Mass Deletion - Collecting Examples to Identify the Problem in Steam Turbine Setups (and more)

[mathmanican]

I'd love to start a discussion about mass deletion, and have people share links to other topics where they know it occurs. Some of the issues are probably not fixable without radically affecting FPS, while others might be. If we can congregate all our ideas together, hopefully as a community we can help the devs deal with some of these issues. 

• The fix to matter conversion (the good old philosopher's stone) may have resulted in an increase in mass deletion. This thread discusses some of these issues, and @Lifegrow clearly shows how a tiny blob of gas, that is in a room full of another gas, gets deleted, even if the vent has nothing coming out of it. 
• You can read a similar bug report from @DaClown here, where he shows how trapped gasses get deleted by vents.
• Steam turbine setups that have more than one gas in them will delete steam (for the same reasons as above).  This is easily seen in this bug report from @DefectiveNu. 
• Many SPOMs delete mass, which I'm guessing is related to the same issue as all the above. Multigas environments with turbulent flow results in mass loss (which I'm guessing is to avoid a recursive loop that could potentially result from a fix - leading a possible huge drop in FPS).  
• Note that phase changes that transform gas into liquid, when the liquid has less than 10g, will result in the liquid being deleted (liquids under 10g disappear from the game).  The phase change also can briefly creates a vacuum tile (more below). 
• I see posts and complaints about mass deletion in steam turbine setups.  @sheaker's bug report is where will be placing my comments as they relate to turbines.  Thanks for adding a save. 
• Steam can remain at 0.0mcg, creating lots of other problems. 

Here is one scenario, that I cannot explain at all, and cannot recreate. A couple weeks ago while playing survival, i built a standard CO2 lock to enter the slime biome. After deconstructing some tiles, one tile remained vacuum and the CO2 would not occupy that tile.  Instead, the CO2 tried pushing 10% of its mass into the tile every tick, which resulted in a rapid loss of CO2.  The vacuum tile was consuming my CO2!  I should have taken a video (unfortunately did not), but instead quickly built a tile over the offending vacuum tile, and then deconstructed the tile. The vacuum instantly got replaced by CO2 (normal behavior), and I moved on my way.  I now regret this choice, as I lost the chance to snag and squash a rather nasty bug.

Note that this survival run was modded, so it may be 100% due to a mod (hence another reason for not putting out a bug report).  Anyone else seen this? Has anyone captured it on video, or posted bug reports on it (been looking but haven't see it yet)?  I'm looking for more details in my search for steam turbine mass deletion. If what happens in the steam turbine chambers is just another reincarnation of vacuum tiles eating nearby gas, then that would make complete sense. Every time a blob of 2000g liquid transforms to steam, we can briefly get a vacuum tile.  If that tile, in one tick, consumes 10% of the surrounding steam, then over time we'd loose a lot of steam. Unfortunately, I don't yet know how to reproduce the bug, and so don't really have a proper bug report to make.

Something is going on, and I'd love to really get it fixed. If you want to join the fray, please do.  Anecdotal stories, bug reports, massive date collection efforts, save files, memes, whatever, are all welcome. If you use mods, please include that info as well.   

[TheMule]

My 2c.

I've seen at least one example of steam deletion, reported as a bug, of which the OP posted a .sav file which I tried and wasn't unable to reproduce the bug on it. So one thing we should do it to share .sav files and confirm it's consistent across different players. There might be more than one issue at play here and we do need to isolate them.

[mathmanican]

Based off @sheaker's save in the bug report listed above, and on my own failed attempts, I wonder if it does have to do with "save/load".  What I noticed upon loading his file is that the steam turbine, even while automation ports is off, briefly starts working upon load. I wonder if the turbine gets into a stuck state, until it is toggled on/off/on again. Don't have time to test now, but that's what I'm going to look at next.  It could be that anytime the turbine is in "off" mode upon a load (whether by automation, or too cold steam), then the bug happens till it cycles on/off/on again.  

[sheaker]

16 minutes ago, mathmanican said:

What I noticed upon loading his file is that the steam turbine, even while automation ports is off, briefly starts working upon load.

In the save I uploaded steam turbines and aquatuners were on. That was the point. Turbines have to work immediately after loading the game (game was saved when turbines were on and pipes were full of 2kg water packages). Maybe You saved the game in the meantime.

[mathmanican]

1 hour ago, sheaker said:

In the save I uploaded

I had no mass deletion when opening your save. I saved as instructed after water had drained. Then I reopened and waited a full on off cycle. That is when mass disappeared.

[psusi]

5 hours ago, mathmanican said:

Every time a blob of 2000g liquid transforms to steam, we can briefly get a vacuum tile. 

What?  Where?  I thought the cell of water turned into a cell of steam... where's the vacuum?

 

[beowulf2010]

33 minutes ago, psusi said:

What?  Where?  I thought the cell of water turned into a cell of steam... where's the vacuum?

 

IIRC, the cell of water disappears creating a vacuum, the new cell of steam is inserted into one of the adjacent cells, then the gas/liquid moves to fill in the vacuum tile where the water used to be. 

[mathmanican]

33 minutes ago, beowulf2010 said:

IIRC

I did say "can" just in case I recalled incorrectly (IIRI). 

In @sheakers save I defintely see steam deletion. Pictures are on the bug thread. It only happened to me on a fresh restart of ONI. I wonder if the game does not see the liquid in some of the pipes. Maybe the difference is directly related to the fact that things load faster on a fresh restart.

I had three different results occur and the differences were basically the amounts in the pipe from the turbine to the vent....  more testing needed.

At some point this bug will fall into the save/load swamp to join a host of others.... I wonder if we can mitigate it with automation that only activates on loading? I also wonder if it affects other buildings.

Has anyone seen steam deletion not on save/load? 

[mathmanican]

I added another example to the bug report. 

Spoiler

Here is another save where it is occurring. This save does not have any mods on it, so it is not the mod you have running. 

The Quarantined League.sav

When I open this file from a fresh restart of ONI, the steam amount drops. The amount it drops changes each time, and doesn't appear to be related to anything I understand. There is definitely something going on.  Does anyone else want to confirm that this save deletes heat?

1. Open ONI. Open this save. 
2. Unpause and let the machine run till it stabilizes, and then toggle the temperature automation. 
3. Let the water drain from the pipes. 
4. Use debug to measure your total steam content in the 10 squares below the turbine. 

From a fresh restart, here is a picture of what I see (the 10.4kg is different each time I load). :

Then reload the game (without restarting ONI). Repeat the above.

You should get pretty close to this amount (43.7kg, plus or minus a couple). 

Same save, but when I reload the game matters. 

I wonder if it is connected to the temp sensor automation.  Gonna test that now. 

 

[mathmanican]

If you want to follow the rambling thought process of a mad man, feel free to head to the bug report and see the last few hours of testing I did.  If you want the summary, here is what I found. 

Conclusions:

• Upon starting the game (restart or reload), steam turbines can delete mass. 
• Mass deletion occurs with the liquid over the vent. The mass of liquid doesn't appear to be related to whether it occurs or not. 
• I was unable to get mass deletion ever without liquid in the steam room. So don't put crude oil on the bottom of your turbine rooms. 
• The issue does not appear to be related to automation at all. It appears to be a save/load issue. 
• The amount of steam can drop a lot, or a little.  The actual amount seems random and has nothing to do with initial conditions (the same save, when opened repeatedly, can result in various amounts of steam loss, or none at all. 
• There does not appear to be any way to mitigate it. If you have liquid at the bottom of your turbine, you may eventually lose steam (or get lucky and never see it...)

Without having access to more about what is going on during save/load, I don't think tracking any more information about this bug is really possible.  I don't know of any way to prevent it. 

[BaloneyOs]

Quote

a tiny blob of gas, that is in a room full of another gas, gets deleted

How proportionately tiny are we talking about here? Whenever I allow my jet suit dupes into my steam chamber they could emit enough CO2 to occupy multiple tiles in the steam chamber set to maintain 20kg/tile of pressure. The CO2 always eventually gets deleted without fail. Same happened with tiny bits of PO.

[SamLogan]

 

[Gurgel]

16 hours ago, TheMule said:

My 2c.

I've seen at least one example of steam deletion, reported as a bug, of which the OP posted a .sav file which I tried and wasn't unable to reproduce the bug on it. So one thing we should do it to share .sav files and confirm it's consistent across different players. There might be more than one issue at play here and we do need to isolate them.

So you suspect system specific behavior? Those bugs are nasty.

[TheMule]

19 minutes ago, Gurgel said:

So you suspect system specific behavior? Those bugs are nasty.

System or mod(s). I did reload it w/o mods, still no bug.

The system affects speed. And speed might affect timing. And I suspect the order of some events at load time matters.
 

Here's the bug report. I believe OP posted also on  this forum but I'm unable to locate the topic.

I tried the .sav file IIRC it ran w/o problems for 10 cycles on my PC.

[psusi]

13 hours ago, beowulf2010 said:

IIRC, the cell of water disappears creating a vacuum, the new cell of steam is inserted into one of the adjacent cells, then the gas/liquid moves to fill in the vacuum tile where the water used to be. 

That is really weird. I thought it turned into steam, and then the liquid pushed the steam up to merge with other steam so it could take its place.

10 hours ago, mathmanican said:

I was unable to get mass deletion ever without liquid in the steam room. So don't put crude oil on the bottom of your turbine rooms. 

But if you don't, then you get heat deletion don't you?  I guess this finally explains why @Saturnus's hydrogen vent tamer needed the water refilled every now and again.

I wonder if this random behavior may be related to multithreading?  You might try restricting the game to only run on one core and see if it changes the behavior.

[TripleM999]

8 minutes ago, psusi said:

But if you don't, then you get heat deletion don't you?

I hope, this will be fixed soon too... Or we have to abuse the cooling power more 

 

[mathmanican]

2 hours ago, psusi said:

But if you don't, then you get heat deletion don't you?

@psusi, by find the actual problem, and not just reducing the symptoms with mitigation strategies, we hope the devs can completely eliminate the heat deletion problem. It all has to do with random horizontal gas temp swapping in an edge case that occurs quite often in a steam turbine. If you prevent cold low mass gas from ever being next to hot high mass gas, then the bug cannot happen.  We found the bug, abused the bug (made a few borg cubes), and found mitigation strategies till the devs address it (with the need to address it crucial or else massive abuse ensues). 

At the current state of the game, the best options I see are (1) don't put vent in or above liquid, (2) design so that newly phased changed steam cannot move horizontally, (3) the heat source is on the opposite side of the steam room as the phase changed steam. 

5 hours ago, SamLogan said:

 

Thanks for the link @SamLogan.  If I read this correctly, it just has to deal with heat deletion, not mass deletion. Luckily we know exactly why the heat deletion occurs now. 

[SamLogan]

On 6/26/2020 at 5:00 PM, mathmanican said:

If I read this correctly, it just has to deal with heat deletion, not mass deletion. Luckily we know exactly why the heat deletion occurs now. 

Yes, it's only the heat deletion problem, and that's a good news you have identified the problem.

[DaClown]

On 6/25/2020 at 4:15 PM, beowulf2010 said:

IIRC, the cell of water disappears creating a vacuum, the new cell of steam is inserted into one of the adjacent cells, then the gas/liquid moves to fill in the vacuum tile where the water used to be. 

 

On 6/26/2020 at 5:39 AM, psusi said:

That is really weird. I thought it turned into steam, and then the liquid pushed the steam up to merge with other steam so it could take its place.

But if you don't, then you get heat deletion don't you?  I guess this finally explains why @Saturnus's hydrogen vent tamer needed the water refilled every now and again.

I wonder if this random behavior may be related to multithreading?  You might try restricting the game to only run on one core and see if it changes the behavior.

The tiles have preference and priority for making valid (dis)placements. Existent tiles displace; they swap positions rather than creating new tiles unless a vacuum tile is present and not bugged.

Gases in particular will split mass to fill a vacuum tile. Liquid tiles will displace gas often forcing mergers of gas tiles with other similar gas tiles. Liquid tiles divide their mass as they spread along the ground; usually the mass deletion from this is small, but it becomes very noticeable when slow low mass flows are spread out over long distances, and it seems they have increased the viscosity of fluids over time to reduce how far liquids will tend to spread and how high (compared to when I tested this back before the oil/rocketry updates).

When a tile of one type transitions to a tile of another type then a creation event occurs. The usual displacement mechanics are not primary in this case; so water transitioning to steam will end up creating a new tile and that tile will be displaced to the same or adjacent tiles with preference and priority for valid placement of the tile; this often means merging with an existing tile of the material like other tiles of steam.

However, if the tile is in a condition of strict confinement or its transition from one phase to another results in the consequent phase being in a state of strict confinement then it is possible at that transformation for the tile to be deleted due to no valid placements. The fact that liquids produce a vacuum tile when transitioning to a gas or gases create a vacuum tile when  transitioning to a liquid often means that there are sufficient degrees of freedom in the transformation for the system to avoid most strict confinement conditions that would result in no valid placement of the resulting consequent.

This is not statistically ensured; there exists generally conditions in which the vacuum tile can reduce the effective degrees of freedom in race conditions between droplets, gases transitioning to liquids, liquids transitioning to gases, or solids transitioning to liquids.

It is compounded in rarefied multi-liquid setups in particular, and in some edge cases for multiple types of rarefied gasses near their phase transition temperature particularly in contact with some heat or cold reservoir depending on the specific configurations. When you have mono-liquids they tend to just stack and stack and stack with the mass of the bottom most tiles increasing first before growing up or out; when you have two or more rarefied liquids mixed with each other then they tend to have different dynamics than liquid-on-solid tiles that tends to spread one or both thin and long. When you add temperature extremes or heat deletion engines to this then you can get very chaotic and rapid vacillation of liquids to their gas states and back resulting in a mix of conditions that make rare events or configuration spaces more frequently occuring

Creation and annihilation operators are generally not mass-conserving in this case whereas displacement can be relatively safe.