirmed that the targeting by size is actually correct.

Well, that generates complications. It means we are actually at a inbetween zone from the extremes “all size 1” and “all by cost” targeting. Example:

  • Currently there are 43.5 LAV InfL going up against one Mammoth Tank by cost basis.
  • However, the actual size difference is only 116/16= 7.25.
  • It follows that while of course all 43.5 LAV would fire if there is nothing but tanks, once the tank gets cover, it would shift to a less pressed matchup still that is however 7.25 times higher(=worse) than the previously assumed cover effect.

That is actually very interesting. It effectively means there is a dampening effect to the effectiveness of cover. It continues to exist as mass differences aren’t as extreme as cost differences, but the cover is not as pronounced anymore.
At first I was shocked, but the experimental results (HVeh being really powerful) have to come out of this somehow, and so they still do.

I will think about how to incorporate this lead, and maybe append the theory. (could take some time)

The effectiveness in terms of mass is much simpler to get, because density of power is easily seen and mostly doesn’t even have to be tested by matchup. ‘Bigger+Heavier is better’ comes as a conclusion with few examples, so only cost effectiveness is a real problem to solve.

Otherwise I agree with skoormit on the tonnage not being as important as cost in most cases, though I understand why you would at first think otherwise. I also at C# dawn thought that it was obvious power concentration was the domain of heavy assault forces for invasion, but actually playing has shown me that I can pretty much always move my entire available attack army (and that despite rarely stopping production), so I say this became a surprise disprove by experiment.

But anyway, the concluding solution is still out until a way to somewhat satisfyingly incorporate the cover effect exists.

1 Like

I will probably correct the cover mechanic example in the post then too. The correct way to explain it would then be:

“A tank of cost 12 might get targeted by 100x0.12 cost enemy infantry at once, being easily destroyed. However, if he appears with a cost equivalent contingent of 100 infantry himself, the enemy also receiving another 100, we have two equal cost forces again. Though an infantry man comes at 1/100th the cost, they still weight 1/10th of a tanks weight. Thus the infantry force of equal cost to the tank will get targeted 10 times more often due to their sprawling size.
The chance that the tank gets targeted suddenly shrinks to 2/11th of what it was before, thus representing about 82% cover. Naively he now has ~5 times as much time to deal with the enemy infantry that is only twice as large, - more than doubling his personal performance. (though, of course, cover will deplete over time) …”

That should make it clear.

Once again you post the solution to a problem I currently mull in my own head and even started to make spreadsheets. I was at “how many tons of what opponent does a ton of this unit kill per round (assuming no fortification)”.

Well played.

Oh, really? When was the the last time, I don’t even remember anything.
I swear we are not surveilling you. The smoke alarm grates because of a defect, no need to look.

1 Like

I shall wait for some good samaritan to compile this into a dedicated calculator so that I may use it years after it becomes outdated, and then ignore it in favor of vibes.

2 Likes

The Excel file provided works for that (it won’t even judge you for ignoring its results), but what kind of calculator do you need?

1 Like

Couple of thoughts/queries:

Was Lancaster’s Laws used in the determinations? In short, the dead don’t fire back…

I’ve found MedVeh to be the backbone of most offensive forces. They kill far more than they are killed, double digit ratios against Inf and most NPR LgtVeh. That said, it does not venture out with Inf in support, or at least LgtSta CAP/HCAP.

When AV is > PEN the amount of damage received by the target is almost insignificant. Death by a thousand cuts still happens, but it is more like death by a couple million cuts when AV > 2*Pen

There is also a supply consideration. In my previous campaign I built some super-heavies that were virtually indestructible and were designed for anti-infantry, but they took far too long to accomplish anything. They didn’t generate enough shots and ran out of supplies. I eventually complemented them with heavy infantry and that combination worked better.

Didn’t know the name for it, but yes, this is solved using similar differential equations. As you can read in the hidden derivation section, the ‘troop count’ actually won’t have a zero solution (->tanh results in error) if that side wins, because nobody is left to shoot the winner.
Which is also btw. why the efficiency calculation is made by counting the survivor percentage of the winner, - even for the loser’s side.

It is still the case that medium and especially heavy vehicles wade through hundreds of infantry before being destroyed if they have enough protection to keep using their guns.
This MBT against PWI for example:


If it could engage infantry 1 on 1 continuously, it would destroy 270 of PWI before on average going down itself. Given the huge cost difference between infantry and tanks, that would still “just” but be 6.5 budget gain, but all in all pretty good.
A HVeh is even more dramatic, going to over 1000 kills before sundown and even better budget performance.:

However, the real battlefield will not have anyone wait on the tank to engage 1v1 with every enemy element of course. That is why, imagining an infinite line of tanks vs an infinite line of infantry, the initial most sensible approach is to have every tank be engaged by its cost equivalent in infantry and see how it fares.

Well, that is what the current “battlefield” equations show.
Due to its cost, for example, the Mammoth Tank above is engaged by 116 PWI, which are cost equivalent to himself. Their kill chance in summary rises to over 11% per turn (you can see it rightmost in screenshot), and while that reduces by ~0.24% per turn due to the tank eliminating enemies, you can quickly guess that since the infantry will take around 10 turns to destroy the tank and the tank requiring almost 50, that the cost equivalent lineup of infantry will win. (calculation says he destroys 11 units before that happens) Then factor in the cost advantage and the win becomes even more decisive.

That being said, I also share the experience that any vehicles of the army tend to survive longer the heavier, while performing really well on kills through all that extra time.
As discussed in the last section of the initial post, the current model of a “battlefield” being just a line of a single unit type against another type, and then testing that for different enemy types to try to get a summary battlefield performance, …it isn’t really realistic. In reality, once the battlefield units mix together, heavier vehicles will be targeted less, so their survivability goes up and their usually stronger weapons get chance to shine more.

That means any real solution will actually land somewhere in between the two shown extremes of
“duel / 1v1 / hundreds of kills for the tanks”
and
“battlefield / cost equivalent forces rush you / tank destroys maybe 10”.
Because in reality only a part of the cost equivalent forces generated by the cost of your tank will actually engage the tank itself, while the rest will instead end up increasing pressure on other field units accompanying him, that then die more than expected.
Thus generates the empirical visual of tanks always being last man standing and also having excellent kill ratios compared to their quickly dying comrades.

The issue with this emergent cover mechanic however is, that the effect is difficult to catch. It could easily be calculated for any individual given fighting force composition (I might at least provide those calculations later on), but what I would really want is a good independant estimation, - like ‘cover benefit vector’ -, that can just tell you how much a certain vehicle will benefit from cover, so you can directly see when it enters ±0 efficiency performance and beyond. That way you could plug in any of the infinite possible combinations of accompanying fighting forces and just know that it will be good without having to test them all in a calculator.
Not sure if it is possible to get there though, but I will keep thinking about it.

You can definitely get me defending the immense importance of TL armor, with the incredible effect that the fourth power (atk/def)^4 survival chance law brings.
That said, it is definitely not one in a million or anything. I mean, the picture of HVeh against PWI above directly lists you the kill chances of the infantry’s guns on the right. It is 0.1%.(hit chance already factored in)
So lets stay with “death by a thousand cuts” then. :slightly_smiling_face:

That is also the case, but since this is another force composition effect, it is impossible to quantify.
When thinking about logistics beforehand, I came to the conclusion that the best that can be demonstrated in theory is a “this works best if it works/{is supplied}” state. So one would have to do the extra work themselves to figure out how to bring that state into being.

I cannot describe how funny I found this, intentional or not, that is math comedy gold.

Now to the poking holes…

Aurora does not calculate ongoing attrition. Instead, combat resolves in discrete ticks, with casualties on both sides applied at the end of each tick.

There is also the matter of overkill. Attacks are directed at elements within formations, and successful kills can exceed the number of units available in the targeted element. For example, if an element contains 100 infantry and suffers 110 successful destruction checks, only 100 casualties can result. The remaining 10 are wasted; they do not carry over to another element or formation.

This brings us to another important distinction: Aurora selects targets by element, rather than distributing individual shots among all units in a formation. Smaller elements can limit casualties from concentrated overkill, although dividing everything into single-unit elements would be an organizational nightmare, not to mention the potential CPU overhead.

Furthermore, ground units do not suffer progressive reductions in combat effectiveness from accumulated damage. A unit either survives a destruction check or it doesn’t; partial damage isn’t retained between combat rounds. I am not saying this effects the calculations for efficiency, but it is often overlooked.

So while I think the original goal of demonstrating unit efficiency was more or less achieved, there is a fundamental limitation in using a continuous attrition model to represent a discrete, simultaneous combat system. The original approach doesn’t fully capture target selection, overkill, or survivability.

I think a modified version of Hughes’ Salvo Model is a better fit for Aurora’s ground combat mechanics, incorporating those considerations into the expected casualty calculations.

So, for consideration… (This may or may not change the efficiency models, working through LaTex with AI “help” has taken more than the time I had for checking that)

Modified Hughes Salvo Model — Expected Casualties

E[K_F]= E\left[ \sum_{e\in F} \min\left( N_e,\, \sum_{g\in G} I_{g,F} \sum_{a\in g} I_{a,e} \sum_{c\in C_a} \sum_{s=1}^{S_c} X_{a,c,s,e} \right) \right]

This formula calculates the expected number of individual unit casualties in defending formation F during one simultaneous salvo.

Where I_g,F is Formation Selection

P(I_{g,F}=1)\approx \frac{\operatorname{round}(S_F\cdot M_{\mathrm{position},F})} {\displaystyle\sum_{f\in\mathcal{F}_g}\operatorname{round}(S_f\cdot M_{\mathrm{position},f})}

This formula calculates the probability that an attacking formation selects a particular defending formation based on its size, adjusted for battlefield position (FL = 1, S = 0.25, RE = 0.05), relative to all eligible opposing formations.

Where I_a,e is Element Selection

P(I_{a,e}=1\mid I_{g,F}=1)= \frac{\left\lfloor S_e\cdot M_{\mathrm{combat},e}\right\rfloor} {\displaystyle\sum_{k\in F}\left\lfloor S_k\cdot M_{\mathrm{combat},k}\right\rfloor}

This formula calculates the probability that an attacking unit selects a particular element within the targeted defending formation, based on the element’s size. Non-combat elements receive a reduced targeting weight

Where X_a,c,s,e is a Kill Check

X_{a,c,s,e}= I_{\mathrm{hit},a,c,s,e}\cdot I_{\mathrm{pen},a,c,s,e}\cdot I_{\mathrm{dest},a,c,s,e}

This formula determines whether an individual shot from a weapon component successfully hits, penetrates armor, and destroys a unit within the targeted element. A successful destruction check counts as one potential casualty; failed checks cause no lasting damage.

Where P_hit is Hit Probability per Component Shot

P_{\mathrm{hit},a,c,s,e}= \min\left(1,\max\left(0,\frac{\lfloor H_{a,c,s,e}\rfloor}{10000}\right)\right)

This formula calculates the probability that an individual weapon-component shot hits the targeted element. The hit threshold carries forward through successive shots and weapon components of the same firing unit, resetting for each new firing unit

Where H_a,c,s,e is the Running Hit Threshold

\begin{aligned} H_{a,0,e} &= H_{\mathrm{initial},a,e}\\ H_{a,c,s,e} &= H_{a,\mathrm{prev}(c,s),e}\cdot M_{\mathrm{offence},a,c,s} \end{aligned}

This formula calculates the running hit threshold for each successive shot, carrying forward across all weapon components of the same firing unit. It resets for each new firing unit, with the initial threshold determined by applicable combat modifiers.

Where I_pen,a,c,s,e is the Armor Penetration Result

P(I_{\mathrm{pen},a,c,s,e}=1\mid I_{\mathrm{hit},a,c,s,e}=1)= \min\left(1, \left( \frac{\operatorname{round}(Pen_c\cdot W,2)} {\operatorname{round}(Armor_e\cdot D,1)} \right)^2 \right)

This formula calculates the probability that a successful hit penetrates the defending element’s armor, accounting for the relative offensive and defensive technology strengths.

Where I_dest,a,c,s,e is the Destruction Result After Penetration

P(I_{\mathrm{dest},a,c,s,e}=1\mid I_{\mathrm{hit},a,c,s,e}=1,\ I_{\mathrm{pen},a,c,s,e}=1)= \min\left(1, \left( \frac{\operatorname{round}(Dam_c\cdot W,2)} {\operatorname{round}(HP_e\cdot D,1)\cdot Q_e} \right)^2 \right)

This formula calculates the probability that a shot destroys a unit in the defending element after successfully hitting and penetrating its armor, accounting for relative technology strengths.

Where W and D are Offensive and Defensive Strength Modifiers

\begin{aligned} Pen_{\mathrm{eff}} &= \operatorname{round}(Pen_c\cdot W,2)\\ Dam_{\mathrm{eff}} &= \operatorname{round}(Dam_c\cdot W,2)\\ Armor_{\mathrm{eff}} &= \operatorname{round}(Armor_e\cdot D,1)\\ HP_{\mathrm{eff}} &= \operatorname{round}(HP_e\cdot D,1)\cdot Q_e \end{aligned}

This formula calculates effective weapon penetration, damage, armor and HP using the attacking unit’s weapon strength (W) and defending unit’s armor strength (D), including Aurora’s intermediate rounding and applicable infantry HP enhancements.

Where P_kill,c,e is Kill Probability per Successful Component Shot

P_{\mathrm{kill},c,e}= P_{\mathrm{pen},c,e}\cdot P_{\mathrm{dest},c,e}

This formula combines the probability of penetrating armor with the probability of destruction after penetration. It is a summary of the preceding two formulas, not an additional combat check.

Where Equal Technology Levels Simplify Kill Probability

P_{\mathrm{kill},c,e}= \min\left(1,\left(\frac{Pen_c}{Armor_e}\right)^2\right) \cdot \min\left(1,\left(\frac{Dam_c}{HP_e}\right)^2\right)

This formula calculates the probability that a successful hit penetrates and destroys a unit when offensive and defensive technology strengths are equal. The technology multipliers cancel, leaving only the weapon’s penetration and damage relative to the target’s armor and HP.

That is of course true, but Aurora combat is also long. Consider I waved a *5 stretch factor that would give plenty of target averaging and reassessment time due to essentially lower/more granual kill throughput; - enough time for other elements to retarget after having overkill minimized. The effect of unit granularity is also suppressed on a large enough battlefield itself where the pixelation steadily approximates away towards the mathematically graspable continuum.

Also true, but a rather fringe minuscule special case influence that is also again minimized further by any scale to the encounter.

Similar for the next point on element targeting. It is interesting, but as you said, unless someone intentionally games that system to make every unit unique, it only ever has a tiny impact on the rare event case where an element class got obliterated entirely.
Add to that that this effect could be bi-directional and cancel out efficiency losses with gains through overkills on your own side, and it seems not worth focusing on when trying to get a real measure of battlefield efficiency.

In other words, if you want to make the point that the influence of these target transition events is significant, I would like to see that demonstrated in what scale of error it really results in.
It is an existing error factor, yes, but even on a smaller battle of 10kt on both sides with perhaps 5 formation segmentations, is it even approaching 1% inaccuracy?

For small unit boarding it might matter, but that is also a situation where you don’t use cost efficiency to begin with.

Hoho, that was a quick conclusion based on assertion without any supporting numbers. Maybe you are right, and I completely misjudge the effects, but without seeing proof I am not just going to follow.

To the formulas:

Yes, some inconsistencies in labeling, like I_{hit} and P_{hit}, and lots of things unexplained (like what is the Q_e on the Destruction Result? why does it vanish later at the kill equation? {which is still then just the same as I did before btw.?}). HP and armor can be rounded to 2 relevant digits, not 1.(hint:genetic engineering)
The Running Hit Threshold I did not understand with so little explanation. Maybe it is simpler than expected, but I see a lot of labeling without hints to why and what it does. It seems like a stepwise/less usable writing of the compact product from the duel-kill% equation on my part before at first, but I don’t know.
Also, if this is meant to measure reducing value from additional shots against the same target, - it is not really just the hit chance that scales in approximation to 100%, but the entire destruction chance. But again, maybe I just got the meaning of this wrong.

Okay, so we went from [atk-formation vs def-formation] over [atk-unit vs def-element] to just [check if shot hit and killed]. So the double sums before the X kill calculation likely take all shots and all weapons of the unit in question. I see from this:

  • If Running Hit Threshold were static, we would be back at my own battlefield equation standard. (in matters of kill chance, - no efficiency calculated here) Every shot seeks a new target.
  • If Running Hit Threshold really reduces multiplicatively, we are looking at a “only one target per unit per turn” / “duels on battlefields” situation that might also be calculated wrongly, because gains from additional shots in duels devalue by the entire elimination chance, not just the hit% part.
    {if ‘all shots on one target’ was actually the intent: is that actually what happens? Would be a huge misunderstanding on my part, and definitely devalue all vehicles even more. Byebye validity of AP-tanks}
  • There was a third view but I lost it.

Anyway, but the part about element and formation targeting is worked in by the extras I_g and I_a. If you just wanted to make a compute of an entire ‘Force 1 v Force 2’ setup, this is how you could expand upon the mere kill chance base indeed. (no efficiency measurement yet though)

As said, I also wanted to add this after thinking about cover some more, though then also solving how to actually extract individual unit-/ and cost-efficiencies from it.
As I explained, doing something like this does automatically include the effects of cover, which is nice, but it comes at the price of only computing a very specific arbitrary “vs these guys” scenario without giving us an objective performance.
It is something, - maybe the best that can be had if you make the enemy force representative -, but as said I wanted to think about possibly getting a numerical extract of how much a unit benefits from cover. Because then, all scenarios are solved, - no need to type in each new theatre or approximate results.

Ah, I should also say that of course your solution (given that the questions about X resolve) of course also corrects for the formation and element granularity error, which was your original intention.
So technically, you could now prove that this is worth doing by comparing some “unit vs theatre” setups after your calculation, and then one that foregoes the granularity limits.

If the result isn’t high enough, I’d argue that it isn’t worth it. Though the above would be more correct, using it would mean any efficiency question can only ever be answered by considering the entire enemy battle force, thus making analyzing formulas like those I gave impossible.
I’d say if the composition effect is <1% error, I’d rather stick to the quick heuristics.

///Oh, and I am traveling tomorrow, so no answer chance until Friday earliest.

The running hit threshold: it isn’t intended to model diminishing returns from shooting an already-destroyed target. It’s Aurora’s actual running to-hit value, modified by the applicable offensive command bonus as successive shots are processed. Separately, each firing unit retains its selected target element for all its shots.

I’ll look at getting some things cleaned up, and I know there was some redundancy in formulas, but did not want to assume the connection, so provided rather than inferred.

Your distinction between mathematical accuracy and useful predictive accuracy is well taken. The old “is the juice worth the squeeze” quandary. Safe travels, I’ll run some comparisons with “actual” forces and see what shapes up between the two and if there is any reason for further consideration.

Ah, that makes it clear about the running hit threshold. I had thought about it on my train drives and it dawned on me that I had confused element with unit (strangely only sometimes) for this part at least, which made me assume the targeting had to be responsible for something more complicated that would have been needed in a v-unit case. But on element levels that is of course gone.

I am looking forward to see numbers on the losses. I will definitely use the formation and element separation should I include a force1 v force2 calculation. Why not if it is more exact? The real challenge would anyway be to turn the incremental nature into formulas. I guess I will still end up simulating that part.