Factorio Forums

I have heard that super jumbo nuclear setups eat up a lot of UPS performance, and I have wondered if it is because of a problem with calculating the heat flow in the giant heat pipe network. Has anyone compared UPS cost of, for example, 10 10X2 completely separate systems compared to one giant 100X2? (Or however many 10X2s equal a 100X2.)

My theory is that perhaps there is some calculation for the heat pipes that is non-linear on the number of interconnected pipes.

This suggestion would take more reactors and so waste some uranium, but it is plentiful, and UPS is not.

Anyone have a giant base where this would be easy to test?

It doesn't really seem like the number of objects in these big nuclear builds should be causing the problems I've heard about, relative to the size of the rest of these big factories, so it seems likely that there are some optimizations possible...

It's "only" linear, but that is bad enough. The ultimate fix is to internally merge all pipes into 1, like they did with solar panels.

But even after the merge, nuclear will have more moving parts than solar.

I did a quick test, my factory is at 25fps/ups

I use a 120 reactor setup, i stole the blueprint some where. I tore down all reactors and heat pipes. As half was losing steam my UPS climed up to ca 30. When everything stoped and no steam it was at 40 UPS. At that point the factory was running at the remaining 6,5gw solar i still have wich is not enough. So some of it might be due to less production, but the factory looked ok at day time, tho acumulators was not chargeing.

I think the water and steam pipes outnumber the heatpipes and unless they are already optimized they should have the same impact on the cpu.

The UPS cost is because factorio is doing a basic simulation of the heat flow, just like it simulates the flow of liquids (afaik both are the same system), if you remove pipes, of course you get more UPS since you're making the network smaller and is less work to simulate.

The other more performant alternative you have is, like Dave said, removing the simulation and making it behave like the electric network, so what happens in one place is felt instantaneously in the other extreme of the factory, thus you can assume everything is connected to a single node/pole/pipe/whatever and just do the math. Whenever you really want or not do that is another discussion though.