The Goal: make a Mark I-O, 2.33 eff, negative heat reactor using only 2 chambers.
It's fairly easy to make a heat neutral 2.33 reactor with 2 chambers, and simply adding a third chamber is a dull way to create negative heat gain, which is nice for having a self-correcting reactor instead of one that just stores heat. So it's a good challenge to make a negative heat, 2-chamber, 2.33 efficiency design. No one-use coolants or redstone timing either; that's cheating.
Also--and I'm probably dead wrong, so please correct me if so--the highest efficiency you can get with a Mark I is 2.33. I'm sure it's the highest you can get at 2 chambers, but my meager math tells me it's the highest for any number of chambers too (maybe a 6 chamber, 2.5 ... maybe not).
My math says that you need 19 cooling cells to make a heat neutral design: 2.33 eff = 48 heat; water cooling + two chambers = 29 external cooling; 48-29 = 19 cooling cells required. That leaves only 8 spaces for other components: 30 total - 3 uranium - 19 cooling = 8 extra spaces.
After fussing around with using the extra space for a 20th cooling cell (-1 heat/tick) and placing 7 HDs, there was just no possible way to do it. The best I could do was heat neutral with one unreachable space. Finally, I settled on a reactor plating for the tiny heat loss and ended up with the design above.
The Problem: Alblaka says that the reactor plating has -0.1 heat/tick, but also that it transfers heat immediately to adjacent components. The order that happens in is critical. If it dissipates the heat before transferring, then the design has a very small negative heat gain. If it transfers before dissipating, it's back to a heat neutral design. The javascript simulator seems to go with the latter. I guess I should just build the thing and test it.
Oh, and the simulator puts more heat on the right side components, even though it's a symmetric design. Maybe there's some nuance I'm missing?
Ideas and corrections are welcome.