Vehicle protection from artillery shells

[A Canadian Cat]

15 minutes ago, Armorgunner said:

Hmmm that makes it very strange. And you are in 4.0? Not that i see any difference since 3.0, but anyway. 

Yes, version 4.0. Well we would have to do some statistical work and note shell landing and proximity and run the test a lot to show either way. Not really a high priority right now...

 

1 hour ago, Sgt.Squarehead said:

Compared to some of the images we've seen, even the most battered tanks in these tests got off fairly lightly, I think the damage could probably be safely dialled up a tad for direct hits. 

Just my 2c. 

Because this is probably where we should focus the discussion.

The thing is we would need to make sure we really knew what calibre shell hit the tank and how many. And I don't think that give us clarity on what kind of damage near by hits would or would not do.

[Armorgunner]

39 minutes ago, IanL said:

Yes, version 4.0. Well we would have to do some statistical work and note shell landing and proximity and run the test a lot to show either way. Not really a high priority right now...

 

Because this is probably where we should focus the discussion.

The thing is we would need to make sure we really knew what calibre shell hit the tank and how many. And I don't think that give us clarity on what kind of damage near by hits would or would not do.

Definitely worth further investigation. With the mind totally free of "i must be the one whose right"

[HerrTom]

I've been hard at work reading a lot of papers dating from the 2000s all the way to the 1940s about penetration mechanics and artillery shell fragments.  It's been quite exciting.  I still have some work I want to do regarding making a probability field and a "lethality plot" of an artillery shell against various armor plates.

So - here are some hopefully interesting plots.  Here is the velocity of fragments from a 152mm shell exploding.

This contour plot shows the velocity of fragments as they get further away from the epicenter.  As one would expect, heavier fragments are able to travel further, faster - since their weight versus their surface area is larger.

Fragments traveling more than about 60 m/s are generally going to be lethally dangerous.  The dangerous speed is faster for smaller fragments and slower for larger fragments.

The following plots use the Mott equations of armor penetration, which are quite old, but still considered somewhat reliable for at least a first guess.  They use data from the top graph as an input.

The Mott equations were created from empirical data from shell penetration tests in the 1940s, and as such has some doubt in my mind to the applicability to artillery shell fragment penetration.  Nonetheless, it remains a common model.

What's worth noting about the Mott equations is how quickly the penetration velocity falls off.

I also created some plots using the Recht and Ipson model - which is supposedly better for penetration of materials with a similar thickness to the projectile - which is more like what we're looking at here - fragments of ~1 inch thickness penetrating steel plates ~1 inch in thickness.

The Recht and Ipson model takes into account 

Note that the penetration on the thinner plate is actually slightly worse using this equation compared to the Mott equation!  This model then shows the fragments capable of penetrating significantly thicker armor compared to the old Mott model.  According to the Recht and Ipson model, you'll need a 3" or 76mm plate to be relatively safe against 152mm shells.

Finally, I hope to get a probability field using the artillery shell fragmentation presented in one of the papers that has already been posted in this thread using the probability of fragment mass:

As well as the angular dispersion of fragments on the cylindrical section of the shell.

That's all I have for now, but I'd be happy to answer questions and I'll try to analyse the data I already have in a bit more detail later.

[DougPhresh]

I am in awe. Excellent work and presentation.

[A Canadian Cat]

Very interesting. I'm still reading that but I am confused about this:

12 hours ago, HerrTom said:

As well as the angular dispersion of fragments on the cylindrical section of the shell.

 

Is that saying that the blast from an artillery shell is directional? Please clarify.

[Muzzleflash1990]

5 minutes ago, IanL said:

Is that saying that the blast from an artillery shell is directional? Please clarify.

Sort of. As I understand it, most fragments are are dispersed to the sides of the direction of travel. So if 0 degrees is front/direction of travel, 180 is backwards, between are to the sides. See the following images: image 1, image 2, image 3 . I believe shells that come down very vertical can also be more dangerous to infantry that way, since more is dispersed parallel to the horizon.

[HerrTom]

Yep, Muzzleflash beat me to it. It seems artillery shells are slightly thinner along the sides so when the blast starts pressurizing the inside, it shatters there first, taking most of the energy sideways. There are also some fragments from the ends, but they are much slower and larger, generally.

It's also worth noting that the spray shown in the histogram is larger than it looks. I'll try to put together a graphic to show it better tonight.

[A Canadian Cat]

That is really interesting - thanks guys.

[cool breeze]

That data dump on artillery tests JK gave a couple pages back was really good and explained/showed that stuff

[c3k]

HerrTom,

Great stuff. I think it'd be fairly easy to use the mass of iron (steel, in its various forms) to come up with a first-order approximation of size of fragment as it relates to mass. Your Recht and Ipson model uses relative size of fragment as it compares to the plate for the equation's output. Let's also remember that most shells do NOT have thicker and thicker walls. There is (as I understand it) a usual range of thickness. (British 5.5" (grr, or was it the 4.5" shell?) was very thick in comparison to the norm. It had a very poor performance against soft targets. Too few fragments.) Therefore, most shell fragments would gain mass by increasing their length. (Shell wall thickness is a maximum value. Metal crystal structure and detonating material/amount would determine width, to a large degree. The only thing left would be length. Obviously, this is HIGHLY variable. I'm thinking only of the larger fragments, since I think we can disregard the small fragments as a penetrating problem.)

It would be interesting to compare the probable energies with the STANAG protection levels of the various external components. Shredding of sights, barrels, comm gear, ERA, etc., could thereby be more closely modeled.

Of course, this is all back of the envelope. It'll be better when you refine it and include a motion gif.  

Great stuff.

Ken

[HerrTom]

I was under the impression that shells were thinner on the cylindrical section line these 152mm shells. Further looking shows this is not always the case. Thanks!

[HerrTom]

Also, a slight correction from my big post with the plots - the first equation set is the THOR equations. I somehow mixed them with the Mott equations which describe the fragmentation pattern (the last two plots).

Damn - and I also meant to say the Recht and Ipson model takes into account the material plug created by the penetrating object.  I think it comes largely from first principles and was later verified against data, compared to the purely empirical THOR equations - which are limited by their very definition.

Edited April 25, 2017 by HerrTom

[John Kettler]

HerrTom,

Fear not the flagrum, for thine mistakes are beyond most of us anyway! Besides, all this going on about apple sauce and printers is confusing. On a more serious note, I think you've done some valuable work which will give BFC and people here a lot to ponder. Your top graph doesn't seem to incorporate the frag masses found in the graph second from the bottom, either. In the first, your upper limit on frag size is 0.1 kg grams, while the other one shows  fragments of as much as 0.5 kg. That, presumably, doesn't account for the whopping chunk of base and shell sidewall Haiduk provided us, either. The angular distribution histogram is, on the face of it, counterintuitive, but flash radiographs, high speed film/video and a thought model provide the answer. The football shaped configuration the shell case assumes before it rips apart puts out far fewer frags perpendicular to the shell's longitudinal axis than people who don't know about shell case prior to and rupture phenomena during detonation would expect. It is no surprise to me the front of the frag pattern is so tiny.

Regards,

John Kettler

[Haiduk]

Some pictures again. Firing range target M60A3 after direct hit with 155 mm HE in the glacis. Source: http://ftr.wot-news.com/2017/05/05/m60a3-hit-by-he/

 

Edited May 13, 2017 by Haiduk

[John Kettler]

Haiduk,

Not one pic you posted currently displays! Would love to know more about what state the tank was in before being fired upon and what state it was assessed as being in, if it was assessed, after being hit. Offhand, I see multiple periscopes/vision blocks for the driver blown in, a driver's hatch possibly blown open, considerable damage to the lower left and right cheeks near the mantlet (can't tell whether there are any turret penetrations), damage below the mantlet, strikes on and near it and associated turret face which might've immobilized the elevation control, the complete destruction of the waterproof protective jacket covering the mantlet-turret interface. The gun barrel has sustained numerous strikes, though no penetrations are evident. Mind, this doesn't mean the gun is fine, for I have zero information on presence or absence of internal bulging, nor can I see the state of the far less tough bore evacuator. In any event, it's all moot, for the tank is 100% M-Killed via left track destruction and destruction of left return sprocket and first idler. Headlights are gone, too, as if that mattered! If that tank started hatches buttoned and all optics in place, then I think it likely the driver (and possibly one or more of the controls) would be a casualty. 

Regards,

John Kettler

Edited May 13, 2017 by John Kettler

[Haiduk]

13 hours ago, John Kettler said:

Haiduk,

Not one pic you posted currently displays!

John Kettler

Alas, this is known problem with this web-page. Pictures are disappearing since some time, but still visible when you watch origin web-page by link. There is no info what conditions have this tank befire turned out a target for artillery practice.

Edited May 14, 2017 by Haiduk

[John Kettler]

Haiduk,

Am relieved, for I thought I'd missed out on something. As for situations like yours with the already online pics, have you tried Copy Image Address? It gives you a URL you can Paste into Insert Other Media. Works great for me.

Regards,

John Kettler

[HerrTom]

Never fear! I haven't forgotten about this.  I ran a few tests with artillery, and it seems direct hits are different than I remember them in v3 a long time ago - I think you may struggle to replicate TFO's paintball Abrams result.

Direct hits seem to be fairly universally fatal to tanks, with a few key exceptions.

This Abrams was hit in the engine compartment, which while only a partial penetration managed to kill the entire crew.  You can see on the right side of the turret it previously survived a hit - but that location is probably the thickest armor on the tank, so understandable.

It received a second hit shortly afterwards which landed directly on the turret top, this one penetrated and would likely kill the crew, too.

Those craters practically under the tank would likely cause more damage than they do, but I can't really quantify that yet.

Same with a Bradley.  A hit to the turret top killed it dead.

Near hits are also capable of killing armored vehicles, too.

This hit to the front of a BTR-70 killed the drive and knocked the vehicle out.

 

The thing currently in my mind from the previous data I collected is that light armored vehicles like BTRs, BMPs, and Bradleys may be more vulnerable to nearby shell impacts than is modeled.  I'm gonna try to make some plots on some of the data I've collected.

[HerrTom]

Grabbed some data from shells landing around the BTR on another test.

12 shells landed before the final one hit the roof of the BTR.

And the fragment density calculated from each impact:

The center black O is the BTR's position, and the right-hand O is the BMP-2's position.  Wasn't testing it, but some shells landed remarkably close, so I logged it anyway!.

the BTR sees approximately 0.55 fragments per square meter, while the BMP sees 0.75 fragments per square meter.  This means the BTR was hit by approximately 10 shell fragments from the barrage, while the BMP-2 was hit by approximately 12.  (It's worth noting here that the fragments that hit the BMP are probably much more lethal owing to the closer impacts it saw!)

We can pull the distribution of fragments I made earlier and generate a probability of any fragment being any specific size:

What I need to do next is compound the probability of the fragment size with the distance from impact of the shell and fragmentation density to generate a probability of a perforation of a certain sized target - then I can apply it here and see if what we're seeing is accurate per the maths!

But it's late... so that's something to do tomorrow!

[c3k]

Great stuff! Keep 'em coming...

[A Canadian Cat]

10 hours ago, HerrTom said:

Never fear! I haven't forgotten about this.  I ran a few tests with artillery, and it seems direct hits are different than I remember them in v3 a long time ago - I think you may struggle to replicate TFO's paintball Abrams result.

I agree with @c3k looking good keep the info coming. I will however just say that v3 was always quite deadly to tanks getting hits from arty. I agree that v4 ups that slightly but I never reproduced the immunity that people reported so what you saw yesterday is what I would have expected and with in range of what I was even seeing in v3 too.

[HerrTom]

Minor correction. In my rushing to finish the post before going to bed I forgot that I had a log scale! I thought the number of fragments seemed low...

It's 10^0.55 per square meter at the BTR and 10^0.75 at the BMP.

3.55 and 5.62 respectively. Leading to 64 fragments hitting the BTR and 90 hitting the BMP. Now we're coming with gas! Makes a lot more intuitive sense haha.

[Muzzleflash1990]

39 minutes ago, HerrTom said:

Minor correction. In my rushing to finish the post before going to bed I forgot that I had a log scale! I thought the number of fragments seemed low...

It's 10^0.55 per square meter at the BTR and 10^0.75 at the BMP.

3.55 and 5.62 respectively. Leading to 64 fragments hitting the BTR and 90 hitting the BMP. Now we're coming with gas! Makes a lot more intuitive sense haha.

But the figure is the sum of all 12 shell right? So it's  10^0.55 =3.55  total right?

[HerrTom]

Correct, Muzzleflash. The 12 shells introduced an average of 3.55 fragments per square meter where the BTR was sitting.

The 12 shells caused absolutely no damage to either vehicle until the BTR ate a 152 aft of the turret, so I've got a fair amount of data on non-damaging strikes.

Edit: Haha, haiduk, muzzle flash, it all looks the same to me before my second cup of coffee! ☕☕

Edited May 23, 2017 by HerrTom
Reading is hard

[Haiduk]

@HerrTom

Did you shell vehicles with precision ammunition or usual ? 

If usual, alas, dispersion of shell impacts still unrealistic dense (especially on screen with Bradley)... If usual rounds could hit with such density, then no need to develop Excalibur %). Sorry for offtop

Edited May 23, 2017 by Haiduk