The leaking is fairly complex. With some difficulty we have established exactly which valves on the manifold are paired on circuits. Obviously the same order on out and return manifolds would have been much too easy! - so the photographs taken before burying were useful, but one can check by putting the ground loop circulating pump to 'continuous on' (menu 5.11). Then when the second valv of a pair is closed nothing happens, whereas when any other return valve is closed the pipework makes a perceptible sound as it relaxes.
But the actual behaviour is confusing - basically the system does not leak if the (5.11) pump is running all the time, presumably because there is some negative pressure on the return half of the loop. So one needs to go round shutting the loops one at a time in succession, for a week or so each to be sure (the fluid level varies a bit due to thermal expansion). This does look to work and we are moderately confident of which loop, and if the conjecture above is right whcih borehole, is involved.
