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Please be aware that the content of this thread may be outdated and no longer applicable.

biopon

Counterflow Heat Exchange

Recommended Posts

biopon    226

Apologies if this is already known, or obvious to some of the more physics-educated of you, but I wanted to put forth a "discovery" I made recently. Quotes because I'm sure it's not a novel concept in the real world - in fact googling the title of this post yields many results. I was happy to discover the behavior in ONI however, and I find it tremendously useful.

I wanted to share, because while the concept of using heat exchangers is well-known in the community (for example when cooking petrol), it's usually not done very efficiently.

This is much better for most applications:

image.thumb.png.137e0b34f89c9f6f1bc79ab17b3d1473.png

The left pumps are pumping 350K H2, 1000g/sec. The right pumps are pumping 250K H2, 1000g/sec. The gas flows in the pipes in opposite directions, to be vented to space in this sandbox setup.

After a few cycles of burn-in period while the relatively heavy tempshift plates stabilize in temperature, the originally 250 Kelvin gas exits the heat exchanger at 331 Kelvins. The originally 350 Kelvin hydrogen exits at 268 Kelvin. This represents an approximately 81-82% effective swap of temperature.

The chamber is filled with 2000g/tile H2, the pipes are wolframite, and the tempshift plates are diamond. Using thermium pipes yields an additional 5% heat exchanged over this distance.

Cutting the heat exchanger's length in half to a more realistic size yields 323 Kelvin and 380 Kelvin outputs, for a 70-73% efficiency. Not much bigger than a normal averaging heat exchanger would be, yet much more useful.

image.png.64e5f2539f8b9973d4f41616eddaf7f3.png

The same technique also works with liquids of course. I'm using this design in my survival game for a petrol cooker:

image.thumb.png.bd6521b56f76f42be6587e8970788466.png

(Ignore the airflow blobs, those are just to house the pumps and massive amounts of liquid needed for the chambers to reach their operating temperature.)

The left input is 402.2C petroleum, the left output is 316.8C crude. The right input is 89.6C crude, the right output is 185.8C petroleum. The "efficiency" is 69% for petroleum, and 72% for crude. The bias towards petroleum can be attributed to its higher SHC.

Note that the piping is plain gold. Thermium, even if just a few bits in the middle of the boxes, should provide a significant boost.

I'm also using this in my survival game for 6 pinchas and 19 wheat plants:

image.thumb.png.d80df7dc051931d045e83044f338729f.png

Right input is 39.7C degree sieved water at 633g/sec, left output is 16C water. Left input is 8.5C pwater at 350g/sec, right output is 39.7 degree pwater. This is very assymetric because of the nearly 2:1 water:pwater ratio, but still a very useful swap that doesn't need much further management.

A few practical notes:

1) The tempshifts are needed and should be placed with 2 tiles between them horizontally, if you are not using multiple small chambers. As per my testing results, multiple small chambers over a long distance do not provide any benefit and just cost extra material and space. If you're short on space however, like in my oil/petrol example, they are superior to a single chamber.

2) The medium you fill the chamber with should be high conductivity. Use high mass (such as oil or petroleum) if you want stability because you expect uneven/periodic flow. Use low mass (such as hydrogen) if you want quick reaction time and/or have a steady flow.

3) Don't expect to counterflow vastly different mass*shc matter and get a symmetric result. See the pincha/wheat example above.

4) The size of the exchanger needed depends on the rate of the flow and the difference in heat content. Experiment in sandbox to find out the optimal size for your use. 

 

 

Share this post


Link to post
Share on other sites
Lilalaunekuh    848

Best usage: A slush geyser and a water sieve.

 

=> Your getting water in the right temperature to grow bristle blossom

(still cold enough to cool the lights and keep up with bad insulation)

Share this post


Link to post
Share on other sites
chemie    654

the main reason I use counter flow is response time to the controller.

By having your cool hit the exiting material and thermal sensor first, the flow is turned off quickly to avoid overshoots.  It is more about thermal mass and controller response versus log mean temperature differences.

Share this post


Link to post
Share on other sites
The Plum Gate    366

You should paint out the hydrogen, instead use a vacuum and see how that goes. Since the plates are touching they should conduct.

If the least thermally reactive element is used, then the side by side touching should confer the heat between two sections of temp shift plates without the need for hydrogen - this adds an element with a relatively low thermal conductivity versus that of a massive temp shift plate. You may be slowing the process and unnecessarily heating or cooling a gas. Have you tried squashing everything down to the point that its touching. Because these systems look more like they will heat up over time.

Furthermore, this would simplify mathematical stuff down to a theoretical system of averages of the liquids to simple weight ratios. i.e. 10kg liquid x vs 10kg liquid x will ignore the thermal conduction properties of the material it resides in unless they are of lower thermal conductivity than those of the liquids. Same goes for gas. The device simply reacts to temperature differences.

Basically it would be like adding liquid to like liquid : ( temp + temp ) / 2 = resulting temp when checking for equilibrium.

When mass is different in a conduction medium, like liquids behave similarly to if you were adding one to another. Where the ratio of mass becomes a factor. I'm a bit rusty on the math here, but when the same conduction mechanisms are at work as described above:

A ratio of masses can be used. (Small mass / large mass = a fractional factor F, where F*{(temp + temp)/2} is your equilibrium temperature. That's esentially worded as: the fractional difference of mass multiplied by the average temperature is equal to the resultant temperature.

These examples assume that there is no continuous inputs but it can be worked like that with careful control of input temperatures - mass dominates the results and the equations when dealing with equilibrium temperatures.

I can't speak for specific heat capacities in mixed material systems. This is beyond my level of wakefulness at the time of posting this. But that would be good to get schooled on - and anything else if I have gotten something wrong. 

Share this post


Link to post
Share on other sites
KittenIsAGeek    1,199
2 hours ago, biopon said:

Apologies if this is already known, or obvious to some of the more physics-educated of you, but I wanted to put forth a "discovery" I made recently. Quotes because I'm sure it's not a novel concept in the real world - in fact googling the title of this post yields many results. I was happy to discover the behavior in ONI however, and I find it tremendously useful.

I've used counter-flow heat exchangers in the past with great success.  I haven't tested recently, but when I built them shortly after Thermal came out (to test thermal exchanges), I found the following:

  • The best exchanges were when you could put the materials in direct contact with each other.  For example, floating clean water above polluted water.
  • If elements differed, sometimes flow rates would have to be adjusted to compensate for differences in thermal capacities.  For example, water and polluted water used to have different heat capacities, so you had to have a larger flow of clean water to make up the difference or you wouldn't get a complete exchange.
  • Thermal shift plates aren't necessary, but if used MUST be spread out.  If they aren't used, the length of the counter-flow must be greater.
  • Liquids transfer heat the best.  Running liquid pipes through a hydrogen room is not nearly as effective as running gas pipes through a pool of water.  This is partially due to mass (liquids are more dense) and partially due to the thermal properties.  You can also use solid materials.  Granite actually works somewhat decently, and I suspect metal tiles or diamond window tiles would work very well.

 

Share this post


Link to post
Share on other sites
biopon    226
41 minutes ago, The Plum Gate said:

You should paint out the hydrogen, instead use a vacuum and see how that goes. Since the plates are touching they should conduct.

That's not how this works in ONI though. Plates don't conduct to another plate, they only conduct to a tile they touch or a gas/liquid that covers their 3x3 AoE. They also won't touch the pipes thermally. 

The closest to what you're recommending is to run the pipes in parallel, touching metal tiles. It's a possibilty, but one that's way too slow to get in balance for gases. Testing with liquids, the approach doesn't work, for the same reason the 2-tile gap is needed between the plates horizontally in the hydrogen setup: horizontal conductivity within the medium (in this case the tiles) really sharply decrease the difference between the end-to-end temperatures.

Taking out every 2nd tile segment works a bit better, like so (gold in gold):

image.png.131a2c4a2b341a7c75fb3d867bab9184.png

But with crude going from 89 to 218, and petroleum going from 402 to 278, this is way behind the segmented heat exchanger variant. Even with thermium tiles we're just starting to approach the gold-piped segmented performance, which was using cheap gold and ceramics.

 

Share this post


Link to post
Share on other sites
The Plum Gate    366

Plates not touching plates - not conducting is an affront to f***ing common sense and physics. I need to get into doing debug more often - so many things are not really explained in the building descriptions.

Metal to metal as a rout is nice to know. Even if it's not as effective. Now I'm wondering about drywall. 

Would I be wrong to assume that the 3x3 invisible wall around the plate is where the mass of the plate is distributed and where conduction occurs?

Share this post


Link to post
Share on other sites
biopon    226
4 minutes ago, KittenIsAGeek said:
  • The best exchanges were when you could put the materials in direct contact with each other.  For example, floating clean water above polluted water.

This depends on the conductivity between the two elements. Water/pwater would be 0.609, water/radiant gold would be whatever the internal formula gives for 0.609 and 120 - clearly something significantly greater. I think with water->radiant gold->medium<-radiant gold<-water we're approaching the best possible result. Cutting out the medium and one pipe (in other words running the pipe in the other liquid) is a bit more cumbersome as you need to freeflow, pump, etc the outer liquid, and can't really guarantee a steady flow rate either.

9 minutes ago, KittenIsAGeek said:

Thermal shift plates aren't necessary, but if used MUST be spread out.  If they aren't used, the length of the counter-flow must be greater.

I found they dramatically increase the efficiency of the transfer. Of course, you need to space them out, otherwise they eliminate the little temperature pockets they were supposed to create. IIRC with the hydrogen gas test they cut the length in half for the same efficiency.

10 minutes ago, KittenIsAGeek said:

Liquids transfer heat the best.  Running liquid pipes through a hydrogen room is not nearly as effective as running gas pipes through a pool of water.  This is partially due to mass (liquids are more dense) and partially due to the thermal properties.  You can also use solid materials.  Granite actually works somewhat decently, and I suspect metal tiles or diamond window tiles would work very well.

I'm pretty sure you can do liquids in gas, gas in liquids, etc. The main difference you're going to see (unless you pick something absolutely non-conductive like chlorine :p) is the reaction speed vs. stability of the setup. I touched on this in the OP but more thermal mass will hold a certain temperature for longer, so for uneven or intermittent flows you will want liquids, for steady you can use gas. I've been experimenting with medium mass earlier on, and when it came to efficiency, it didn't seem to matter if I used 2kg, 4kg or 20kg H2 per tile. The medium needs to be able to hold the difference in DTUs at the given temperature (if this makes sense...) but the conductivity is expressed as DTU/(m*s)/Celsius - which for our purposes is DTU/Celsius, given that length is 1 and time is 1. Mass as shown in the game doesn't seem to play a role in terms of heat exchange capability. 

 

5 minutes ago, The Plum Gate said:

Plates not touching plates - not conducting is an affront to f***ing common sense and physics. 

Welcome to ONI! But yeah if you have a gas/liquid covering the target area, a checkerboard pattern of plates is just as effective for heat distribution as a full wall of them. (You are going to get less thermal buffering ability, I would *guess*, if you leave holes, but it wouldn't shock me if even that wasn't the case.)

9 minutes ago, The Plum Gate said:

Would I be wrong to assume that the 3x3 invisible wall around the plate is where the mass of the plate is distributed and where conduction occurs?

Well they have a 3x3 AoE as far as thermal transfer is concerned, centered on the actual placement of the building. I don't know how mass is handled; is it divided among the 9 tiles, or it's just in the middle... They seem to behave as it is in all 9 tiles at once. They're weird. 

Share this post


Link to post
Share on other sites
wachunga    194

If you are going to be using free floating gas, build the exchanger vertically with hot on top and cold on bottom. There is a mechanic which simulates heat rising. Hot above cold is much more stable than cold above hot, or hot and cold horizontally next to each other. Your temperature gradient will be better and your exchanger will be better. Last I checked, this mechanic did not exist for free floating liquids.

If you are limiting yourself to gold/copper/iron, use iron or copper for the "hot" pipes. Some heat mechanics include a factor of the hotter thing's SHC. This make iron about the same as copper (better for pipe to surroundings, worse for pipe to contents) and both better than gold. This isn't a factor in "cold" pipes so copper or gold are better than iron.

 

Share this post


Link to post
Share on other sites
The Plum Gate    366

I think I'll do some more investigation on them, I feel like I need to know where the mass and AoE is.

Sounds like it's a hollow 3x3 cube.

Share this post


Link to post
Share on other sites
biopon    226
3 minutes ago, wachunga said:

There is a mechanic which simulates heat rising. 

Heat mechanics include a factor of the hotter thing's SHC. .

This is fascinating. Do you have a source for it? Or does this knowledge come from the code?

How do you define hot, with regards to iron? Where would be the cutoff be?

I've been able to start using thermium (sparingly) in my survival builds but this testing was done a while back when I didn't yet have access to it, hence gold being the standard.

Share this post


Link to post
Share on other sites
wachunga    194
39 minutes ago, biopon said:

This is fascinating. Do you have a source for it? Or does this knowledge come from the code?

How do you define hot, with regards to iron? Where would be the cutoff be?

I've been able to start using thermium (sparingly) in my survival builds but this testing was done a while back when I didn't yet have access to it, hence gold being the standard.

This knowledge comes from many hours of empirical testing. Here's a brief summary. I haven't retested since that post, but I doubt anything has changed. If you happen to do some testing that shows otherwise do say so, I'd hate to be leading people astray.

By hot I mean when the pipe is hotter than it's surroundings. By cold I mean when the pipe is colder than it's surroundings.

As for the heat rising, use debug to paint a couple tiles of hot hydrogen above cold. Compare it to the same tiles but with cold above hot. The cold above hot should randomly flip positions to make it hot above cold. The end result being that what started as hot above cold maintains a better temperature differential. For completeness, gas tiles horizontally next to each other should randomly average.

I just rechecked the heat rising and it's still there. I'm not going to recheck the other stuff because it's rather time consuming.

 

Some test setups demonstrating the SHC bit with relation to pipes.

Vent Pipe Test.sav

Pipe Test.sav

Share this post


Link to post
Share on other sites
Gurgel    1,188

I think the only real fix here is Klei putting in heat-exchangers, like they put in tanks. You basically would have a tank with two inputs and two outputs (and 3 variants: liquid, liquid-gas, gas) and the suff that comes out has the average of the stuff that comes in in temperature. The device also would store some to get a buffering effect (which I think tanks already have).

Share this post


Link to post
Share on other sites
KittenIsAGeek    1,199
3 hours ago, wachunga said:

If you are going to be using free floating gas, build the exchanger vertically with hot on top and cold on bottom. There is a mechanic which simulates heat rising. Hot above cold is much more stable than cold above hot, or hot and cold horizontally next to each other. Your temperature gradient will be better and your exchanger will be better. 

Correct.  The liquid starting out hot should be on the bottom and the liquid starting out cold should be on the top.  ONI's thermal transfer mechanics prefers this sort of orientation.  Its also why counter-current heat exchangers don't work as well when orientated vertically.  If you must use a vertical space, run your exchangers in an S pattern.

3 hours ago, The Plum Gate said:

Plates not touching plates - not conducting is an affront to f***ing common sense and physics. I need to get into doing debug more often - so many things are not really explained in the building descriptions.

Part of the reason for spacing out the thermal plates is because of how they work in ONI.  Each plate encourages thermal transfers in a 3x3 area which can be counter-productive to the counter-current heat exchange system because they encourage horizontal thermal transfers.  You actually get reduced efficiency running a line of thermal plates compared with alternating every two spaces as shown in the OP.  Ordinarily thermal transfers prefer to move vertically -- hot changes with the tile above, cold with the tile below.  There is SOME horizontal drift, but not as much as you would expect.  If you throw a thermal transfer plate into the mix, then all directions have an equal thermal heat distribution within that 3x3 area.

This means that some of your HOT will be transferred into the HOT that has already been cooled by your COLD line. With a thermal plate, this effect is much more pronounced and if you don't have your flows perfectly balanced, you can run into issues like your HOT line staying hot for the whole run.

Share this post


Link to post
Share on other sites
MorsDux    163
6 hours ago, biopon said:

Apologies if this is already known, or obvious to some of the more physics-educated of you, but I wanted to put forth a "discovery" I made recently. Quotes because I'm sure it's not a novel concept in the real world - in fact googling the title of this post yields many results. I was happy to discover the behavior in ONI however, and I find it tremendously useful.

I wanted to share, because while the concept of using heat exchangers is well-known in the community (for example when cooking petrol), it's usually not done very efficiently.

This is much better for most applications:

image.thumb.png.137e0b34f89c9f6f1bc79ab17b3d1473.png

The left pumps are pumping 350K H2, 1000g/sec. The right pumps are pumping 250K H2, 1000g/sec. The gas flows in the pipes in opposite directions, to be vented to space in this sandbox setup.

After a few cycles of burn-in period while the relatively heavy tempshift plates stabilize in temperature, the originally 250 Kelvin gas exits the heat exchanger at 331 Kelvins. The originally 350 Kelvin hydrogen exits at 268 Kelvin. This represents an approximately 81-82% effective swap of temperature.

The chamber is filled with 2000g/tile H2, the pipes are wolframite, and the tempshift plates are diamond. Using thermium pipes yields an additional 5% heat exchanged over this distance.

Cutting the heat exchanger's length in half to a more realistic size yields 323 Kelvin and 380 Kelvin outputs, for a 70-73% efficiency. Not much bigger than a normal averaging heat exchanger would be, yet much more useful.

image.png.64e5f2539f8b9973d4f41616eddaf7f3.png

The same technique also works with liquids of course. I'm using this design in my survival game for a petrol cooker:

image.thumb.png.bd6521b56f76f42be6587e8970788466.png

(Ignore the airflow blobs, those are just to house the pumps and massive amounts of liquid needed for the chambers to reach their operating temperature.)

The left input is 402.2C petroleum, the left output is 316.8C crude. The right input is 89.6C crude, the right output is 185.8C petroleum. The "efficiency" is 69% for petroleum, and 72% for crude. The bias towards petroleum can be attributed to its higher SHC.

Note that the piping is plain gold. Thermium, even if just a few bits in the middle of the boxes, should provide a significant boost.

I'm also using this in my survival game for 6 pinchas and 19 wheat plants:

image.thumb.png.d80df7dc051931d045e83044f338729f.png

Right input is 39.7C degree sieved water at 633g/sec, left output is 16C water. Left input is 8.5C pwater at 350g/sec, right output is 39.7 degree pwater. This is very assymetric because of the nearly 2:1 water:pwater ratio, but still a very useful swap that doesn't need much further management.

A few practical notes:

1) The tempshifts are needed and should be placed with 2 tiles between them horizontally, if you are not using multiple small chambers. As per my testing results, multiple small chambers over a long distance do not provide any benefit and just cost extra material and space. If you're short on space however, like in my oil/petrol example, they are superior to a single chamber.

2) The medium you fill the chamber with should be high conductivity. Use high mass (such as oil or petroleum) if you want stability because you expect uneven/periodic flow. Use low mass (such as hydrogen) if you want quick reaction time and/or have a steady flow.

3) Don't expect to counterflow vastly different mass*shc matter and get a symmetric result. See the pincha/wheat example above.

4) The size of the exchanger needed depends on the rate of the flow and the difference in heat content. Experiment in sandbox to find out the optimal size for your use. 

 

 

Put diamond window tiles on the middle spots of tempshift plates and make it a vacuum. That should increase effectiveness tremendously. 

Share this post


Link to post
Share on other sites
biopon    226
17 hours ago, MorsDux said:

Put diamond window tiles on the middle spots of tempshift plates and make it a vacuum. That should increase effectiveness tremendously. 

You mean where the gaps are between the tempshift plates? I've tested that, at least with H2 being the medium. For me it did nothing.

Share this post


Link to post
Share on other sites
chemie    654
20 hours ago, cpy said:

Radiant pipes in metal tiles = win.

Yup.  Here is my base air conditioned oxygen design

 

image.thumb.png.b67fe53778a0b29cef3d42e0e401810d.png

(does not have to have H2; that is left over)

 

image.thumb.png.356a0f4354a7c94187134f8c1fcf3a38.png

60C inlet and 20 C outlet

image.thumb.png.a31bde16da9bce3b81abd8f9a918268d.png

700 g/s slush works well for entire base (which is split in 2 zones; one left and one right).  Same slush flows works for both.  Bypass for temp control.

This controls to within 1C

Share this post


Link to post
Share on other sites
biopon    226
41 minutes ago, MorsDux said:

No i mean the middle of each temp shift plate. And vacuum it out. :) try it. 

Ok so you're suggesting this, but with diamond instead of metal.

image.png.131a2c4a2b341a7c75fb3d867bab9184.png

This has tested significantly worse than tempshifts and a gas as a medium, except when using thermium for the metal tiles. Diamond is marginally better than metal. Or am I missing something?

26 minutes ago, chemie said:

700 g/s slush works well for entire base (which is split in 2 zones; one left and one right).  Same slush flows works for both.  Bypass for temp control.

It's neat but you'd see a marked improvement in the amount of slush used (and temperature removed) if you vacuumed that space out and removed every second segment from the metal tiles. The horizontal conductivity in the metal row is working against the counterflow idea.

In your case it doesn't matter much, as you probably don't care about the output slush temperature or even the amount used, and just want to cool gas. My tests have shown that metal tiles (unless thermium) are worse than gas/liquid and tempshifts, but your case is different because it uses 1 vertical tile to do "everything" so the conductivity between an "above" and "below" metal tile doesn't come into play. I haven't tested a gas/liquid heat exchange within one tile; maybe I should as it can be useful.

Share this post


Link to post
Share on other sites
chemie    654
12 minutes ago, biopon said:

Ok so you're suggesting this, but with diamond instead of metal.

image.png.131a2c4a2b341a7c75fb3d867bab9184.png

This has tested significantly worse than tempshifts and a gas as a medium, except when using thermium for the metal tiles. Diamond is marginally better than metal. Or am I missing something?

It's neat but you'd see a marked improvement in the amount of slush used (and temperature removed) if you vacuumed that space out and removed every second segment from the metal tiles. The horizontal conductivity in the metal row is working against the counterflow idea.

In your case it doesn't matter much, as you probably don't care about the output slush temperature or even the amount used, and just want to cool gas. My tests have shown that metal tiles (unless thermium) are worse than gas/liquid and tempshifts, but your case is different because it uses 1 vertical tile to do "everything" so the conductivity between an "above" and "below" metal tile doesn't come into play. I haven't tested a gas/liquid heat exchange within one tile; maybe I should as it can be useful.

I end up heating the slush to 60C with geyser water since it goes to peppers.  Flow is actually set to top off pepper demand. Two metal tiles works fine here too.image.thumb.png.01b5a8a2b54a57a9bbf8d23896079660.png

Yes, I use a water tank for thermal mass when flow is variable.

image.thumb.png.c5a1bc1f5bf43b1d9cf6689d13f189cf.png

Share this post


Link to post
Share on other sites
Saturnus    3,325
On 02/02/2019 at 2:38 PM, biopon said:

I wanted to share, because while the concept of using heat exchangers is well-known in the community (for example when cooking petrol), it's usually not done very efficiently.

I would disagree. Counter current heat exchangers have been known and used by the vast majority of forum users ever since the very first days after the alpha release where we started exploring and testing the special ONI physics in a scientific manor, and posting the findings on the forum.

It should also be noted that in some cases co-current heat exchangers are a better solution. It depends on the purpose.

Share this post


Link to post
Share on other sites
biopon    226
1 minute ago, Saturnus said:

Counter current heat exchangers have been known and used by the vast majority of forum users ever since the very first days

Maybe so, but that's not apparent from recent new topics. I've seen many averaging exchangers but none that attempt to swap heat. It was also news to everyone active on the discord. But then again, I've only been playing the game for a couple of months, so I can't speak for what it's been like before.

Share this post


Link to post
Share on other sites
Saturnus    3,325
11 minutes ago, biopon said:

Maybe so, but that's not apparent from recent new topics. I've seen many averaging exchangers but none that attempt to swap heat. It was also news to everyone active on the discord. But then again, I've only been playing the game for a couple of months, so I can't speak for what it's been like before.

Here's a picture from early September 2017 of a polluted water distiller I built that could process up 5500g/s. This was before geysers was introduced and when sand was only available in very limited amount in the starting biome. I later made a version that could process 10kg/s but this one is more interesting in the sense that it uses both a counter current and a co-current heat exchanger in the situation where each type is the best option.
5c572a52e9c80_2017-09-07(1).thumb.png.11c2e5dcb1f0cd353200925d2cc30ab1.png
 

Share this post


Link to post
Share on other sites