Is this gas vent tamer design broken?

[natek53]

I was trying to recreate this bypass pump + Escher waterfall from the Compendium of Amazing Designs that uses natural tiles to reduce the size. I wonder if it relies on gas displacement behavior that has changed since then, if I'm doing something wrong, or if it was always an unstable design. There are no overlays in the original, so I'm just guessing that neither of the two gases at the bottom should be hydrogen (the pump breaks almost instantly if it is).

Here are the overlays for the version I made that kind of works. N.B. the diagonal gas placement of hydrogen/chlorine is usually avoided for gas pumps, so I imagine that's why this one tends to fail.

Spoiler

Gas overlay: hydrogen and chlorine are diagonal. Petroleum lies on the geyser's TOI. When it works, the hydrogen is displaced upwards and collapses the liquid lock, allowing the hydrogen to move diagonally into the space above with the rest of the hydrogen.

Liquid overlay showing petroleum and oil.

This design seems to last for a random amount of time before a hydrogen packet eventually goes the wrong direction, leading to the broken state shown below.

Spoiler

Chlorine moves instead of hydrogen, gets pushed up by collapsing liquid lock, then blocks hydrogen from moving, causing the vent to eventually overpressurize.

Lack of chlorine breaks the liquid pump, pulling petroleum down.

If oxygen is used in place of chlorine, the pump breaks almost instantly, even if it's just a few grams. Maybe that was just a coincidence, or maybe the relative densities matter a lot for pump stability, even though both should be more dense than petroleum.

After finding @Zarquan's similar design and trying it out myself, I noticed that it's much more stable. The design at the top of my post lasts 1-2 cycles at best, and breaks immediately if you save while the geyser is erupting and reload. Whereas I ran the one in this next image for several cycles and haven't seen it break once. Notably, there's no liquid in the tile of interest and no need for multiple gas elements, so there's basically nothing to break. With the valve set to 300 g, the average power should be 3.6 W, which is basically nothing.

Nevertheless, I wanted to see if I could convert this into an unpowered version without making it too much bigger. The one below seems to work well.

The bounding box is the same size, but it still feels bulkier. Overlays below.

Spoiler

No valid tiles for chlorine & co2 to swap with. The liquid lock toggles quickly enough that I haven't seen any packets get blocked.

It's also possible to make the waterfall go under the neutroneum and fall from the right, but I think that takes up more space.

IMO it's a lot uglier than the first one since it doesn't fit neatly into a rectangle, but it doesn't matter if it doesn't work.

I'd be interested if anyone knows of a smaller version.

[myxal]

I've also had the small bubbling design fail pretty much the same way, as I was trying to use it on a hot steam vent (in sandbox, failed very quickly, within 2 cycles IIRC). A reddit user I was discussing this with at the time insisted that it still works, but I didn't pursue the issue further, because:

• For steam vents, other infrastructure is needed to tame the geyser (turbines/pumps), and the bubbling extractor is too bulky compared to more conventional corner-bypass, or boiling water displacement extractor. By nature of the tamer, piped water is easy to come by. This doesn't apply to other gas vents
• For other vents, I use a few designs that use door-corner compression. The uncooled version can handle almost any of the non-steam vents and is not much bigger than the compendium's original bubbling design.

Spoiler

 

[Zarquan]

It looks fine to me, but you might have some weirdness with having the petroleum cell above the cell where the hydrogen is being bypassed to..  If you dropped the petroleum from 1 higher, so that it formed droplets, it would mitigate this potential problem.

I bet your original died because the petroleum on the hydrogren vent spawn tile bypassed with your escher waterfall gas.  Moving your original escher waterfall one square to the right should fix it, be smaller, and have a nice rectangular shape.  You will need one more natural tile to make it work.

[natek53]

6 hours ago, Zarquan said:

I bet your original died because the petroleum on the hydrogren vent spawn tile bypassed with your escher waterfall gas.  Moving your original escher waterfall one square to the right should fix it, be smaller, and have a nice rectangular shape.  You will need one more natural tile to make it work.

So, like this?

That definitely has a cleaner profile than the other working design I gave, and is smaller (7x7 vs. 8x9 w/ empty spaces in awkward spots). Obviously it should be stable since there's no opportunity for gas/liquid swapping. I ran it for ~3 cycles just in case and measured the output for the last cycle. At the geyser's eruption rate (293.2 g/s for 316 s) I should've measured 18.5 kg in each of the 5 tiles above (92.6 kg total), and I saw 18.4 kg. So there is some mass deletion, but it's < 1%. During eruption, the swap sometimes brings two petroleum tiles down instead of 1, which I assume kills the gas instead of swapping with it. Maybe I need less mass at the top, but I'll still say it's good as it is.

Overlays:

Spoiler

I really appreciate the input!

 

14 hours ago, myxal said:

For other vents, I use a few designs that use door-corner compression. The uncooled version can handle almost any of the non-steam vents and is not much bigger than the compendium's original bubbling design.

I'm looking forward to trying these out. The blueprints are very detailed.

[6Havok9]

1 hour ago, natek53 said:

So, like this?

I'd make it one tile taller, just for comfort. I don't like when the flowing liquid sits directly above the heavy liquid tile, and I bet that it's the cause of mass deletion you are reporting. As Zarquan said, droplets are safer.

[Zarquan]

18 hours ago, natek53 said:

So, like this?

That looks good.  But you might want to measure how much hydrogen you get in an eruption, which should be pretty much exactly the eruption amount times the time.  In my design, the bypass point is on top of the critical tile where the gas spawns, removing the gas in the first tick it exists, then gets the gas out of the way in the next tick, leaving it wide open for 2 or 3 ticks for the next instance of eruption. 

I would not be surprised if you were getting all the gas, but I would also not be surprised if it was getting stifled every once and a while.

[natek53]

3 hours ago, Zarquan said:

you might want to measure how much hydrogen you get in an eruption, which should be pretty much exactly the eruption amount times the time.

Yep, I did that:

22 hours ago, natek53 said:

At the geyser's eruption rate (293.2 g/s for 316 s) I should've measured 18.5 kg in each of the 5 tiles above (92.6 kg total), and I saw 18.4 kg. So there is some mass deletion, but it's < 1%. During eruption, the swap sometimes brings two petroleum tiles down instead of 1, which I assume kills the gas instead of swapping with it.

The problem I ended up with was getting the drips to happen too often, which could mean I used too much liquid, but I am also interested in trying out this:

3 hours ago, Zarquan said:

In my design, the bypass point is on top of the critical tile where the gas spawns, removing the gas in the first tick it exists

because that sounds more consistent.

It'll be some time before I can try that and compare with the droplets version you & @6Havok9 mentioned.