I now understand better the 'impeller' bit - it's not just the face of the 'cups' - it's the radial clearances there, and with stacked-impellers as fit the HP of the motor used; and wear in a 'diffuser' above - whose function is apparently, to properly load the (I'd call it a "ducted fan") process.

So the leakage can be internal wear there, or that + combo of (as Box suggested) some leak in the exit pipe. No practical way to tell, apparently.

I'll be running mine until the %on-time curve crosses the [this is now a stupid waste of electrical power!] line. :-/

Next I have to find out what sort of graft has 'forced' via 'code': these guys to use plastic-pipe costing $3/FOOT (for an under-30# weight as tensile force ~= trivial.) Ya can't argue physics, probabilities VS #$^*% biznes$ gerrymandering... but I may try for an arabesque anyway. The curse of experience with 'strength of materials'.

(And I don't see any layered-mesh composition, either.)

Oh well.. 30 years ain't bad.