Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Sarmatian123]

Unfortunately the double slit method on wiki is cumbersome and dangerous enough to kill dwarves with annoying frequency.

I use a more robust and obviously larger scale (3/6 pumps) double slit method to avoid drowning casualties among dwarves, get 3x3 up/down stairs passage right away and skip playing with "." to peek into bellow layer. As bonus I get 4 wells with fresh water, 2 on each side of stairs, right in the middle of my fortress.

[PatrikLundell]

The only danger I've had with the double-slit method (once I learned to perform it properly) is when the pumper goes on a break, but the worker below usually manages to get up before drowning. I'd expect there to be less than one death per 10 fortresses.

It can also be mentioned that I don't use the single step peeking method to determine whether the layer below is an aquifer or not, but I rather check the material to see if it's an aquifer bearing one or not (the wiki is helpful if you're uncertain).

[mikekchar]

Not that Loci and PatrikLundell need backing up, but I've also successfully used the double slit method.  My guess (if Notere is not wandering out of scope from the instructions) is that they are actually at the last level and didn't check/realise it.  The order of the blocks you remove is important on the last level.  For the other levels, the order is not so important and it's actually trivial once you realise how it works (which, let me be clear, is *not* trivial ;-) ).

I was going to write a detailed description, but I discovered that the math doesn't work.  Maybe someone can correct it for me:

An aquifer block leaks water on each of it's faces.  Think of it like a cube.  On each face it leaks water.  However, if the water is under pressure, it instead *absorbs* water.  Water is under pressure if there is other water in the tile above it (simplistic explanation that will work for this discussion).  An aquifer face can produce an infinite amount of water and absorb and infinite amount of water, but it can only do it at a certain rate.  An aquifer face absorbs water at the same rate that it produces it (possibly incorrect, but I think it's true).  I think a pump can pump water slightly faster than an aquifer face produces water, but I'm not exactly sure what rate it is (and the wiki doesn't seem to say).  Easy enough to test, though.

So imagine digging down 1 tile into the aquifer.  That pit has 4 aquifer faces, one in each cardinal direction, leaking water into it.  At first, the water is not under pressure, so it fills up the pit.  As soon as the pit is about to overflow, the water comes under pressure and the faces *absorb* water.  So basically it instantly absorbs the water until there is no water above the pit.  This leaves it in a steady state with water in the pit and none flowing out of the pit.

You might be wondering about the face on the bottom of the pit (assuming it is also an aquifer block).  This block does *not* leak water upward and it does not *absorb* water downward, so you can effectively ignore it... I think.

If we add a single pump to pump out the water, we pump a certain amount out (which must be a bit more than 1 aquifer face worth of water -- more on that in a bit).  I think a single pump will not be able to remove the water out of this pump because there are 4 faces supplying water and only 1 pump removing water.  I'm not sure how many pumps you need to empty a single tile, but see below when we talk about double slit.

Now consider mining out 2 tiles next to each other in the aquifer.  How many faces will be leaking water?  You might think 8 because it is 2 x 4, but it's actually 6 -- count them.  We can set up 2 pumps and pump out enough water to empty this pit (so I guess each pump must pump at least 3 faces worth of water).  But what about the waste water?  If we dig a similar pit on the other side, then we will have 6 aquifer faces.  The pump deposits the water on top of that pit.  The pit is full of water, so that water becomes under pressure.  This allows the aquifer faces to absorb the water.

Note: It takes time to absorb the water.  Since you are pumping over top of the pit, if you don't have walls around the pit, the water *will* flow off into neighbouring tiles!  You need to contain the water, but the tiles below it *will* absorb all of the water.  Possibly this was one of the original problems.

Once we have the first slit empty of water, we can dig a single tile below.  How many aquifer faces are there (assuming the layer below is also aquifer)?  The answer is 10.  We have the original 6 and then we get 4 more by digging that single tile.  If we stop pumping, the hole below will fill up with water and then the hole above it will start filling up.  This puts the water below under pressure.  However, we have 6 tiles producing water and only 4 tiles absorbing water.  Because of this, we need to dig out both floor tiles before we stop the pumps.  Having done so, we could stop pumping and the pit would remain with only the lower level being under water (because 6 tiles are producing water and 6 are absorbing water).  This is the secret of the double slit method.

That's not so useful, of course.  So instead, we can dig out only *one* of the floors, which gives us 4 faces of draining power.  If we dig out one more tile on the top layer, how many faces are producing water? There are 3 tiles dug out, which gives us 8 faces producing water.  We have 6 faces worth of water being pumped out (I guess) and we have 4 faces of water being absorbed in the single tile pit on the second level.

By building a wall, you stop the water from flowing out of 3 of the faces.  You keep doing it like this, just making sure that you have more drainage power than water leaking in.

The problem comes when you are on the bottom layer because you can't drain water below.  That's why you have to get very tricky on the last layer.

I hope that explanation helps (and that someone can fill in the details with the actual pumping rates, or any other mistakes that I made).

[Sarmatian123]

The only danger I've had with the double-slit method (once I learned to perform it properly) is when the pumper goes on a break, but the worker below usually manages to get up before drowning. I'd expect there to be less than one death per 10 fortresses.

My first issue with wiki's double-slit method is the 2x2 up/down stairs design. This is why I formed my own method. Fit for 3x3 up/down stairs.

1 death per 10 fortresses? How deep were those aquifers? You ever landed on deep soil aquifer embark? There is quite few aquifer levels in sand, before you even reach the porous aquifer stone levels.

On aquifer embarks, first thing you set up, together with dormitory is hospital with beds and well. So with wiki's double-slit method, if your drowning dwarf first loses in the struggle consciousness, then who will be next to drown? His brave medic/miner rescue team mate. Suddenly there are 6 dwarven corpses in watery grave and accidentally you are also out of all your picks until merchant caravan arrives next year...  Yeah, 1 out of 10 fortress, yeah! 

My method needs few more dwarves, so this is not 1st year project I do. However hard scripted 2 waves bring enough labor for manning 3/6 pumps + some spares mechanic/pump operators/architects to fill eventual vacancies. It is a solid 2nd year project. 10 mechanics/architects/pump operators is enough labor for breaching aquifer and later for taking care of 100+ trap cages.

With my double-slit method, I can excavate 2 aquifer wall chunks or even 4 aquifer wall chunks (depends how far from drainage is that aquifer wall) at the same time, which is 100%-300% faster, then the double-slit method on wiki. I can also do it on 2 sides simultaneously with 6 pumps too. This means 200%-600% faster then wiki's double-slit method.

The wiki's double-slit method is a deadly trap, VERY SLOW and slams down your fortress design with odd 2x2 up/down stairs. You won't be able to dig massive mass pit in no time in aquifer with wiki's method either, because it is inefficient.

[PatrikLundell]

@mikechar: Unfortunately, I believe your logic doesn't quite match up, even if the overall description is sound.
- A single pump is able to drain 6 faces of water (two tiles side by side) so a dorf can work down there.
- As noted, a floor does neither absorb nor produce water, even if the level below is an aquifer. However, a down stair allows the water to pour out downwards, allowing you to turn off the pump (assuming the level below is an aquifer, of course). It's better to dig downstairs on both tiles (and in practice, it's an up/down stair to the level below, as it allows access to that level later), however, as the water drains off immediately rather than flow over the floor tile to reach the down stair.
- I suspect the absorption rate of aquifers to be unlimited in themselves, but limited in practice by the rate at which water can flow down from above. Any water that's flowing to the side on the level above will have to flow back to get down, and that won't happen if there's water on top at the flow check moment.

@123: If a dorf DOES drown occasionally, no rescuers will kill themselves, because at the time they'd try to collect the corpse there's more than 3/7 water covering it (or the dorf wouldn't have died in the first place), so they won't be tasked to go there. A 1:10 ratio is probably a significant exaggeration as well, as deaths are fairly rare (2?, 3?) since 0.40.X. The pick (and corpse) is easily collected once the pump is dwarfed again. With an extra effort you can use powered pumps, and thus no drowning risk at all, but I haven't found the risk reduction to be worth the extra time and effort (with dead civs, where every dorf I'd ever have came from the embark team, 2 migrant waves, and locally produced offspring).

You may not like the double slit method because you want a pit and a nicer staircase, and yes, the double slit method is not intended to be used to build pits, but to get through the aquifer as quickly as possible with the minimum use of dorf power. You're free to use your own method (but don't call it "double-slit", as that would cause confusion, since that name is established), while I (and probably quite a few other old buggers) prefer to be done within a year and a season (regardless of aquifer depth: I'm not sure what my maximum is, but it's well over 10 levels. However, once you get down to the stone layers you can switch to stone smoothing, which is a lot faster than digging and building walls).
I've used alternative designs to build nicer staircases in the past, but eventually decided the better looks weren't worth the extra effort and time required. However, looks aren't my primary concern, and others may have different priorities.

[Sarmatian123]

True. On my old cpu, DF was running constantly at 10 fps. Now, around 40-50 fps. So, I was rather devoted passionately to open wall-less architecture and broad passages.

[Loci]

The wiki's double-slit method is a deadly trap, VERY SLOW...

If your dwarves are drowning then you're doing it wrong. Using the double-slit method, I punched through a 2-layer test aquifer in less than 4 months with zero casualties and zero "mild endangerments".

1 death per 10 fortresses? How deep were those aquifers? You ever landed on deep soil aquifer embark? There is quite few aquifer levels in sand, before you even reach the porous aquifer stone levels.

This demonstrates a lack of understanding of the double-slit method. The "upper layers" are essentially risk-free because they are drained directly into lower layers. After digging a drain (which is the first step), pumping is no longer necessary and that level cannot *ever* flood.

An aquifer face absorbs water at the same rate that it produces it (possibly incorrect, but I think it's true).  I think a pump can pump water slightly faster than an aquifer face produces water, but I'm not exactly sure what rate it is (and the wiki doesn't seem to say). 

Empirically, aquifer-adjacent tiles gain 1/7 water every 14 ticks on average--this rate does not appear to change significantly based on the number of adjacent aquifer tiles. Meanwhile, a pump can move 7/7 water per tick. So, theoretically, a single pump could keep pace with 98 aquifer-adjacent tiles (though the water level in many of those tiles would remain too high for any practical use). However, a single aquifer tile can absorb *any* amount of pressurized water each tick. In a test, I drained 50,000+ tiles of ocean water into a single aquifer tile in 32 ticks (the time it took for the water to "fall" and check for a pressure-teleportation destination).
 

[Sanctume]

Double-slit method is for those challenging embarks with minimal resources. 

I prefer the pump-stack method since it is driven by water reactor to handle all the pump operation. 

But you can also just link a water reactor to the double-slit method.  The caviat is, you need stones or metals to make mechanisms for gears.

[Notere]

So, in the time I've let this thread sit, it's gotten far more educational. This is now a post-mortem on the attempts.

I should preface this, at this point, with the backstory. I spent a considerable amount of dwarfpower doing exploratory dig checks. You have no conception of how much horizontal space I dug just to get around this goddamned aquifer, and it didn't work out. I resorted to the double slit method for one simple reason: I didn't have a lot of usable space with a significant amount of the map already consumed by caverns of exploratory down stair digs.

The portion that confused me the most was the portion where it simply trails off with "and so repeat" in terms of digging out the inner blocked with UD stairs because I'd been to that point before and the upwash from the pumps was causing too many problems to even be reasonable. I lost one dwarf in a slit attempt prior, and two in the second.

The second slit was actually a "failed" attempt from before when I didn't understand transitioning from down stairs to UD stairs and back again via construction. No tutorial I read even mentioned that, and as far as I knew, once a tile was consumed by any given z-transition item, it was usually stuck that way. Eh. Basics biting me in the ass, but I go on.

Transitioning the center four blocks to UD stairs was biting me in the ass, so I decided that an attempt would be made to wall off the right section as standard, take the center portions and build wall there too, just to see if the slit could handle the flow. Worst case scenario, I thought, was that I could change it out again after the fact. This was wrong.

While the right slit would get dry as a bone, once I tried to do anything to pump water into it, it didn't have anything like the absorbing power needed to handle the production from the left slit. Even with the UD stairs carved in as a drain, the whole thing failed miserably. So, I decided to cut down the wall on the right and carve the UD stairs proper. This didn't help. Both sides ended up making the pump operator's life utter misery as he sat in a tidepool no matter what. I even tried a simultaneous transfer from an upper level with another pump into a trench (which I then pumped out onto the ground to keep it from potentially causing the same problem or spilling back into the slit stairs) as an attempt to salvage that dig. Nah. No dice.

Honestly, the whole thing has been very educational on the function and use of pumps against aquifer barriers as an exercise, but after getting that close to doing it right and still being baffled by what the water physics were doing to me, I said "to hell with it" and grabbed DFhack.

It wasn't deep, is all I can really say. 10 layers? Nah. Four in its deepest spots, maybe. But it was wide. There were overlapping aquifer tiles everywhere. If I was sure I could have done it without drowning a platoon, I might have been able to force a mining dive past the aquifer tiles and just carve them all out before letting the water evaporate. If I knew where that was possible. But I didn't. That and it's always a good idea to practice advanced mechanics, just in case. The details and specifics of it just left me scratching my head in certain parts.

Call me a poor sport, but I was launching this fort to check out other new mechanics, and I was tired of hearing the word "aquifer" and being warned about damp tiles every two seconds. So I nuked it.

[Notere]

Oh, and while I'm here, this was the state of my last attempt before I flipped the table, so to speak. https://imgur.com/a/EZPib55

[anewaname]

In that image, you see how there is a path for the water to travel through the diagonal where the two up/down stairs are? If the water in those center tiles could not drain into the z-level below (because the lower z-level either did not have up/down stairs or or did not have aquifer tiles), then your pump operator would stand on the z-level above, moving water from one side to the other, and then the water would push through that diagonal, instead of draining down.

But, I get it that you might not want to think about it further. I avoided aquifers for a few embarks after having similar issues.

[Sarmatian123]

Blame it all on my poor understanding of the double-slit wiki's method. All Dwarves in my attempt, who died got washed to this 1 level down drain ditch. Water falling from pump one level down, goes not only right down into drain hole, but splash around on tiles too. So, it is dwarf-life crucial to clear the hole from stones and loose items, before punching drain hole and defining zone on some other place as preferred water source. However, maybe the method itself is little too poorly explained on DF wikipedia to all those people with no knowledge of DF fluids mechanics. I think this double-slit method on DF wikipedia leads to exaggerated conception that aquifer embark is way too hard, way too dangerous and too cumbersome. While, when you know your basic DF fluids, punching 3x3 up/down stairs is simple, easy, safe and relatively fast. No matter how many levels deep is aquifer. Specially if you employ more robust methods, like the one one used by me. Also, it is easy, though little time consuming to dig out entire 11x11 hole for mass pitting in aquifer. I once cleared entire 25x25 space for entire level of my fortress, when I was using less spacious per level and more fps intensive design for my fortress. Little messy due the mud, which needs clearing by smoothing or constructing floor. You can safely dig out entire aquifer levels from the map even, while using magma on edges of map, when you terraform too. Though terraforming takes years even without aquifer on even more or less leveled already terrain.

[mikekchar]

@Notere It's hard to understand what you were doing.  Without understanding the mechanics of water dynamics in DF, it's pretty hard to find a solution.  I'll try to explain by pointing out some bits that look like they may have tripped you up.  When you start, here is what the last non-aquifer layer should look like:

Spoiler (click to show/hide)

Here's what the first aquifer layer should look like

Spoiler (click to show/hide)

You build floors over the lower 2 stair tiles in the non-aquifer layer  so you can pump from left to right.  But you have to remove the down ramp in the lower left tile (only!) by channeling down.  This is because you can't use the pump over the down ramp.  Don't channel out the upper left one because otherwise your dwarf won't be able to get back out.  Note: you are only using 1 pump -- from the lower left to the lower right!

Once most of the water is gone, send a dwarf down to create up/down stairs in the two *aquifer* layer tiles on the *left hand side only* (Note that in the picture they use a slightly different hashing for the non-aquifer layer and the first aquifer layer).

When you are done, your aquifer layer will look like this:

Spoiler (click to show/hide)

Err... well, pretend that you mined those out...  Important:  When mining out the first one of these stair cases, step using the "." until the layer *below* is revealed.  Then use "k" to check to see if the layer is damp or not.  If it is not damp, then there is no aquifer below that level and you have to skip down to the last section.  If it *is* damp, then you have at least 2 aquifer layers and you can drain into the bottom one.

Once you've mined out those two stairs, disassemble the pump and reassemble it the other way around -- on the bottom only!  Again, we are using only 1 pump!  Do not put it on the top!  Follow the exact same instructions as before, replacing "left" with "right".  Keep in mind that we are working only on the *bottom* right here.  Do *not* mine out the down ramp in the upper right corner!

At this point you can dismantle the pump.  You do not need it because the aquifer is draining into the level below.

Your non-aquifer layer should look like this (well, instead of a pump there, you'll have floors -- you can disassemble the floors now too and be left with your original staircases):

Spoiler (click to show/hide)

First aquifer layer should look like this:

Spoiler (click to show/hide)

The aquifer layer below that should look like this:

Spoiler (click to show/hide)

In other words, you have *2 aquifer layers* which have up/down stair in them and you are going to work on the first of those.

Next dig out sections *in the first aquifer layer* and add walls until you get this:

Spoiler (click to show/hide)

I recommend doing it one at a time (dig a space, add the wall), even though the wiki says to do it 4 at a time.  Just until you get used to the process.  It doesn't matter what order you do it in.

Next dig an up/down staircase in the centre and then an empty space above/below it and add a wall:

Spoiler (click to show/hide)

Do top/right, bottom/right, bottom/left, top/left (I don't think the order matters) until you get:

Spoiler (click to show/hide)

Remember, there is no pumping involved!  It is just draining into the aquifer below.

After that is done, you are essentially back to the starting place.  Your first aquifer layer is the same as what you had in your last non-aquifer layer when you started except the downramps are up/down staircases (it really doesn't matter).  You can go down as many layers as you want until you get to the bottom.

It gets tricky once you get to the point where there is one more aquifer layer to wall up, but there is no aquifer layer below it.  Then you *must* follow the instructions in the wiki *exactly* from http://dwarffortresswiki.org/index.php/DF2014:Double-slit_method#Walling_off_the_last_aquifer_level.

Hope that was helpful.  I know I'm just restating the wiki, but I hope that stressing different things will get you into an "Aha!" moment. The aquifer layers before the last one are really very easy.  The last one is very tricky.