North-east of z0 screw pump is constructed wall, right?
The path outside the trench is 1-wide with trench being 3-wide Ramp & Ramp & Updown+Hatch, right?
As I understand,the floodgates are 1 z higher than trench - resulting in trench getting at most 5 units of water, not 7, as 1/7 would remain on floodgate and between-floodgate squares. However, past the first one toggle, there will be 2/7 water remaining down below.
Now, testing it out in 43.03, with giving one initial flow before installing hatch, for the 3/2/2 setup as described...And it works:
z99:
z100:
z101:
With windmill powering the two pumps.
(Blood provided thanks to minecarts)
What's interesting here is that just regular 3-wide trench with 2/2/3 water will have blood remain on the tile it was (when I've stationed dwarf inside trench and then pow'd them with a minecart).
As such, I suspect this is caused by either aquifer and/or flow remaining down after winding down to 1/7s and refilling. For testing those in that order, I will use the following setup:
(Stair squares are restricted, water was added via liquids plugin, minecart moved with autodump as needed)
A: Dwarf passed through center tile, and stored Fish Cleaner in the hospital
A and B both: Dwarf carried fisherdwarf through a trench, and blood got washed into the wall.
B: Dwarf passed through westmost tile to carry miner to hospital, and..remained still.
End result of trenches:
Additional hypothesis: It might be that while flow and drain is both required, portable hole would work instead of aquifer:
Spattering deposited in...And remained still.
Well, until I toggled hatch. Then it got washed nicely into wall.
Retesting, with re-injuring dwarf (running out of them)...I think it didn't work.
(The water buffalo blood in center tile remained there.)
...Though side note: seems rain does wash it, unreliable as that may be. Reinjured buffalo and covered it up to double-check.