This could work

[masterada]

Now we are getting somewhere

Ye, i totally messed up that ABC model.
I finally understand what A_inner,external means to you. (For some strange reason i thought it was the area in the outer container outside of the inner container).

So the result is A_inner,external = A_inner,internal. Makes much more sense. Could that mean the area of the inner container can be almost as big as the outer?
Now onto the next problem at hand. Lets make a hole!
printersketch.jpg
I will pull up some math on it tomorrow or during the weekend, after 36hours of not sleeping im not really up to the task now.

Edit: might actually be able to switch back the above tank, so theres no need for that long pipe

Edit2:
Thats not enough:

h_sink = M_inner / (d*A_inner,external)
          = (M_inner,initial + V_inner * d) / (d*A_inner,external)
          = M_inner,initial/(d*A_inner,external) + h_inner*(A_innier,internal/A_inner,external)

Meaning that (A_innier,internal/A_inner,external) must be 1 for it to work.
So lets switch back to the tank above:

printersketchv2.jpg
And lets put less salt in the water in the outer container. (d_outer < d_inner)

h_sink = M_inner / (d_outer*A_inner,external)
          = (M_inner,initial + V_inner * d_inner) / (d_outer*A_inner,external)
          = M_inner,initial/(d_outer*A_inner,external) + h_inner*(A_innier,internal/A_inner,external)*(d_inner/d_outer)

If the inner container is a 700mm*700mm base column with 5mm wall thickness, then A_innier,internal/A_inner,external = 1,014, so its highly likely is possible to get the right densities. (Dont know the density of resin).
PS: im not fully statisfied with the design yet, especially the "some system to avoid waves from drops part". The long tubes are too vulnerable.