Testing various Counterflow Heat Exchangers designs

[Fradow]

Thanks for the details and the efficiency numbers. I've corrected my previous post to reflect the right 1.5K number.

It seems like the number I was aiming for (600KDTU/s used) is around 90% efficiency. That's good to know.

[Saturnus]

33 minutes ago, nakomaru said:

Personally, I don't think the stability is usually too big of a concern. At high efficiencies you recover almost all of the extra work you accidentally do in your boiler, and your output will always stabilize in your tank. Still a nice thing to be aware of when going for the best design possible.

Stability directly dictates efficiency because the crude oil must be kept under the boiling point. So the maximum temperature in the variance determines the maximum efficiency you can achieve. The greater the variance, the lower average temperature of the crude oil you can safely process is.

Assuming the crude oil is from oil wells that means the crude oil is between 90C and 100C The upper limit set by not allow the water to actually boil. Typically we use the midpoint, 95C, for testing but most people would use a 95C water source for their oil wells, so the actually crude oil temperature output of the oil wells will be between 98C and 100C for an oil well set up with a buffer pool of crude oil to absorb the heat from the natural gas output. But we have to allow for the crude oil to be at least slightly hotter than that, say 102C, at the input and still not have crude oil boiling in the pipes at any point.

So in a test set up with 95C input crude, you cannot allow the crude oil to be hotter than 96C at any point before the boiler.

If we go back to @Fradow's original numbers we see that a 50 pipe length waterfall heat exchanger has a temperature variance of 20C, and even if we made that significantly longer to get closer to optimal efficiency it wouldn't affect stability that much, so naturally a waterfall heat exchanger would have significantly lower maximum efficiency than for example a Z design with a variance of just 2C.

[mathmanican]

3 hours ago, Saturnus said:

The greater the variance, the lower average temperature of the crude oil you can safely process is

And this is precisely why I brought up using the bead flaker design You can design things so your outgoing petro has < 0.1K variability, and the incoming crude temp does not matter (provided its 3K below phase change because of flaking limits). The boiler outputs petro at an essentially constant temperature, and you can throw whatever crude you want at it (observing flaking conditions).  

I agree completely that just dumping liquid straight of a ledge with a waterfall design is terrible for variability, and will mostly likely result in all kinds of busted pipes if you're not careful. But if you swap the boiler, then feel free to send 110C crude.  Things won't break.  The Z at the top isn't needed for stability. You can use a stair, or waterfall, or whatever.   The Z is one way to answer stability.  There are others. 

4 hours ago, nakomaru said:

(The new material spawns in 1.5K less than the temperature of the previous material.) But this is the same for everyone except maybe for flaking boilers (?).

Correct. You don't loose the 1.5K when you flake, which is another plus if you are trying to conserve magma.  

[nakomaru]

I've slept on it a day and re-read all the comments on stability again. Maybe this is a matter of just not having built a particular boiler in a particular way, but it's still hard for me to see the problem with stability as caused by the exchanger.

Of course, if your boiler is sending out petroleum between 405-425 I can see where this is going to boil oil in your pipes when pushing high efficiency. I assume this is not the problem you guys are talking about, because this thread is mostly about comparing heat exchangers, not boilers. If you have a wacky variable boiler output, this can't be blamed on your waterfall exchanger design, and you should fix the boiler design.

One part of the stability we are talking about is the output petroleum varying between say 110-130C, right? But this is your final petroleum output which goes into storage, and never touches the oil in your extraction room. The only thing it touches is the oil which you have already placed into pipes and sent into the exchanger.

At this point, whether your setup sends 95C or 102C or 300C crude oil into the exchanger doesn't matter, and any fluctuation on the input crude oil also can't be blamed on the exchanger. Also, thanks to the exponential nature of heat exchange, any large fluctuations at the oil input side are going to be greatly smoothed out at the oil output side.

For example,
for inputs of 95C and 405C at 95% efficiency, your output oil will reach 389.5 C at steady state
for inputs of 120C and 405C at 95% efficiency, your output oil will reach 390.8 C at steady state

In a real exchanger varying 25 degrees on the input oil (worse than 20C variability on the output petroleum), the output will probably be nearly stable at about 390C. (n.b. the same exchanger might have a slightly different efficiency at different temperatures, but I assume the difference is negligible at these numbers)

The major concern of stability is oil coming out of the exchanger. If it fluctuated 20 degrees, between say 391-411C, then we'd have a big problem. But this basically can't happen because of your heat exchanger because of the heat equation. Fradow's second tests don't show the variation because it was too small to measure when actually isolating the exchanger. This is proving that the issue of stability, which you ostensibly want so that you can make a precise boiler, is only caused by an imprecise boiler. (and furthermore, can be trivially solved even in that case by adding some bridges in the exchanger)

[mathmanican]

50 minutes ago, nakomaru said:

I assume this is not the problem you guys are talking about, because this thread is mostly about comparing heat exchangers, not boilers. If you have a wacky variable boiler output, this can't be blamed on your waterfall exchanger design, and you should fix the boiler design.

I completely agree. Either I missed something crucial, or the Z shape design is there to help smooth out the variability of stuff going into the boiler.  As such, I don't really see any gain with the Z design for counterflow heat exchange. This is why I kept pushing the topic towards boilers, as that appears to be the topic addressed by the Z shape.  What I see is a mismatch between the title of the thread, and the topic of the discussion.  

I'm right with you @nakomaru.  I loved your first suggestion, and played around with constant input/output temps till I was happy with what I saw.  I gained from this thread that right-to-left stairs and waterfalls are essentially the same efficiency per pipe length when it comes to counterflow heat exchanging. The original claim (which @Fradow since refuted) was that waterfalls were significantly worse.  Having seen the numbers myself, and seeing @Fradow's numbers, has been reassuring. 

With the new wiki page dedicated to flaking (thanks to @Fradow), I think it's time to make a proper post for "The Bead Flaker." 

[Saturnus]

1 hour ago, nakomaru said:

The major concern of stability is oil coming out of the exchanger. If it fluctuated 20 degrees, between say 391-411C, then we'd have a big problem. But this basically can't happen because of your heat exchanger because of the heat equation. 

The variance is intrinsic in the heat exchanger. It is not due to boiler variance. You can observe the variance with completely stable temperature petroleum dumped in at the top, and completely stable temperature crude oil pumped in at the bottom of the heat exchanger.

[nakomaru]

I'm having trouble detecting any meaningful variation.
 

The beads generate the highest variation because a packet will encounter 2.5 beads per segment on average, but only a whole number of beads. In fact, I think any single packet will encounter either 2 per segment or 3 per segment across the entire path.

Waterfall oil outputs have no variation at all for these lengths.

[Fradow]

3 hours ago, nakomaru said:

Of course, if your boiler is sending out petroleum between 405-425 I can see where this is going to boil oil in your pipes when pushing high efficiency. I assume this is not the problem you guys are talking about, because this thread is mostly about comparing heat exchangers, not boilers. If you have a wacky variable boiler output, this can't be blamed on your waterfall exchanger design, and you should fix the boiler design.

That's exactly the problem I was talking about. You have to remember that my angle of approach is not to find the perfect heat exchanger, but to teach beginners how to do a petroleum boiler. This means I have to assume their boiler design will be sub-par (I really don't want to recommend a perfect boiler that uses all game mechanisms, but rather one you can complete easily with minimal chances of messing up). Thus I have to recommend a heat exchanger accordingly. Waterfall and staircase falls short with that approach.

3 hours ago, nakomaru said:

This is proving that the issue of stability, which you ostensibly want so that you can make a precise boiler, is only caused by an imprecise boiler.

Indeed, I also edited my original post to mention it, the relevant line is: "Instead of working on stability in the heat exchanger, you can also improve your heat bridge and boiling chamber to avoid any heat fluctuation, rendering the heat exchanger stability meaningless.". Let me know if it's not clear enough!

On the other hand, this also means I should investigate a bit boiler design and at least show and link to other, better designs. So far, I know of:

• the burping boiler  the basic one I showed and that's shown nearly everywhere. The worst one, but also the easiest to do because it has very few assumptions.
• the submerged boiler, linked by @Craigjw. It fixes the most glaring issues of the burping boiler, and is good enough for most people.
• the flaking boiler, explained by @mathmanican (I'd love an up-to-date write-up to link to on it). The advanced super-experienced players design (since flaking is still so obscure).

Is there any other design I should be aware of?

 

[nakomaru]

1 hour ago, Fradow said:

Thus I have to recommend a heat exchanger accordingly. Waterfall and staircase falls short with that approach.

I mentioned modifying it before but you wanted to keep the test simple. Before you throw it out, let me try one more time - you may solve stability in a waterfall exchanger for variable boiler temperature very simply. Behold, the ladder. If you want more, add drywall.

Regarding boilers, even using the first "bad" approach can be very precise for temperature. It greatly depends on the exact arrangement of tiles, materials, heat source and especially heat capacity in that area, and no special materials are needed.

[Fradow]

23 hours ago, nakomaru said:

I mentioned modifying it before but you wanted to keep the test simple. Before you throw it out, let me try one more time - you may solve stability in a waterfall exchanger for variable boiler temperature very simply. Behold, the ladder. If you want more, add drywall.

Regarding boilers, even using the first "bad" approach can be very precise for temperature. It greatly depends on the exact arrangement of tiles, materials, heat source and especially heat capacity in that area, and no special materials are needed.

That's very interesting. I'll run some more stability tests using ladders and ladders + drywall.

As for materials, it's either Granite (highest TC) or Igneous Rock (highest SHC). Since you mention heat capacity, is Igneous the way to go?

 

After thinking about it a bit more, so far I've come to the conclusion that, as a player, you should progress through 3 stages:

• blurping boiler + layers: the easiest to build and repair if anything goes wrong, and it minimizes the chance of building something wrong because you lack knowledge. Of course it's far from being optimized. Use a core tap for heat source because it's, in my opinion, the easiest to use.
• submerged boiler (edit: or another better boiler like shown below) + Z-shape: works reasonably well, have a few gotchas, but nothing too complicated for an experienced player. Dead-spaces can be used by a volcano and part of a petroleum power plant for example.
• flaking boiler + waterfall: the perfect petroleum boiler, at the edge of what's currently known of game mechanisms, since it uses flaking, which is currently still poorly known (let's hope it gains some visibility).

If tests shows that adding ladders + drywall add significant stability, I would consider waterfall a good option instead of Z-shape for the second setup (but not as a starter one: waterfalls are still "experienced player" stuff in my opinion).

I'd be interested to know if those "stages" seem wrong to someone, and if so for what reason.

 

[nakomaru]

I'm not sure if granite or igneous rock would be better. Granite has okay HC too, but much higher TC, so it may help to absorb the shock quicker.

Here's a 4th idea about boilers - only about 3C max fluctuation of petrol (401-404) and fairly steady if you tune the temps. Oil boils in time thanks to a counterflow bead. Input is the earlier 384.1C oil from the aluminum waterfall.

If you have much better efficiency in your exchanger, it will be much steadier output with a normal drop (I can get 403.5-404.5C with extremely regular flow in that case).

[wachunga]

This is the style of boiler that I'm currently partial to:

The one on the left is probably more forgiving to player sloppiness at startup since it has a well to collect unboiled crude. In my testing, both gracefully survive a loss of heat event during operation and restart just fine. The one on the left has a slightly burpier mass flow with the incoming crude displacing ~130kg of petroleum once every second. This results in a flow down the waterfall that varies between about 1920g to 2080g per cell. The boiler on the right only displaces ~30kg of petroleum, resulting in a tighter variation of about 1990g to 2010g per cell. That said, the greater mass flow variation of the left boiler isn't particularly meaningful.

Temperature fluctuations in the crude are almost entirely due to variations in the temperature output of the boiler. Outgoing petroleum varies by about 2.5C which results in crude variation of about 1C. This can be lowered by adding more thermal mass to the boiler or by reducing the heat exchange at the door. There is a bug with doors that creates or deletes heat, I usually avoid this by flanking the door with shift plates. However that results in heat transfer that is much too high. Typically I would advocate not going directly from magma to boiler as a best practice for that reason. But for the sake of continuity with previous tests I went without shift plates and directly from the magma.

The insulated and airflow tiles are to avoid a different bug.

Save with a stair design for the hell of it. boilers.sav

I should add that @nakomaru's point about the boiler being thermally connected to the exchanger is a very good one. It's the biggest and most common issue I've seen relating to petroleum boilers as a whole. Deliberately separating the boiling chamber from the exchanger with a drop that creates droplets instead of beads is highly advised. I didn't do that in this case because I placed the last radiant pipe segment low enough so that waterfall mechanics prevent the heat flow. 

[JRup]

2 hours ago, wachunga said:

The insulated and airflow tiles are to avoid a different bug.

Is there a report for that one I could read? I was thinking of using a metal tile for boiling petro but now this has got me reconsidering stuff.

2 hours ago, wachunga said:

Deliberately separating the boiling chamber from the exchanger

I feel this should be standard for any build. Or at least done as much as possible.

[nakomaru]

20 minutes ago, JRup said:

Is there a report for that one I could read? I was thinking of using a metal tile for boiling petro but now this has got me reconsidering stuff.

Metal should be okay, as long as it's the only conduction happening in your cell and you don't mind the amount it overshoots your target. The bug wachunga is accounting for is probably the one where 1.5K is deleted multiple times upon phase change if conducting with multiple cells. That's why you see insulation/airflow completely surrounding my boiler as well (bridges conduct heat vertically instead - the bug does not happen for any number of building conductions).

wachunga's boiler is probably flat for this reason as well - mine can cause the bug by interacting with the petroleum above/below it.

[wachunga]

37 minutes ago, JRup said:

Is there a report for that one I could read?

I've lost track if there is an official bug report for it. Or where the best explanation is, but this post is probably good enough.

The essence is that when something changes state, the resulting material gains or loses 1.5C as a rebound. Boiling loses heat and freezing gains heat. I think of it as a simulated "heat of vaporization" but dunno if that was the intention. The bug is that each cell exchanging heat with the material causes an additional 1.5C loss. If all 4 surrounding cells exchange heat then 6C is lost, which seems like a bug to me. Airflow tiles exchange no heat and avoid the bug. Insulated ceramic within several hundred degrees don't exchange heat either and should also avoid the bug. Apparently I forgot the specifics, insulated ceramic triggers the 1.5C loss even though the heat exchange is so small that the game effectively discards it. So the insulated ceramic tiles aren't helpful with regard to the bug. I might be forgetting something else as I seem to remember using them in that capacity but it's been awhile.

I think it's insulated ceramic triggers the additional loss when the material is superheated like in the linked post. But not when conduction from some other source brings it over the boiling point.

[JRup]

@nakomaru & @wachunga Thanks for the breadcrumb trail, I've also found this one for follow up reasons:

 

This helps avoid skulls full of aches.

[Occam Blazer]

Since we're talking about boilers, here's the design I'm currently using.

Spoiler

I maaaay have been ignoring the problem for a few hundred cycles. Since I only have about 50 cycles of natural gas left I think it's time to clean up this mess.

 

[tnankie]

I am really confused....are you testing heat exchangers or boilers?

If it is heat exchangers why didn't you use the same liquid in both flow directions? Using things with the same heat capacity greatly simplifies the calculations. I also suspect there may be some weird behavior with the 1 degree C cap on heat transfer.

[Fradow]

I was testing heat exchangers, applied to a petroleum boiler. The conversation went off-topic with boilers.

I'm not interested in "perfect" results, but to how it will work for a petroleum boiler. It makes sense to stay as close as possible to the expected operational parameters of that (and that includes counter-flowing Petroleum/Crude Oil), to have results that will as closely as possible reflect survival builds of Petroleum boilers.

If someone more theory-inclined want to explore the heat transfer weird behaviors applied to heat exchangers using different liquids, I'll be interested about it, but that's not my goal.

[Fradow]

On 5/18/2021 at 8:55 AM, nakomaru said:

I mentioned modifying it before but you wanted to keep the test simple. Before you throw it out, let me try one more time - you may solve stability in a waterfall exchanger for variable boiler temperature very simply. Behold, the ladder. If you want more, add drywall.

I finally got around testing how muc that helps. Here are the results (G for Granite, I for Igneous Rock):

I don't want to draw too many conclusions, because experiments are imperfect, and some slight variations could be attributed to flaws in the experiment rather than a real reproductible difference.

On the other hand, one thing is clear: putting ladder + drywall does help with stability, but not by a lot. If you are serious about stability because you want to push efficiency up, either use a top layer or a better boiling chamber design, like the ones you guys posted.

There seems to be a slight difference in heat used, when not having any additional mass vs having at least ladders. For waterfalls, it's around 5 to 10%. More experimentation would be needed to confirm that, I don't have a high confidence in that.

There doesn't seem to be a big difference between Igneous Rock and Granite. With that data, it's impossible to tell which will perform better. I suspect it would be Granite because of its higher TC, since adding twice the mass with drywall only result in a very slight improvement.

 

And that's about it! As I said before, I won't be getting too much into boiling chamber design. There are great example here, but having a top layer rather than a good boiling chamber design totally solve stability issues, which is good enough for me, and the impact of a bad boiling chamber on efficiency is minimal according to my 2 previous experiments.

[Saturnus]

The bottom line is that Z-based designs still reign Zupreme (), or alternatively a 7-shaped design skipping the bottom layer, for people that wants something that just works, is easy to build, and have almost the best possible efficiency of any design regardless of complexity. In addition, it's far more resilient to different boiler set ups and variances that may occur than more complicated set ups.

I think that's a good place to draw the line on the discussion and start making a few easy to build designs with various heat input types to replace many of the popular but inefficient and/or unstable knee-jerk designs that is often given people in help channels.

[mathmanican]

If stability is the issue @Fradow, then I think we've all missed an important option. Rather than build a Z (a rather poor heat exchanger), just build one of these at the top of your build right before you vent liquid into your boiler. 

Connect the automation port to a shutoff, and done. Your oil will come out with almost zero variability. If you struggle to keep petro from vaporizing your pipes, then that's a boiler stability problem, with various ways to fix that as well. 

I still see no reason for the long Z, other than it might look cool. If you don't want to build a reservoir because of the costly materials, then how about a pit that holds 6 tiles of oil.  Just have a lip that lets the oil spill into your boiler.

I created a thread on using flaking for a boiler, with a liquid storage tank for stability.