Statistics on mines

[Joeri]

Hi,

Continuing my quest to understand the inner workings of CM I did some tests on mine detonation chances. Both for AT mines and AP mines. What I am interested in is the chance of a minefield detonating, the effect this has and also the effect of stacking two minefields on top of each other (does this increase the chance of detonation and how). Here are the results for those interested.

--------------------------------------------

AT mines

Setup

Flat dry open terrain.

16 minefield in a row.

16 M4 shermans just in front of the mines.

I Fast moved the tanks over there minefield and I recorded either: no detonation (N), immobilization (I) or kill (K). Did this for 128 tanks (8 times 16).

I also stacked 2 minefields on top of each other and did the same test only with 56 tanks (8 times 8).

Results:

single minefield:

total 128

N 47

I 35

K 46

double minefield:

total 56

N 10

I 24

K 22

Discussion

For the detonation chance of a tank driving over a minefield, this comes down to 63% ((35+46)/128*100%). For a double minefield this would in theory result in a chance of

(1.0-(1.0-0.63)^2) * 100% = 86% chance ( lost anyone). The experiment gives a 82% chance which is close enough in my opinion to conclude that both minefields work independent. Including the double minefield result the chance would be approx 60% per minefield.

Once detonated a M4 Sherman has approx 46% chance of an immobilization only.

I didn't check if some tanks did set of a mine and survived it without getting immobilized because ... well I didn't want to spent too much time on it.

---------------------------------------------

AP mines

Setup

Flat dry open terrain.

16 minefield in a row.

4 platoons facing the minefields 4 * (3 squads + 1 hq) 12 man squads (US rifle)

run the platoons over the minefield. Recorded the results of the hq separate to see if number of man has anything to do with it. I recorded the number of detonations (det) and also the number of casualties (cas). I did this for 6 times. so 6 * 4 * 3 = 72 squads and 6 * 4 * 1 = 24 hq's

Results

For the 12 man squads

Total 72

det 69

cas 162

For the 4 man hq's

Total 24

det 23

cas 40

Discussion

It seems like the chance of a detonation is independent on the nr. of men walking over the mine 96% for squads as well as hq's. This could be a coincidence because of the high chance and the low nr. of tests (especially with hq's). Average casualties number for a squad detonating a minefield is 2.3 and 1.7 for a hq. Casualties ranged between 1 and 6 for 12 man squads.

Note that sometimes the squads cause multiple detonations when travelling through the minefield. I counted this as one detonation and I took the total number of casualties.

I also tried this with double minefields but that got kind of complex. They detonated separately both causing casualties and often routing the squads who would than redetonate the first minefield or others which were next to them. From the things I observed from this test it seems that the two minefield work independent just like the AT mines.

By the way, double minefield are very effective against squads (half dead, rest routed).

--------------------------------------------

In contradiction to the results of my bogging experiments the mine results were pretty close to my expectations. Except of course the no-mine-detonation-bug-when-reversing which I previously reported in another thread.

Kind of boring results actually ,

Joeri

[This message has been edited by Joeri (edited 12-05-2000).]

[This message has been edited by Joeri (edited 12-05-2000).]

[Subvet]

I appreciate the work you put in. It was very informative to me, so thanks!

[Pak40]

Good work. Thanks for the info.

[Peterk]

I've been wondering about mines for a while. Thanks for the research.

p.

[Tris]

Joeri, in your above test you made a mistake with the number of Shermans rolling over double mine tiles. It was actaully 64 Shermans, not 56, so the to-detonate% would be more like 72, not 83. I think you should use a larger sample with another nine tests, also perhaps note the standard deviation.

[Joeri]

Oops,

Tris you were partly right. I did make a mistake but it were not 64 test it were 56 but I did the test only 7 times in stead of 8. Hmm, let me think 7*8 = 56 correct?

I don't think it's necessary to do more test because these tests made it pretty clear that both minefields work independent of each other. Therefore, the two test (single and double minefield) can be put together to calculate the chance for a single minefield to detonate. However this involves some statistics I can't reproduce without looking it up first.

I did it the simple way:

single minefield gives 63% chance

double minefield gives 82% chance.

This is (1-sqrt(1-0.82))*100% = 58% chance per minefield (assuming they work independent).

So combining the results would give somewhere between 58% and 63% chance. Lets make it 61%. This is the chance based on a total of 128 single test and 56 double tests

so its quite a large number of tests imo.

If anyone can do the exact calculation of the chance based on the test results, I would be interested to read it.

Joeri

[Tris]

I also stacked 2 minefields on top of each other and did the same test only with 56 tanks (8 times 8).

Joeri, you presented the test criteria as it appears above; this is what I took issue with. What you actually did I had no further idea, but it was reasonable to assume your tank constant had not changed from the 8 tanks you used in the test immediately prior, so I assumed your error was one of math (you clearly gave the function "8 times 8" as equal to "56 tanks") and not a common typo. My confusion does not seem unwarranted.

All I knew is that with 64 tanks versus the 56 you cited in the test the result would have been considerably altered, thus my feedback.

Now, based on your test there does seem to be a correlation between stacking minefields and a greater to-detonate%, but the sample given is kind of small--a larger one giving a similar result would be more conclusive.

I suggested that you might well derive a standard deviation based on the results of each test taken in aggregate (or at least note the specific deviation from one test to another) as a means for you to either better corroborate your data during presentation (i.e., given small deviation the results would look better, larger deviation would point to more randomization) or render something more substantial to chew on.

That's all.

[This message has been edited by Tris (edited 12-06-2000).]

[mensch]

I wonder if these test would be different with veteran troops or elite troops? I mean a tank is a tank and a trooper is a trooper but as for infantry would they have a higher chance locating the minefield?

I knoticed one game I ran a elite unit over a AP mine field took some losses but no routing and they made it through.. my following Regular unit faired worse and sat in the middle of the field crying for help as I figure it out... I tried moving them out they took a few meters distance then paniced and bolted for the nearest woods.

so the question states, would troop quality reduce the effectiveness of a mine field due to thier experience finding the little suckers and avoid them.

well thats my 2 Pf.

mensch

----------

Der Kessel Home of „Die Sturmgruppe“; Scenario Design Group for Combat Mission.

[Joeri]

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

Joeri, you presented the test criteria as it appears above; this is what I took issue with. What you actually did I had no further idea, but it was reasonable to assume your tank constant had not changed from the 8 tanks you used in the test immediately prior, so I assumed your error was one of math (you clearly gave the function "8 times 8" as equal to "56 tanks") and not a common typo. My confusion does not seem unwarranted.

<HR></BLOCKQUOTE>

You are absolutely right. I made a mistake which caused your confusion. Only the mistake was not in the 56 but in the 8 times 8, hence my comment.

About your other comments (more tests, standard deviation). I just want to know some relevant info to better judge situations in games. For example:

- If I mine a road can I be absolutely certain that nothing is coming through or only 60% sure.

- Is it useful to put 2 minefields on top of each other or does this change nothing and is it just a waste.

- I know my opponent has mined this area do I order my inf. through it anyway or go through another area where I know he has a couple of MG's.

That sort of things. I'm not really interested in knowing the exact details just the relevant info for playing. Besides statistics never was one of my favorite subjects.

However, feel free to use my results for a nice statistical analyses

Joeri

[This message has been edited by Joeri (edited 12-06-2000).]

[Tris]

Joeri: I mentioned deviation as a factor only because it would tend to more strongly present the case one way or the other, this with an eye to BTS taking notice on something which seems amiss.

Mensch: that's a good idea, to run some troop travellers over mine and double mine tiles at different experience levels and see if that changes the issue. If fact I wish you'd go out and do that right now.

[Olle Petersson]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Joeri:

1) - If I mine a road can I be absolutely certain that nothing is coming through or only 60% sure.

2) - Is it useful to put 2 minefields on top of each other or does this change nothing and is it just a waste.

3) - I know my opponent has mined this area do I order my inf. through it anyway or go through another area where I know he has a couple of MG's.<HR></BLOCKQUOTE>1) If you're using Daisy-chains, the opponent most likely won't go through there.

2) Two minefields each of AT and AP, bomined with barbed wire and covered by a MG pillbox and a PaK40 pillbox will stop most advances.

3) You just have to ask yourself: "Do I feel lucky?" Well, do you, punk?

Cheers

Olle

[mensch]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Tris:

Mensch: that's a good idea, to run some troop travellers over mine and double mine tiles at different experience levels and see if that changes the issue. If fact I wish you'd go out and do that right now. <HR></BLOCKQUOTE>

no problem my old chap but not before I do some test with varrious ammunition rounds on your shaagy butt! muhaha... infact I would not mind testing my PitBull Fluffy on your arm, leg, head, neck, and those soft musshy parts. Ask MACE oh how that feels hes always being attcked by pitbulls, girl scouts, old grandmas on the street.. I've done some tests if you want the results drop me a line.