NH₃ vs CO₂ Refrigeration: Total Cost of Ownership Comparison for Large Cold Storage

Ammonia (NH₃, R717) and carbon dioxide (CO₂, R744) are the two natural refrigerants that dominate serious industrial cold storage decisions. Both have negligible global warming potential, both are exempt from F-gas phase-down risk, and both are supported by mature compressor, valve and control supply chains. The question facing procurement directors is not which refrigerant is 'better' in the abstract, but which one delivers the lowest risk-adjusted total cost of ownership for a specific load profile, climate, jurisdiction and operating team.

This comparison is deliberately vendor-neutral. It is written for plant managers, refrigeration engineers and procurement directors preparing an RFQ for a facility in the 500 kW to 10 MW range, where the refrigerant decision has a 20-year financial consequence.

The short answer

For large, single-site, 24/7 frozen storage with an in-house or contracted industrial refrigeration team, low-charge ammonia usually wins on lifetime energy and lifetime cost. For multi-zone distribution centres, sites in cool climates, sites where a machinery room or safety zone is difficult to permit, and sites with useful heat-recovery demand, CO₂ transcritical is often equal or better once opex and compliance overhead are included. For very large frozen facilities in hot climates, NH₃/CO₂ cascade — ammonia on the high stage, CO₂ pumped on the low stage — frequently beats either fluid used alone.

Thermodynamics and energy efficiency

Ammonia has the highest latent heat of vaporisation of any common refrigerant and excellent heat-transfer characteristics, which translates into small charge, compact piping and high cycle efficiency. At typical frozen-store conditions (−32 °C evaporating, +35 °C condensing) a well-designed two-stage or economised screw ammonia plant will comfortably operate in the region of COP 1.6–2.0, with modern low-charge packaged plants at the upper end when combined with variable-speed compressors and floating head pressure.

CO₂ behaves differently because of its low critical temperature (31.0 °C). Below that ambient the cycle is subcritical and highly efficient; above it the system runs transcritical, and discharge pressures reach 90–120 bar with a marked efficiency penalty unless mitigation is designed in. In cool and temperate climates, transcritical CO₂ with parallel compression is broadly competitive with ammonia. In hot climates (design ambients above roughly 32–35 °C) it needs adiabatic gas coolers, ejectors, parallel compression or mechanical subcooling to close the gap — each of which adds capex and maintenance scope.

The practical rule: model energy for your own bin-hour weather data, not a headline COP. Two identical plants in Rotterdam and Riyadh can differ by 25–40 % in annual kWh on CO₂ transcritical, and by far less on ammonia.

Safety, standards and compliance

Ammonia is toxic (Group B2L under ASHRAE 34 — toxic and mildly flammable). Design and operation are governed by EN 378 and ISO 5149 in Europe, ASHRAE 15 and IIAR 2/6/7/9 in North America, plus the Pressure Equipment Directive or ASME B31.5 for pipework. Above jurisdictional charge thresholds — commonly 5,000 kg in the EU under the Seveso III Directive, and 10,000 lb (≈4,536 kg) under US OSHA Process Safety Management — the site enters a major-accident regime with formal hazard studies, emergency planning and inspection obligations. Practical requirements include a dedicated machinery room with gas detection and mechanical extract, emergency ventilation and stop controls, trained operators, and written safety procedures.

Low-charge ammonia packages have changed this calculus materially. By moving to CO₂ or glycol secondary loops, charge on modern plants can drop from several thousand kilograms to 20–60 kg per 100 kW, which in many jurisdictions keeps the facility below the major-accident thresholds and shrinks the compliance burden.

CO₂ is A1 — non-toxic and non-flammable — which removes the toxic-release scenario and normally the safety-zone and major-accident obligations. It is not risk-free: CO₂ is an asphyxiant at concentrations above roughly 5 %, so occupied spaces still require detection and ventilation, and the high standstill pressures (typically 60–90 bar design) demand pressure-rated components, trained brazers and welders, and a standstill strategy — auxiliary condensing unit or controlled venting — for power outages. EN 378 and ISO 5149 still apply.

For procurement, the compliance difference usually shows up as permitting time and insurance premium rather than pure hardware cost. In dense urban distribution sites and rented facilities, that difference often decides the project.

Capex: what actually differs

For a 1 MW frozen facility, the two architectures rarely differ by more than 10–20 % in installed cost, but the composition differs. Ammonia carries the cost of a machinery room, gas detection, emergency ventilation, safety documentation and steel or stainless pipework, offset by relatively low compressor and valve costs and small pipe diameters. CO₂ carries the cost of high-pressure components, stainless or high-grade copper pipework, gas cooler with adiabatic pads in hot climates, standstill protection and specialist installation labour, offset by no machinery room requirement and simpler permitting.

As a planning envelope, indicative installed refrigeration cost for large industrial projects lands around USD 900–1,500 per kW for ammonia and USD 1,000–1,700 per kW for CO₂ transcritical in hot climates, with NH₃/CO₂ cascade in between. Treat these as order-of-magnitude inputs to a screening model, not a quotation — actual pricing is driven by site conditions, local labour and scope split.

Opex over 15 years

Energy dominates. A 1 MW plant running 6,000 equivalent full-load hours at USD 0.12/kWh spends roughly USD 430,000 per year at COP 2.0 and roughly USD 540,000 at COP 1.6 — a 15-year difference of about USD 1.6 million, which dwarfs any plausible capex gap between the two technologies. This is why the refrigerant decision must be made on modelled annual kWh, not headline plant price.

Maintenance sits at roughly 2–4 % of installed cost per year for both, but the skills differ: ammonia needs certified industrial refrigeration technicians and periodic pressure-vessel inspection; CO₂ needs high-pressure competence and disciplined leak management. Refrigerant top-up is negligible for both (ammonia around USD 1–3/kg, CO₂ around USD 1–2/kg) — a decisive contrast with HFC systems facing phase-down pricing.

Heat recovery is where CO₂ can reverse the ranking. Transcritical discharge gas at 90–120 °C is directly usable for hot water, wash-down, defrost and space heating. Where a site has a genuine, year-round heat demand, recovered heat can cut net energy cost by 10–25 % and turn a marginal CO₂ case into the winning one. Ammonia can also recover heat, but at lower delivered temperatures without additional equipment.

Building a defensible TCO model

The structured refrigerant decision method, including the HFC/HFO alternative and the traps that catch buyers in hot climates, is set out in Ammonia vs. CO₂ vs. HFC/HFO: a vendor-neutral refrigerant decision framework. To put numbers against your own site, use the refrigeration load calculator and the electricity cost calculator.

How to put this into an RFQ

Do not specify the refrigerant in the RFQ unless a regulatory or site constraint forces it. Specify the duty: capacity at design evaporating and ambient temperatures, temperature zones, throughput and pull-down duties, availability and redundancy, noise limits, heat-recovery demand and the maximum acceptable annual energy consumption. Then ask each supplier to price their best-fit architecture and to state guaranteed annual kWh under a defined operating profile. Comparing guaranteed energy against a common baseline is the single most effective way to make NH₃ and CO₂ offers genuinely comparable.

ColdMatch Group runs this comparison for buyers as part of a vendor-neutral sourcing process — the same duty specification is issued to qualified ammonia and CO₂ manufacturers, and offers are normalised on total cost of ownership rather than headline price.