Vertical infinite liquid storage broken

[Gurgel]

Update: Only the mouse-over display is broken, see my update later in the thread.

I did one because it fitted well. Stores about 1.2t of Ethanol per cycle. Now I observe massive liquid creation and eventually corruption of the liquid amounth to a very small negative (!) number.

Does anybody have similar observations or an idea what seems to make vertical different from horizontal for infinite liquid storage?

My current plan to deal with this is to just destroy the liquid, but it would be nice to know that this is due to the thing being vertical. The numbers it overflows at should be hit at somewhere like 1M cycles (about 6.5 years real time at 3x and not even I will be able to get there ) and hence not be a problem in the real world. But I have been running this only around 5k cycles.

 

 

[Simonova]

I wonder if the inflation couldn't be a result of high pressure liquid's tendency to oscillate (and do so more aggressively the higher the pressure gets), and rounding errors; it's speculation, but if the shifting of pressure causes rounding errors that cause pressure to go up a small amount, and higher pressures means the pressure shifting happens more often, it could result it growing pressure that suddenly explodes?

The only association I see with the vertical orientation is the tendency of high pressure liquids to oscillate pressure vertically. That's something I noticed on the water asteroid: ceilings and floors will end up broken by pressure, but walls will not. I'm speculating again, but maybe pushing down additional fluid pressure from the top is making the oscillations stronger, creating more openings for rounding errors (or whatever is is that's causing the liquid mass to go nuts).

[Gurgel]

55 minutes ago, Simonova said:

The only association I see with the vertical orientation is the tendency of high pressure liquids to oscillate pressure vertically. That's something I noticed on the water asteroid: ceilings and floors will end up broken by pressure, but walls will not.

That is a definite possibility. In vertical infinite storage, the vertical measure of the liquid in the storage chamber is periodically 2, 3, and 4 tiles high, while in a horizontal infinite storage the liquid in the storage chamber is always a static 2 tiles high while running. 

[KonfigSys]

I think there is a difference in the game mechanics between vertical and horizontal pressure. 

I decided to go without any infinite storages this time but could not resist building one on the ocean asteroid to speed up hydrogen rocket port construction there.

This small pocket holds almost all water of ocean asteroid.

Spoiler

 

[Gurgel]

33 minutes ago, KonfigSys said:

I think there is a difference in the game mechanics between vertical and horizontal pressure. 

Very likely, yes. I have 5 or 6 other infinite liquid storages running, all horizontal, all with no problems. I do not really need any of them though, it is just something I started using last year for the first time.

[hbarudi]

I prefer airflow tiles over doors. I build a set of normal tiles that will hold the gas tile with the liquid output and pipes to it, then build conveyors and chute underneath along with the first row of airflow tiles and remove and rebuild the floor with airflow tile to drop all other materials. Then build the liquid pump in the area with pipes, wires, and automation wire. Then close up the area with airflow tile. Now once the spot fills up, a single gas tile will be placed on top and 5 tiles will become infinite storage and you can deliver ice to melt there for water.

[Gurgel]

Here is an update: The cell does not corrupt, it is the mouse-over display that does, see link below.

As an experiment, I have run this for longer. There is a point where the material in the storage cell just stops increasing. At 4'789'561'000'000t the amount stored becomes static. That is in line with floating-point number behavior, were adding a comparatively very small number to a large one leaves the large one unchanged. This is with around 3t added per door-cycle. Note that the other 3 cells store almost nothing in comparison. Also note that there is a 2nd bug that creates liquid in this process, so the storage chamber fills up a lot faster than it should. But, in essence, vertical door compressors very likely work for liquids and will not eventually corrupt your storage. 

 

New bug report: