Land Unit Movement

[el cid again]

You said it. Your words make certain you never looked at the database. I assume you are looking at the two books (Numbers, Predictions and War and its sequil) - which amount to outlines - not the database. The entire database is vast - and took decades to assemble and evaluate. You have to combine a limited exposure with unwarranted assumptions (if you read any of the books or the related literature you could not assume what you are saying). Nor would it have been possible to create a model that does what this one does - never mind so accurately or so broadly. This may be the greatest feat of statistical analysis of all time - and if it is not - I have no clue what is?

[herwin]

ORIGINAL: el cid again

You said it. Your words make certain you never looked at the database. I assume you are looking at the two books (Numbers, Predictions and War and its sequil) - which amount to outlines - not the database. The entire database is vast - and took decades to assemble and evaluate. You have to combine a limited exposure with unwarranted assumptions (if you read any of the books or the related literature you could not assume what you are saying). Nor would it have been possible to create a model that does what this one does - never mind so accurately or so broadly. This may be the greatest feat of statistical analysis of all time - and if it is not - I have no clue what is?

My exposure to their work was in the mid-late 1980s. They did release their data then--I was interested in comparing it with mine--and it didn't support their conclusions. Very disappointing.

[herwin]

I've been having an off-line discussion that's relevant to my criticisms of the Dupuy work.

Interactions matter most in combat, but to get the interactions right, you have to get the first (and second) order effects, too. That's a lot of parameters (many of which are not in the Dupuy model) and three times as many samples at least.

Basically, the most effective combat force of a given size is a task-organised team. You can model the combat power of a force as the product of the combat powers of the various individual arms, Ei, each to some power, pi (p sub i, not 3.1415926...), with the sum of the pi over all the elements, Ei, being 1. The combat power of an individual element is a constant, Ki, times its weapons count, Wi. A tactical/operational competence rating describes how quickly the combat power degrades as a function of casualty percentage. A rating of 2 (elite) indicates that the unit loses 2% of its combat power for every 1% casualties. 2.5 is professional; 3 is well-trained; 3.5 is second-line; and 4 is rabble. The elements that contribute significantly are situational, with the pi also being situational.

The goal of the defense is to keep the offense from closing with its positions. A WWII platoon was designed to close with and defeat an isolated squad or section. The three LMGs of the platoon allowed the PL to suppress the target with two and a bit more squads and maneuver with the remainder. That was still not enough for an attack on a good position, and he needed a light mortar to take out the squad MG--retired MG gunners tend to set off the bomb detectors at airports. The defense added support weapons at platoon, and so the company needed more and longer ranged support weapons to suppress them, and so on. That introduced combined arms interactions, and you can't model that by adding up weapons values. Look at the defensive position in detail and decide what you need to deal with all its elements. That's how a staff plans a deliberate attack--if something is left unsuppressed, it can ruin your whole day.

There's a phenomenon in combat--targets are randomly distributed except where you're aiming, where they've taken cover. You had to suppress infantry to be able to advance against it. You could do that with lots of indirect fire artillery, about half as much direct-fire artillery, supporting armoured vehicles, or by bringing about 2.3+ times the direct fire assets to the gunfight. (The figure of 2.3+ was to account for the 5-fold difference in vulnerability to fire between moving and dug-in troops organised for a hasty defense. You needed 3-1 superiority to allow 25% of your force to be maneuvering at any given time. You needed 4-1 to advance against a prepared defense for similar reasons.) An armoured battalion could advance against a dug-in infantry battalion, but it needed supporting infantry to provide eyes for target acquisition and local defense. In open terrain, a tank platoon needed an infantry squad for this; in rolling terrain, it needed an infantry platoon; and in three-dimensional terrain (mountains, urban, jungle, etc.), it needed an infantry company.

The goal of fire was to prevent the enemy from closing with the defended position. This could be done by direct fire on a front or indirect fire on an area. Indirect fire involved multiple guns in a line parallel with the front, each one covering its frontage.

You can unify artillery, rifle, and MG lethality by considering lethal area. The lethal area of an artillery round is an region defined at the boundary by a 50% chance of resulting in a casualty. If you assume targets are uniformly distributed over an area, the expected number of casualties caused by the shell is the expected number of targets in the lethal area. Dug-in troops reduce this by reducing the lethal area of a round by a factor of 5 or more. Being prone was good for x3. A mortar shell of a given caliber has about 3x the lethal area of a gun shell of the same caliber, and a howitzer shell has about 2x the lethal area of the gun shell. The lethal area of a shell in a given category is proportional to the square of the caliber, and you can calibrate the table by knowing the lethal area for a WWII 105mm howitzer shell was 600 square meters (30 meters wide) and for a 155mm howitzer was 1500 square meters (50 meters wide).

Let x be the caliber in centimeters. Then the lethal area of a WWII howitzer shell of that caliber is (approximately) 6*x*x square meters.
A WWII mortar shell had a lethal area of about 9*x*x, and a gun shell was 3*x*x.

75 mm gun--7.5 meters x 22.5 meters
25 lb--17.6x26.4
105 mm--20x30
155 mm--30x50
81 mm mortar--24x24
60 mm mortar--18x18

When you take rate of fire into account, you get lethal area per minute.

Now consider direct fire. Basically, a direct fire weapon has a narrow zone over which it dominates. The effective width dominated is proportional to the sustained rate of fire (10 for bolt action rifle, 20 for semiautomatic, 60 for BAR, 120 for light MG, 240 for tripod-mounted MG). Infantry exists to serve and protect its automatic weapons. As a calibration point, a full-strength UK infantry platoon was able to dominate 250 meters of front in rolling terrain to a depth of about 300 meters using three Bren guns and a light mortar. That's a continuous lethal area of about 75,000 square meters/minute.

Antitank weapons can be analysed the same way.

Deployment matters, by the way. A battalion advancing in column has the forward firepower of a couple of platoons. That's why infantry has to deploy when it encounters resistance, and why fast mobile troops can slow down an advance so much.

[Dili]

Years ago i read that a 120mm mortar round was considered equivalent of a 155mm (M109/M198)round.

[herwin]

ORIGINAL: Dili

Years ago i read that a 120mm mortar round was considered equivalent of a 155mm (M109/M198)round.

Close. The mortar round contains a larger bursting charge due to the reduced stresses it has to handle.

[el cid again]

This analysis appears far more tactical than the Dupuy's work was. And it appears to consider infantry and artillery but not AFVs. It isn't bad - although I don't think IRL targets are really "randomly distributed." I have more experience on the defense - and I can say that the placement of teams and individuals is anything but random in the case of a prepared defense. Indeed, Dupuy found that time to prepair (for larger units) made a very significant difference in the situation - and surely that is the case. Nevertheless, it is a serious analysis - and any model must be a simiplification. I don't see, however, how this approach is going to be applicable in all eras (no MG in the earlier ones) or to larger situations - or to more mobile (or indeed more fixed) situations? [I have had to fight INSIDE a ship, or a tunnel/building complex: it is not at all like fighting on the ground usually is]. You cannot use any mortars or artillery - and even machine guns are almost worthless (unless light - in which case they are still not worth as much as they normally would be). Building a general model is difficult - and I don't mean to discourage anyone from trying to do it well. But in general it is better to be positive and constructive rather than dismissive of the work of giants who came before us.

[JWE]

This is so bizarre.

I was in DC about a week ago, in Gen. Ornstein's office (the DCSOPS), talking about my company's battlefield tactical surveilance system (hoping to get a prime contract). We were chatting about this and that and the subject of DuPuy came up.

The thinking of the military establishment (or, at least, the Abrams/Ornstein wing) is that DuPuy is prescient. There is (and has been for about two decades or so) a 'feeling' within the DOD establishment that the specific professionalism of the German National Arms ought to be institutionalized in our Armed Forces.

Needless to say, I am a true believer.

I've never met Trevor DuPuy, nor his son, but I have read their books and have been impressed with their insight and perception. Seeing as how the US Army School System was redirected according to their precepts, I'm assuming they did something right.

[herwin]

ORIGINAL: el cid again

This analysis appears far more tactical than the Dupuy's work was. And it appears to consider infantry and artillery but not AFVs. It isn't bad - although I don't think IRL targets are really "randomly distributed." I have more experience on the defense - and I can say that the placement of teams and individuals is anything but random in the case of a prepared defense. Indeed, Dupuy found that time to prepair (for larger units) made a very significant difference in the situation - and surely that is the case. Nevertheless, it is a serious analysis - and any model must be a simiplification. I don't see, however, how this approach is going to be applicable in all eras (no MG in the earlier ones) or to larger situations - or to more mobile (or indeed more fixed) situations? [I have had to fight INSIDE a ship, or a tunnel/building complex: it is not at all like fighting on the ground usually is]. You cannot use any mortars or artillery - and even machine guns are almost worthless (unless light - in which case they are still not worth as much as they normally would be). Building a general model is difficult - and I don't mean to discourage anyone from trying to do it well. But in general it is better to be positive and constructive rather than dismissive of the work of giants who came before us.

I've left out the AFV analysis (which was the reason for the whole exercise in the beginning). In the context of WWII armor versus infantry operations, a tank is a tank is a tank. Almost any AFV would suppress a LMG and allow its supporting infantry to maneuver, and almost any AFV was vulnerable to short-range infantry attack from above, below, the side or the rear. The complicated stuff started getting important about 1955.

Planning is very important. It's a lot of the difference between a deliberate and a hasty attack and between prepared and hasty defenses.

The comment about target distribution was actually meant to be humorous. It's the reason proving ground numbers aren't really relevant. Troops respond differently to artillery and air bombardment. Under artillery bombardment, they go to ground; under air bombardment, they scatter away from the impact point of the bombs headed at them. It takes a unit a lot longer to recover from an air bombardment, since the leaders have to go around a larger area to round their men up.

The applicability of the approach to other eras is quite good. I've used it in analyses of Napoleonic and Civil War operations.

Remember CityFight? I used to answer questions on it. 8)

[herwin]

ORIGINAL: JWE

This is so bizarre.

I was in DC about a week ago, in Gen. Ornstein's office (the DCSOPS), talking about my company's battlefield tactical surveilance system (hoping to get a prime contract). We were chatting about this and that and the subject of DuPuy came up.

The thinking of the military establishment (or, at least, the Abrams/Ornstein wing) is that DuPuy is prescient. There is (and has been for about two decades or so) a 'feeling' within the DOD establishment that the specific professionalism of the German National Arms ought to be institutionalized in our Armed Forces.

Needless to say, I am a true believer.

I've never met Trevor DuPuy, nor his son, but I have read their books and have been impressed with their insight and perception. Seeing as how the US Army School System was redirected according to their precepts, I'm assuming they did something right.

I'm not arguing that they were wrong; It's just that their statistical expertise was sorely lacking.

I'm a neuroscientist, lecturing in a school of computing--I was an adjunct assistant professor of computer science in the states when I was doing my neuroscience PhD. There are a number of statisticians on the faculty, but I end up teaching the classes on practical statistics--not because I'm a statistician, but because I've used statistics in my work and because I've consulted in statistical programming. It's easy to get the statistics wrong if you lack the practical experience of doing it in the real world.

begin(rant) The other thing--and my colleagues made sure I understood this from day one back when I was a wet-behind-the-ears mathematician in my first job as a system engineer--is that interactions matter (a lot) in combat. It's rock, scissors, paper, all the time. To understand an interaction between three explanatory variables, A:B:C (for example, rifle caliber weapons, AFVs, and indirect fire artillery), you have to understand A:B, A:C, B:C, A, B, and C as well. If there's non-linearity, you also have to understand A^2, B^2, and C^2. That means a simple game model would need ten parameters. Now add tempo, weather, terrain, mission, and troop quality to the mix--they also matter a lot--now you have ~250 parameters in your model, and you need a minimum of 750-800 samples to calibrate the model. That's a *lot* of samples given the available literature, and I haven't included air power yet, which doubles things yet again. The QJM modeling approached the problem using a Bayesian approach, by taking account of professional expertise, but all that gave you was ~250 guesses. Until you have a *lot* of data, all you have is a bunch of guesses. Yet they were operating like those guesses were rock-solid--I had independently done much of the same work, and their values, especially the interaction terms involving air power, tempo, and troop quality, were not very good. A good model provides you insight into the processes you're modelling. That means you should try to incorporate what you already know into the model, and you should think about what the model is telling you. QJM didn't have that property. It didn't tell you *why* it gave you good results when it gave you good results. Their interaction terms just didn't make sense. An opaque model is not a useful model. It's just an oracle that is likely to fail you when you need it to work. end(rant)

Now I have to get back to thinking about why my generalised additive model of neurone spiking is blowing up in my face.

[Dili]

Close. The mortar round contains a larger bursting charge due to the reduced stresses it has to handle

 
One of reasons i remember was the trajectory, because that the fragmentation patern in a mortar round is better than in an howitzer.

[herwin]

ORIGINAL: Dili

Close. The mortar round contains a larger bursting charge due to the reduced stresses it has to handle

One of reasons i remember was the trajectory, because that the fragmentation patern in a mortar round is better than in an howitzer.

The trajectory is a problem--the shell is vulnerable to wind currents and loses accuracy. The larger bursting charge is because the muzzle velocity is lower.

[hueglin]

Now consider direct fire. Basically, a direct fire weapon has a narrow zone over which it dominates. The effective width dominated is proportional to the sustained rate of fire (10 for bolt action rifle, 20 for semiautomatic, 60 for BAR, 120 for light MG, 240 for tripod-mounted MG). Infantry exists to serve and protect its automatic weapons. As a calibration point, a full-strength UK infantry platoon was able to dominate 250 meters of front in rolling terrain to a depth of about 300 meters using three Bren guns and a light mortar. That's a continuous lethal area of about 75,000 square meters/minute.

How was the value of 250 m by 300 m arrived at? Is it possible to break it down into smaller elements? For example, how would you calculate the area dominated by a 10 man section firing bolt action rifles?

[JeffroK]

Divide by 3 and make a SWAG about the effect of the Bren & Mortar.

Probably about 50 metres, but to a much lesser depth

But the section had a Cpl with a Thompson or Sten, 6 men with SMLE's and a Bren group with a Bren and 2 SMLE.

[el cid again]

ORIGINAL: hueglin

Now consider direct fire. Basically, a direct fire weapon has a narrow zone over which it dominates. The effective width dominated is proportional to the sustained rate of fire (10 for bolt action rifle, 20 for semiautomatic, 60 for BAR, 120 for light MG, 240 for tripod-mounted MG). Infantry exists to serve and protect its automatic weapons. As a calibration point, a full-strength UK infantry platoon was able to dominate 250 meters of front in rolling terrain to a depth of about 300 meters using three Bren guns and a light mortar. That's a continuous lethal area of about 75,000 square meters/minute.

How was the value of 250 m by 300 m arrived at? Is it possible to break it down into smaller elements? For example, how would you calculate the area dominated by a 10 man section firing bolt action rifles?

Skilled bolt action riflemen may be about as effective as infantry as American infantry with LMGs are. There is an incident at Anzio indicating virtual equality. The problem with bolt action rifles (which I personally prefer in infantry combat and would use today if required) is psychological. A US Army historian, independently confirmed by German and Soviet studies, found that men with an LMG believed they were effective - which is a big part of the switch to using them (note, however, that an M-16 no longer fires full auto). If you train your bolt action riflemen properly - they will dominate the tactical space. Germans facing professional British infantry in 1914 believed they faced motor machine gun battalions. And in the Balkans British infantry fired counterbattery effectively against mountain guns (when the Britis had no artillery at all) - using bolts. Not many armies train to that standard - but I was taught it by the Marines - and I wasn't even a Marine (but a sailor in a Marine school). I trained Vietnamese militia to USMC standards - and they integrated perfectly. Since their modern weapons had not arrived when we had to fight - we used WWII rifles instead.

[hueglin]

ORIGINAL: JeffK

Divide by 3 and make a SWAG about the effect of the Bren & Mortar.

Probably about 50 metres, but to a much lesser depth

But the section had a Cpl with a Thompson or Sten, 6 men with SMLE's and a Bren group with a Bren and 2 SMLE.

Thanks for the input. I'm referring to a section from 1914 though, rather than 1941. I'm doing a mod for WPO, but I'm posting here because the game engines are the same, but WITP has a much more active community.

[JeffroK]

It doesnt have to be bolt action, the Self Loading Rifle (FN FAL) used by Australians in Vietnam and up to the Steyr being used was a far better Infantry weapon than the M16.
 
Firing single shots makes the soldier think and aim, have auto fire sees them (natuarally) want to get off as many rounds as they can.
 
Plus, IMHO, the 30cal/7.62mm/.303 round is far more lethal than the 5.66mm. So what if I carry less ammo, I'll be more careful with it!!

[JeffroK]

ORIGINAL: hueglin

ORIGINAL: JeffK

Divide by 3 and make a SWAG about the effect of the Bren & Mortar.

Probably about 50 metres, but to a much lesser depth

But the section had a Cpl with a Thompson or Sten, 6 men with SMLE's and a Bren group with a Bren and 2 SMLE.

Thanks for the input. I'm referring to a section from 1914 though, rather than 1941. I'm doing a mod for WPO, but I'm posting here because the game engines are the same, but WITP has a much more active community.

Now you tells me[8D]

Are you talking theoretical or in practise??

[hueglin]

You definitely feel like you have more firepower in your hands with the FN. The Canadian Forces had a version also, until converting to the Canadian version of the M16 - the C7. Being a rather small framed person, I must confess I prefer the C7 for its lightness and ease of handling. Even though it has automatic fire capability, the CF trains to fight with it as a SLR - aimed shot, using the scope with which our weapons are now fitted

[hueglin]

I'm trying to come up with some standard formulas to use to create the anti-soft values for the mod. herwin has given me some excellent info for calculating artillery - which he posted on this thread. I am thinking of having a device that is a basic 10 man bolt action section (then a variation for each nation (e.g German Infantry Section, British Infantry Section). I am going to add MMGs as separate devices (crew served weapons). When LMGs are introduced later in the war I am just going to have them appear as other separate devices (even though they were integrated into some of the sections). I think that will be the simplest and most effective way to go.

[Dili]

Self Loading Rifle (FN FAL) used by Australians in Vietnam and up to the Steyr being used was a far better Infantry weapon than the M16

 
In most instances i dont think so for the FAL. The 5.56 Cal(despite its various shortcomings) improved marksmanship that in some armies the score levels had to be remade. One of them was the British Army with the unreliable(still?) but precise SA80.