HE Effectiveness

[rexford]

British reports WO 291/113 and WO 291/146 are presented on the John Salt site and provide the following results:

1. Ricochet fire effectiveness

A. 105mm with delay

100 rounds fired, 95 effective, 65 wooden targets hit and incapacitated (65% for effective target hits/rounds fired)

B. 75mm with delay

38 rounds fired, 29 effective, 33 wooden targets hit and incapacitated (87% for effective target hits/rounds fired)

C. 25 pounder

100 rounds fired, 82 effective, 26 wooden targets hit and incapacitated (26% for effective target hits/rounds fired)

D. 25 pounder (timed airburst)

100 rounds fired, 85 effective, 9 wooden targets hit and incapacitated (9%)

2. Lethal areas for each round

25 pounder timed airburst 240 sq ft

105mm ricochet 3000 sq ft

75mm ricochet 2500 sq ft

25 pounder ricochet 720 sq ft

3. Impact of Velocity Upon HE Accuracy

Normal 75mm HE has 90% horizontal zone of 35.5 yards for distance dispersion

Super 75mm HE has 90% horizontal zone of 59 yards for distance dispersion

Increasing 75mm HE muzzle velocity increases the ground scatter, which reduces accuracy against ground point targets.

4. 25 pounder performance increased when instantaneous DA was used, 200 rounds, 159 effective, 61 targets incapacitated (30.5%) and lethal area of 2300 sq ft.

5. Firing test used wooden targets in a 9 x 9 grid 100 yards wide and 150 yards deep.

6. 75mm HE outperforms 105mm HE during ricochet firing tests in terms of casualties per round fired.

7. Slower 75mm HE appears to be more accurate against ground points in terms of getting a higher percentage of the shots within a given distance of the target point.

[JasonC]

The conclusion about 75s vs. 105s does not follow from the data. The test had 100 shells 105 fired and only 38 by 75. The measure of "effectiveness" is how many wooden targets get penetrated by shrapnel, correct? Nothing about how many times each. The 105 case will obviously show more "overkill" because more rounds were fired in absolute terms.

The first round has the most chances of piercing several previously unhit boards. Each subsequent one will show diminishing returns, even if the round is accurate and pierces some boards in the target zone - because they happen to have already been hit. This makes different sample sizes non-comparable. You can't divide overall effect by rounds fired and get effect per round, because the effect of each round is dependent on its order in the diminish-returns sequence. (Mathematically speaking, it is an integral rather than a multiplication).

In addition, once the portion of the hit boards exceeds about 50%, the remaining data will be very "noisy", in terms of any relevance for questions like HE effectiveness of different rounds. Because by then large portions of the target area are saturated. You are mostly measuring how evenly spaced the fall of shot is in a particular barrage, by that point, not how much each round does. Since perfect spacing can cover the ground with less effective rounds, once near saturation, and overkill barrages will still leave some areas unhit if the rounds just land "clumpy" enough.

If comparing barrages, the comparison should be made between samples of the same size, same number of rounds. If comparing single rounds, one would have to chance the boards after each shot (impractical in impact zones). Smaller samples farther away from the saturation point will be more meaningful for shell comparisons than large, overkill barrages that go far into diminishing returns, and mostly measure achieved scatter-clumpiness, not shell effect.

Interesting data, but they don't bear the weight of that part of the conclusions.

What stands out far more than the 105-75 question is the poor performance of the 25-lber compared to the 105, with equal numbers of shells fired. Not entirely surprising since the 105 carries 2 1/2 times the HE burster, but a useful corrective to sometimes inflated claims made for the 25-lber.

As for the poor performance of 25 lber VT, I would suspect the small burster requires a low height to be at all effective, and limits the any possible gain from VT. In the test case, probably they were set to go off too high. The benefit of being up in the air is obviously related to the absolute blast of the round.

A big shell wastes more from a ground detonation, while small ones in higher numbers may achieve most of the desired dispersion effect without (much or any) elevation, from scattered impact points. To see this, imagine the same HE burst weight in one 155mm shell or in 8 scattered 25-lber shells (~14 lbs HE in either case). Lifting the one biggy up in the air will improve the coverage far more than lifting up the 8 little guys.

I hope this helps.

[ 08-27-2001: Message edited by: JasonC ]

[Username]

The test seems to be comparing HE effectiveness from ricochet (delay fuze) and airburst (time fuze).

I agree that maybe you can compare the 105mm and 25 lbr because they both fired 100 shells. even then it depends on some data being explained.

Part of the problem at looking at data like this is that all the test parameters arent present. Effectives is not defined.

Did they, fire one round. Stop the test and walk around to all the targets? Did they circle the hits so that the next round wasnt counted as a loser because it hit an already hit target? Were fresh targets put down once they were hit?

What is an effective? A round that got at least one hit? Whats that say about the 75mm then. Is an ineffective a fuze that does not detonate the shell?

Ricochet firing is optimized by having the shell angle up. The side of the shell ideally is pointing down at the targets like some flying claymore. It will then deliver a decisive blast. Unfortunately, they wont ricochet after a certain angle.

In WWI, the real shrapnel shells were like shotguns and you wanted the front of the shell to be pointing at what you wanted to kill. The timed airburst does not want this but rather to fly past the infantry targets and go off actually behind them.

Interesting but needs more work. Is this from your book?

Lewis

[rexford]

We just came across this HE test data.

Granted that the test parameters are barely mentioned, but the results still show 75mm HE being very effective on ricochet shots.

If new targets put out for each gun and perforated targets are identified after all the rounds are fired, guns with more rounds may penetrate targets many times. But results still show 75mm hitting more targets per round than 105mm.

If all guns had fired 100 rounds the results would probably be more comparable, but even then we don't know how range estimates and ground aim was varied.

Still, 75mm HE may be equal to or better than 105mm HE at casualty causation due to factors we have not identified.

25 pounder HE about the same HE filler weight as 75mm HE, 1.75 vs 1.70 pounds, but 75mm has lower total weight. So 75mm probably puts out a greater number of effective fragments (equal explosion blows 75mm shell into more small pieces travelling at higher velocities).

The fact that 75mm HE outperformed 105mm HE in a test suggests much.

Also note that lethal radius of 75mm is only slightly smaller than 105mm HE, 28' vs 31'. Lethal radius data suggests that each shot may have been analyzed separately.

I'll try to obtain entire report to see if key data is presented.

[JasonC]

I'm pretty darn sure they didn't walk the range after every round. You have to see the artillery practicalities here. The impact zone is an off limits, dangerous place, and they naturally would set aside a time to examine it. A time, not 238 times for 238 shells fired. It is not like rifle accuracy scoring, where men in the pits can look at the target safely as often as desired just by retracting them. Effectives probably refers to non-duds that landed within the targeted zone, as opposed to "shorts" and "overs", etc.

And no, the data do not say "greater effectiveness per round for 75", they say a random pattern of 38 casualty zones covered 15000 square yards about half as well as a different random pattern of 100 casualty zones. The 105 shells hit nearly twice as many targets. If you place the 75 shells in a perfect array, the 38 will hit every target with CZs of only 11 yards radius. In fact, they hit 33 boards.

Artillery rounds do not fall in perfect arrays, they fall in random patterns of clumpiness. Above any significant amount of saturation that is all you are measuring - how close that particular random pattern of impact points was to a perfect array. Which is just noise.

You have no idea whether the first 38 105 rounds pierced moreor less than the 33 boards pierced by the 75 rounds. You now all 100 pierced 65 or whatever. But the last 30 may have pierced all of 4 new ones, while putting many more holes in already holed boards. And the first 38 may have pierced 50 of the 65 for the first time. The data simply do not say.

You cannot divide a diminishing returns integral by its length and get a meaningful average, expecting it to be true for the first shell and for the last. The more boards have already been pierced, the fewer *new* ones each subsequent shell will pierce. The 105s pierced twice as many with a longer barrage; you have no idea if a 75 barrage of equal length would have pierced the same number, less, or more. You can't assume it would have pierced 100/38 = 2.63 times as many as 38 shells did, because the later ones would pierce fewer fresh boards and more already pierced ones.

[ 08-27-2001: Message edited by: JasonC ]

[Simon Fox]

<BLOCKQUOTE>quote:</font><HR>JasonC wrote:

What stands out far more than the 105-75 question is the poor performance of the 25-lber compared to the 105, with equal numbers of shells fired. Not entirely surprising since the 105 carries 2 1/2 times the HE burster, but a useful corrective to sometimes inflated claims made for the 25-lber.<HR></BLOCKQUOTE>Not at all. Lorrin's comments are directed at the part of the study related to ricochet fire vs timed airburst which he has selected for presentation here. These and other studies presented at the same site give data for other fusing systems including VT. As far as I can see John Salt does not present the conclusions of the study with regard to relative performance so I assume rexford has provided his own. As you point out it is a little difficult to discern exactly how they collected the data. By reducing the effectiveness of a round to the single determinant of burster charge weight you are ignoring the effect of fuse type and effectiveness which is dramatically shown by this study. It seems that at the time of the study the 25pdr fuse types relevant to timed airburst and ricochet fire were hopeless.

[FFE]

Correct me if I'm wrong. It appears the 75mm test used 40 wooden targets. The rest of the tests used 100 wooden targets.

[rexford]

It is true that one can't say for certain that 100 75mm HE shots would penetrate "100/38" times as many boards.

The interesting point is that the Sherman 75mm is so close to 105mm HE. Note the lethal area for 75mm and 105mm HE using ricochet fire is very close, a few feet difference in radius.

Lethal radius would have to be determined from a single shot, more or less. Or so it would seem.

Note also that 75mm HE gets more wooden targets incapacitated per effective round than 105mm. So 75mm seems to be spraying fragments around better (29 effective rounds, 33 targets hit with effective fragments).

Since we don't know the ranges of the tests, the projectile height above ground, and range settings and variations, etc., definite conclusions are impossible to make.

But the fact that 75mm HE is close to or superior to 105mm HE in any respect, even with fewer shots and extrapolation of results, is very surprising. This suggests that ricochet fire may introduce some factors that we did not previously consider, since everything else says that 105mm HE is worlds above 75mm HE.

Statistics aside for the moment, when 75mm HE outperforms 105mm HE in a ricochet fire test it seems important enough to take note of. This is a highly unusual result.

The other test result, where slower 75mm HE is more accurate than higher speed HE against ground point targets (less horizontal dispersion), confirms what the Germans and our group calculated.

Higher velocity HE has greater ground scatter, which can sometimes be good but can also be less effective against a small group far from others.

[Username]

I think that theres alot that needs to be known about the test. Rex is going off on his well known tangents to stretch anything to back his assumptions.

The height of the ricochet might not matter as much as he thinks. The targets, if square, flat wood might be lined up like grave stones. The shell might be fired at the broad sides of the lot and when it bursts, the sides of the shell is spraying its lethality at 'thin' targets. To be precise, the targets are not dimensionally correct and dont approximate a human body well.

Again

A. 105mm with delay

100 rounds fired, 95 effective, 65 wooden targets hit and incapacitated (65% for effective target hits/rounds fired)

B. 75mm with delay

38 rounds fired, 29 effective, 33 wooden targets hit and incapacitated (87% for effective target hits/rounds fired)

95/100 vs 29/38?

The data is too vague and it is worth persuing. It isnt worth making conclusions about.

Lewis

PS put a link to the site if you can

[ 08-28-2001: Message edited by: Username ]

[rexford]

Ah, but it is worth making conclusions about, which is what the scientific method is all about, isn't it?

Get some data, draw a hypothesis even if data is limited, and go after more to see if original theory holds water.

We have made some considerable discoveries by taking limited data that really doesn't appear to support a theory with any real mathematical certainty, producing a theory and mathematical relationship and then going after data to test the theory.

In some cases the original theory is crap, in many others the original conclusion holds really well. Drawing a conclusion from limited data is a research aid that focuses attention and is a valuable approach.

I have been doing research for over thirty years and use the above noted approach all the time. I guess that the approach we use may not come across as valid to some readers, but it works well from our angle. Unconventional, but has paid dividends.

It is also more beneficial to explore possibilities than to gripe over limited data.

[rexford]

Ricochet fire data shows that 105mm HE is better than 75mm at hitting things, 95% to 80%, but 75mm is better at incapacitating targets, 87% to 65%.

Means that 75mm frags have more penetration than 105mm, and more velocity at impact.

75mm HE is simply the best!

[Username]

Simple things being the scientists specialty.

[JasonC]

Rexford, it doesn't matter how many times you repeat the false inference "the 75 is more effective per round in this test", it still will not make it so.

Believe me, I am not merely being contentious. There is an actual principle of statistical reasoning involved, and a well known one. I think you are perfectly capable of seeing and understanding the principle, and that when you do so, you will not regard this test as evidence of higher HE effectiveness for 75mm vs. 105mm. You might still think that for other reasons - that is not the point. The issue is whether -this test- shows that, or doesn't.

I will explain the reason the inference is a false one with a parallel example, that commits the same fallacy.

One Sherman tank is set up, and two panzerfausts are fired at its front, each penetrating the tank and destroying it.

Then two Sherman tanks are set up, and 50 rounds of 75L70 AP are fired through each one, leaving metal splatter and twisted junk where each tank used to be.

The tester then proclaims that 2 Panzerfaust were as effective as 50 rounds of 75L70 AP. He reaches this conclusion by taking the number of dead tanks (1 vs 2) and dividing by the total rounds fired (2 vs. 100), arriving at the "conclusion" that it take 25 75L70 AP to kill a Sherman, while Panzerfausts only need two shots to do so.

What is the fallacy in this example? There is no correction for saturation, for overkill. (In either case). The last 75 AP hit a tank that was already dead. This does not mean that it could not have killed the tank if it were still alive. But the act of simply dividing the measured dead tanks by the number of rounds fired pretends that it wouldn't have, since it awards no -new- kill to that round.

In the artillery test, the count was the number of boards penetrated. If they were penetrated 30 times over it made no difference. Once a large portion of the boards were already penetrated, the remaining shells were firing at "dead Shermans" (pierced boards). They could be as effective as you please against them, just as the 75L70 AP is as effective as you please against a plain Sherman. But no new hit awards were allocated to them, because they pierced boards already pierced, just as the dead Shermans were already dead.

The test tells us that 38 75mm rounds pierced 33 boards, -not- that the chance of each round penetrating -one- board is 33/38. And the test tells us that 100 105mm rounds pierced 66 boards, -not- that the chance of each round penetrating one board is 66/100.

It is likely that the first round fired in both cases pierced the largest number of boards, because it had the largest number of virgin boards to hit. The area it could impact with fragments and get a new virgin board for it, was very large.

But by the time 30 boards have already been pierced, the area in which a round could fall and produce no new pierced boards, but only overkill on ones already pierced, has become important. And it obviously gets higher the later in the sequence of shells you go. The amount of overkill rises with each shell in the count.

This is a well understood mathematical relationship. The same sort of thing happens in economics and even in physics all the time, in various ways. It takes a general name of "diminishing returns" from its economic manifestation. Probability statistics refers to it as "saturation".

The way in which it must be handled mathematically is also well known. You need a -measure- of the fall-off of the "return" with each subsequent element. Then you must -integrate- over that measure, instead of adding identical numbers for each term.

Adding an identical number for each term is merely a special case of this, using a "flat" measure of 1 for all terms. Which is the same as multiplying. An average obtained by dividing a result by the number of terms is only meaningful in that "flat" case.

Otherwise effectiveness is meaningful only for an ensemble of a given size (e.g. 100 rounds fired) - and is then measured by the integral - not "per item". (I.e. 66 boards for 100 rounds, not .66 boards per round).

In other words, the damage expected from each subsequent shell would have to be the same, and independent of the number of shells already fired and what they had already done, for a simple division of dead things by shells fired to give a meaningful average.

And we know that is not the case in this trial. Pretending the "average boards hit per shell" is meaningful with different amounts of overkill from barrages of different sizes, is simply a statistical mistake.

Rexford noted that the result is "surprising", that it is out of line with other data that shows the greater effectivenss of 105s in most cases. But in fact there is nothing whatever in the data that is out of line with those statements. The 105 barrage broke twice as many boards. There is only something in Rexford's conclusions that is, and it is something we know, as a matter of math and statistics, is an illegitimate inference from the data it was based on.

He only got something "surprising" by dividing 2 dead Shermans by 100 AP shells (a big barrage with lots of overkill for the later shells) and comparing it with 1 dead Sherman from a few Panzerfausts (a small barrage with much less overkill in the sequence of shells).

I have nothing against trying to reason out of the data what the data can support, and nothing against considering a surprising conclusion. But the data have to legitimately support the conclusion. And here (for the 75 - 105 inference alone, I mean) they just don't.

[rexford]

We really don't know how shots were distributed over the target area with regard to changes in ground aim point or lateral distribution, so definite conclusions are not possible.

If 38 rounds from 75mm gun are distributed in same proportion and area as 105mm, then 75mm HE averages a greater number of effective hits per shot than 105mm.

So, if shots in field during combat are distributed in accordance with test results and targets are distributed in same manner as wooden targets, 75mm HE using ricochet fire has higher probability of successfully penetrating an infantry target.

105mm has higher chance of hitting someone, but 75mm has higher chance of doing something bad to someone.

It doesn't matter much if targets are hit repeatedly, since the test measures how many different targets were hit and/or incapacitated, which is a function of statistical area coverage.

75mm HE incapacitated 33 targets with 38 shots, 105mm HE did in 65 targets with 100 shots. If distribution of shots is similar, 75mm has higher statistical probability of doing something useful.

I would assume that British were smart enough to make sure that 38 rounds from 75mm HE were distributed in similar fashion to 100 rounds from 105mm. If that is case, test results can be compared and analyzed statistically.

And 75mm HE outperforms 105mm.

Responding to an earlier comment:

Regarding ricochet height and aim point, these are very important since number of effective fragments is very dependent on distance and angle from explosion.

[rexford]

I would also note that the previous comments on my analysis overlooked the similarity of the lethal areas for 75mm and 105mm HE using ricochet fire. This is very surprising and is not consistent with test results for ground blasts, so is something new that the statistical naysayers overlook in their griping.

And it is surprising.

Definition of lethal blast area requires that results from individual shots be analyzed, a point I made in past posts which seems to have been passed over.

Ricochet fire is not ground blast, and it is quite possible that 75mm HE has advantages over 105mm that show up in ricochet tests and do not show up in ground blast tests. So the surprise is due to different conditions for explosion location, and it may not be proper to use ground blast data to predict performance of ricochet fire.

If 75mm has smaller lethal area than 105mm and 75mm critically hits more targets per round, than this suggests that 75mm is putting out a higher number of effective fragments per round during ricochet fire.

Which suggests that ground blast data does not hold during ricochet fire. Which is a very important and surprising result/theory, and places some common assumptions in doubt.

That is why I made such a big deal out of 75mm being more effective than 105mm, because it is totally out of line with what we expect. I expected 75mm to be inferior to 105mm, to tell the truth.

[rexford]

The Panther vs Panzerfaust analogy is way off base, is not correct and misses the point.

We're talking area coverage here, for 75mm vs 105mm HE, so hitting the same point repeatedly is expected. And is okay.

75mm ricochet HE hits a larger number of targets effectively than 105mm, per each shot, which implies greater area coverage for effective shots.

105mm may hit more targets per shot, but 75mm hits more targets effectively per shot. A critical distinction.

[JasonC]

Nope, sorry Rexford, in any stat class that last answer would rate you an "F". The distributions could be exactly the same, and it won't account for the overkill at all.

You can see this easily if you imagine the following case. There are 66 targets exactly, and 100 105 rounds get them all. 10,000 shots would also get them all. The same number of hits from 100 rounds as from 10,000 rounds would be perfectly compatible with the exact same distribution in the scatter of the shots. Yet one would leave your "average per shot" measure at 1/100th of the other one. Without reflecting any underlying difference in the shells, what they typically do, their distribution, or anything else. The subsequent 9,900 shots are only breaking already broken boards. They have no business being including in the denominator of a supposed "per shell average".

Saturation is a seperate additional issue from how wide the scatter of shots is. In my AP vs. faust example, both could have 100% hit and kill probabilities. But one just has more overkill than the other, as a direct result of more shots fired. The added overkill does not allow any conclusions about a supposed lower average effectiveness for a single round fired, as claims about per round effectiveness try to pretend.

You simply can't divide a declining curve (-virgin- boards hit, which declines as the number already hit increases) linearly by a number of shells and get a meaningful per shell average, that will hold ("allow comparison") across barrages of different sizes. What the last shell can do is simply not independent of what the shells before it have done.

To see the "not independent" point, imagine only 1 board in the upper right corner of the impact area remains unpierced, after some length of barrage. Then any subsequent shell can only gain a single additional pierced board, and can only do so in the low probability event that it lands close to that corner. Whereas the first shell could gain many pierced boards for an impact anywhere in the target area.

[rexford]

The key question when it comes to area coverage for ricochet fire test is whether 75mm HE would hit at least 65 targets effectively if it were given 62 more shots.

Were the 33 targets that were critically hit by 75mm HE the most that could be hit, and would adding more shots add to the hit total?

[JasonC]

No rexford, the 75 does not hit a larger number of targets. It hits a smaller number of targets, right there in the data. 33 vs. 66. Your error enters when you effectively assume everything is independent and linear (as a matter of the math), to justify -dividing- those results from -collections- of shells, into a supposed -per shell average-.

But the first shell does not hit the same number of boards as the last. If you repeat the trial 100 times with 100 shells each time, and look after each one, the average hits of all the first shells will be consistently higher than the average from all the last ones. Comparing the first shot into all virgin boards with the last one into a region with 65 boards already pierced, is not an "apples to apples" comparison.

Imagine the diminishing returns goes as a power law, say 2nd power, as in many distributions of cumulative random processes, OK? Then what added effect should one expect from doubling the number of shells fired, well out into the diminishing returns progression? Not twice as many hits. Because their are fewer virgin boards left for the second half of the shells. The overkill effects they will face will be large than those in the first half. If that diminishing returns goes as 2nd power, you'd expect doubling the size of the barrage not to double the number of pierced boards, but to raise it 41% (square root of double).

How could one tell if that is how it works? You determine the rate of fall off, by comparing the results of firing more and more shells of exactly the same kind. 10 105 rounds averages 20 hits, say, then 40 averages 40 hits, or however it goes. It will not go linearly, beyond the first ones, and especially not after a significant portion of the targets are already hit.

Suppose the diminishing returns did go as a square root power law. Then how many additional hits would you expect from increasing a barrage size from 38 to 100 shells, if the shells were exactly the same? Not 100/38 = 2.63. But that ^.5, or 1.62 times. If the 100 hit 66, you'd then expect 38 to hit 66/1.62 = 41.

If only 33 are seen hit by another shell types, you'd conclude -only for barrages of that size- that the first shell types was 23% more effective. But that relative effectiveness factor would -change- as the size of the barrages changed, and as each approached saturation.

The difference would in fact be -largest- for a single shell of each type, and would go away entirely as the number of shells in each compared barrage goes to infinity. If you fired 2m shells at the target area, you'd hit the same number of boards in either case - all of them, and not one board more.

[JasonC]

Certainly the 75 would hit more than the 33 it did if it had the 62 extra shots. But how many more? All 33 more to match the 105s? 67 more to exceed them? Or 15 more to underperform them? The issue that will determine which would happen is the way in which the "returns" fall off as the number of shells increase. That is an empirical question, and dependent on how close the number fired are to saturating the target area.

But there is no good reason to assume it will be linear, and in fact there is excellent reasons to assume it will be less than linear. Because we know the later rounds face greater chances of overkill "waste" than the early ones. If I had to guess without additional data, I'd assume some sort of power law, and start with a squaring one as a first approximation. But it would be just a guess, without additional empirical data. It could go as the log of the shells fired. Or the cube root. Or the 2/3rds power.

But there will be some diminishing returns function, and it will become significant as soon as additional shells face a significant chance of overkill waste. Without knowing what it is for the number of boards and number of shells and shell types used (since the degree of saturation will vary with those things), one is left comparing barrages of the same size.

Incidentally, this issue was well known to the British artillerist during the war. And because of it, when they came up with their barrage planning 25-lber equivalences for shell types and estimated volumes of fire needed to achieve this or that tactical result, they avoided trying to say anything about cases with high saturation. They restricted themselves to an estimate of the volume of fire needed to produce ~10% losses for exposed infantry, and left everything above that level as "unknown".

They put in that cut off because they knew that saturation considerations come in strongly above that, and they wanted to stick to the "close enough to linear" region. Where each shell is about as useful as the first, because most of the places it might come down are "virgin" for that barrage. The whole set of estimates was meant to be used by artillery planners to add (linear) firing guns and lengths of barrages to reach desired levels of effect. And for adding to give meaningful numbers, they had to stick to the region before saturation effects become large.

[rexford]

It took awhile but it came through the daze I have been in these last few days.

If 75mm HE rounds were fired under same conditions as 105mm, but only 33 targets were effectively hit, then there is no reason to believe that taking more shots would add targets.

In other words, first 38 shots by 105mm probably penetrated around 65 targets, and next 62 shots kept hitting same targets over and over again.

(if 105mm took 1000 shots, number of impacted targets would still be about 65 but percentage would be so low 20mm HE would look more effective)

So 105mm might impact twice the targets as 75mm, which would seem reasonable.

Got it, and appreciate your patience.

Now, 75mm has about the same lethal radius as 105mm, which is interesting. This suggests that 75mm effective fragment density is less than 105mm (about half) but reaches out to almost the same distance.

Finally, British reports show that slower HE rounds are more accurate.

Thanks for comments, which I will be sure to add to my HE comments on other sites. My grandmother called me a hardhead, and this may be true more than I care to admit.

[rexford]

I wouldn't be surprised if British didn't stop 75mm tests at 38 shots because increase in targets damaged with additional shots had stopped.

If logic in last few posts holds, then ricochet 105mm impacts about twice the targets as 75mm, and instantaneous burst 25 pounder HE impacts almost as many as ricochet 105.

British report also adds that airburst rounds are no more effective than groundburst against targets in slit trenches despite theoretical 2:1 advantage in effective fragments.

[JasonC]

Right, glad it helped.

"This suggests that 75mm effective fragment density is less than 105mm (about half) but reaches out to almost the same distance."

That is somewhat plausible. The 105 has a large amount of HE, while the 75 is a comparatively thick-walled round (though still TNT-heavy by British standards, certainly). I'd expect fragments from the thicker-walled shell, with less explosive blowing them apart, to be somewhat larger (discounting largest baseplate pieces perhaps). Which ought to give thinner coverage, but might easily see at least some fragments carry about as far as smaller pieces driving with more initial force.

2.8 times as much TNT is still a large difference, however. The Brits wartime approximation was to use the square root of the HE burster weight times the number of shells fired - for barrages well below saturation levels. That would mean roughly 3 105mm shells to 5 75mm shells. Particular variations might narrow that gap slightly I suppose, but I'd expect it to be reasonable close for typical point-detonating, small barrages. Say, 24 rounds 105 or 40 rounds 75, typical "sheaf" for the area covered. Maybe 32-36 75s would do the same job, but something in that range.

As for airburst effects, I think a key point is the absolute size of the shell, and the number of impacts. A few big impacts get worse figures for "best angle" to this or that spot. And the shells with the largest amounts of HE each probably have the densest shrapnel showers, nearby anyway. Raising them in the air will have more of an effect at getting around cover than lots o littles will.

Lots o littles will tend to have one round quite close, or not, and that is what the effect will turn on. The few big will have a larger average distance to a given point. If the shell goes off on the ground, that provides more obstructions. Raising the shell detonation overcomes that. While smaller rounds (1) aren't as likely to get a sliver headed in the right direction (fewer of them), and (2) are more likely to have "solved" the cover problem "already" as it were, by another hit on the right side of whatever cover.

So, I'd expect the difference between air burst and not to be relatively slight at the smaller shell end, and much bigger up at the 155mm and up end. But that is just something to look for, not data, so take it with salt.

[Username]

Thanks for clueing in rex.

Fragment "quality" is usually a function of size and velocity. High velocity small fragments usually scrub their speed quickly and are not much good beyond the explosives own deadly effects. they are deadly dust and soon fall away.

This brings up a good point. Ricochet firing a shell to an height will also reduce the explosive 'inverse-cube' effect. The killing ability of the HE falls off quickly. Hence, another reason that being shoulder high is better than 30 feet off the ground.

I have data on the US 75mm by the way. Number of fragments (the author describes the ideal ones as being a 1/4 inch I beleive) as well as a picture of a 75mm next to the fragments piled in groups by size.

Ideally, you would want the shell to have a minimal mass of powerful HE that fractures a maximum amount of metal into these 1/4 inch shapes with maximum velocity. Reality is different. The back of the shell as well as the front are stronger material and there is a distribution of weights/velocitys. Having very strong case metals leads to very powerful HE being needed and large splinters. case shape can lead to long splinters. Most shells fracture along the sides and thats where the killing is done, orthoganal to the axis of the shell.

Perhaps rex can post something from his book about this. Didnt you include something in your book similar to this thread? 75mm beats out everything! (Hypothesis are usually posed in the form of a question).

Lewis

[ 08-29-2001: Message edited by: Username ]

[Username]

http://www.milparade.com/1996/18/50-52.htm

http://www.milparade.ru/security/24/131.htm

http://www.mja.com.au/public/issues/may19/nocera/nocera.html#refbody4

http://www.mja.com.au/public/issues/may19/nocera/nocbox.html

http://www.winterwar.com/Weapons/artyinfo.htm#Artillery%20projectile%20blast%20effect

http://srd.yahoo.com/goo/Shell+fragmentation/19/*http://www.army.mil/cmh-pg/faq/shrapnel.htm

http://www.history.navy.mil/faqs/faq96-1.htm

http://www.westviewpress.com/fireinthesky/bombercombat.html

http://www.fas.org/man/dod-101/navy/docs/es310/dam_crit/dam_crit.htm

http://www.cfonews.com/alr/d040297z.txt

Heres some good reading. I especially like the russian inovation in tank shells.

Oh rexford, this is a bump, how about some of that book preview?