Propylene vs Ammonia vs Ethylene Refrigeration: How to Select the Right Refrigerant for Petrochemical Low-Temperature Duties

Why 2026 Puts Refrigerant Selection on the Critical Path

China's ethylene capacity is entering a concentrated commissioning window. The BASF Zhanjiang Verbund cracker started up in January 2026; the SINOPEC–SABIC Gulei 1.8 Mt/y mixed-feed cracker is scheduled for second-half startup; Hujin Panjin (Aramco JV) 1.65 Mt/y and the Dushanzi Tarim Phase II 1.2 Mt/y ethylene trains are both slotted inside the year. Industry trackers put total 2026 Chinese chemical megaproject investment above RMB 830 billion.

For anyone specifying rotating and static equipment, that wave translates into a single repeated engineering question. Every cracker, and every satellite PDH, polyolefin, methanol-to-olefins and cryogenic tank-farm project that hangs off it, has to nominate a refrigerant for its low-temperature duties. Propylene, ammonia or ethylene — and, increasingly, a cascade of two of them.

The decision looks deceptively simple in a datasheet. In practice it drives compressor frame size, hazardous-area classification, material compatibility, lubricant strategy, insulation thickness, the relief philosophy and, eventually, the operator training programme. Change it after the process package is frozen and the ripple runs through the entire equipment list.

This article sets out the selection boundaries the way an equipment supplier sees them: where each fluid genuinely wins, where the physics stops it, and which mistakes we see repeatedly in enquiry documents.

Rule One: Evaporating Temperature Selects the Refrigerant

If you remember one heuristic from this article, make it this one: start from the required evaporating temperature and let it eliminate candidates before you argue about anything else.

• −5 °C to −45 °C — propylene (R1270) is the workhorse of petrochemical plants.
• −10 °C to −50 °C — ammonia (R717) is thermodynamically superior, provided the site can carry its toxicity management burden.
• Below −60 °C — ethylene (R1150) in cascade with propylene. There is no single-fluid alternative in this range for conventional petrochemical service.

Everything else — efficiency, GWP, availability, maintenance culture — is a tie-breaker applied after the temperature cut. Engineers who start from "which refrigerant is best" instead of "what evaporating temperature do I actually need" tend to generate long meetings and short-lived conclusions.

Property Comparison: The Three Fluids Side by Side

The physical properties explain the boundaries above. This table is the one worth pasting into a selection report.

Parameter	Propylene R1270	Ammonia R717	Ethylene R1150
Normal boiling point	−47.6 °C	−33.3 °C	−103.7 °C
Critical temperature	91.1 °C	132.4 °C	9.2 °C
Practical evaporating range	−5 to −45 °C	−10 to −50 °C	−60 to −100 °C (cascade low stage)
ODP / GWP	0 / ~2	0 / 0	0 / ~4
Safety group (ASHRAE 34)	A3 — flammable	B2L — toxic, lower flammability	A3 — flammable
Volumetric capacity	Moderate	High	High at low temperature
Mineral oil miscibility	Miscible	Immiscible — oil drainage required	Miscible
Copper compatibility	Compatible	Copper and copper alloys prohibited	Compatible

Propylene: the default, with a hard floor at about −45 °C

A normal boiling point of −47.6 °C means that at an evaporating temperature of −45 °C the suction pressure is already close to atmospheric. Push further and the low side goes into vacuum. Three problems arrive together: air and moisture ingress through shaft seals and flanged joints, a steep rise in suction specific volume that inflates the required displacement, and leak detection that becomes an operational chore rather than a periodic check. We have delivered a deep-cooling propylene package at −45 °C for a domestic PTMEG unit; that is realistically the economic edge of the fluid.

Ammonia: best thermodynamics, heaviest safety envelope

Ammonia's latent heat gives roughly three times the mass-specific refrigerating effect of propylene, and its GWP is genuinely zero. In a world steadily retiring high-GWP synthetic refrigerants, its regulatory position only strengthens. The cost is that R717 is a B2L fluid with a low permissible exposure limit. Compressor houses need designed ventilation, gas detection, water curtains and emergency provisions, and not one copper-bearing component may appear anywhere in the circuit — instruments, valve internals and gaskets all need checking.

Crucially, this burden is not new to petrochemical operators. Rectisol (low-temperature methanol wash) units and ammonia synthesis complexes already handle ammonia in bulk. On those sites choosing R717 for the associated refrigeration duty is the path of least resistance, not the exotic option.

Ethylene: irreplaceable below −60 °C, but never independent

The 9.2 °C critical temperature is the whole story. Cooling water at 32 °C cannot condense ethylene at any pressure, so an ethylene circuit has no way to reject its condensing heat to ambient. It must reject that heat into a higher-temperature refrigerant. In practice the propylene high stage condenses the ethylene at around −30 °C, and the ethylene low stage then delivers −60 °C to −100 °C. Cracker cold-box deep-cooling sections are the canonical application.

Matching the Refrigerant to the Process Unit

Properties are theory. Here is how the choice lands on real units, with reference duties we have delivered.

Unit / duty	Typical temperature	Refrigerant	Reference duty delivered
PDH (propane dehydrogenation)	−30 to −40 °C	Propylene	Zhenhai Refining & Chemical PDH refrigeration system, 8,330 kW @ −32 °C (2022)
Polypropylene / polyethylene brine station	−10 to −35 °C	Propylene with glycol secondary loop	Hengli Petrochemical HDPE brine units, 1,034 / 2,200 kW, 3 units (2018)
Rectisol / low-temperature methanol wash	−38 to −45 °C	Ammonia (direct expansion)	Fujian Shenyuan New Materials, 5,600 kW @ −38 °C, 4 units (2020); Wanhua Chemical R717 condensing unit 1,700 kW (2021)
Ethylene tank farm BOG re-liquefaction	−30 to −45 °C	Propylene (condensing the ethylene)	Lianyungang Petrochemical ethylene tank farm, 1,000 kW (2019); 50,000 m³ cryogenic ethylene tank farm, 1,200 kW (2018)
MTO ethylene liquefaction	−35 to −45 °C	Propylene	Nanjing Chengzhi Clean Energy MTO ethylene liquefaction, 1,848 kW, 3 units (2017)
Cracker cold box, deep-cooling section	−60 to −100 °C	Ethylene / propylene cascade	Multiple propylene high-stage packages in cascade service

The tank-farm trap worth flagging in every FEED review

One recurring source of confusion deserves its own paragraph. When a specification says "ethylene refrigeration" for an atmospheric ethylene storage tank, the refrigerant is usually propylene, not ethylene. An atmospheric ethylene tank holds around −104 °C through a boil-off gas loop: vapour is compressed, condensed against a propylene refrigeration system, and returned to the tank as liquid. Ethylene here is the process fluid being condensed, not the working fluid of the refrigeration cycle.

Ethylene genuinely acts as the refrigerant only where process streams must be driven to −100 °C class temperatures — cold-box duties. Separating these two cases cleanly in the process description removes several clarification rounds during bid evaluation, and prevents a tank-farm enquiry from being priced against cold-box assumptions.

Five Screens to Apply After the Temperature Cut

Once the evaporating temperature has narrowed the field, run the candidates through these five filters.

1. Safety and fire philosophy. Propylene and ethylene are A3 flammable fluids: the package area falls inside a hazardous zone and all electrical equipment must be certified accordingly. Ammonia is B2L, so explosion-protection requirements are actually lighter — but toxicity management (detection, ventilation, water curtain, emergency eyewash) is non-negotiable. Because a petrochemical plant is already a classified area, the "extra cost" of A3 fluids is far smaller here than it would be in a cold-storage setting. That is a large part of why propylene is the industry default.

2. Refrigerant availability on site. Propylene and ethylene are already process materials inside the battery limits — easy to draw, easy to top up, purity assured. Ammonia needs a dedicated storage and charging regime unless the site already handles it. This unglamorous logistics point settles perhaps half of all selection arguments in practice.

3. Compressor configuration and oil strategy. Oil-injected screw compressor packages dominate these duties; an economiser-equipped single-machine two-stage arrangement reaches down to about −45 °C. Ammonia's immiscibility with mineral oil means the evaporator side needs oil-drainage provisions, or a miscible synthetic lubricant. Propylene and ethylene are miscible with oil, so oil return is simpler — but dilution effects on viscosity must be checked at the design duty, not just at nameplate conditions.

4. Maintenance and spares reality. Ammonia's copper prohibition narrows instrument and valve selection. Hydrocarbon systems impose stricter hot-work and mechanical-seal management. Neither is maintenance-free; the honest question is which discipline your operations team already lives with every day.

5. Lifecycle and regulation. All three fluids have zero ODP and near-zero GWP, and none appears on a phase-down schedule under the Kigali Amendment. For plants still running R22 legacy packages, migrating straight to a natural refrigerant during revamp avoids the "modify now, modify again in five years" trap.

Cascade Design Points That Bite Late

Cascade systems fail commercially more often than technically — usually because interface responsibilities were vague at enquiry stage. Three items are worth writing explicitly into the specification.

The cascade condenser is the contractual seam. It is simultaneously the ethylene condenser and the propylene evaporator. Whoever supplies it owns the approach temperature that sets both stages' efficiency. Define the design margin, the fouling allowance and the applicable code — typically ASME BPVC Section VIII together with TEMA mechanical class — in the enquiry, not in a later technical query.

Standstill pressure governs the design pressure. An ethylene circuit warming to ambient during a shutdown will pressurise well beyond its operating envelope. The low-stage design pressure and the relief philosophy have to be set for the parked condition, not the running one. This single item has driven more than one late vessel re-rating.

Startup and pull-down sequencing. The high stage must be established before the low stage can be loaded. Whether that logic sits in the package control panel or in the plant DCS is a scope question — agree it early, because retrofitting the interlock after FAT is expensive and slow.

Two Reminders from Project Sites

Do not let habit make the decision. We have seen more than one project specify propylene simply because the previous unit on the same site used propylene. In one case the duty ran year-round at −20 °C, where ammonia would have saved a compressor frame size. The mirror-image error is more painful: forcing propylene onto a genuine −48 °C duty, then living with a permanently sub-atmospheric low side and leak hunting as a routine shift task. A week of duty verification during process design is cheaper than either outcome.

Check single-train capability early, because units keep getting bigger. The 2026 ethylene projects are typically 1.2–1.8 Mt/y, and their refrigeration loads run into the tens of megawatts. The largest single-project refrigerating capacity we have delivered is 23,300 kW (brine packages for an acrylonitrile unit, 2022). At that scale, unit count, sparing philosophy and part-load control strategy have to be frozen in the concept phase — every later change propagates through foundations, piping and electrical distribution.

What Lmart Supplies — and What It Does Not

Suzhou Lmart Energy Equipment has supplied more than 130 industrial refrigeration packages to the petrochemical sector since 2016, covering propylene, ammonia, ethylene and mixed-refrigerant duties across PDH, polyolefin, Rectisol, ethylene tank-farm and MTO applications.

It is worth being precise about our role, because equipment scope is often blurred in this market. Lmart is an equipment supplier and package integrator, not a compressor manufacturer. Compressor blocks are sourced from established original equipment manufacturers. What we design and build in our own works are the pressure vessels and shell-and-tube heat exchangers — chillers, condensers, economisers, oil coolers, separators and receivers — and what we deliver as a system is the complete screw compressor package: unit integration, skid fabrication, piping and instrumentation, control panel interface, and works testing before shipment.

On the certification side, Lmart holds ASME U-Stamp (Section VIII Division 1), PED 2014/68/EU (CE) and ISO 9001, together with CCS Type & Works Approval and works approval from DNV, LR, BV, NK and RINA, plus KGS certification for the Korean market. Equipment is built to TEMA, EN 13445 and GB/T 150 as the project requires.

For the projects in the current commissioning wave, the practical offer is straightforward: send us the duty, the evaporating temperature, the available cooling medium and the site's hazardous-area and toxicity constraints, and we will tell you which of the three refrigerants the physics actually allows — including the cases where the answer is a cascade you had not budgeted for.

Conclusion

There is no universally superior refrigerant among propylene, ammonia and ethylene — only a fluid that suits a given temperature band, unit type and site culture. Propylene is the default answer down to −45 °C. Ammonia wins on thermodynamics and regulation but has to clear the toxicity gate. Below −60 °C, an ethylene cascade is the only route. In a year when a large slice of global new ethylene capacity comes online at once, doing that arithmetic properly during process design is the cheapest engineering investment on the project.