#### Archived

This topic is now archived and is closed to further replies.

Please be aware that the content of this thread may be outdated and no longer applicable.

# Let's finally figure out the Steam Turbine Heat Deletion Issue

## Recommended Posts

mathmanican    3410

Title is the point of this thread.  I think it's time to figure this thing out completely (if possible), and hopefully the mass deletion bug as well.  @Tonyroid's solution is  here, but only addresses the symptom, rather than the exact cause. It is the most perpetuated solution on all the forums.  I'll be using this thread over the next week to share ideas, and hope to have others contribute so we can track it down.

@wachunga gave some great advice here (quoted below) as well as showed that we can mitigate the bug 100% by using the 1kg/s phase change mechanics to change the water to steam in the pipes.

"For anyone who doesn't know, adjacent gases of the same material will randomly engage in a temperature transference of a different nature than the normal conduction. Hot gas below cold gas will sometimes swap which creates a pseudo "heat rises" effect. Horizontal gases of the same (similar?) mass will average out their temperatures. Different masses and they swap temps instead. So high mass hot gas swapping temps with low mass cold gas deletes heat. Vice versa creates heat."

I wanted to know if the pressure was the issue, so I built this.

The top turbines have 100kg/tile steam.  The bottom have 20kg/tile steam.  The watts produced by all the turbines, regardless of design or pressure, is 541.5. The tiny differences are explained by the length of the pipes that the water has to go through. If 95C water were heated up and consumed, then the energy should be around 566.66666. However, changing water to steam requires that when the water hits 3C above the vaporization point, then the temp of the steam drops by 1.5C.  The percent wattage I'm getting is 541.6/566.666 = 0.956, so 95.6%. You can't get 100% exactly if the water leaves the pipes (you have to pay extra for phasing up).  What I'm seeing is the wattage I should get if I had to raise the temp 3C for phasing, not 1.5C.  My incoming steam is 161.9C on all the turbines (and I loose about .1C during transit to the pipes). Note 161.9-95.1 = 66.8, and then 66.8/(66.8+1.5) = 0.978 and 66.9/(66.9+3) = 0.957.

So, it looks like I'm missing the heat for a single phase transition with these designs. Did I somehow miss an extra 1.5C that I'm not account for?  Once I can get a baseline build (venting the water out), I plan to make lots of changes.  Or maybe this gives me exactly how much hot high mass gas is swapping vertically with cold low mass gas, though if that were the issue I assume that the 100kg/tile build would have a lower wattage, but it does not.

Feel free to jump into the fray and post anything you find. I'll use this space as my thought board. Lots of stuff will probably be wrong to start with, so we can make a recap post when done.

I bumped the pressure up to 1000kg/tile of steam, and the temperature equiibrium didn't really change.  I'm still missing 1.5C worth of energy.

##### Share on other sites
wachunga    473

I think each cell to cell heat transfer is deleting the 1.5C worth of heat even though it should only happen once. Try changing the diamond window tile to insulated insulation and see what happens, assuming the steam to water cell transfer and/or conveyor bridge is sufficient to boil the water.

My test:

95C water interacting with 141C thermium, transfer is capped at 1/4 of the delta or 11.5C. Minus the 1.5C for phase change and the resulting steam should be 10C higher or 105C, which it is. Each thermium cell in contact adds this 10C while losing the same amount of heat, so the 1.5C deletion is happening with each.

The extra blocks are included so you can use easily the "total joules" from the selection tool as a quick reference. It's not very precise, but the 3 and 4 thermium cells show 23.6 MJ after one tick while the 1 and 2 cells show 23.7 MJ. Which means more heat deletion.

I'm sure you know, but for others checking it out load the save and advance one tick with alt-.

##### Share on other sites
mathmanican    3410
26 minutes ago, wachunga said:

Try changing the diamond window tile to insulated insulation and see what happens

That gave me the extra 1.5C I was missing, exactly. So now this build is getting me exactly what I would expect.

I did need to add conductive wire bridge behind things to get heat to transfer, but it works.  This is enough evidence that I don' think the steam turbine heat deletion comes from vertical differences in pressure and temperature.  Your findings are interesting and make me question how much heat deletion occurs whenever solids next to liquids cause them to phase change (I wonder if it's always a 1.5C loss).

Thanks for the suggestion, as it worked perfectly.

##### Share on other sites
TripleM999    285

My first experiment is a semifail. As @wachunga said, hot gases only swap to the top, but not vice versa, so they are trapped there, and only exchange heat via conduction.

Starting conditions:

- 100kg Hydrogen per tile

- warm section +100°C every hydrogen tile

- cold section -100°C every hydrogen tile

first exchanger reached a swap, but not final equilibrium:

##### Share on other sites
TripleM999    285

Found something wonky, while creating my control group.

Steps to reproduce:

- create the chamber with insulation made of insulation

- build a thermium metal tile in the middle

- modify the temp of the metal tile to 0°C/273.15K ( i painted used super coolant at 273.15K several times)

- paint 100kg hydrogen at 373.15K into one chamber

- paint 100kg oxygen at 173.15K into the other

- wait

System will settle at 67.2°C hydrogen, 66.2°C metal, -63.1°C oxygen. Why? Probably cause oxygen and the metal tile have the same absolute heat at 21110.5 kDTU, so can not exchange heat despite quite a temp difference.

##### Share on other sites
TripleM999    285

So, now to the experiment. First the control group.

I used 10000kg of hydrogen and oxygen to test for conductivity, at either + or - 100°C, with only 100kg i had some issue, like too heavy influence of SHC of transfer materials. Final temp with optimal conductivity should be either at -40.97°C or 40.97°C, my game is limited to show only one decimal digit without rounding, so +-40.9°C

1. thermium metal tile, hydrogen 100°C, oxygen -100°C, result: hydrogen 41.4°C, metal tile 40.7°C, oxygen 39.7°C , it converges the fastest though with 2.

2. thermium metal tile, hydrogen -100°C, oxygen 100°C, result: hydrogen -41.4°C, metal tile -40.7°C, oxygen -39.7°C , it converges the fastest though with 1.

3. thermium wire bridge, hydrogen 100°C, oxygen -100°C, result: hydrogen 40.9°C, bridge 40.9°C, oxygen 40.9°C , it converges slow

4. thermium conveyor bridge, hydrogen 100°C, oxygen -100°C, result: hydrogen 40.9°C, bridge 40.9°C, oxygen 40.9°C , it converges fast

5. thermium wire bridge, hydrogen -100°C, oxygen 100°C, result: hydrogen -40.9°C, bridge -40.9°C, oxygen -40.9°C , it converges slow

6. thermium conveyor bridge, hydrogen -100°C, oxygen 100°C, result: hydrogen -40.9°C, bridge -40.9°C, oxygen -40.9°C , it converges fast

7. and 9. hydrogen -100°C, oxygen 100°C, result: hydrogen -41.2°C, oxygen -40.2°, it converges the slowest with 8. and 0.

8. and 0. hydrogen 100°C, oxygen -100°C, result: hydrogen 41.2°C, oxygen 40.2°, it converges the slowest with 7. and 9.

Only 3-6 are able to fully equalize the temps, and quite to point, so i use these for the final experiment.

Both chambers now contain hydrogen, at the beginning only on one tile at 900kg, the rest is vacuum, temperature is either -100°C or 100°C. Heat transfer bridges are at 0°C. I have done both counter flow and parallel flow experiment, with both wire bridge and conveyor bridge.

1. conveyor bridge, counterflow

2. conveyor bridge, parallel flow

3. wire bridge, counter flow

4. wire bridge, parallel flow

If i made no mistake, there is definitely some heat deletion present.

Btw. vertical exchange has not settle yet with hot at the top. Hot at bottom has equalized pretty good

##### Share on other sites
psusi    308
14 hours ago, mathmanican said:

Horizontal gases of the same (similar?) mass will average out their temperatures. Different masses and they swap temps instead. So high mass hot gas swapping temps with low mass cold gas deletes heat. Vice versa creates heat.

That sure sounds like the issue right there doesn't it?  New cell of cold, low pressure steam swaps temperature with high pressure, high temperature steam and deletes heat.  Let the water drop into petrol and when it turns to steam, it can't swap heat with the steam next to it ( since the sides are petrol ) and problem solved.

##### Share on other sites
mathmanican    3410
12 hours ago, wachunga said:

I think each cell to cell heat transfer is deleting the 1.5C worth of heat even though it should only happen once.

I decided to test this even further, and see if using soilds or liquids made a difference. The 4 turbines below, from left to right, have incoming steam temps of 158.9C, 160.4C, 161.9C, and 163.4C (each differing by 1.5C, the cost to boil water to steam).  Before reading on, make sure you can spot the differences at the boiling plate. I know the experiment is not controlled, but all 4 designs have the same output when the boiling plate is the same (and regardless of pressure).

The right most turbine gets up to 163.4C (the theoretical max possible). The one with crude exposed got up to 161.9C. Conjecture, the water-crude contact causes the crude to loose exactly 1.5C worth of steam energy upon boiling, which means if we have a really large blob of water and decent amount of  crude, I wonder if we can freeze the crude oil by boling the water (that's for another test, and more abuse....). Add heat to crude to freeze it, mwhahaha. There is probably a clamping effect in place that prevents this, but who knows.

The two left turbines have 3 and 2 contact diamond tiles. Conveyor bridges don't appear to affect the temps, rather just speed things up.  The conjecture was that each additional tile would drop the temp by 1.5C (based off @wachunga's experiment).  And sure enough, the left most turbine (3 contact points) settled at 158.9C steam, and the one with 2 contact points settled at 160.4C steam.

These conclusions suggest that the best boiling plates should have no liquid or solid surfaces touching the boiling point. (Seems counter intuitive, and a bug in the code). Instead, boiling should be accomplished either via conduction with a gas from above, or through conduction with a bridge of some sort. The more solid/liquid contact points you have, the more heat you loose. This has quite a lot of implications for all kinds of boilers, and should get a thread of it's own.  I'll come back to it if no one else picks it up.

One thing this shows for sure is that dropping your water into a layer of petro on the bottom of a steam turbine room is guaranteed to loose you energy (precisely 1.5C worth of steam energy, maybe more). If you want zero heat deletion, then don't use this strategy.  This does not address the unequal pressures horizontal swapping issues, which is probably what the petro solution accomplishes, but I'm sure we can do better.  All of the designs above work just fine.

Here is a turbine design the fully mitigates heat deletion.

There are no tempshift plates or bridges. Regular conduction does everything.  The mass of steam is 20kg/tile, and I used the same mass for crude and petro. I would strongly suggest using a bridge on the bottom layer, otherwise starting/stopping could result in water condensing. The max temp reached by the aquatuner is under 230C, around 219-221C, for this example (cooling water in the pipes). You can get better results by spliting the 2kg output into 1kg chungs, and then transforming the water to steam in pipes.

You can stop reading if all you want is a design to mitigate heat deletion. This still addresses the symptom, not the cause, so the work continues.

17 minutes ago, psusi said:

and problem solved

Symptoms hopefully circumvented. The problem has not been solve, rather the symptoms have been reduced and mitigated. It's like what happens when you go to the doctor because of pain, and they just give you pain medication (rather than finding the cause of the pain).  Sure, you might have less pain, but the problem is still there.

My post above actually shows that the solution of using petro on the bottom guarantees you will delete heat. Please spread the word.

##### Share on other sites
OxCD    752

I wonder if there's any relation (speaking about code) between this bug and the flaking one.

And I wonder if devs will be able to isolate the failure inside the code to solve it in the end, as it seems to be very fuzzy, and could involve so many things... Despite the determination you all 4 put into, trying to figure out the Why and the How.

@mathmanican are pressure (water, oil, petrol amount) important when using your last design above ?

##### Share on other sites
psusi    308
19 minutes ago, mathmanican said:

Symptoms hopefully circumvented. The problem has not been solve, rather the symptoms have been reduced and mitigated. It's like what happens when you go to the doctor because of pain, and they just give you pain medication (rather than finding the cause of the pain).  Sure, you might have less pain, but the problem is still there.

But you described the pathology: cooler, lower mass tile of gas swaps temperature with hotter, higher mass tile of same gas to the side.  The reason that the petrol on the floor works is because it prevents this pathology.

##### Share on other sites
mathmanican    3410
5 minutes ago, OxCD said:

@mathmanican are pressure (water, oil, petrol amount) important when using your last design above ?

Not that I'm aware of (i'm currently testing). The design just reaches equilibrium much faster with lower masses, though fluctuates above and below the 163.4C by .2C as expected for a low mass setup. Higher mass gets you to 163.4C, and then visibly stops changing.  I could have used all crude, rather than crude and petro, in the tuner room (just tested this).  I just wanted quick results so 20kg of each was easy. Personally, I would probably just dump 1 load of 200 crude, one load of 200 petro, and one load of 200kg water. Igneous or ceramic insulation should be fine. The tuner needs to be steel for this. Seal it all up, call it done, and never worry again.

5 minutes ago, psusi said:

The reason that the petrol on the floor works is because it prevents this pathology

It only helps reduce it, yet we still don't know exactly when, and under what circumstances, this swapping occurs. The part you quoted is my current working hypothesis, which if we can track down and fully explain, then we can use to our advantage.  That is the goal.

In addition, if the water appears above the petro (which is often true despite being vented into the petro), then it makes the problem worse. If it appears in the petro, then you have even more liquid contact points, and even more heat deletion.

##### Share on other sites
psusi    308
5 minutes ago, mathmanican said:

In addition, if the water appears above the petro (which is often true despite being vented into the petro), then it makes the problem worse. If it appears in the petro, then you have even more liquid contact points, and even more heat deletion.

Wait, where is the heat deletion coming from when the water is dropped into the petrol?  Are we agreed that at least this eliminates the heat deletion from low pressure cool steam swapping temperature with horizontal high pressure hot steam?  Would it not go a long way to fixing this if they removed this ridiculous mechanism of two tiles of gas at different masses simply swapping temperature?

##### Share on other sites
mathmanican    3410
10 minutes ago, psusi said:

where is the heat deletion coming from when the water is dropped into the petrol?

See @wachunga comment early on, and then response.

45 minutes ago, mathmanican said:

These conclusions suggest that the best boiling plates should have no liquid or solid surfaces touching the boiling point. (Seems counter intuitive, and a bug in the code). Instead, boiling should be accomplished either via conduction with a gas from above, or through conduction with a bridge of some sort. The more solid/liquid contact points you have, the more heat you loose.

##### Share on other sites
TripleM999    285

One additional experiment, two tiles of hydrogen at 10000kg and +-100°C, inbetween is vacuum, when more than two tiles.

Horizontal the game tries to equalize the temperatures quickly. The two neighboring tiles of hydrogen equalize in one tick to 0°C. The temp exchange with more tiles does not follow a given sequence, at least i don't see any. The resulting temperature is thus highly random.

Vertically it simple exchange place, when the hotter gas is below, but else, normal conduction takes place.

##### Share on other sites
mathmanican    3410

@TripleM999, if you don't make this assumption already as you try to model, it might help. The game appears to perform all computations from left to right, top to bottom.  So an entire row gets computed, then the one below, and so on.  This is the assumption I made when analyzing liquid flow, and it also appears to be the assumption I've seen others post about. With liquid flow it's a simple 1/4 of the liquid from the right most tile gets pulled left (if possible, and then a few other modifications). There may be some simple flow issue causing this (though the random nature you suggest with no pattern may prove tough).

Does anyone know for certain if the temp exchange code has a random number generator call in it?  If so, any idea which random number generator they use.   If it gets that tough, I will probably give up as I don't want to predict the seed.  Though I guess we could try and do something akin to figuring out how to beat a casino that uses a machine to shuffle cards...

##### Share on other sites
mathmanican    3410

Is it crucial to block the crude oil from interacting thermally vertically?

This design is crap (too much fluctuation in pressure the tile above the vent).

You definitely want to make sure the tile above the vent (where large pressure changes occur) cannot exchange temp horizontally). In the picture below, the second tile above the vent has a very small fluctuation in pressure. however the fact that steam splits to two tiles tall on the left above the tuner is my guess a reason why this one does not perform as well.

Adding a layer of petro on top of the crude keeps a 1 tile long flow under the turbine. Pressure grade actually is largest on the right (where it's coldest), and lowest on the left (where it gets hot, even hotter if I put the tuner one left). So if things are horizontally temp swapping (and that is the only issue), then the design below could actually get better results than theoretical max.

It's too close to call without lots of cycles of testing and only 200kg total water (too much fluctuation).  I'll try both increasing the steam content, and running it for hours to compare the two. Also, a couple carefully chosen tempshifts, and maybe a 1 tile taller vertical column, would help me create more heat than is possible, on purpose).  Gotta head for now.

##### Share on other sites
TripleM999    285

I tried to follow the simulation tick by tick, and i believe, gas dynamic in uniform gases is something like this, which happens every 0.2s tick:

- two adjacent tiles adjust pressure, 10% of the difference is added to the lower, substracted from the higher, temperature/heat of the resulting masses is calculated correct, total heat stays the same. This happens horizontally and vertically. (In a stable flow this is 50% or 1/2 of the pressure difference from tile to tile per second )

- tiles exchange heat through conduction. This happens horizontally and vertically.

- 10% random cells are choosen. I think, it is 10%, cause i observed through multiple runs, that most of the time a tile is modified at most once every ten ticks this way, but it also can happen multiple times in a row every tick. (Maybe only 5% are choosen, as every interaction involves 2 tiles.)

Things, that are checked for these tiles, and randomly executed:

1. if tile above is colder, or tile below is hotter, tiles are swapped, total heat stays the same. this is responsible for the occasional pressure fluctuation in systems, where the heat is generated below.

2. the tile equalizes heat with it's left or right neighbor, total heat stays the same, exact formula is for now unknown to me, goal is to bring the two temperatures closer to each other.

3. the tile swaps it's temperature with the left or right neighbor, total heat changes, if the mass of the tiles is different

In low pressure environments this does not play a significant role. it becomes a problem though in high pressure environments, or better high pressure gradient environments.

If 3 would be similar to 1, it would make for some disturbances in gas flow, but overall would be the better choice.

##### Share on other sites
psusi    308
5 hours ago, mathmanican said:

See @wachunga comment early on, and then response.

So I call that rounding error.  The +/- 1.5 C hysteresis is not being accounted for properly.  But still, isn't the majority of the problem caused by horizontally adjacent gasses swapping temperature without regard to their mass?  And that nicely explains why 2 kg of water being dripped into 20kg/tile of steam without any petrol on the ground seems to delete the most heat?  And why petrol on the ground has such a significant reduction in lost heat?

##### Share on other sites
mathmanican    3410
4 hours ago, TripleM999 said:

10% of the difference is added to the lower, substracted from the higher,

There is a flow variable that controls this. You can find it on oni-db.com . Oxygen and PO2 have 0.12. All others are 0.1.

4 hours ago, TripleM999 said:

10% random cells are choosen. I

This means gas calculations have another check on every cell. This would help explain even more why lag increases with more gas cells. I'll check this as well soon. Thanks for the estimate. Maybe it's actually the flow variable as well and more wild with oxygen.

4 hours ago, TripleM999 said:

. the tile equalizes heat with it's left or right neighbor,

This one I've definitely seen. Instant equalization. However I noticed it won't happen unless cells are close enough in pressure.

5 hours ago, TripleM999 said:

3. the tile swaps it's temperature with the left or right neighbor, total heat changes, if the mass of the tiles is different

This is the big one. Questions. No pattern at all? Do masses have to be close? Does conduction happen before swap?

Thanks for the finds.

##### Share on other sites
TripleM999    285
3 hours ago, mathmanican said:

This one I've definitely seen. Instant equalization. However I noticed it won't happen unless cells are close enough in pressure

Then the question would be, how close is close? I observed this behavior too between neighboring tiles in a flow, so pressure gradient present. Two tiles at same pressure will have same temperature after this, that is easy to reproduce and observe, two tiles at different pressure will have different temps, but are more near to each other in temperature after this operation. I think, there is some cap involved, or the formula is in a way, so tiles with lower pressure will not exceed the temperature of the tile with the higher pressure.

3 hours ago, mathmanican said:

No pattern at all? Do masses have to be close? Does conduction happen before swap?

No pattern, i could observe, i even think, it happens for tiles of same temperature too, but is then unobservable. Masses don't need to be close, it happens in the gradient of a stable flow too. I think, conduction happens as part of the regular steps, so before swap, but i didn't checked this by experiment, the pressures in my test setup were too high for conduction to play a major or even observable role.

##### Share on other sites
13 hours ago, mathmanican said:

So what is the best way to scale up this design? Especially w.r.t the 2:3 / 3:2 AT/ST ratios for water/supercoolant.

I'm guessing its pretty close to the limit on how far the steam can spread horizontally so you couldnt just merge the AT's into one block, merge the ST 95C water outputs to one vent and call it a day?

##### Share on other sites
Manarz    36
7 hours ago, mathmanican said:
12 hours ago, TripleM999 said:

3. the tile swaps it's temperature with the left or right neighbor, total heat changes, if the mass of the tiles is different

This is the big one. Questions. No pattern at all? Do masses have to be close? Does conduction happen before swap?

in all my tests horizontal swapping of temperatures only occured from the higher pressure to the lower pressure tile. If thats a rule you could only ever delete energy with it. The amount of temperature and pressure diffrence doesnt seem to be a condition to the swapping.

##### Share on other sites
TripleM999    285

I think, i found something quite interesting.

My Testbed:

First Test:

hydrogen at 20000kg, 100°C on the left, at 10000kg, 0°C on the right

Most cells show normal flow behavior with 19000kg, 100°C on the left and 11000kg, 9.1°C on the right. Cells 04, 21, 39, 43, 47, 59, 81, 87, 88 and 94 have swapped temperature.

Now the, in my opinion, more interesting test with hydrogen at 20000kg, 100°C on the left, at 15000kg, 0°C on the right.

As before most cells show normal flow behavior with 19500kg, 100°C on the left and 15500kg, 3.2°C on the right. Of much interest are the abnormal cells. They have indeed equalized heat to the extent, that the left cell is now at 19500kg, 23.3°C and the right cell is at 15500kg, 99.8°C.

I think, the temperature swapping is just the solution for the case, that equalizing the heat of two cell would exceed initial temperatures. A bad one, i might say. High pressure operations are especially susceptible for this corner case.

Btw. affected cells are indeed random, even to the point, that a reload will change the pattern. 10% seems to be ok as an estimate though. I will test soon with oxygen.

##### Share on other sites
mathmanican    3410
8 hours ago, TheOneFinn said:

I'm guessing its pretty close to the limit on how far the steam can spread horizontally so you couldnt just merge the AT's into one block, merge the ST 95C water outputs to one vent and call it a day?

That will do it. Things spread just fine, though you will need a tad more liquid. You can also put them on different sides. If the uninsolated version turns out to be just as good, then just make sure all your boiling is done in a single tile wide vertical column, with only one tile of horizontal height.  That should be it.

##### Share on other sites
Manarz    36

Here's the design i came up with:

I used the extra space between the ATs to make the Turbines self sustaining through preheating. this setup produces around 400W continious power and some free cooling.
The heatswapping cannot occur in this design as it is guaranteed that all tiles with a higher temperature always have lower pressure due to flow forced by the turbines.

Also no unnecassary heat should be lost through phasechanges.

If you want to run the turbines with higher output you just inject heat into the steamchambers with the ATs through the doors beneath them.
The Tepidizer is also in a steamchamber and tricked by automation to reach 125°C.

The amount of steam in the turbine steamchamber should be finetuned so that the Tiles in wich the AT injects heat have a steam pressure of around 50g. This tile should also be the only one to reach temperatures above 125°C to trigger the turbines.

This can be built with gold amalgam only. So basically as soon as u reach plastic.