Magma powered petro boiler

[nakomaru]

Basically, the minimum vertical flow is too big for its compressive strength.

These are the properties of water:

Here's what happens when you paint 1000kg tiles.
We end up with 999.10→1009.10→1020.0→1030.2→1045.5→1050.9→1061.4
Basically, the cap is Max Tile Mass. Each tile below the cap will be the tile above times Max Compression
There's a little bit of error with this prediction and that's because 999.10 is a display bug (see what happens when you paint 999.98kg in debug).

These are the properties of magma:

Magma should like to stack like this:
1840→1858.4→1877.0→1895.8

Let's see how magma flows vertically.

Some observations/theory:

• If below Max Tile Mass - 2×Minimum Vertical Flow (1840-2×20 = 1800), then any increment of 20kg can be accepted into the tile from natural gravity from above.
• Once above 1800kg, there needs to be at least a 20kg differential to have ordinary flow.
• Compressive flow is a different process. The top tiles act as if they have mass ×1.01 when checking for downwards compression, and requires a 2×MinVertFlow (40kg) differential against this value and the tile below.
• A top tile of 1840kg has 1858.4kg compressive strength, but the bottom 1840kg tile can be no bigger than 1818.4kg to accept downwards compressive flow. No flow occurs.
  • The water equivalent of two 1000kg tiles: the top tile has 1010kg compressive strength, an effective differential of 10kg, and a minimum effective differential of 0.02kg. No problem.

If this model is correct, two tiles of 3999kg will therefore remain stable, and two tiles of 4000kg will show downwards compression. Well:

[psusi]

Interesting... I didn't realize liquids had a separate minimum *vertical* flow or max compression.  I just thought that they tended to flow down, and if the tile below was over pressure, it would try to flow anywhere else and that could sometimes be to the side, then if the tile to the side reached max pressure, it would try to flow somewhere else, and that could be up.

[TheMule]

On 12/17/2020 at 2:22 PM, nakomaru said:

It's nice that you can put a ladder in and have everything be dupe accessible for modifications later too. Just start with the waterfall and add whatever when you want.

If you care about boiler energy, 18K is a big difference. With an optimal heater and accounting for phase change loss, the total boiler heat required for the wide exchanger is 22.6J/g. The required heat for the straight drip shot is 52.9J/g.

  Reveal hidden contents

Kinda jumping in late, I have no idea how I missed this thread, but great design @nakomaru!

For a 10kg/s boiler that would be 226 kJ/s and 529 kJ/s? I've build boiler with a straighforward waterfall and an aluminum pipe I believe less than 15 segments, and I remember the thermium AT (with supercoolant) running 1/3 of the time (roughly eyeballed), so it's something in between (1181/3 = 393 kDTU/s).

But now that I think of it, oil was around 368K.

And the AT cooled down the outgoing petroleum (the only source of heat for the AT), which is a way to artificially increase the efficiency of the exchanger. And now that I think of it even more, if you consider the AT part of the exchanger, that's 100% efficiency (heat wise, of course you pay a price in electrical energy to achieve that).
I don't remember at what temperature petroleum came out of the 1st stage of the exchanger. I remember that after the 2nd stage (with the supercoolant from the AT), it was something around 377-378K.

[nakomaru]

Thanks. Yes that would be the heat flow required at those numbers.

First, two notes for myself.

Regarding thermal transfer in a waterfall design. Heat cannot flow upwards whatsoever.

However, there is inefficiency while falling still. For example 100C 1kg petroleum falling above 0C 1kg petroleum tile will become 87.5C and 12.5C the next frame.

So it makes sense that a beading design can outperform a continuous waterfall if the sizes are right. This is not new info.

Second, my previous understanding and calculations about phase change hysteresis were wrong.

• Crude oil's vaporization point is exactly 673K = 399.85C (listed as 399.9C).
• If there is ever heat transfer which results in it becoming >+3K (not ≥) this number, it becomes petroleum (even if it is e.g. 1500C)
• Once it boils, it loses 1.5K.

I was also using aluminum. I tested a straight 10kg/s waterfall for 15 tiles of aluminum.

• Oil 94.8C→366.5C (100% efficiency would be 401.4C)
• Petroleum 401.4C→140.5C (100% efficiency would be 107C)
• I consider this exchange to be 88.6% efficient.

Spoiler

• To bring the oil to 402.9C in order to boil, we need the boiler to give it +36.4K, which is 615kJ/s at 10kg/s.
• An aquatuner on supercoolant supplies 1182kJ/s.
• It needs to run 52% of the time in order to do this.

Of course when you actually couple your boiler and aquatuner to the system things get a little more complicated.

• As you point out, your petroleum will be colder than expected as a result of cooling it, but that won't help with the heat exchanger's efficiency.
• It's almost impossible to not overshoot your temperature, slightly reducing efficiency.
  • About 90% of the extra heat will be recovered anyway.
• I dunno, there is lots of weird stuff going on in the game.
  • You can create heat by certain boiler tile arrangements.
  • You can delete heat at high petroleum storage masses.

If you have a design that operates significantly different from these predictions I would be interested to see it.

[Jann5s]

This thread is inspiring, now i'm doubting to build a petro boiler even if my sour gas distiller is running very well and I have absolutely no need for petrolium, nor power.

[Nedix]

This thread is so advanced (for me) that it belongs into a completely different category entitled "How to overthink and over-engineer this game"

[KittenIsAGeek]

On 12/16/2020 at 8:05 PM, nakomaru said:

You can see the math

I love to see the math. More people need to include math for their builds. Rock on!

[FIXBUGFIXBUGFIX]

On 12/18/2020 at 3:16 AM, psusi said:

Weird... what property of magma causes this strange behavior?  Why will it not push up, it is fine to top off a partially pumped cell?

  - elementId: Magma
    maxMass: 1840
    liquidCompression: 1.01
    minVerticalFlow: 20

 

Mass of a full tile: 1840kg
Ideal mass difference between two layers: 1.01*1840-1840=18.4kg
Min mass to start vertical movement: 20kg
20kg>18.4kg

 

Spoiler

Similar exploit

  - elementId: ViscoGel
    maxMass: 100
    liquidCompression: 1.01
    minVerticalFlow: 10

[nakomaru]

5 hours ago, FIXBUGFIXBUGFIX said:

Mass of a full tile: 1840kg
Ideal mass difference between two layers: 1.01*1840-1840=18.4kg
Min mass to start vertical movement: 20kg
20kg>18.4kg

Close, but not quite.

Spoiler

On 12/18/2020 at 6:50 AM, nakomaru said:

• Compressive flow is a different process. The top tiles act as if they have mass ×1.01 when checking for downwards compression, and requires a 2×MinVertFlow (40kg) differential against this value and the tile below.
• A top tile of 1840kg has 1858.4kg compressive strength, but the bottom 1840kg tile can be no bigger than 1818.4kg to accept downwards compressive flow. No flow occurs.
  • The water equivalent of two 1000kg tiles: the top tile has 1010kg compressive strength, an effective differential of 10kg, and a minimum effective differential of 0.02kg. No problem.

If this model is correct, two tiles of 3999kg will therefore remain stable, and two tiles of 4000kg will show downwards compression. Well:

 

(Your model predicts the transition happens at 2000kg, not 4000kg.)

[Adirelle]

Noob question to HeatEngine: in your magma pump setup, won't the bit of oil in the bottom right corner of the pump be pumped ? (Or why it is not ?)

[nakomaru]

@Adirelle That's naphtha, and it is pumped. It's also vented back 2 tiles above the pump.

Incidentally, it is also cooled and provides cooling for the pump via the tempshift plate near it.

[HeatEngine]

20 hours ago, Adirelle said:

Noob question to HeatEngine: in your magma pump setup, won't the bit of oil in the bottom right corner of the pump be pumped ? (Or why it is not ?)

It does get sucked in but then gets ejected via the liquid vent two tiles above. The liquid pipe element sensor opens the vent when naphtha passes it and closes the vent otherwise allowing magma to flow further along the pipe.

The whole thing is required to trick the pump into pumping magma without actually being submerged in it. More details in

 

 

 

[TheMule]

On 1/8/2021 at 5:37 AM, nakomaru said:

If you have a design that operates significantly different from these predictions I would be interested to see it.

Oh, I doubt it. The only interesting part of that design it was the lack of a tepidizer. Using the outgoing petroleum to heat up supercoolant is something I came up independently (and I'm proud of), but as many things in ONI, it's most likely I've rediscovered something already known to the community.

[Jann5s]

@nakomaru, can you please explain how to automate the door such as to initiate the waterfall?

[nakomaru]

9 hours ago, Jann5s said:

can you please explain how to automate the door such as to initiate the waterfall?

On 12/18/2020 at 1:31 AM, nakomaru said:

The minimum you need is the hydro sensor directly above the door.

• When there's no liquid in the tile above the door, close the door.
• When there is liquid, open it.
  • Opening the door when there is liquid above it will initiate a waterfall.

It looks like I used a buffer to delay closing the door, but that's probably not needed.