20.1 NH3/CO2 Cascade Designs

Key Takeaways

  • In an NH3/CO2 cascade, the CO2 (R-744) low-temp circuit rejects heat by condensing in a cascade heat exchanger against boiling ammonia; ammonia stays in the machinery room and CO2 stays in production piping and evaporators.
  • Cascade ΔT is the approach across that exchanger: CO2 must condense warmer than ammonia evaporates, so the ammonia suction is colder than a two-stage ammonia plant would need for the same process temperature.
  • CO2 standstill pressure climbs toward hundreds of psig if an isolated circuit warms (about 491 psig at 32°F and about 838 psig at 70°F on a saturated teaching table); every isolatable volume needs relief, including hydrostatic protection for trapped liquid.
  • Liquid CO2 is often denser than refrigeration oil, so oil tends to float; CO2 compressors use OEM POE or PAG oil, never the ammonia mineral/PAO charge, and a cascade-HX leak can form solid ammonium carbamate.
  • ANSI/IIAR CO2-2021 is IIAR’s published closed-circuit CO2 safety standard, and IIAR also publishes a CO2 handbook; ASHRAE 15 remains the general refrigeration safety standard. Do not invent IIAR CO2 paragraph numbers.
Last updated: September 2026

A cascade refrigeration plant is two closed circuits that exchange heat at one exchanger and do not share refrigerant. The low-temperature circuit uses carbon dioxide, R-744. That circuit absorbs heat from blast cells, IQF tunnels, ice cream freezers, and other low-temp loads and rejects it by condensing CO2 against boiling ammonia. The high-stage ammonia (R-717) circuit then rejects heat to the evaporative condenser the way any ammonia high side does.

CIRO Content Area 7 groups cascade with two-stage ammonia and secondary coolants. The exam distinction is not that cascade is always more efficient. It is why you split refrigerants, what the cascade ΔT costs you, and which CO2-only hazards—standstill pressure, relief, dry ice at the triple point, and oil that does not sink—do not exist on a two-stage ammonia plant.

Why CO2 is in the rooms and ammonia is not

Ammonia is thermodynamically excellent, but it is toxic, it carries a 10,000 lb OSHA PSM / EPA RMP threshold as anhydrous ammonia, and a leak in a production room is a people-and-product event. Carbon dioxide is classified A1 under ASHRAE 34 (nonflammable, lower toxicity class). It is still an asphyxiant: OSHA PEL 5,000 ppm (8-hour TWA), NIOSH IDLH 40,000 ppm. Those numbers are not ammonia’s 50 ppm PEL / 300 ppm IDLH. A dense CO2 release can pool and displace oxygen without the sharp odor that sends ammonia operators running.

The architecture uses that pairing:

  • CO2 evaporators and process piping stay in occupied and production areas.
  • Ammonia stays in the machinery room—compressors, vessels, condenser, and the ammonia side of the cascade heat exchanger.

A production-room leak is then a high-pressure CO2 event, not an ammonia release over food. The ammonia leak path is still real; it is concentrated where ANSI/IIAR 2 machinery-room detection, ventilation, and emergency shutdown already apply.

ANSI/IIAR CO2-2021, Safety Standard for Closed-Circuit Carbon Dioxide Refrigeration Systems, is IIAR’s published CO2 safety standard. It sets minimum requirements for design, installation, start-up, and inspection, testing, and maintenance of closed-circuit CO2 plants. IIAR also publishes a CO2 handbook for theory and application. ASHRAE 15 remains the general refrigeration safety standard; IIAR CO2 is the CO2-specific companion the way IIAR 2 is ammonia-specific. Model mechanical and fire codes have been adopting IIAR CO2. Do not invent IIAR CO2 paragraph numbers on the exam. Name the documents, use the copy in process safety information, and apply the same mechanical-integrity discipline you already use on ammonia relief and vessels.

IIAR CO2 and ASHRAE 15 also address occupancy, equipment location, and concentration limits if a circuit discharges. Industry briefings on the 2021 standard have cited a complete-discharge concentration limit of 15,000 ppm in rooms that contain CO2 equipment. Treat that as a pointer to the standard in PSI, not as a number to size a room from memory unless the exam item gives you the figure.

The cascade heat exchanger

Heat flows only downhill. CO2 must condense warmer than ammonia evaporates, by the exchanger approach—the cascade ΔT. Typical industrial cascade exchangers are welded-plate, plate-and-shell, or shell-and-tube units: CO2 condenses on one side, ammonia boils on the other. A clean, well-selected unit often holds approach in the 5–10°F range. Oil film, fouling, noncondensable gas on the CO2 side, or a starved ammonia feed opens that ΔT. CO2 condensing pressure then rises, CO2 discharge temperature and compression ratio climb, and the freezer loses capacity while the ammonia high stage can look oddly lightly loaded.

Put numbers on the penalty. Process needs −40°F CO2 evaporating. If CO2 condenses at +15°F on the cascade exchanger and approach is 8°F, ammonia must evaporate near +7°F. The ammonia machines are then a modest-lift high stage to the evaporative condenser. That extra 8°F is compressor work a two-stage ammonia plant would not spend, because two-stage ammonia evaporates at the process temperature. Cascade ΔT is the thermodynamic price of keeping the fluids—and the ammonia charge—apart.

If the cascade exchanger leaks, the circuits contaminate each other. Ammonia and CO2 can form solid ammonium carbamate, especially with moisture, and that solid plugs valves, orifices, and exchangers. Oil chemistries also mix. Isolate per the SOP; do not keep running to finish a production run on a known cascade-HX leak. Both charges are suspect until the OEM and lab say otherwise.

Cascade versus two-stage ammonia

Two-stage ammonia (Chapter 19) uses one refrigerant. Low-stage compressors take cold suction; an open or closed intercooler cools discharge gas and usually subcools liquid; high-stage machines reject to the condenser. Interstage pressure is often near the geometric mean of suction and discharge absolute pressures (psia, not psig). Ammonia piping and evaporators are in the production space. Mineral oil sinks in liquid ammonia and is drained from low-side pots.

NH3/CO2 cascade uses two refrigerants and two oil systems. There is no ammonia interstage in the two-stage sense. The cascade exchanger is the heat sink for the entire CO2 circuit.

DecisionTwo-stage NH3NH3/CO2 cascade
Refrigerant in production evaporatorsAmmoniaCO2
Ammonia locationEngine room and production piping/coilsMachinery room and cascade HX ammonia side
Extra heat-transfer penaltyIntercooler, not a cascade HXCascade HX approach (CO2 condensing warmer than NH3 evaporating)
Low-temp compressor sizeLarge (high NH3 specific volume at −40°F)Smaller (dense CO2 suction vapor)
Standstill pressure in roomsAmmonia (may go to vacuum on LT)CO2 rises toward hundreds of psig if the circuit warms
Oil in flooded vesselsSinksOften floats on dense liquid CO2

Choose cascade when production must not contain ammonia; when very low temperature makes CO2’s dense suction vapor attractive; when the owner wants ammonia charge confined to the engine room (possibly as part of a PSM inventory strategy—count the high-stage charge; do not assume you are under 10,000 lb); or when an HFC phase-down freezer will be a natural refrigerant without ammonia in the rooms.

Stay with two-stage ammonia when evaporators in the space are already accepted, you want to avoid CO2 high-pressure vessels and standstill design in production, and the extra cascade exchanger and second oil system are not worth a dual-circuit leak.

Both architectures beat single-stage ammonia at deep freeze because they cut compression ratio and discharge temperature on the coldest suction. Cascade does it with a second fluid; two-stage does it with two ammonia stages.

CO2 pressure, warm-up, and relief

CO2’s critical temperature is 87.8°F and critical pressure about 1,070 psia. An industrial cascade low stage is normally subcritical: the cascade HX is a true condenser, not a transcritical gas cooler. The operator fact is simpler: low-temp CO2 is still high pressure compared with ammonia.

Approximate saturated pressures for teaching (use a published R-744 P-T chart on the job):

Saturated temperatureApproximate CO2 pressure
−40°F~131 psig
−20°F~200 psig
0°F~291 psig
32°F~491 psig
70°F (warm room / idle)~838 psig

Ammonia at −40°F is in vacuum. CO2 at −40°F is already a high-pressure system. If that circuit warms toward room temperature while isolated—power loss, an idle freezer equalized, liquid trapped between valves—pressure follows the saturation curve unless the charge is small enough to go all-vapor below the relief setting. Warm-up toward 70°F is a relief case on a liquid-full or nearly liquid-full circuit.

Operator rules:

  • Every isolatable CO2 volume needs overpressure protection. Hydrostatic (trapped-liquid) relief is as important as vapor relief. CO2 liquid is nearly incompressible; two closed valves on a liquid line make a pump.
  • High-pressure cutout must stop compressors before relief valves lift. Relief is not capacity control.
  • Rapid depressurization toward the triple point (−69.9°F, 75.1 psia) can form solid CO2 (dry ice) that plugs valves, orifices, and relief paths. Pump-out and vent procedures must not create a solid plug in a relief line.
  • Some plants provide a standstill condenser, water-cooled dump, or transfer into a designed holding volume so idle warm-up does not lift relief. Know which you have. Never isolate a liquid-full CO2 coil for a minute without a relief path.

IIAR has continued work on CO2 overpressure-relief guidance after the 2021 standard. Use the current published IIAR CO2 standard, vessel nameplates, and relief-device data in PSI.

Oil: two circuits, opposite gravity

Ammonia mineral oil sinks. Liquid CO2 in the industrial low-temp range is often denser than refrigeration oil, so oil tends to float on liquid CO2 in flooded vessels. If you drain the bottom of a CO2 vessel looking for oil the way you do on ammonia, you may dump refrigerant and leave the lubricant. Take oil from the connection the designer put on the oil layer—often high in a flooded CO2 vessel.

CO2 compressors typically use POE or PAG oils specified by the OEM—not the mineral or PAO charge from the ammonia screws. Never equalize, drain, or top up so the two oils mix. Oil return on CO2 depends on velocity, separators, stills, and those vessel connections. Cold, viscous oil plus a wide cascade ΔT is a classic CO2 compressor-starvation mode.

On a CIRO-style screen, read CO2 suction and cascade (CO2 condensing) pressure converted to saturation temperature, ammonia evaporating temperature on the other side of the HX, cascade approach, and oil pressure/temperature on each compressor family. Rising CO2 condensing pressure with a falling or unexpectedly cold ammonia suction on the same exchanger is a dirty, oil-filmed, or underfed cascade HX, not a condenser wet-bulb problem.

Loading diagram...
NH3/CO2 cascade: separate circuits meet only at the cascade HX
Approximate saturated CO2 pressure vs temperature (teaching values, psig)
Test Your Knowledge

In an industrial NH3/CO2 cascade, what is the cascade heat exchanger doing?

A
B
C
D
Test Your Knowledge

A food plant wants ammonia out of occupied production but still wants a natural-refrigerant freezer at −40°F. Which statement matches NH3/CO2 cascade practice?

A
B
C
D
Test Your Knowledge

A CO2 freezer circuit is isolated liquid-full and warms toward a 70°F machinery-room ambient. Which operator concern is correct?

A
B
C
D
Test Your Knowledge

Compared with two-stage ammonia at the same process temperature, why does an NH3/CO2 cascade usually run the ammonia evaporating temperature colder than the process?

A
B
C
D