Bismarck vs. Iowa

[delcyros]

I share Byron´s perspective in this.

If You have contact with mr. Saxton, can You ask him to explain this and the relevant references to us?

thanks in advance,
delc

[alecsandros]

Sorry for a such a long reply; I'll try to contact David today, hope he can still shed light over this matter.

=============

Hello dear Byron,

As you already know, I spent the better part of the last year working on the basics of a shell/plate simulator, accounting for the majority of battleship hardware created after the Washington Treaty. To this end, I have used Nathan's works almost daily, adn his programs quite often. Naturaly, the bibliography I've touched was quite diverse, and also, naturaly, as you predicted, the project failed for lack of information (concerning specific perforating processes)

The "advantage" I think I had was the vast online library created in the last 5-10 years, covering British, German, American and even some Italian and Japanese primary and secondary (translated) documents. Without these vast library, practicaly always available, I wouldn't have learned about the adverse effects of larger crystaline grains in stoping a perforarting shell (case: French armor), nor about the insufficient cap-hardening of most Japanese APC shells.

And, thanks to you and others members of the forum, I slowly began to understadn many issues which were previously unknown or incomprehensible to me.

One of these issues is cross-country shell/plate interaction, which Delycros has nicely unearthed in a separate thread. This is, I think, the root of all real problems when trying to simulate teh effects of Shell X attacking Plate Y, in given parameters (speed, tri-axial angles of incidence, motion of the plate, yaw etc). Well, this particular issue is irreconcilible, for lack of sufficient practical tests. How can we create a program which accurately simulates the effects of a Japanese 18.1" shell striking 310mm US Class A armor at, say, (35*, 0*, 0*) angle of fall ? We can not. Because the hardware - the 18.1 inch shell and gun are no longer available. Invariably, they must have had specific details which are not taken into consideration by a program developed around American empirical tests. Take, for instance, the 500-510 Brinell value of the 460mm shell cap hardness. Exactly how would this low Brinell value affect penetretation/perforation of a 660-680 Brinell FHA Class A plate ?

How exactly would the 460mm shell behave during a diving hit ? Would it perform as designed, and its trajectory would resemble a parabole, with the "tip" towards the surface, or would it simply bounce around in the water harmlessly ? What kind of perforation should we expect from a 460mm diving shell striking at, say, 20* from the water plane, and at about 40m from the target ship ... ?

And don't get me started on teh differences between real shells/plates and simulated (hypothetical) shells/plates. The Italian 15" shells come to mind, and they rarely "behaved" similarly during battles. The differences in cap weight were up to 1-2%, and same for filler weight. This made hiting a real target a game of luck more than a game of precision.

=========

PLease excuse the large preamble, but I wanted you to know where I'm coming from. With this in mind, the task of creating a realistic shell/plate simulator seems a very steep chalenge. Imagination must play a huge role, and, unfortunately, not only in envisioning the processes which have to be simulated, but also in filling the gaps between known and un-known phenomenons. And, again, unfortunately, most of the un-known phenomenons are, today, un-knowable. They can at best be speculated starting from various other empirical evidence and physical properties which are already known.

Nathan explicitly mentions several times this problem: the lack of available primary sources to draw upon so as to replicate accurate results. Say British plates attacked by German shells. No ? Well then, how about some Italian plates attacked by Japanese shells ? Still nothing ? French versus German ? Nothing... ?

Consider the Massachussets vs Jean Bart incident. How can you replicate the result (2700pds shell fired at 20-22km perforating 22+150+40mm of homogenous armor and exploding) without drasticaly altering the quality factor of the plate... ?
And why alter the quality of the plate ? On what source should we rely upon ? Why not asume the shell was of considerable superior quality than the "average" 2700pds shell and out-performed its "genus" in a singular engagement ? Exactly what should be modified ? The quality of the plate or of the shell ? Why ? No one knows...

Consider also KGV hiting Bismarck's Caesar turret's frontal armor at less than 10km. The shell bounced off! Why ? The expected perforation at this range shoudl have been over 450mm of British CA plate. And the target angle wasn't very bad, because Caesar was tured towards KGV when the shell hit, IIRC.

Those and many other problems make me skeptical about the usefullness of modeling the shell/plate interactions, and I tend to favor firign trials, even if they do not fit a pre-determined pattern.

==============

[Dave Saxton]

It took me awhile to determine what was being talked about here. I included this data in a rough draft which has not been finalized or gone through final editing. I'm not really sure what to do with this at this time actually. I have also been puzzled by this data. The notes on the data mention that the plate was 70* errect. Obviously it could not be tilted the one way such that the total obliquity was very little. It appears to be set up so as simulate the effect of a sloped belt. I personally think that the 20* slope implied was propbably simply a mistaken listing and that it was more like a 10* to 15* slope. However, It may have been one of the proposed scenarios by delcroys. It's possible and perhaps even likely. It's also possible that the implied data is correct despite being well outside expections. It happens.

The significance to me of these tests isn't the exact amount of penetration attained, but that the L/4.4 performed well during relatively highly oblique attack. This has bearing on the armouring scheme design considered by the designers in order to meet the IZ specfications given by the KM of 20km- 30km vs 38cm attack. An armouring scheme with a scarp triangle is a better way of dealing with such oblique attack perfomance than a sloped belt design should a potential enemy also develop shells of similar capability. The L/4.4 was proofed at 30*, 45*, and 60* and performed well.

The Germans adopted the 0.93 cr head for the L/4.4 naval AP shell prior to the war during the 1930s. This specification is for the head of the shell itself- not the cap. Did this adoption give a much better result during oblique attack than indicated by the established curves? I don't know?

[Thorsten Wahl]

So 310 mm US Class A seem to be equivalent to 294 mm KC (nA). This raw thickness seem to be penetrable at around 20000 m ~ 22 kyard (Obliqity 14°+19° ~34 degrees, impact speed ~510 m/s). maybe somwhat less if you add around 10m/s for hull plating and other non ballistic plate

Krupp testing showed perforation of 308mm KC n/A declined at 20* for scaled 25km shootings. This implies a 23.7* angle of fall, and a 43.7* final obliquity angle.

I see no reason why Nathan's computations should take precedence over Krupp's ballistic trials.

Alexandros this pentration expectation is based on the original Unterlagen zur Bestimmung der Hauptkampfentfernung Durschlagsangaben Schlachtschiffe Bismarck und Tirpitz
I adopted only the relative thickness of Class A compared to KC nA from Nathan Okuns Data

and Byron according to the Unterlagen the projectile didnt break at impact angles of about 45°

we are confronted by a projectile capable of penetrating 4/5 caliber KC at 43 deg net obliquity i a fully intact condition - a completely mind-boggling performance.

The standard of proof condition for 7,5 cm Ap projectile
was Heil durch 100 mm at 45 degrees
for 8,8 cm projectile
heil durch 140 mm at 45 degrees until mid 1945
heil durch 180 mm at 45 degrees from mid 1945

The 8,8 cm AP-projectile was testet up to 1800 m/s impact velocity
showing best performance at 1280m/s through 305 mm plate Angle of incidence 60 degrees (90 degrees = normal impact)

these projectiles are belonging to the same family of projectiles and were manufactured in a roughgly equal way

if its possible to put a 8,8 cm projectile in a condition fit to burst through a plate roughly 2 times of caliberthickness it seems plausible to me that the upscaled projectile is able to pass through 0,8 caliber plate at 30 degrees

[Byron Angel]

So 310 mm US Class A seem to be equivalent to 294 mm KC (nA). This raw thickness seem to be penetrable at around 20000 m ~ 22 kyard (Obliqity 14°+19° ~34 degrees, impact speed ~510 m/s). maybe somwhat less if you add around 10m/s for hull plating and other non ballistic plate

Krupp testing showed perforation of 308mm KC n/A declined at 20* for scaled 25km shootings. This implies a 23.7* angle of fall, and a 43.7* final obliquity angle.

I see no reason why Nathan's computations should take precedence over Krupp's ballistic trials.

Alexandros this pentration expectation is based on the original Unterlagen zur Bestimmung der Hauptkampfentfernung Durschlagsangaben Schlachtschiffe Bismarck und Tirpitz
I adopted only the relative thickness of Class A compared to KC nA from Nathan Okuns Data

and Byron according to the Unterlagen the projectile didnt break at impact angles of about 45°

we are confronted by a projectile capable of penetrating 4/5 caliber KC at 43 deg net obliquity i a fully intact condition - a completely mind-boggling performance.

The standard of proof condition for 7,5 cm Ap projectile
was Heil durch 100 mm at 45 degrees
for 8,8 cm projectile
heil durch 140 mm at 45 degrees until mid 1945
heil durch 180 mm at 45 degrees from mid 1945

The 8,8 cm AP-projectile was testet up to 1800 m/s impact velocity
showing best performance at 1280m/s through 305 mm plate Angle of incidence 60 degrees (90 degrees = normal impact)

these projectiles are belonging to the same family of projectiles and were manufactured in a roughgly equal way

if its possible to put a 8,8 cm projectile in a condition fit to burst through a plate roughly 2 times of caliber thickness it seems plausible to me that the upscaled projectile is able to pass through 0,8 caliber plate at 30 degrees

..... I'm sure your information is correct. I'm equally confident that the proofing tests to which you refer involved the standard German PzGR39 APHECBC a/o APCBC and not the PzGr40 tungsten core projectile type. However, if you don't mind me asking, what type of armor were these projectiles tested against? I'm guessing rolled homogeneous plate, which is understood to be rather more tolerant of high obliquity attack than KC. And this is what concerns me a little, as Nathan Okun has gone into some eye-glazing detail on the warship1 discussion forum about the dangers inherent in extrapolating or inferring AP performance too far from one case to another. I'm not necessarily saying that your argument is wrong; only that my lack of mastery of the topic and Okun's above-mentioned commentary oblige me to remain skeptical until more convincing proof reaches the scene.

B

[Thorsten Wahl]

standard steel core projectile each
the projectiles were testet against homogenous plate

there is also a difference between the projectiles the cavity of the 7,5 and 8,8 cm shots was relatively small, compared to the naval projectiles. From this point of view the likehood of Heilbleiben is potentially increased. so the Grenzschußkurve(penetration with projectile broken) was moved to higher obliquities.

The information obtained by der Unterlagen zur Bestimmung der Hauptkampfentfernung give evidence that all german naval projectiles usually didnt shatter against homogenous plate.

I dont have problems to accept shatter and therfore reduced performance if someone gives a reference that this usually happens for a certain projectile.
But if this shatter is derived from a projectile wich shatters on homogenous plate at 0.5 calibethicknes of plate at 45 degrees, ther is strong evidenc that this projectile shouldnt be used to predict performance of german projectiles.

[RobertsonN]

I am surprised by the high performance of the 8.8 cm gun. Of course, it was likely to be well developed since the Germans made more of this type than any other and it was a true multipurpose weapon. Were these shells uncapped? They fall below the 15 cm limit of Gercke's formula but they are similar sized to the 3 in on which the Americans did many tests after the war. The ususal scaling would result in much higher penetrations for battleship sized shells than were ever achieved, and so seems to imply that the performance of 38 cm shells cannot be predicted by looking at the results for 8.8 cm shells. Or was it that the Germans had obviously given up tests with heavy naval artillery by the late War period, but had they continued they might have considerably improved on their earlier shells?

[dunmunro]

Consider also KGV hiting Bismarck's Caesar turret's frontal armor at less than 10km. The shell bounced off! Why ? The expected perforation at this range shoudl have been over 450mm of British CA plate. And the target angle wasn't very bad, because Caesar was tured towards KGV when the shell hit, IIRC.

What's your source?

[Byron Angel]

Consider also KGV hiting Bismarck's Caesar turret's frontal armor at less than 10km. The shell bounced off! Why ? The expected perforation at this range shoudl have been over 450mm of British CA plate. And the target angle wasn't very bad, because Caesar was tured towards KGV when the shell hit, IIRC.

What's your source?

..... There is in the Naval Review archive a provocative article about inter-war British naval projectile manufacturing and proofing, titled "The End of an Era". It was written by an individual who claimed an intimate familiarity with naval ordnance affairs during that time and through WW2. He describes a sequence of re-organization and economizing meansures taken after WW1 which ultimately surrendered all responsibility for the design, testing and manufacture of heavy armor piercing naval projectile to the two giant survivors of the WW1 British armaments industrial complex - Firth and Hadfield. The author goes on to argue that this arrangement, only reversed after a lengthy and tortuous bureaucratic struggle, resulted in the same sort of poor quality and proofing failures that had afflicted British AP projectiles in WW1 - a situation only redressed late in the war by the climactic establishment of the Cardonald factory and their "clean sheet of paper" projectile design work. It's an article well worth perusing and may well explain the BISMARCK episode mentioned by Alex.

BTW, many thanks for the insights on Friedman's British cruiser book.

B

[RobertsonN]

Regarding the 20 deg issue, some well known books have quoted data for 20 deg. Whitley's German Capital Ships in WWII does for all the data it gives and, from memory, Brown's (1980) Tirpitz: The Floating Fortress gives 350 mm at 21000 m, which is also a 20 deg figure. The meaning of the 20 deg was never clear to me, but I tended to the view that it meant 20 deg angle of fall at 90 deg target angle. However, now I think it is probably 20 deg (from the normal) target angle, because the fighting instructions said target angle should ideally be at least 20 deg to give immunity down to 10000 m to the Bismarck from shells penetrating the belt plus sloped deck combination.

[delcyros]

The notes on the data mention that the plate was 70* errect. Obviously it could not be tilted the one way such that the total obliquity was very little. It appears to be set up so as simulate the effect of a sloped belt. I personally think that the 20* slope implied was propbably simply a mistaken listing and that it was more like a 10* to 15* slope. However, It may have been one of the proposed scenarios by delcroys. It's possible and perhaps even likely. It's also possible that the implied data is correct despite being well outside expections. It happens.

Thanks for replying to my questions, Dave.
What was the exact term used in the primary source: "gedreht", "im Winkel zu" or maybe "geneigt"?

Since we have a set of data, we can be more precise in this question with our hypothesises:

A 38cm L/4.4 fired from the 38cm/L52 gun, against KC plates also declined by 20* obtained the following results:

20km-364mm
21.8km-350mm
25km-308mm

I will ignore the 21.8km data, as it looks a bit fishy to me (rounded armour penetration but fractured distance measure)

for 20km the striking velocity is 510m/s and the obliquity 16 deg
for 25km the striking velocity is 470m/s and the obliquity 24 deg (source: ballistic graph for 38cm SK/C38)

the possible obliquities and penetrations according to the GKdos-100 figures for the four aforementioned scenarios are:

[A 20km]: 36 deg; =intact penetration 280mm KC; penetration broken: 305mm KC
[A 25km): 44 deg; =intact penetration 223mm KC; penetration broken: 240mm KC
: 4 deg; =intact penetration 455mm KC; penetration broken: 460mm KC
: 4 deg; =intact penetration 390mm KC; penetration broken: 395mm KC
[C 20km]: 25.41 deg; = intact penetration 355mm KC; penetration broken: 370mm KC
[C 25km]: 30.86 deg; = intact penetration 320mm KC; penetration broken: 340mm KC
[D 20km]: 20 deg; = intact penetration 380mm KC; penetration broken: 400mm KC
[D 25km]: 20 deg; = intact penetration 355mm KC; penetration broken: 370mm KC

This is a sizable sample. The result seems to confirm my previous estimate. The figures for intact penetration in a condition fit to burst from the GKdos-100 are in less than 4% deviation to the reported figures and thus agreeing to both, the 364mm KC figure at 20km & the 308mm KC figure at 25km.
0.81 to 0.84 cal KC/n.A. at 30.86 deg obliquity are reasonably good performance.

I know that plates were also erected for trials in order to simulate a certain target angle as described by Hoyer.

[dunmunro]

Consider also KGV hiting Bismarck's Caesar turret's frontal armor at less than 10km. The shell bounced off! Why ? The expected perforation at this range shoudl have been over 450mm of British CA plate. And the target angle wasn't very bad, because Caesar was tured towards KGV when the shell hit, IIRC.

What's your source?

..... There is in the Naval Review archive a provocative article about inter-war British naval projectile manufacturing and proofing, titled "The End of an Era". It was written by an individual who claimed an intimate familiarity with naval ordnance affairs during that time and through WW2. He describes a sequence of re-organization and economizing meansures taken after WW1 which ultimately surrendered all responsibility for the design, testing and manufacture of heavy armor piercing naval projectile to the two giant survivors of the WW1 British armaments industrial complex - Firth and Hadfield. The author goes on to argue that this arrangement, only reversed after a lengthy and tortuous bureaucratic struggle, resulted in the same sort of poor quality and proofing failures that had afflicted British AP projectiles in WW1 - a situation only redressed late in the war by the climactic establishment of the Cardonald factory and their "clean sheet of paper" projectile design work. It's an article well worth perusing and may well explain the BISMARCK episode mentioned by Alex.

BTW, many thanks for the insights on Friedman's British cruiser book.

B

No problem about Friedman.

However, a 14in shell fired at less than 10km (SV = ~1900fps), striking 350mm KM plate at 20degs, will not "bounce off". It might shatter or burst upon impact but a simple ricochet is not going to happen.

[Byron Angel]

However, a 14in shell fired at less than 10km (SV = ~1900fps), striking 350mm KM plate at 20degs, will not "bounce off". It might shatter or burst upon impact but a simple ricochet is not going to happen.

..... Which leads me to wonder whether the apparently observed ricochet was not the result of a hit upon the turret face, but rather a glancing strike upon its sloping forward roof plate. Even with binoculars, observation from a moving platform through smoky chaotic action conditions at 4-5 miles distance may well have been misleading.

B

[Byron Angel]

Hello dear Byron,
the task of creating a realistic shell/plate simulator seems a very steep chalenge. Imagination must play a huge role, and, unfortunately, not only in envisioning the processes which have to be simulated, but also in filling the gaps between known and un-known phenomenons. And, again, unfortunately, most of the un-known phenomenons are, today, un-knowable. They can at best be speculated starting from various other empirical evidence and physical properties which are already known.

Nathan explicitly mentions several times this problem: the lack of available primary sources to draw upon so as to replicate accurate results. Say British plates attacked by German shells. No ? Well then, how about some Italian plates attacked by Japanese shells ? Still nothing ? French versus German ? Nothing... ?

Consider the Massachussets vs Jean Bart incident. How can you replicate the result (2700pds shell fired at 20-22km perforating 22+150+40mm of homogenous armor and exploding) without drasticaly altering the quality factor of the plate... ?
And why alter the quality of the plate ? On what source should we rely upon ? Why not asume the shell was of considerable superior quality than the "average" 2700pds shell and out-performed its "genus" in a singular engagement ? Exactly what should be modified ? The quality of the plate or of the shell ? Why ? No one knows...

Consider also KGV hiting Bismarck's Caesar turret's frontal armor at less than 10km. The shell bounced off! Why ? The expected perforation at this range shoudl have been over 450mm of British CA plate. And the target angle wasn't very bad, because Caesar was tured towards KGV when the shell hit, IIRC.

Those and many other problems make me skeptical about the usefullness of modeling the shell/plate interactions, and I tend to favor firign trials, even if they do not fit a pre-determined pattern.

==============

..... I broadly agree with you assessment. Most of us here are amateurs (I count myself as one such) engaged in the pursuit of scientific archaeology. I have been friendly with Nathan Okun and Bill Jurens for some time and pretty much concur with their fundamental assessments, i.e. -

Nathan: we really have too little empirical data to draw a fully comprehensive and accurate picture of the armor versus projectile phenomenon.

Bill: attempting to accurately predict complicated phenomena by means of nearly century old mathematical functions, which necessarily do not allow for later science or the flexibility offered by modern computing systems, can at very best only provide windows of clarity as opposed to any sort of broad landscape - if you take my metaphor.

My interest in ballistics offers parallel testimony. Ingalls ballistic formulae and tables, developed at the turn of the 20th century for calculation of range tables, were quite accurate for gun elevations up to about 15 degrees. But, it was known even then that they rapidly deteriorated in value at elevations beyond 15 degrees. If they knew why this was the case, they would have dealt with it, but the causes of this deterioration in accuracy was at the time unknown and therefore insoluble. The ALVF range tables, developed in France in 1918 for use by the railroad artillery, was the very first range table incorporating mathemical integration. It's a large book - about 100 pages 20 x 30 cm, filled with column after column after column of numbers and factors which allowed for ballistic performance >>approximations<< to be made in 5 degree increments from 15 to 40 degrees. Anything in between had to be interpolated. Every bit of these math integrations was performed by hand calculation.

IMO, any search for complete and perfect exactitude is a lot like trying to make bricks without straw.

BRgds / B

[alecsandros]

Byron: