20.3 Why Secondary Coolants and Glycol Testing/Maintenance

Direct ammonia in production coils is efficient: one ΔT from air to evaporating ammonia. Plants still install secondary coolant systems—and CIRO tests them—because ammonia in occupied space is the hazard that drives architecture, not the extra pump. This section is the why, the energy and temperature penalties, plate-heat-exchanger glycol chillers, and how glycol testing and maintenance (the chemistry in 20.2) become an operating discipline instead of a forgotten drum on the mezzanine.

Why plants accept a secondary loop

Keep ammonia out of occupied and production areas. A coil leak in a packaging room is then glycol or brine, not an anhydrous ammonia release over people and product. Detectors, evacuation, and product hold still exist for the engine room; they are not triggered by every pinhole in a production coil. That is the same logic as NH3/CO2 cascade, with a pumped liquid instead of a second vapor-compression circuit.

Smaller ammonia charge. The ammonia inventory lives in the chiller package, vessels, and high side in the machinery room. Production coils are full of glycol, not R-717. Charge reduction can be part of staying closer to—or in some designs under—the 10,000 lb PSM/RMP threshold. Count pounds. A large flooded chiller and condenser still add up. Secondary is not a magic exemption from IIAR 2 or PSM if you remain above threshold.

Simpler coil maintenance. Production air units can be washed, gasketed, or isolated and drained without hot work and ammonia pump-out on every coil job. Valve stations in the room are glycol valves. That matters in food plants that open coolers for sanitation constantly.

Freeze protection. Water coils freeze and split. Inhibited glycol (or brine) lets room coils and piping sit below 32°F. Blast cells and dock coils that see below-freezing air need a freeze point below the lowest fluid temperature, not a 20% mix that only burst-protects a rooftop coil in winter.

Glycol coils typically defrost with water, electric, or warm glycol, not ammonia hot gas in the occupied space. Defrost sequence details are Chapter 21; the architectural point is that you moved the ammonia phase-change out of the room.

Cascade, two-stage, and secondary are three answers to the same cold-product problem. Cascade puts CO2 in the rooms. Two-stage puts ammonia in the rooms with a better compression ratio. Secondary puts pumped liquid in the rooms and keeps the phase-change in the engine room. CIRO expects you to pick the architecture that matches the hazard, then operate the extra ΔT, the extra pumps, and the extra chemistry as if they were as important as suction pressure—because they are.

The penalties you pay

Secondary is not free refrigeration.

Extra ΔT (colder ammonia). Heat must cross two exchangers: process air to glycol in the room coil, then glycol to ammonia (or CO2) in the chiller. Each has an approach. If the room needs 20°F air, glycol may leave the coil at 12°F, and ammonia may need to evaporate at 5°F or lower depending on the plate chiller. Direct ammonia might have evaporated at 10–12°F for the same air. Colder suction is more compressor work (higher specific volume, more BHP/ton). Operators who fix a warm room by dropping the chiller setpoint without checking glycol flow, approach, and freeze point are often hiding a dirty plate pack or a weak mix.

Work a stacked-approach example (illustrative). Air off the coil must be 0°F. Coil TD is 10°F, so glycol enters the coil near −10°F. The return glycol is −4°F. The plate chiller holds 6°F approach, so ammonia evaporates near −16°F. A direct ammonia coil with the same 10°F TD might have evaporated near −10°F. That extra 6°F of ammonia lift is the secondary penalty before you even turn the pumps on. If the pack fouls and approach opens to 12°F, ammonia is at −22°F and the screw is doing blast-freezer work to hold a cooler.

Pump energy. Circulating pumps add kW on the electric bill and dump that energy into the glycol as heat. Oversized pumps, missing VFDs, balancing valves wide open, and 50% PG at −10°F all inflate that penalty. Pump heat is load the chiller must remove. A round that ignores pump amps while staring at compressor kW/ton is missing half the secondary plant.

Brine and glycol maintenance. Chemistry, strainers, expansion tanks, air elimination, and leak cleanup are a second utility. A neglected loop fouls the chiller, corrodes pumps, and eventually freezes a coil. The maintenance cost is why some owners stay with direct ammonia and accept IIAR 2 evaporators in the rooms.

Plate heat-exchanger glycol chillers

The engine-room machine that makes secondary work is usually an ammonia-to-glycol chiller. Gasketed plate heat exchangers (PHEs) dominate new installs because they are compact and can hold close approach (a few degrees Fahrenheit when clean and correctly selected).

Common industrial arrangements:

• Gasketed PHE: both sides gasketed; can be opened for cleaning. Elastomers must be compatible with ammonia and the glycol. Wrong gaskets swell or crack.
• Semi-welded (cassette) PHE: ammonia in welded channels, glycol in gasketed channels. A gasket failure dumps glycol, not ammonia, into the area around the pack—the usual reason food plants pay for cassettes.
• Shell-and-tube flooded chiller: still common; larger ammonia charge in the shell; oil drainage like any flooded evaporator (oil sinks in ammonia).

Close approach is the sales pitch and the operating KPI. Leaving glycol temperature minus ammonia evaporating temperature is the chiller ΔT. If that number opens over months, suspect fouling, oil on the ammonia plates, air in the glycol, low glycol flow, or a freeze-point mix so viscous the plates cannot wet. Do not crank ammonia suction down as the first move; you risk freezing glycol in the pack if local plate temperature goes below the freeze point. A glycol freeze in a PHE can split plates—the same catastrophe as a water-chiller freeze, except you also dump glycol into the ammonia side or ammonia into the glycol side depending on which channel fails.

Plate chillers need strainers ahead of the pack, air separators, a closed expansion tank sized for the loop, and flow proving so the ammonia feed cannot pull the pack below freeze without flow. Low-flow is a freeze interlock, not a suggestion. Hydrostatic relief belongs on any isolatable liquid-full glycol volume and on the ammonia side if valves can trap liquid. Opening a pack that still has ammonia pressure is an incident, not a cleaning method.

On a CIRO operating screen, a secondary plant should show glycol supply and return temperatures, pump status or differential, chiller approach, and ammonia suction. A warm room with small glycol ΔT and high pump ΔP is a flow or air-bound problem. A warm room with large glycol ΔT and wide chiller approach is a dirty or oil-filmed pack, weak mix, or starved ammonia feed. Those two pictures are not the same setpoint chase.

Testing and maintenance as operations, not chemistry class

Section 20.2 is the lab sheet. The operator job is to run the program.

• Keep a log: freeze point, pH, inhibitor, conductivity, makeup gallons, and chiller approach. CIRO-style what is wrong with this screen items are often a rising chiller approach with a falling freeze point after weeks of silent city-water makeup.
• Sample from a flowing header, not a dead leg. Label EG vs PG so the lab uses the right table.
• Never mix leftover glycol types to correct freeze point. Dump and recharge if the loop was cross-contaminated.
• Treat expansion-tank level and automatic makeup as mechanical-integrity items. Blind city-water makeup is how freeze protection disappears.
• After a production-coil leak, glycol in the drain is both a sanitation event and a concentration event: you lost fluid, then added water. Retest before you declare the freezer ready.
• PHE service: isolate, relieve trapped liquid (glycol hydrostatic and ammonia hydrostatic are both real), follow lockout, and do not open a pack that still has ammonia pressure.

If inhibitor is depleted but freeze point is still correct, add inhibitor only as the vendor directs—many packages are not a grocery-store additive you dump until pH looks pretty. If chlorides, iron, or biological counts are out of band, the loop may need a controlled dump and recharge, not another drum of neat glycol. Premix at the specified concentration. Record what you did. The next operator, and the next PHA, should be able to see that the secondary system was treated as process equipment, not as a plumbing afterthought.

Secondary does not retire ammonia skill. The chiller is still an ammonia evaporator with oil, level, expansion devices, and relief. What you gained is that a failed production coil is a glycol event. What you must not give back is a frozen plate pack, a mixed-glycol science experiment, or an unmanned makeup line that quietly turns a −20°F freezer loop into 15% water.