[Bug] Radiators on projectiles cause ridiculous damage

It definitely does feel like the physics handle radiator damage, but it feels like something is done wrong. It almost seems like penetration is calculated from the craft's hull cross-section without the radiators and the total mass including radiators, except it even penetrates more than osmium rods of that cross-section. I have no idea why it punches that size of holes, but they seem way too large for the central RTG and (except for Whipple shield damage) way too small for the radiators.

The holes in the first screenshot do go through all the layers of armor. The ship is 49m wide and 185m long so 2.5m of armor is actually pretty thin. There were big black patches on the armor overlay and a bunch of orange from spalling/plasma, but I didn't show that because it made everything too bright in the screenshot. I could also easily see the internals through the holes from a different angle.
I've done a fair bit of experimentation with penetrators based on roughly cylindrical armor and/or radiation shields, and this thing blows them all out of the water. I can make a long rod that can punch a tiny hole through as much armor as this, a pancake that can punch as big of a hole through very thin armor, or something in between, but none of those designs can destroy even 1% this volume of armor. Everything in my experience is saying this lithium waffle should just go splat on the surface and not even penetrate the Gunship's inner armor, yet it can rip a Gunship in half or go through multiple meters of VCS like a 2mm needle while making a 30cm hole.

I tried 98g lithium flak from 1-25 pieces with very narrow spread. It's fairly powerful but nothing like the radiator projectiles. It takes 30-50 hits in the same spot to punch through the target dummy instead of almost every shot going through. Against a Gunship it tends to take 1-2 hits in the same place to get through the front armor, and 2-5 total hits to go through the whole ship. By the time flak can do significant internal damage to a Gunship, the radiator projectiles would've already broken it into 3-5 pieces. An interesting note is that the flak impacts leave armor glowing around the impact and blast off Whipple shields on the way out but not the way in, compared to the radiator projectiles that blast off the front Whipple shield and don't heat anything.

I was originally going for a cutting effect like a broadhead arrow, but the test results drove me in this direction. Sturdy, blade-like radiators didn't actually cut lines in the target. They just expanded the central hole and somehow increased penetration. I progressively made the radiators longer and weaker, then eventually less dense, for consistently better results. With this design at this speed, it seems like the radiators just do slight damage to Whipple shields and weak armor, then somehow increase the penetration and size the core hole to ridiculous levels. The radiators are nearly the size of the giant holes in the Whipple shields, so they're clearly not going into the target intact, yet their mass seems to be doing something because a 1.28g RTG definitely doesn't make a hole like that.

Quite a few of my weapon tests have involved long rods at over 25km/s and two things become apparent: Density and strength are almost irrelevant on impact, and the narrower the rod, the greater the penetration. My scariest weapon so far is a railgun that shoots a 10g 1.72mm x 5m potassium needle at 130km/s and can effortlessly penetrate 20 meters of VCS armor behind Whipple shields. I tried denser materials than potassium but they either make the game stutter like crazy on impact or make the rod too short to shoot accurately. For heavier rods I prefer osmium because shorter rods cause less slowdown from collision detection and they don't go fast enough to need accuracy at 1.5Mm anyway.

According to my math (which may be flawed) these projectiles have 25032mm² of impact area. That's equivalent to a 178.5mm x 7.5mm lithium plate, which happens to be a terrible projectile. It only made 8 holes in a Gunship after 150-300 hits and couldn't even reliably get through the Target Dummy's 1mm osmium/2mm gold/3mm platinum heavy Whipple shield.