How to calculate shaped-charge damage for theoretical explosives?

[Sam Baughn]

I've been trying to make sense of ultra-tech warheads, and figure out what the assumptions are behind them.

High-Tech gives standard TL7 40mm grenade launcher shells (I think that's the right terminology) 4d of damage. This seems to be consistent with 32g of Composition-B filler, which seems to have been standard for 1970s shells.
At TL8 that goes up to 6d+2, which is about 163% of the TL7 damage, which implies the TNT-equivalent of explosive in the shell is 264% of what is in the TL7 one. I can't find a source for what filler is used in modern shells, but some internet research suggests that the explosive used isn't much more powerful, probably just having more RDX in the mix, giving it about 10% better REF. That means the main difference must be packing more explosive into the shell. RDX is a bit denser than TNT, so replacing most of the TNT with RDX would help somewhat, but not nearly enough. I guess the difference is a thinner casing and smaller fuse, allowing a bit more material to be jammed in.
The TL9 HE 40mm warhead in Ultra-Tech does 8d. That's only about 23% more damage than the TL8 one, or about 51% more TNT-equivalent. Assuming the TL8 warhead uses an explosive with REF of about 1.6, and no improvements from density, that implies an REF of around 2.4, which seems pretty conservative.
High-Tech lists CL20 as a TL8 explosive with REF 2.3. It seems that might be outdated, because Wikipedia lists it as REF 1.9. It also seems to need mixing with about 5% binder for practical use, which would bring it down to around 1.8.
Octanitrocubane apparently has an REF of 2.38 according to some sources (although others suggest it might be closer to 2 and it could require mixing with less powerful explosives or inert binding agents to be practical, but it's also a bit denser than most explosives). It's possible to make it now, but very difficult. If a manufacturing process were invented to make it cheap, or a similar but easier to make substance discovered, that would be a good fit for the TL9 warhead filler.

Looking at shaped-charge warheads, things seem more complicated.
The TL7 HEDP shell does 4d(10) with its jet, and 4d+2 explosive. That implies a bit more explosive filler than the HE shell, around 27% more TNT-equivalent, so likely something like 40g of Composition-B.
The TL8 HEDP does 6d explosive, so probably around 78% more TNT-equivalent. Like the HE shell, this seems to be mostly a case of packing more explosive into it than significant improvements in REF.
The jet damage goes up to 7d(10), or 75% more than the TL7 one. That looks a lot like the jet's penetration goes up in a linear fashion with the force of the explosion.
That seems counter-intuitive to me. Most sources (including High-Tech) suggest that explosive force needed to penetrate something goes up with the square of its thickness; that is, penetration depth (directly linked to damage in GURPS in a linear fashion) should scale with the square root of explosive energy (or TNT-equivalent weight, which is basically the same thing as far as I can tell).
Finding formulas for expected results of different explosives in shaped charges is difficult. The only thing I was able to find was that 'jet length' seems to be the crucial figure, assuming size, shape, liner material, and target material all stay the same. It looks like penetration is simply related to jet length (i.e. twice as long a jet means twice the penetration), but I can't find anything that says how to calculate expected jet length. I did find one document which said an LX-19 (CL-20 with binder) hemispherical shaped charge generated a 50% longer jet than an LX-14 (HMX with the same binder in similar proportions), but also said this may be 'an experimental anomaly', so it's hard to draw conclusions from that.
It seems possible that much of the difference between the TL7 HEDP shell and the TL8 one is not due to the increased force of the added explosives, but other factors like higher detonation velocity (which seems to be related to REF, but not in a simple way), better liner materials (I think the transition from copper to tantalum, copper-tungsten or molybdenum would have been in the late 70s or 80s), and better understanding of the physics leading to more efficient geometry.
However, the TL9 shaped charge warheads in Ultra-Tech make a massive jump in effectiveness, with the 40mm one doing 6d×4(10) with 4d explosive. The relatively low explosive damage implies that this isn't a super-high REF explosive, but I find it hard to believe that other factors could account for more than triple the penetration, even if you use uranium or something as the liner material. Even if we ignore the explosion damage and make the generous assumption that explosive force translates directly into damage, that's ~3.43 more damage than a TL8 40mm shell, implying an REF of about 5.49, which is on the high-end of estimates for Octaazacubane, a completely theoretical metastable form of nitrogen, which seems to be about the limit for practical near-future explosives without getting into weird forms of matter or atomics.

So, it seems that the shaped charge damage in Ultra-Tech is way off and the HE / HEC damage might be a bit low. Adjusting normal explosive effects (e.g. HE damage) to account for different effectiveness in future explosives seems fairly straightforward; just figure out the REF and weight of explosive, then multiply them together and use the usual explosion rules.
But for shaped charges, I have no idea what appropriate values are. My instinct is to make them scale at the same rate as normal explosives, i.e. at the square root of mass × REF, but since I'm not a mathematician, scientist, or engineer, I don't think my instincts are to be trusted on this.

Can anyone who actually knows what they are talking about comment?