Mark Borgerson wrote:

SNIP

With the proper placement of the check valves, I think you could
start with the initial boiling happening in the freshwater side---
after all it is going to boil at a lower temperature.

Well, except that 1) unless you fill the entire apparatus with fresh
water (ignoring for a moment how much fresh water that might take, and
how much system capacity is lost in re-distilling the fresh water), you
haven't eliminated the carryover contamination issue, since you still
have a contiguous water stream, and 2) the freshwater side of the system
is configured for *cooling* and the seawater for heating (passive system
remember), so boiling will always be initiated on the seawater side.

The procedure might look like this:

1 Pump both fresh and salt water to near the top.
2. Shut offf the salt water side pump, but keep the
tube closed at the bottom.
3. Pump a bit more fresh water into the tube---where
it overflows to the sal****er side, displacing
the rest of the air out the check valve.

And where is the barrier layer that keeps salt from moving into the
freshwater?

You now have no air in the tube and a small layer of
fresh water on top of the salt water.

But it won't stay that way. As soon as you begin to heat the seawater,
you'll almost certainly have seawater rising into the fresh (do to the
density change with heating) before you get boiling going on.

4 Release the pressure at the bottom, and the fresh
water at the top will boil and create your head space
with little or no contamination of the freshwater
side.

Yeah, but "little" is not the goal. And you'd have to quantify that
"little" empirically, since there are many factors that contribute to
the process.

5 Apply your heat differential and remove distilled
fresh water as it overflows the reservoir at the
bottom.

This should work until the dissolved gas problem lengthens
the vapor path to the point where you have to start over
at step 1.

Again, the design complexity involved in being able to heat the fresh
side to initiate the boiling there *first*, and then switching to
cooling mode when there is sufficient column separation puts paid to any
thoughts of this being a simple system. And then, you have a very
complex, and Horribly inefficient system.

There are lots of ways that you could make the system work, but why?
The *only* feature this concept has going for it to start with is
simplicity, and basically a passive (save for some human work input) system.

Keith Hughes