[2.0.16] Holmium Plate not producing at stated rate

[Stevetrov]

I have a legendary foundry with 10 legendary beacons each with 2 legendary speed 3 modules, this gives the foundry a speed of 201.6. this means it should produce 503 holmium plates per second according to the tooltip.



Some of the beacons are below the production window.

As you can see from the production window its producing exactly 300 plates / s. As there is sufficient fluid flow I would expect it to produce 503 as per the tooltip.

I see that the copper & iron plate casting recipes produce 2 plates at a time meaning that its impossible for those recipes to complete so quickly. Perphaps the solution is to make 2 holmium plates from 80 solution in 2 seconds.

[boskid]

Ok, i looked at this setup and it looks it was input fluid starved. For 2.0.19 i made the input fluid box for this specific recipe to hold 10x ingredient fluid amount (compared to 2x from before) to account for insane crafting speed and short recipe time of this specific recipe.

[MacFeejj]

Since whether a recipe is limited by pipe throughput / fluidbox size is entirely computable at runtime - check if the maximal per-tick consumption outpaces the maximal per-tick flow - it could be worth adjusting the GUI / adding some sort of warning to deal with this, especially for modded cases.

[Nemoricus]

Ok, i looked at this setup and it looks it was input fluid starved. For 2.0.19 i made the input fluid box for this specific recipe to hold 10x ingredient fluid amount (compared to 2x from before) to account for insane crafting speed and short recipe time of this specific recipe.

Would it be possible to apply the same sort of logic as inserter input limits to the fluid inputs?

[Xeinaemm]

A similar problem is with acid neutralisation and steam condensation. Cryogenic plants cannot produce more than 2k water/s. There should be more proper implementation like functionality where machine-to-machine connection means 'unlimited' throughput. I assume that when a machine can produce 110k steam/s it can transfer it internally somehow because otherwise pipes need redesigning to handle such outputs but I don't see value in it.


PS: Recipe buffers should scale with machine results after applied bonuses, not vanilla recipes. Most heavy beaconized setups starve on vanilla settings.

[Theragus]

A similar problem is with acid neutralisation and steam condensation. Cryogenic plants cannot produce more than 2k water/s. There should be more proper implementation like functionality where machine-to-machine connection means 'unlimited' throughput. I assume that when a machine can produce 110k steam/s it can transfer it internally somehow because otherwise pipes need redesigning to handle such outputs but I don't see value in it.

Sorry for revival of older post, I noticed the same issue, I think this is related to the fact that fluid operations are limited to 6k fluids per second as described here -> https://factorio.com/blog/post/fff-430

[Yoyobuae]

This also happens to max speed Foundry with molten metal from lava recipes. This foundry can never reach the nominal 9600k/s lava consumption the tooltip claims:

Screenshot_2024-12-01_13-40-18.jpg (825.86 KiB) Viewed 921 times

Again this is a problem of the input fluid box being too small. As this fluidbox fills up it slows down the flow to the point where it limits the throughput.

[Yoyobuae]

Ok, someone else mentioned that all machines are limited to 4300 fluid/s input, and they are mostly correct. But it's not a limit of the machine but a limit of the fluidboxes for recipes. Holmium plates recipe doesn't show that limit.

There needs to be a more generalized solution than what was applied to just holmium plates recipe. Either modify the fluidbox sizes for all recipes that could possibly consume more than 4300 fluid/s. Or change the rules of how machine input fluidboxes work to not slow down as the fluidbox fills up.

Something similar happens with the emptying of output fluidboxes. As the output fluidbox empties out, the max outflow lowers according to the fill level of the fluidbox. This is problematic because the machine will not work until the fluidbox empties out enough for the recipe output. So it becomes impossible for any machine to achieve 6000/s fluid output.

Productivity can make this doubly worse. A productivity cycle can finish just before or just after the normal craft cycle, thus filling up the fluidbox buffer and causing the machine to go into "Output full" state even when the output flow is well under the fluid flow limit for the the output pipes:

Screenshot_2024-12-05_00-47-57.jpg (382.08 KiB) Viewed 799 times

[Xeinaemm]

Ok, someone else mentioned that all machines are limited to 4300 fluid/s input, and they are mostly correct.

I wrote automated stress tests for fluid blueprints and this is partially true. Depending on the number of inputs and outputs I have hit the machine's pipeline limits such as 6k/s or 9k/s. The tribal knowledge, in this case, is that you need 10-15% more ingredients produced because the machines are not fast enough to fill the fluid buffer without stalling cycles because of low buffer limits. In CPU architecture it's called 'pipeline stall' or 'air bubble' in real fluid systems.