There are a few reasons for the rework. The original problem is that the multiblock system used for the current solid-fuel reactor is very old and really limiting - almost four years old, from back when I hardly knew what I was doing with Java. Over time the performance has got better, but it's still not great and the API I use for the MSR, heat exchanger, turbine, etc. is so much more flexible, so I want to move to that for the solid-fuel fission and current stellarator fusion reactors too. This would already not be compatible with old builds, so I decided it would be a good time to rejig things.

The old system was not very realistic - although there have been cooler placement rule changes here and there, and some tweaks to the effects of moderators, the basic system has been as it is for, again, about four years, and was originally inspired by IC2's mechanics. The cells have pretty much the exact same power and heat gen rules as the uranium fuel cells in an IC2 reactor. The main problems are the following:

1) Adjacent cells/rods do not really care about each other to first order, as the energy of the neutrons released in fission are far too high to be absorbed with high probability by the fissile material to continue the chain reaction. This is why moderating material, which slows down the neutrons, is placed between rods in thermal reactors - it's to actually bring the energy of the neutron down to where the fission cross-section, i.e. probability of absorption -> fission, is high. In the overhauled mechanics, cells need to be given at least some critical level of incident moderated neutron flux, based on the fuel, to remain active and maintain the chain reaction. The chain reaction also needs to be primed by neutron pulses from the wall of the reactor, with is more interesting and realistic that magically initiating it with a lever.

2) Moderators just sitting next to a cell would do nothing. Why would they? They would just moderate the neutron flux coming through it for, well, no reason. Moderators should always be between cells. So now moderators don't just raise the efficiency of nearby cells, but instead their mechanics are purely concerned with the moderation of the neutrons between the cells.

3) In the current system, the power produced is far from proportional to the heat needed to be dealt with by the coolers. Again, it's a game mechanic, but one that can be done without for realism without ruining anything. In the new system, the efficiency and heat multiplier are still not quite equal for various reasons, but at least there's no more fundamental quadratic relationships anywhere.

4) Currently, there is no concept of heat flow from the cells to the casing - the fission produces heat which needs to be taken away to some sort of coolant, in most cases. That is dealt with by the new 'cluster' system, where cells must be able to offload their heat through some series or group of heat sinks to the casing to heat water into high-pressure steam.

5) The cooler rules are not perfect. The idea behind the mechanics of the placement rules is pretty much a staple of NC now, but has evolved quite a lot over time. In 1.7.10, originally coolers could be placed anywhere, but there was only one type (originally fission reactors were super limited when it came to casing dimensions cell arrangement). Then when things got more flexible, i.e. variable reactor size and cell-structure, more coolers were added, but they could still work anywhere, with bonuses if placed in certain positions.

   It was when moving to 1.10.2 that I decided the mechanics would be stricter - they would have to be placed in a correct configuration to do anything at all, and that was the start of the standard form of the puzzle today. However, although the rules were 'balanced' to an extent (i.e. the ways in which different coolers could support others, etc.), some of the placement rules were rather exploitable, and so in v2.9 they were slightly tweaked again, along with some dodgy moderator efficiency maths, in order to remove the exploitable parts. There were also five more cooler types added.

Unfortunately, I did this without quite so carefully considering the balancing of the rules and the relationships between coolers' support for each other, which means that the current rules now are not as balanced as they could be. For example, water and redstone coolers support two other types, yet quartz only supports one. So having learned how to avoid exploitable mechanics and to be more careful in considering the tree of supporting rules, the new 'heat sink' rules should be more fun. There are also many more sink types, at 24, and the tree goes much deeper than it did before. Currently, there's only one rule at a depth of 2, which is for the iron cooler, while in the new system there are many. The actual cooling rates are also a little more streamlined now, incrementing by 5 starting from 50 H/t for water all the way to 165 H/t for liquid helium.

   I also really do not like active coolers. Their use in cooling fusion reactors is also a very improvable mechanic (and that will be dealt with!) but I have always seen active coolers used in a way which sorta just ruins the puzzle aspect of the reactors - the cooling rates are just so high that it becomes easy to deal with any reactor with very little thought going into the design.

6) A thought I’d had for a while, but eventually convinced of very recently, is the addition of neutron reflectors. These would not really work very well with the current mechanics due to the weird moderators, but with the new moderation and neutron flux mechanics, they will probably have an interesting dual role: as a tool for helping cooler-running fuels reach criticality, and then as an ‘afterthought’ of sorts when trying to beef up the power output of a reactor.

So all in all, the new mechanics are designed to be more realistic, more in-depth, more flexible, less susceptible to cheesing, and also just a nice refresh :)