Water plumbing heat experiment

[Sevio]

@Cilya That's a perfect example of the heat destruction bug in action. If a small layer of water forms on top of another layer for any reason, and it's colder than the lower layer, it will basically make everything below it more or less its own (lower) temperature and destroy a bunch of heat. During the startup phase of the boiler it will take some time to get everything heated up and some steam may condense back down onto the aquatuner, but it shouldn't be happening once the aquatuner basin and everything above it is heated up properly. If you're still having trouble, can you post a screenshot?

[Sevio]

Posting my final boiler build here, with a separated flow heat exchanger. Save file attached at the bottom.

Overlays in the spoiler:

Spoiler

• Input temperature: 26.9 C - vaporization temperature: 122.4 C -  thermal gradient: ~ 95.5 K
• Two modes available: Cold and Power
• Throughput: 2700 g/s polluted water in Cold Mode, 4500 g/s in Power Mode (Power Mode not fully tested)
• Power usage: 1.71 kW in Cold Mode, 2.08 kW in Power Mode. (Power usage tracked with a large battery bank and averaged over 1 cycle, includes remote polluted water input pump) (Power Mode not fully tested)
• Water output temperature: ~ 2 C in Cold Mode, ~ untested in Power Mode (depends somewhat on environment temperature around the cooling tank) (Power Mode not fully tested)
• Maximum Aquatuner effectiveness through use of a polluted water cooling loop
• Power mode also doubles as Efficiency Mode due to Tepidizer preheating being more efficient than relying mostly on the Aquatuner.
• Polluted oxygen siphon
• Aquatuner feedback loop that can save on pumping power when running in Power mode

Caveat: Running it in power mode will break the gas and liquid bridges in the heat exchanger and put a permanent broken building message on your screen. It will work fine otherwise.

It seems my earlier report of 2800 g/s may have been a fluke. I could run 2800 g/s occasionally but it tends to destabilize during an unfortunate evaporation cycle where the aquatuner feed pump cools down too much.

I've also spent quite some time trying to tweak and improve the boiler for cold operation. With granite liquid/gas bridges and wire bridges already in place, the only area left to try and improve was the horizontal conduction of heat through the heat exchanger, which is a bad thing as it reduces efficiency. As we know, conduction between water and granite uses the lower conductivity (that of water), but heat within the granite tiles can conduct to the colder areas more easily. So the idea was to try and replace it with a tile that had the lowest conductivity as possible. Gold Amalgam Hydroponic Farm tiles fit this, but it had no noticable effect or even made things worse. They each put up annoying red markers as well that you can't really get rid of, so I settled for the next best thing, which is Igneous Rock tiles.

Another thing I learned was that in cold mode, the temperature of the water from the Aquatuner feed pump can fluctuate quite strongly when there is very little polluted water in the heat exchanger. I had to set the feed pump's hydro switch to 200 kg to reduce those fluctuations.

The First Law Cleaned.sav

[Cilya]

I also build one boiler with the same design as you. I built it a bit bigger, with two aquatuners and two input geysers as well.

I'm near 8kg/s with 40°C output water.

 

[Sevio]

@Cilya If that's excluding geyser water, impressive throughput! And I should stop saying each boiler I post is my final build because I'm working on another design of my own as well. I'm going (slightly) bigger as well but sticking to a single aquatuner. I have a couple other useful surprises for this one though.

So far it's looking good, I'm getting a throughput of about 3800 g/s with about 1.4 C output water for 1.87 kW and lots of free cooling to spare. I'll post a full presentation once I'm happy with its performance and have tested all the relevant stats.

[Saturnus]

@Sevio I've naturally been toying around a little with a few other optimizations but my main concern I've been addressing is to get rid of the reliance of external cooling and heating. The bottom part in previous designs is thermally active, and that's my main problem with the design. It's difficult to just copy the design and use for others due to this. So my main focus has been to enclose it fully with abyssalite outer case so that it's thermally detached from the environment. The most obvious consequence is having clean water output go all the way round the outside instead. And moving all thermally active parts, like transformers and batteries into a hydrogen filled chamber that also helps heating up incoming water. The last thing I'm still working on is making it fully autonomous. Feed it any amount of water up to it's maximum and it'll process it, stop the feed and it'll go into an idling state that still makes it possible to start up with relatively low start up time. So no dupe controlled valves that needs to be adjusted. Fully plug-n-play.

[Cilya]

49 minutes ago, Sevio said:

@Cilya If that's excluding geyser water, impressive throughput!

It's actually a bit more than 8kg/s (of input polluted water), the effectiveness is slowly increasing as the temperature before the aquatuner pool increases. But that's not very impressive; since I use two aquatuners, this means only a little more than 4kg/s per aquatuner. I'm still not getting the efficiency you have.

This causes another problem. Since the rate of vaporization is near 10 kg/s, a single pump is almost running at 100% when refiling. In fact, when the polluted water starts to evaporate, the two upper tiles of the aquatuner are outside of liquid and stay like this for a long time : each 10kg packet brought by the pump to the upper tiles is instantaneously transformed to steam. The temperature of the lower tiles slowly decrease until it reaches around 110°C where the pool can fill again.

The problem is that during this long period of time where the water doesn't go above one tile high, the aquatuners overheat.

I have two pumps for the clean water since I add 8kg/s of geyser water to the 8kg/s of cleaned water.

[Sevio]

@Saturnus For me being thermally self-contained isn't a goal in itself, a base that's large enough to run an Aquatuner boiler should have enough heat generation that free cooling should be a welcome thing. But the design I'm working on does work fine if you insulate the cooling tank with the right tepidizer setting.

Whether you run the cooling tank insulated or not, the thermal deficit or surplus will eventually exit the boiler into your base in the output water. Of course you can choose whether to accept the heat or cold in your base with your choice of output pipes and holding tank material.

@Cilya That's indeed a problem, how about adding two pumps for input polluted water so they can stop the escalating rate of evaporation as the water level drops?

[Cilya]

27 minutes ago, Sevio said:

 

@Cilya That's indeed a problem, how about adding two pumps for input polluted water so they can stop the escalating rate of evaporation as the water level drops?

I had the same idea. I tested something else though: I let the water level rise to three tiles. Surprisingly, the steam generation is smoother.

[Saturnus]

@Sevio I forgot to mention that one of the major improvements I did was to have the aquatuner input water come from the boiler feed pump. That gives an extremely large temperature gradient to play with so you can run with very high feedback ratio. At high throughput down to 1250g/s and then I have a second liquid filter controlled valve with another 500g/s when it's running low input or idling. That also means there's a constant low volume of water flowing through the system equalizing temperatures making it a lot more consistently running instead of large eruptions, stops and restarts. 

[Sevio]

Welcome to Final Boiler Fantasy V.

Overlays in the spoiler:

Spoiler

New features:

• I've made the boiler completely self-regulating so there is no input valve to tweak. This means it cannot accidentally overflow with too much polluted water. With the default settings and when primed with some starting water  it should be "plug and play" from the start. Additionally, as long as there is fresh polluted water to boil the aquatuner will have no downtime as with my previous designs.
• A freezing protection switch in the cooling tank keeps the clean water or the Aquatuner's cooling medium from freezing in case the tepidizer is set too low.
• A much larger heat exchanger that wraps around allows for much better heat reclamation, making the throughput in "cold" mode much better.
• Battery and Transformer heat reclamation by moving them inside the boiler to replace the statues. They serve as steam condensers, provide extra energy storage and contribute their passive heat to the heat exchanger.
• Integrated polluted water tank with separate tepidizer preheating. This adds an extra pump to power if you're feeding the boiler from another external polluted water tank but it does improve the efficiency of the heat exchanger and its proximity enables the boiler to self-regulate its throughput.
• Minor addition is an access airlock on the right side of the boiler. This can be used to vent troublesome CO2 if any gets in the narrow steam condensing area, since CO2 will not rise above steam into the polluted oxygen siphon.

It can still run in cold or power mode depending on how much you let the tepidizer do preheating, but as with the previous design that relies on liquid bridges, you'll have broken buildings in your boiler. Even the cold mode ends up with a few broken buildings because the heat exchanger works so well.

A minor problem I had is that because the channel wraps around and is quite a bit larger, it suffers from pressure damage if a pump inputs 10 kg/s. That's why there is a valve inside the polluted water tank limiting the input to a maximum of 5500 g/s. That prevents pressure damage from happening at the cold mode's throughput rate, I'm not sure yet if longer sustained flows could still damage it. Either way, the boiler won't be able to boil more than 5500 g/s even with power mode unless some tiles near the liquid vent are replaced with airflow tiles that cannot break.

The abyssalite insulated tiles around the cooling tank are just an addition to quickly test stability when it was insulated, they're not necessary at all if you'd like to provide some free cooling to your base with a conductive tank.

Stats recap:

• Input temperature: 26.9 C - vaporization temperature: 122.4 C -  thermal gradient: ~ 95.5 K
• Can be finely tuned between two modes: Cold and Power
• Throughput: 3750 g/s polluted water in Cold Mode (Power Mode not yet tested)
• Power usage: 1.43 kW in Cold Mode (Power Mode not yet tested) (Power usage tracked with a large battery bank and averaged over 1 cycle)
• Water output temperature: ~ 1.4 C in Cold Mode, (Power Mode not yet tested) (tested with an insulated cooling tank)
• Extra cooling available with uninsulated cooling tank walls in a warm environment and a lower tepidizer setting at the cost of some throughput.
• 240 kJ of battery storage sharable with your main power network if you hook it up directly.
• Electrical equipment heat reclamation.
• (optional) Aquatuner feedback loop can be added that can save on pumping power when running in Power mode. Warranty void if feedback loop is installed while running in cold mode. (pipe will break if the aquatuner feed pump ever stops due to low temperature in the tank)
• Polluted oxygen siphon
• Maximum Aquatuner effectiveness through use of a polluted water cooling loop

Overall I'm quite happy and surprised at the same time that the larger heat exchanger did so much to improve throughput for cold mode. Putting all the electrical equipment inside the boiler to reclaim the free heat they generate while also serving as steam condensers seems like something I should have thought of sooner. And focusing on self-regulation from the start would have saved a lot of wasted time testing with my previous boiler iterations. But that's all in hindsight

Have a look at the save file if you'd like to see it running.

The First Law Cleaned.sav

[Cilya]

Very nice !

At some point, I tried using batteries. But it felt like cheating. Batteries emits a constant heat without consuming any power. The heating is greater with the smaller batteries.

[Sevio]

8 minutes ago, Cilya said:

At some point, I tried using batteries. But it felt like cheating. Batteries emits a constant heat without consuming any power. The heating is greater with the smaller batteries.

It's a very small heat contribution (about 1% of the aquatuner for all the transformers and batteries combined), so the main reason is actually to improve space usage. And since any base that can run this can use a big battery bank, having some of the batteries in the boiler avoids having to create the waste heat somewhere else where it is harder to remove.

Edit: I didn't think of using the small batteries but even now that I remember they produce as much heat as big ones, I think I still prefer the big ones for the extra power storage.

[Cilya]

12 minutes ago, Sevio said:

It's a very small heat contribution (about 1% of the aquatuner for all the transformers and batteries combined), so the main reason is actually to improve space usage. And since any base that can run this can use a big battery bank, having some of the batteries in the boiler avoids having to create the waste heat somewhere else where it is harder to remove.

Good point.

 

13 minutes ago, Sevio said:

I didn't think of using the small batteries but even now that I remember they produce as much heat as big ones, I think I still prefer the big ones for the extra power storage.

If I remember correctly, the takes half the space for a quarter the capacity and the same heat. So, for the same space, twice the heat and half the capacity. I believe the large batteries may have a better conductivity. Since  it doesn't produce much heat anyway as you stated, I'm convinced the greater capacity the better.

[Saturnus]

I've been running some experiments to see if it was at all needed to separate water flows in the condenser. I'm of the opinion it's not. And I think the main reason is that I run the tepidizer to full capacity. The most infested water I could find in a real game scenario was pure lavatory output and it has 10000 germs per kilo. That's 8 million per tile. At this level absolutely no germs made it to the output water. It was touch and go a few time but adding a second tepidizer, which is almost never on I might add, made sure of it. I could now have 14 million germs per 800kg tile, which is a completely unrealistic level in actual game play before it got even close to contaminating the output water.

[Sevio]

With tepidizers being used for maximum preheating, you can get away with not separating flows but that's where our preferences differ, since I do value exploiting the free cooling that this type of boiler can give you.

That said, with the large increase in throughput that Boiler Fantasy V gave while still producing very cold output water, I'm fairly sure that the bigger your heat exchanger gets, the smaller the throughput difference gets between hot and cold mode. Cold mode benefited about 38% from it, I haven't had time to test hot mode yet but I expect hot mode to benefit much less from the larger exchanger as the peak temperature was already very close to the condensation point of steam..

[Saturnus]

Btw, if you want to dramatically increase the efficiency of your heat exchanger I've found that making the polluted water channel 2 tiles tall and have it zig-zag have a profound effect. Liquid and gas movement is always to adjacent squares, not diagonally, so it increases the length of the channel and you can still plaster the whole thing with bridges.

As you can see the straight channel is 16 tiles long while the zig-zag channel is 24 tiles long.

As a side note, our difference in approach is also that you have given up on making it the most compact design it can be. Once you go big a lot of things becomes more efficient and easier. That's not a challenge I see worthwhile pursuing.

I have never worried about water output temperature. There's much more efficient water cooler designs that can easily take care of that. And besides, if it's mainly sewage then that would be recycled in lavs/showers and come out at 30C no matter what. My main goal is take the largest amount of sewage water and convert it into clean germ free water in the most compact space possible while using less than 0.5W/g to do so.

[Cilya]

1 hour ago, Saturnus said:

 0.5W/g

Did you mean J/g ?

[Sevio]

30 minutes ago, Saturnus said:

As a side note, our difference in approach is also that you have given up on making it the most compact design it can be. Once you go big a lot of things becomes more efficient and easier.

What changed is that I now consider Cold mode performance as the guiding metric. That is necessarily a balancing act between heat exchanger size and effectiveness, what a "good enough" throughput is and how complex and large the build is to set up. I'm not trying to make it the most compact boiler design possible anymore, that would be a tepidizer + aquatuner with a minimum of space used for heat exchange.

I've not given up on overall space efficiency though. The integrated polluted water tank that is needed for fast response with the boiler's self-regulation system can save you space elsewhere since you can get rid of a pre-existing polluted water tank once the boiler is running. And the batteries inside the boiler room can save you space on a battery bank elsewhere.

57 minutes ago, Saturnus said:

Btw, if you want to dramatically increase the efficiency of your heat exchanger I've found that making the polluted water channel 2 tiles tall and have it zig-zag have a profound effect.

That zig-zag heat exchanger idea is definitely very interesting, will do some comparison testing soon to see how much better it performs compared to straight channels.

[Saturnus]

@Sevio What I see as the problem is that it competes with a water purifier, a tepidizer and a pump running at 50% of the time. There's enough sand on any map to keep a water purifier going at full tilt for at least 5000 cycles which is longer than you can realistically play as lag would get immense. I'm willing to pay double the power cost per treated g of water and twice the space used just to have it maintenance free and not having to bother about the polluted dirt output. That's how I see it at least.

Edit: it is worth noting that water purifiers destroys or creates heat as the output water is 40C regardless of input water temperature.

[Sevio]

3 hours ago, Saturnus said:

What I see as the problem is that it competes with a water purifier, a tepidizer and a pump running at 50% of the time. There's enough sand on any map to keep a water purifier going at full tilt for at least 5000 cycles which is longer than you can realistically play as lag would get immense.

My interest in boilers started back in AU when sand was relatively scarce. If you really can run a purifier that long off the sand in a map in OU, I suppose that makes all boilers a fairly niche class of build.

It looks like the water purifier outputs polluted dirt every 10 kg, with 200 g produced per second that's a new dupe job every 50 seconds. Sand is delivered up to 200 kg at maximum, with a request for more at 50 kg. So that's roughly a dupe job every 150 seconds. To avoid the constant dupe jobs generated by the polluted dirt, one would have to dedicate low priority containers for it and accept the outgasing or build several purifiers behind locked doors with enough sand delivered to the purifiers in rounds to last a while, the rooms could be CO2 filled so polluted oxygen can be siphoned from the top.

[Sevio]

Well... I made an experiment out of the zig-zag heat exchanger that @Saturnus suggested.

Since you can have the zig-zag on either the bottom or the top of an exchanger line, I decided to compare the two variants with each other and a same-length straight channel exchanger. Each exchanger is 16 tiles long and the zig-zaggers have 4 teeth each. The exchangers are plastered with granite gas bridges at any position where a bridge would touch the hot and cold side or at least the granite wall in the middle. Each channel gets 3000 g/s of polluted water. Cold (6.9 C, 280 K) flows from left to right, hot (106.9 C, 380 K) flows from right to left.

Spoiler

What I can immediately say is that the zig-zag channels require all their tiles except the very end to be filled completely before they start giving any throughput, as it took a long time to fill them up and also a long time to get into a steady state. The longer your zig-zag channel gets, the higher the pressure, so beyond a certain length or throughput this will start damaging tiles.

To recap, Cold starts at 6.9 C and hot starts at 106.9 C. Since the mass flows and heat capacities in both directions are the same, the average temperature is 56.9 C, which we should also see when we average the two output temperatures of each experiment.

From top to bottom, results are:

1. Hot cooled down to 52.5 C, Cold heated up to 53.1 C, average temperature 52.8 C
2. Hot cooled down to 26 C, Cold heated up to 38.7 C, average temperature 32.35 C
3. Hot cooled down to 59.8 C, Cold heated up to 53.9 C, average temperature 56.85 C

Both of the zig-zag experiments cooled down the hot water better than the straight channel, but they also heated up the cold water less than the straight channel did. I think we're mainly seeing heat destruction from dropping water on top of other water, especially in experiment 2. Hard to say if there would have been any benefit if that bug were not present, but as-is I doubt this is useful as there are probably better ways to use the water drop exploit to destroy lots of heat.

Save file included in case you'd like to examine it more closely.

The Zig Zag Law.sav

[Saturnus]

3 hours ago, Sevio said:

Well... I made an experiment out of the zig-zag heat exchanger that @Saturnus suggested.

From top to bottom, results are:

1. Hot cooled down to 52.5 C, Cold heated up to 53.1 C, average temperature 52.8 C
2. Hot cooled down to 26 C, Cold heated up to 38.7 C, average temperature 32.35 C
3. Hot cooled down to 59.8 C, Cold heated up to 53.9 C, average temperature 56.85 C

I think you're not read the results correctly. In a cross streamed heat exchanger it's the total temperature change that is interesting to gauge efficiency.

Both your results was hampered by the fact you'd placed the pump on the zig-zag line one tile too low on both, so the upper channel isn't filled correctly. There were some other changes that could be made like putting in unconnected iron tepedizers in the straight channels and unconnected iron liquid filters in the ziggy-zaggy channels.

So here's the temperature changes and efficiency index assuming the way it was done before is the baseline. And taking your own results:
1.  54.4C + 46.2C = 100.6C = 107%
2.  80.9C + 31.8C = 112.7C = 120%
3. 47.1C + 47.0C = 94.1C = 100%

[Saturnus]

@Sevio I corrected the mistakes you made in your set up and ran the test again at 600kg/tile pressure.

The main reason I did this was because in the test the difference between the 2 top ones should be minimal so with this large discrepancy, there had to be a major flaw in the test set up.

Please remember the goal is the maximum temperature change in total. Perfect equalization at 56.9C is only half way to the goal. The ultimate goal is that the hot input is cooled down to cold input temperature, and vice versa.

The corrected results are:

1. Hot cooled down to 51.7 C, Cold heated up to 54.4 C. t(delta)=55.2C+47.5C=102.7C
2. Hot cooled down to 40.1 C, Cold heated up to 41.8 C. t(delta)=66.8C+34.9C=101.7C
3. Hot cooled down to 59.2 C, Cold heated up to 55.9 C. t(delta)=47.7C+48.8C=96.5C

And with tepidizers and liquid filters added the results were:

1. Hot cooled down to 51.7 C, Cold heated up to 54.5 C. t(delta)=55.2C+47.6C=102.8C
2. Hot cooled down to 42.8 C, Cold heated up to 43.5 C. t(delta)=64.1C+36.6C=100.7C
3. Hot cooled down to 59.2 C, Cold heated up to 55.9 C. t(delta)=47.7C+48.8C=96.5C

So contrary to what I thought adding extra hardware in the channels actually had no effect at all, and even a small negative effect on set up 2. I should also note that in the heat exchanger it's set up 1 that is used so that's why I saw the about 6.5% efficiency increase as these test results also show.

Here's the save files for reference:

ZZ_corrected.sav

ZZ_hardware.sav

[Sevio]

19 hours ago, Saturnus said:

I think you're not read the results correctly. In a cross streamed heat exchanger it's the total temperature change that is interesting to gauge efficiency.

You're correct that the total temperature change affected on both streams is a gauge of the exchanger's efficiency, but I saw so much heat being lost due to the heat destruction bug I didn't want to make any statements about efficiency as that number is unavoidably affected by the bug.

10 hours ago, Saturnus said:

The corrected results are:

1. Hot cooled down to 51.7 C, Cold heated up to 54.4 C. t(delta)=55.2C+47.5C=102.7C
2. Hot cooled down to 40.1 C, Cold heated up to 41.8 C. t(delta)=66.8C+34.9C=101.7C
3. Hot cooled down to 59.2 C, Cold heated up to 55.9 C. t(delta)=47.7C+48.8C=96.5C

I loaded up your corrected saves to have a look and from your results it looks like there is a lot less heat being lost now on the cold -> hot side compared to my previous test but the straight channel is still heating the cold water up the most, which would be the most beneficial for the aquatuner in a boiler.

I think the tepidizers and liquid filters are having no (or a negative) effect because they increase horizontal heat conduction within a channel, but the goal in a heat exchanger is to minimize that horizontal conduction as it makes heat flow back through the channel from the hot side to the cold side.

I do have one possible small improvement to offer on your corrected version which is to replace the teeth with iron mechanized airlock doors. This should accelerate vertical heat conduction as the Steel Door <DO NOT TRANSLATE> tiles conduct heat very quickly between themselves and/or via the airlock building over top of them.

After letting it run for a while with the airlock doors on 10x super speed I ended up with the following results:

1. Hot cooled down to 53.9 C, Cold heated up to 55.9 C. t(delta)=53.5C+49.0C=102.5 C - t(avg) = 54.9 C (96% heat retention)
2. Hot cooled down to 49.4 C, Cold heated up to 49.3 C. t(delta)=57.5C+42.4C=99.9 C - t(avg) = 49.35 C (86% heat retention)
3. Hot cooled down to 59.5 C, Cold heated up to 54.8 C. t(delta)=47.4C+47.9C=95.3 C - t(avg) = 57.15 C (100% heat retention)

What I found interesting here is that experiment 2 still has a significant heat loss but it seems to be almost gone in experiment 1. Also interesting is that experiment 1 it is now able to heat up the cold -> hot channel slightly better than the straight channel, which does make it a possible throughput improvement for feeding an aquatuner. When calculating the average temperatures of the output liquids, (At the void tank endpoints) I also noticed that the straight channel outputs add up to a little over the expected temperature of 56.9 C, perhaps this is due to the heat the pumps are outputting.

ZZ Airlocks.sav

[Saturnus]

Maybe I should mention that one of the things I noticed in the original heat exchanger design was that they have equal direction of flow, ie. the hot and cold comes the same side so that means the best you can achieve is equalization to the median. I changed the direction so it's cross streamed like in the test so it's possible for the hot to be cooled to colder than the median temperature, and the cold to be heated to hotter than the median.

If you do not use cross streamed flow, any heat exchanger design will have significantly worse performance.