The Chiller That Lives in a Vacuum

The machine is pulling air and water in, not pushing refrigerant out. A low-pressure R-123 chiller runs below atmospheric pressure on the low side (about 26 in. Hg vacuum at 40°F saturation, roughly 1.9 psia). Any leak below atmospheric is an ingress path: air and moisture come in, refrigerant does not go out. That single fact explains every symptom:

• Non-condensables collect in the condenser, raising the approach temperature from 2°F to 9°F.
• The purge unit runs almost continuously because it is the only thing removing the air; its climbing counter is the diagnostic signal, not the fault.
• The machine cannot hold vacuum when off - drifting from 26 to 18 in. Hg vacuum overnight is a large, fast in-leak.

Where to look. The low-side boundary: evaporator tube sheets and water-box gaskets, the rupture disc and its flange, the charging valve, the purge unit's own seals and piping, and the compressor shaft seal. The high-side (condenser) components at 100+ psig are not the likely source of an ingress leak - pressurizing the whole machine to chase a high-side leak would miss the point.

Detection method. Use an electronic detector suitable for R-123, or pressure the isolated section to a few psig with dry nitrogen and use soap solution. R-123 is a B1 refrigerant (higher toxicity, 50 ppm acceptable exposure limit); ventilate the machine room and monitor while working.

It is a defensible idea executed dangerously. A nitrogen pressure test is the right tool for a machine that normally sits in a vacuum, but three specifics make this plan unsafe:

1. The margin is 3 psi. The rupture disc is stamped 15 psig. A regulator that creeps, a bottle warming in the sun, or a slug of liquid in a section can lift the disc and dump 900 lb of R-123 into a hospital machine room. Test at the lowest pressure that will reveal the leak - a few psig - isolate sections with valves where possible, use a nitrogen regulator with a relief device downstream, and watch the gauge continuously.
2. Nitrogen only. Never use oxygen or compressed air. Oil plus oxygen under pressure is a fire/explosion hazard, and shop air supplies exactly the moisture the machine is already fighting.
3. Never use the chiller's own compressor to build pressure. It is not a test compressor, and running it against a pressurized, non-condensable-filled circuit risks damage.

Also note the sequence problem: the machine still contains R-123. You cannot pressure-test a charged machine meaningfully, and you must not open it to atmosphere with refrigerant in it. Recover first to 25 mm Hg absolute with low-pressure (Type III) equipment, then leak-test, then repair.

Liquid into the evaporator: no. The evaporator is full of water at about 40°F. Dumping liquid R-123 into it flashes and can chill the water below freezing; a split tube then puts water into the refrigerant circuit and turns gasket work into a full rebuild. Charge R-123 as a vapor into the evaporator or suction side, with the chilled-water pump running, weighing the charge or following the manufacturer's charging chart. Verify the result with the machine's normal indicators, not by watching the low-side gauge climb.

Deeper vacuum on a running machine: impossible and harmful. You cannot evacuate a vessel that contains refrigerant - the pump would simply pull refrigerant out of the machine. Pulling below the manufacturer's specified limit can also damage motor windings and overload the purge. Moisture is removed by:

• recovering to 25 mm Hg absolute with low-pressure recovery equipment,
• replacing the dehydrator/filter-driers (moisture in an R-123 machine forms acids that attack the windings - the dehydrator is a maintenance item, not a decoration),
• evacuating with a vacuum pump and a micron gauge, then performing a standing vacuum test before recharging.

The purge unit should be serviced and its efficiency verified, but it is a symptom-treater here; the leak is the problem.