Failed experiment: Molten Glass super-cooler

[Fradow]

Thanks to @JRup Chilly Glass Forge I got interested in the Molten Glass => Glass state change, and its astounding 4.2x SHC multiplier. That's huge. By comparison, the well-known Crying Crab uses Ethanol, which only has a 1.15x SHC multiplier.

Spoiler alert: as mentioned in the title, this is a failed experiment, and I cannot see any circumstance where this could be an optimal choice. That said, I found it interesting enough to post about it, and let it be a cautionary tale for anyone trying to find a use out of it.

Now a few machines already exist that exploit SHC differences between different states: Salt boilers, Regolith melters, the Crying Crab (ethanol), or my own Petroleum Well. While the crying crab amplifies cooling, it does so around 80°C. All the others are about amplifying heating.

The core of the concept is to use Molten Glass (out of a Glass Forge, heated Sand or re-heated Glass), use a Valve to get 10% packets, then supercool it to as cool as possible (near -270°C), and let it state change to Glass.
That Glass can now be used to cool whatever you need and will cool 4.2x as much as what you used to cool it.

The main application is LOX/LH where you need a lot of cooling, and that inherently requires quite a bit of infrastructure.
Unfortunately, it doesn’t make any sense to do that with Super Coolant ATs, since they are almost free with STs return.
There is a potential use though: a LOX/LH build without space materials, using the 10% trick to get to Liquid Hydrogen temperature, which inherently is going to use a lot of power and not have lots of return from STs.

That’s what my prototype focused on: having a low-power pre-space LOX/LH build, in the context of a Tear speedrun (on hiatus), to avoid having to gather Supercoolant, a lengthy step I wanted to bypass.

Here is the prototype:

I didn’t use all of JRup tricks for the Glass Forge because I use the Stock Bug Fix mod, which makes the Glass Forge internal storage insulated. That’s a mostly irrelevant detail.

1. Molten Glass is produced at 1750°C in the Glass Forge
2. It immediately get split into 2 1kg/s pipes
3. It gets passively cooled down to ~125°C in the Steam Chamber
4. It gets actively cooled down to ~-250°C in a series of 4 chamber of decreasing efficiency (PWater AT => Ethanol AT => Hydrogen Reg => 10% PWater AT)
5. It gets vented and immediately transform into Glass
6. Glass gets counterflowed against Oxygen/Hydrogen

In theory, this brings the power cost from 2600W to 620W (STs return excluded), or a saving of 1980W for a 0.5kg/s throughput of both Oxygen and Hydrogen (each 10% Molten Glass pipe only packs enough DTU for ~0.25kg/s throughput of both).

Now this seems great in theory … Until you have a look at the power consumption to get a steady 2kg/s of Molten Glass.

The easiest option is of course the Glass Forge, and a really short test with a 0 Machinery Dupe shows that it’s able to fill about half a 1kg/s pipe with constant use and lighting. Not great. What about a 9 Machinery Dupe? Almost one pipe.
That means, with a back-of-the-napkin calculation, you need 2 full-time +10 Machinery Dupes operating 2 Glass Forge to power the build, or roughly 2400W. To save 1980W. There you have it, the main issue: it doesn’t actually save power. Plus you have to feed it tons of Sand.
What about +20 Machinery Dupes? It should save an underwhelming 780W if it’s able to fill both pipes alone. But considering how much training is required (unless you game the pod to get a perfect dupe…), you are better off going for Super Coolant and saving on construction time and headaches.

What about using re-heated Glass? It would save the constant Dupe time and glass forge power use, but it requires having a volcano near the surface and even more complicated building to pull off, and cross the threshold into “too much hassle, too little gain”, even for me. It’s also going to use a humongous amount of heat, about 2MDTU/s, while an average volcano can only provide about 300kDTU/s for that temperature. That’s not even close, and you cannot minimize the heat use much by counter flowing high-temperature Molten Glass either, because of the aforementioned SHC differential we are after that works against you in that way. Even a perfect counter flow would only save you a quarter of the heat requirement, and 1.5MDTU/s is still way too much.

The last option, which I did not pursue, is to use the Liquid Duplication exploit to duplicate 1750°C Molten Glass in a nearly power-free way (it just needs to be pumped with a tricked pump). That method also cross the threshold into “too much hassle”, as you need to preheat the area to 1750°C to have another liquid at the temperature (a non-trivial operation) then design a tricked pump for a low-viscosity liquid (which probably involves double waterfall and an escher waterfall to pump at max efficiency) and integrate both the duplicator and the pump together. And it’s also susceptible to be fixed any day, though the bug report leads to belief efforts to get it fixed got dropped.

There you have it, a funny but ultimately failed experiment. And I didn’t even get into the big issue of the 2 right cooling chambers priming, which are an entire other issue making the build non-viable (it takes ages, especially with the huge thermal mass of tiles, to get down to temperature). Or the solidifying liquids to debris temperature bug, which is also another issue to solve when initializing the build.

Is there anything that I overlooked that would make the build viable? Let me know! Especially if you find a cheap way to get Molten Glass.

[JRup]

Ah, wonderful!

This is a great ode to the terror of molten glass's siren song.

I had entertained the idea of using solidified glass as a heat source at some point; but then the reality of having a dupe task for that hit. It was somewhat of a hassle.

Then the bright idea of adapting a regolith melter to also process sand bubbled up... I thought "glass floats on magma anyway..." Another siren was singing.

Glass cannot be turned back into sand in the game so those avenues of thought were, eh, crushed.

The Chilly Glass Forge was then built as a way to appease the sirens and set sail to other mini projects before they came up with another song. I will treasure this song nonetheless.

[babba]

I like the topic title , more people should be encouraged to post "fails". Its not all glory in the game, in life and anyways !

Life is not like LinkedIn and Instagram, with all the fake stated achievements and CVs. According to LinkedIn I have 30 former work colleges which praise that they invented the atomic bomb and such, but I know they all only made coffee for the staff. Someday we all will be dust

Lets print more dupes, that's (of) more substance !

[ghkbrew]

3 hours ago, JRup said:

Glass cannot be turned back into sand in the game

Sure it can. "All" you have to do is vaporize it to rock gas then solidify it to igneous rock, then crush it with a rock crusher.

[JRup]

2 hours ago, babba said:

According to LinkedIn I have 30 former work colleges which praise that they invented the atomic bomb and such, but I know they all only made coffee for the staff.

Which begs the question: Have you tasted their coffee?

2 hours ago, ghkbrew said:

Sure it can. "All" you have to do is vaporize it to rock gas then solidify it to igneous rock, then crush it with a rock crusher.

Ok, I give... Let's amend it to "having a recipe for glass to sand in the rock crusher".

[babba]

@JRup Nothing better than a nice coffee with a mountain of cream on top and a nice fat marshmallow as iceberg swimming in the middle Mhhhhhh, need to go to the kitchen and investigate things...Jummy in the tummy !

[Fradow]

9 hours ago, JRup said:

Ah, wonderful!

This is a great ode to the terror of molten glass's siren song.

I had entertained the idea of using solidified glass as a heat source at some point; but then the reality of having a dupe task for that hit. It was somewhat of a hassle.

Then the bright idea of adapting a regolith melter to also process sand bubbled up... I thought "glass floats on magma anyway..." Another siren was singing.

Glass cannot be turned back into sand in the game so those avenues of thought were, eh, crushed.

The Chilly Glass Forge was then built as a way to appease the sirens and set sail to other mini projects before they came up with another song. I will treasure this song nonetheless.

I've found another siren song to my quest of pre-space LOX/LH build! Without Molten Glass unfortunately.

I'll resurrect @ghkbrew Debris Fridge™ and make it into a survival LOX/LH build.

I had totally forgotten about that exploit, and got reminded it was still a thing. I got mistaken reports that it was fixed a few months ago while I was away. I'm pleased to announce it's still very much alive and annoyed me with this build.

Let's see how it goes...

[wachunga]

Fun idea!

I was going to mention the debris bug but you are already there. Note that there is/was a mod to fix that bug.

 https://steamcommunity.com/sharedfiles/filedetails/?id=1891393570.

It seems like it was an easy fix from decompiling the mod, but who knows if that's still true for SO.

FYI, you can make the 10% loop robust by using a valve for your bypass instead of a bridge and sticking a small buffer on the AT automation. With a valve, your AT and bypass paths will be of the same length and the packets will march along nicely. The other issue being that when the AT is disabled a packet can get stuck inside and ruin your day when the AT is enabled. A 0.5 second buffer solves that for me, but that might be different for your particular framerate.

[Fradow]

7 hours ago, wachunga said:

Fun idea!

I was going to mention the debris bug but you are already there. Note that there is/was a mod to fix that bug.

 https://steamcommunity.com/sharedfiles/filedetails/?id=1891393570.

It seems like it was an easy fix from decompiling the mod, but who knows if that's still true for SO.

FYI, you can make the 10% loop robust by using a valve for your bypass instead of a bridge and sticking a small buffer on the AT automation. With a valve, your AT and bypass paths will be of the same length and the packets will march along nicely. The other issue being that when the AT is disabled a packet can get stuck inside and ruin your day when the AT is enabled. A 0.5 second buffer solves that for me, but that might be different for your particular framerate.

I'm definitely not going to use that mod when doing a full-exploit speedrun to the tear! Provided I actually manage to have a viable build that's not more complicated than the non-exploit variant. I started having a go at it, and it's not going super well.

I managed to get the initial super-cold debris to around -265°C: a bit of piping shenanigan, but otherwise easy, albeit slow.

I managed to find a nice way to increase this debris mass without loosing any heat: that was a bit more complicated, dripping water like in the exploit example loose a few degrees, and it adds up super fast, so that's no-go, you gotta have water already past freezing point with the 10% trick to work correctly. Using conveyor rails and solid tiles (gold metal tiles are my go-to because they have low SHC, relatively high TC, and it's easy to produce too), 2 pipes at 10% work well enough (and it's going to provide way more cooling than needed anyway).

The last step which is giving me trouble is the Hydrogen condensation, because of the low temperature differential. I'm pretty sure it's going to require a big-ass heat-exchanger, which is not great because having a low build time is one of the goals. At least we got the conveyor meter since Mergedown, which simplifies things. The Oxygen condensation will be an issue too, to avoid bursting pipes.

Anyway, I'll probably open a new thread once everything is figured out and the build optimized (unless it's failed, then it won't be optimized).

 

Thanks a lot for the valve trick, it's indeed working well, those pipes bursting were driving me mad. I tried the 0.5s buffer before, but it didn't work at all for me, and I didn't care to try since it's working so well since adding the valve (and I don't care that much if an odd packet solidify).

[GreezyHammer]

On 12/30/2021 at 7:56 PM, ghkbrew said:

Sure it can. "All" you have to do is vaporize it to rock gas then solidify it to igneous rock, then crush it with a rock crusher.

Enter glass boiler

 

 

[JRup]

3 hours ago, GreezyHammer said:

Enter glass boiler

Have you considered leaving a dab of liquid steel at these spots? It would help in minimizing heat leak (if any) through the insulated tile right next to them. (It would actually enable the use of an airflow tile in between liquids instead...)

[Fradow]

On 12/31/2021 at 8:03 AM, Fradow said:

I've found another siren song to my quest of pre-space LOX/LH build! Without Molten Glass unfortunately.

I'll resurrect @ghkbrew Debris Fridge™ and make it into a survival LOX/LH build.

I had totally forgotten about that exploit, and got reminded it was still a thing. I got mistaken reports that it was fixed a few months ago while I was away. I'm pleased to announce it's still very much alive and annoyed me with this build.

Let's see how it goes...

For anyone wondering, this was a rather uninteresting failure.

Producing very cold debris is quite easy, with a simple 10% water loop that you cool down near absolute zero and vent at around -270°C. Some of that is lost to the Liquid vent, but enough not to care (less than 1°C). If the Liquid Vent gets colder than 0°C, Ice tiles tend to form on it, though that's not an issue because you only need 1 debris, and can just rebuild the vent to reset the temperature.

Adding mass to the debris is easy enough: either you drip pre-cooled water with 10% pipes (which use power but won't ever take any heat from the debris) or you drip water that has to solidify in a single tick, like shown in ghkbrew debris fridge. It's a bit tricky, and you must avoid dripping more than that, which is quite a low threshold. A liquid blade should do wonder to smooth the flow from Ice melting at random points and avoid the larger drip amounts if ice melts too close to the dripping edge.

Then comes the hard-part: priming and having a working liquedizer based on Ice. ghkbrew and myself tried different somewhat techniques, both riddled with issues. At that point, I gave up because the build was already bulky, hard to prime, hard to construct, and adding even more complexity to fix those issues wasn't attractive. If anyone want to have a look, our exchange is on the unofficial Discord in #advanced-machine-design.

 

But ghkbrew suggested a new very simple, but awesome idea: since a Tear speedrun only requires 2800kg, why not get rid of all the infrastructure needed for a constant throughput, and instead vent 2800kg of the stuff in a small area, and slowly cool it down all at once. It greatly reduce the infrastructure needed, remove any priming required (especially since it has very little additional mass to cool down) and should keep the overall power cost rather low using a 4 stage cooler (pwater => ethanol => hydrogen => 10% pwater) for just those 2800kg.

Overall it seems like a winning speedrun strategy. A bit more boring than molten glass or debris exploit shenanigan, but crazy fast, since the actual time needed to liquefy everything can be the time needed to prime some builds (especially the bulky ones with lots of mass). ghkbrew calculated a 18 cycle period for 2800kg of LH, using 1 pwater AT, then 1 ethanol AT, then 5 hydrogen regs, then 1 10%pwater AT (half of the time being spent on the regulators), which is going to be a hard to beat number.