Early metal refinery on magma

[NikiHammerforce]

Hey guys!

I would like to share my project of copper/gold refinery (without metal refinery building) using magma that can be built early.

Update 11/01/2022: more efficient version in replies.

Update 12/01/2022: after running long tests I realized that 14 steps is absolute minimum for this design

Disclaimer: I built and tested it in sandbox only, seems to work fine, no guarantees though. It takes lots of space, considerable amount of materials and is hard to build. Most players just don't need this. It is meant for early game only. Later there are much better options available.

Goal:

• No plastic, steel, gold amalgam
• No conveyor sensors (they are not that fast to research in DLC)
• Temperature of refined metal must be <100°С (I have no crude oil on my current map, so all liquid "drop" locks on my base are made of brine)
• No external cooling used
• Low energy consumption

Required to build:

• Wolframite, tungsten (can be steel or even iron if magma temperature is low enough, but I don't have any iron on map) for left chamber
• Ceramic (can be obsidian) for left chamber
• Mechatronic engineering skill (btw I have master's degree in that, heh)

Results:

• 1200 kg/cycle of copper/gold (can be much more if you don't care about temperature)
• Output temperature is ~75°С (input ore was 27 °С)
• ~20 W average power consumption
• No dupes labor (well, they need to bring ore to loader)

I will briefly explain what is going on. If somebody is interested I will give more details.
Idea is to heat ore from magma and then to cool resulting refined metal with incoming ore.
Here we have 10 stages counter heat exchanger. Metal moves towards ore on conveyors and stays in each stage enough time to almost fully exchange heat. With copper tiles I found that 10 seconds is enough. Optimal time varies for different materials. So the whole thing works in time chunks: 1 second for moving conveyors and 9 seconds to wait.
There are 2 chutes in melting chamber. Left is used when we need to add heat: ore falls on weight plate and liquid metal flows down after melting. Right is used when we need to cool liquid metal to solid state. After we have good amount of solid metal autosweeper grabs it. To know if we need more heat or not I used hydro sensor (not thermo sensor). While metal is liquid we should add ore to it. When there is no liquid metal more is melted with magma.
I believe left chamber may be built in survival without atmosuits considering good planning and luck. But I haven't tried it yet.
Loader is used as buffer as well, it shouldn't be empty. So it must be in vacuum. Otherwise it is heating from hot metal. But cooling of self induced heat is needed. So I have brine there (I would prefer crude oil) which is cooled from the last stage of heat exchanger by means of pipes with polluted water.
Output had to be done with 2 chutes as well. I haven't found any other way to ensure that input is equal to output without conveyor sensors/meters/shutoffs which I deliberately do not use here.
Right plate is just for automatic shut down after we have 2 tons of metal there. Switches are mainly used for starting this thing up.
Timer is used to make movement every 10th second (may be set faster or slower depending on what metal temperature and quantity we want). Even with this "slow" speed we have 1200 kg/cycle. That's really good for me. After all metal refinery building gives 1500 kg/cycle (with untrained dupes and no light, this is early game), eating 1200 W of power and generating tremendous amount of heat. Yep, we will make a good use of it later, but not in early game.
I tested with only copper, only gold and mixed copper/gold. I suspect that mixing with iron would be not very good idea. For mono metal left chamber can be 2 tiles lower. As there is space left for 2 liquids above each plate. For aluminium this will not work due to huge difference in melting points of ore and metal.

What I really do not like here:

• Cooling of loader and sweeper. There is a good chance that sweeper drops hot metal to brine. It haven't occured with my current settings of automation, but I haven't tested it for hundreds cycles. And in the case of losing power during sweeping this scenario is very plausible.
• Bulky difficult build using considerable amount of resources, space and lots of automation but is good only before having turbines. With them this space will be used much much better.
• Much testing must be done to ensure fail safe operation and heat leakage.

Therefore I am actually in doubt if I will build this in survival..
Any thoughts and justified criticism is welcomed! Thanks for the reading!

[ghkbrew]

Nice, I wonder how well this could be repurposed as a rust melter.  They generally need pretty crazy heat exchangers, but running at 1/10th the speed makes the exchanger effectively 10x longer. Though obviously you pay for it in throughput.

[NikiHammerforce]

13 minutes ago, ghkbrew said:

Nice, I wonder how well this could be repurposed as a rust melter.  They generally need pretty crazy heat exchangers, but running at 1/10th the speed makes the exchanger effectively 10x longer. Though obviously you pay for it in throughput.

I don't see why it wouldn't work for rust. Considering magma is hot enough. But if aiming for low output temperature is critical there will be even more sections. Temperature differential is ~1.5 times more than with copper/gold. Some speed increase can come with diamond windows.

There may be some speed boost with using vertical heat exchanger (same as mine, but 90 degrees rotated). Materials on conveyor rails seem to exchange heat with tiles under them if behind isolated tiles. Seems to be a bug. But I haven't tried to use it so far.

Also I thought of additional cooling by dumping rocks (sandstone, granite etc.) to magma with third line of conveyor. But this would increase amount of magma which is no good here.

But after all I believe much more effective way would be cooling by turbines and then just 1-2 stages of this heat exchanger to get below 100 degrees. This way it will generate good amount of power. So if you have turbines don't build this! =)

[nakomaru]

Thank you for sharing. It's quite straightforward and does what it's built for, and the throttle is a great way to improve efficiency/size.

15 hours ago, NikiHammerforce said:

Cooling of loader and sweeper.

A setup like this will allow you to cool both with perfect insulation from the debris, even in failure. Currently your debris interacts with the brine directly, although briefly.

You could be dumping the magma into the heating chamber directly instead of cooling down the entire pool, which will eventually solidify, but I believe this is just for a temporary early game build so that issue is ignored.

Regarding bulk - you don't need much/any insulation besides vacuum, so you can (and should) remove the insulated tiles between exchanger stages.

I can't remember if this was changed or not, but having imperfect insulation tiles touching brine can eventually boil it to steam, if it heats up to something like 130C from the melter/rails. (flash boiling/partial boiling/flaking mechanic). I recommend any tiles you see that touch brine and aren't perfectly insulated be changed to airflow. The melting side seems okay (liquid is a weak conductor to insulated tiles) but I'm worried about the exchanger side.

Debris will indeed exchange at 1/4th rate (wiki says 6.25% rate?) with the tile below if not exchanged with its current tile - don't think this is a bug, you can add rails above the exchanger for a little bonus - you already have space planned out there anyway.

4 hours ago, NikiHammerforce said:

Some speed increase can come with diamond windows.

Maybe not - debris interaction takes the lowest thermal conductivity value, so anything higher than what you're using won't change that (just inter-tile transfer and maybe HC, which you can improve by adding all sorts of bridges also).

[NikiHammerforce]

  

1 hour ago, nakomaru said:

A setup like this will allow you to cool both with perfect insulation from the debris, even in failure. Currently your debris interacts with the brine directly, although briefly.

Thanks a lot for all good advise! I haven't thought of it. This is what I need!

1 hour ago, nakomaru said:

You could be dumping the magma into the heating chamber directly instead of cooling down the entire pool, which will eventually solidify, but I believe this is just for a temporary early game build so that issue is ignored.

It would be good idea to design so that I could easily switch heat exchanger to turbines in the future leaving chamber as is. So solidifying all magma is not what I really want. But I haven't thought of any easy way to get magma inside by portions that will not solidify as solid block. Also I don't know what would I do with hot igneous rock before I have turbines. But I will think more about it.

1 hour ago, nakomaru said:

Regarding bulk - you don't need much/any insulation besides vacuum, so you can (and should) remove the insulated tiles between exchanger stages.

Yes, good idea! I realized that already.

1 hour ago, nakomaru said:

I can't remember if this was changed or not, but having imperfect insulation tiles touching brine can eventually boil it to steam, if it heats up to something like 130C from the melter/rails. (flash boiling/partial boiling/flaking mechanic). I recommend any tiles you see that touch brine and aren't perfectly insulated be changed to airflow. The melting side seems okay (liquid is a weak conductor to insulated tiles) but I'm worried about the exchanger side.

Good point as well. I have seen this already on another run with obsidian insulation. That's why I tried to put 2 tiles of insulation where possible. But most designs I see use only one layer.. so I gave it up too..

1 hour ago, nakomaru said:

Debris will indeed exchange at 1/4th rate (wiki says 6.25% rate?) with the tile below if not exchanged with its current tile - don't think this is a bug, you can add rails above the exchanger for a little bonus - you already have space planned out there anyway.

Already done that. And it is really big improvement. Thermal conductivity of ore is bottleneck here. So with extra exchange due to this game mechanic (very strange if you ask me, but understandable) it seems that exchange time can be reduced from 9 to 5 seconds. (More testing should be done to be sure).

2 hours ago, nakomaru said:

Maybe not - debris interaction takes the lowest thermal conductivity value, so anything higher than what you're using won't change that (just inter-tile transfer and maybe HC, which you can improve by adding all sorts of bridges also).

I was curious and tested with diamond tiles and it seemed to be better. But I haven't done much testing. Also I tried bridges but I gave them up because I switched to just 2 tiles for each stage (there were 4 before, but that didn't gave benefit for the cost of refined metal). Now I realize that I can just place bridges to increase heat capacity of each stage. Why not if they are not of refined metal =)

[nakomaru]

4 minutes ago, NikiHammerforce said:

I haven't thought of any easy way to get magma inside by portions that will not solidify as solid block.

I've used something like this or this.

[NikiHammerforce]

5 minutes ago, nakomaru said:

I've used something like this or this.

Oh,, I am that newbie!.. I've used same thermal isolation method (tile with vacuum) in current design for liquid metal, but couldn't realize that I can use the same for magma. Thank you!! =)

What would I do with rock is a question though. It can't be left inside chamber or it will solidify to solid block eventually as I understand.

[nakomaru]

As I understand, you are melting materials below the freezing point of magma, so just ship out the igneous rock after taking its useful heat on the same rails. See the ceramifier v1 for details - it does this with ceramic + igneous rock. The rock won't be so hot once we use most of its heat. This won't work for iron/rust - so a tricked pump will be best. You could even include its output in the exchanger at a low flow rate.

[NikiHammerforce]

58 minutes ago, nakomaru said:

As I understand, you are melting materials below the freezing point of magma, so just ship out the igneous rock after taking its useful heat on the same rails. See the ceramifier v1 for details - it does this with ceramic + igneous rock. The rock won't be so hot once we use most of its heat. This won't work for iron/rust - so a tricked pump will be best. You could even include its output in the exchanger at a low flow rate.

Thanks! I like this idea very much, but now I will have to think how to design all this early with as low refined metal and wolframite as possible.. I begin to realize that using magma heat to drive turbines may be not the best use of it's limited heat capacity. And if I end up with using magma for power I can build separate independent power station for that. While this refinery will do its job effectively enough. So back to sandbox..

[NikiHammerforce]

Here is updated version of.. Mafinery? =)
v1.1

Update: after running longer tests different results obtained. 14 steps of heat exchanger is absolute minimum for this design: 

Spoiler

 

 

It is still work in progress. I place it here because I am going to redesign melting chamber completely to get magma inside instead of cooling down the whole pool (as @nakomaru pointed out). And I am not sure that I will return to this one. But it works basically and it is more efficient.

Main differences from v1.0:

• ~1700 kg/cycle ~75C ~123C (input ore 27C)
• Conveyor rail for ore is longer so debris on rail exchange additional heat when above metal tiles
• All bridges and tubes/wires (except automation to save on refined metal) thrown in on each exchanger stage
• Vacuum instead of insulated tiles
• Additional liquid sensor and slightly different automation logic
• Reworked sweeper/loader chamber

What should be done:

• Sweeper/loader chamber failed once. I cannot understand why. I couldn't repeat it, but it is not reliable.
• Space optimization
• Automation optimization
• Safety measures
• Building process and rework optimization

Thanks @nakomaru for design ideas!

[NikiHammerforce]

23 hours ago, nakomaru said:

As I understand, you are melting materials below the freezing point of magma, so just ship out the igneous rock after taking its useful heat on the same rails. See the ceramifier v1 for details - it does this with ceramic + igneous rock. The rock won't be so hot once we use most of its heat. This won't work for iron/rust - so a tricked pump will be best. You could even include its output in the exchanger at a low flow rate.

I've tried this approach.. But since igneous rock has much higher heat capacity than ore it comes out with too high temperature. Bigger heat exchanger is needed. Or different design..

Update: I've checked heat capacity of clay and ceramic. It is not a surprise that you successfully managed igneous rock output. Furthermore ceramic comes out only at 50% of clay mass..

[nakomaru]

6 hours ago, NikiHammerforce said:

But since igneous rock has much higher heat capacity than ore it comes out with too high temperature. Bigger heat exchanger is needed. Or different design..

The better exchanger we build, the less rock we have to worry about. But the differing HCs is still a problem if we want to minimize rock temperature. It would be nice to fine tune each element's packet size to correct for differences in HC and TC using a conveyor valve, but that results in bugged packets, sadly. The other option already mentioned, which doesn't bug, is a tricked pump with <10% throughput on the pipes, which will conduct very well and won't take up rail space. This has the bonus of being able to work with rust/iron in a single design.

You've already mentioned a bigger exchanger which is always viable, and probably the easiest solution. I've already done that before though, so I think I'm going to play around with this design keeping your goals in mind.

[ghkbrew]

1 hour ago, nakomaru said:

using a conveyor valve, but that results in bugged packets, sadly.

I'm pretty sure that was fixed, actually.

[NikiHammerforce]

1 hour ago, nakomaru said:

The better exchanger we build, the less rock we have to worry about.

Sure. There is much space for improvement. I tried just basic approach: to feed rock on the same rail with metal, which is very far from perfect. But I've just realized that even with just taking heat from my current magma pool with my current imperfect exchanger it will solidify only in a few thousand cycles. It is a big one, more than 2000 cells. And it will refine millions kg of copper (which I will never have). And even after it solidifies I can still use it for thousands cycles. So it seems impractical to make design more difficult. Just for the sake of curiosity and perfection may be.

Btw, could you please explain why is this happening?

No steam though. And water is around 75-80C. Can't imagine how is that.. Thanks!

[JRup]

1 hour ago, nakomaru said:

It would be nice to fine tune each element's packet size to correct for differences in HC and TC using a conveyor valve, but that results in bugged packets, sadly.

39 minutes ago, ghkbrew said:

I'm pretty sure that was fixed, actually.

 

It should have been fixed (I don't believe it is bugged anymore), but there is a situation that will arise from partially full packets that enter the conveyor meter (let's shorten to CM) or from certain combinations of divisions of the packets.

For the case of partially full packets (I.E.: non-20 kg packets that arrive at the CM) the CM will still ship/divide appropriately and retain any missing/surplus amount in memory. This causes fractional amounts to be stored and will affect all future shipments.

For the case of oddball amounts to be split off 20 kg shipments arriving at the CM it will do as requested and will store any fractional amounts as above.

Both types of the previously mentioned cases will have the risk of causing amounts lower than 1g traveling on the rails. This will gum up any kind of conveyor rail loop that depends on temperature exchange.

The way to prevent this kind of tragedy is to ensure the following:

• Baskets should always arrive full (20 kg shipments only) at the CM before splitting.
• For larger amounts, splitting should respect the factorization of 20 (1, 2, 5) so you can comfortably split the shipment into 1kg, 2kg, 4kg, 5kg baskets...
• For fractional amounts lower than 1 "unit" then we need to do some additional "mafs". I'd done some commenting on that a while ago and it still applies, it's inside the spoiler:

Spoiler

On 10/28/2021 at 10:34 PM, JRup said:

Hi!

I did want to do a quick write-up on the subject of shipping below 1 unit, but... oh well here goes.

You will eventually get amounts below 1 gram on account of rules for binary numbers and this being a simulation in a computer and such...

A couple of wikipedia entries on the subject:

• https://en.wikipedia.org/wiki/Binary_number#Fractions
• https://en.wikipedia.org/wiki/Repeating_decimal This talks about "repeating decimals", but it can be summed up as "repeating numbers after the 'dot'"
• A cutesy video that mostly illustrates this for base 10 numbers in the spoiler:

  Reveal hidden contents

 

If you really have to use less than 1 unit measurements and want to avoid mcg amounts clogging up your rails (especially when using the conveyor meter) I could recommend looking into fractional amounts that are a power of 2, such as 0.5 units, 0.25 units, 0.125 units. Here, in unicode character glory, one half: ½, one quarter: ¼, one eighth: ⅛.

(These would correspond to 500g, 250g & 125g respectively. Quick mafs!)

Edit: @LadenSwallow kindly pointed out a not so minor mistake with the amounts given. This has been fixed.

 

This is capped by the game. Do bear in mind the thermal conductivity rules documented by previous generations of players... (Also mentioned above. Not like it's in the in-game database...)

Again, I believe it to be a good decision for such minor material amounts. Just imagine the computational overhead that is saved that way. (My opinion, anyways...)

If in doubt chock it onto ONI physics.

If your loop is filled with those mini packages then there's work to be done and that can be accomplished with a conveyor chute and some automation. Additionally, it can be a good idea to set a conveyor shutoff right before the CM to stop the shipments safely and manually reset the inner counter of the CM to correct any lingering fraction in its memory...

I hope this covers most of the problems with splitting shipments in conveyors and provides some help.

[nakomaru]

Oh, please feel free to ignore me when I keep rambling about more ideas for improvements - it's for my own curiosity mostly. I've been working on an iron/rust capable variant for several hours now...

1 hour ago, ghkbrew said:

I'm pretty sure that was fixed, actually.

Awesome!

47 minutes ago, NikiHammerforce said:

No steam though. And water is around 75-80C. Can't imagine how is that.. Thanks!

If I understand right, it looks like you have a Notifier hooked up to an Atmo Sensor, which somehow went off despite it being vacuum? That's really strange... no ideas right now.

[NikiHammerforce]

4 minutes ago, nakomaru said:

If I understand right, it looks like you have a Notifier hooked up to an Atmo Sensor, which somehow went off despite it being vacuum? That's really strange... no ideas right now.

Notifier is there to know if brine was boiled. Strange thing is that lower brine sometimes (very rarely, only seen it twice in ~100 cycles) becomes water and salt, but no steam in chamber after that. And salt is not where brine was. Temperature is not even near to 102C. Some micro boil mechanic or something?

12 minutes ago, nakomaru said:

Oh, please feel free to ignore me when I keep rambling about more ideas for improvements - it's for my own curiosity mostly. I've been working on an iron/rust capable variant for several hours now...

I am only happy with new ideas. Your help is much appreciated. I made real improvement thanks to you. So feel free to suggest.
I believe rust melter is a good thing, considering that building do not preserve mass. May be I would try to design as well if I had tons of rust here. But I haven't. As for iron I think it is better processed through building as it is power positive process. But if dupe labor is avoidable, why not..

[nakomaru]

Partial boiling only works in increments of 5kg with a minimum above that, so it can't be that. I think partial boiling doesn't give off secondary elements like salt as well.

I suppose what happened is thermal clamping preventing the brine from properly cooling the equipment. It needs some minimum difference in temperature which is bigger with small amounts. Once it crossed that threshold, one moment of heat transfer was enough to boil it. Steam is much more conductive than liquid waters, so the below 100C tiles instantly cooled it again. Small (mcg) amounts of stream that escaped turned to vacuum. Well, that's my guess.

So keep your water colder (thermoplate or bridge seems nice) and you should be good.

[NikiHammerforce]

6 hours ago, nakomaru said:

Partial boiling only works in increments of 5kg with a minimum above that, so it can't be that. I think partial boiling doesn't give off secondary elements like salt as well.

I suppose what happened is thermal clamping preventing the brine from properly cooling the equipment. It needs some minimum difference in temperature which is bigger with small amounts. Once it crossed that threshold, one moment of heat transfer was enough to boil it. Steam is much more conductive than liquid waters, so the below 100C tiles instantly cooled it again. Small (mcg) amounts of stream that escaped turned to vacuum. Well, that's my guess.

So keep your water colder (thermoplate or bridge seems nice) and you should be good.

I doubt that it is due to thermal clamping. Same thing occured once again with thermal plate and water being ~30C. Salt is 291C though. It is very rare occasion. Hard to investigate. I left water there for now and will check if it will boil or go away with time.

[NikiHammerforce]

Considering heat exchanger design.. After running tests for 100+ cycles I realized my mistake. After coming near equilibrium things are not that fine anymore. Somehow I thought that 10 stages is enough. Actually it is not.

Calculation (which I believe is right) for copper:
(1083C - 27C)/(steps+1) = 96C
That means output is 96+27=123C with perfect heat exchange

To have 97C copper 14 steps needed. More is safer since heat exchange is not perfect (it is near perfect with good rail throttling and mono metal) and 27C ore is not always available. With 50C ore things go as bad as 20 steps..
Throttling just slows coming to equilibrium and increases heat exchange quality. But sooner or later all initial magic goes away..
Also mixing copper and gold makes heat exchange somewhat worse, so for mixed refinery more stages would be better as well.

[nikk]

@NikiHammerforce, can you post a video version too as this thing isn't trivial to build?

I would love to see dimensions and a blueprint for "Blueprints" mod too. Tag this thread with blueprint keyword then.

You can try https://blueprintnotincluded.com.