How to Select an Industrial Refrigeration Unit in 2026: Evaporating Temperature, Refrigerant and Compressor in Three Steps

Most industrial refrigeration enquiries that reach our desk open the same way: "We need a 1,000 kW screw chiller package. Please quote." One number, one machine type, and the single parameter that actually determines the design is missing — the evaporating temperature.

A 1,000 kW duty at −10 °C and a 1,000 kW duty at −40 °C are not variants of the same package. They are different machines: different pressure ratio, motor power differing by more than a factor of two, possibly single-stage versus two-stage, different refrigerant, different vessel design pressure, and a delivered price that can differ by well over 50 %. Quoting the first without knowing the second is guesswork dressed up as a proposal.

This guide sets out the selection sequence we use when supporting design institutes and EPC contractors on process refrigeration duties: evaporating temperature first, refrigerant second, compressor type third. It is a methodology piece rather than a vendor comparison — if you have already settled the duty and want to know who builds these packages in China, that ground is covered separately in our 2026 buyer's guide to industrial refrigeration package manufacturers.

Why the Order of Selection Matters

The three variables are not independent. Evaporating temperature sets the pressure ratio the cycle must deliver, and pressure ratio narrows the refrigerant field. Refrigerant properties then set suction volume flow, discharge temperature and the number of compression stages, which in turn narrows the compressor type. Work in that direction and each decision closes off options cleanly. Work backwards — picking a machine first — and you spend the rest of the project bending the process to suit hardware that was chosen before anyone knew what the process needed.

This matters commercially as well as technically. In tenders where the datasheet leaves evaporating temperature open, bidders fill the gap with their own assumptions. The resulting quotations look comparable and are not. We routinely see two bids separated by 15–20 % on price where one supplier assumed −20 °C and the other −25 °C — a difference of roughly a stage of compression, not of commercial sharpness.

Step 1: Fix the Evaporating Temperature

The rule is straightforward:

Evaporating temperature = outlet temperature of the cooled medium − heat transfer approach

The approach normally sits between 4 and 8 K. Flooded evaporators, with their higher shell-side coefficients, work comfortably at the tight end (4–5 K); direct-expansion evaporators need the wider end (6–8 K). If a process demands glycol solution leaving at −15 °C through a flooded evaporator, the evaporating temperature lands around −20 °C.

The 2–3 % rule and where margin belongs

Here is the number worth writing on the datasheet cover: every 1 °C reduction in evaporating temperature raises compressor power consumption by roughly 2–3 %. That penalty is permanent. It is paid every operating hour for the twenty-year life of the plant.

Which is why the most expensive habit in refrigeration specification is the "safety margin" applied to temperature. A process engineer needs −15 °C, writes −20 °C on the datasheet to feel comfortable, and quietly commits the owner to 10–15 % higher energy cost forever. Margin belongs in capacity — around 10 % is normal and defensible — and in fouling allowance on the exchangers. It does not belong in temperature.

The counter-argument is usually heat exchanger cost: widening the approach by 2 K shrinks the surface area and the capital cost of the evaporator. Run the numbers over the operating life and the electricity bill wins almost every time on a continuously running process duty. It is a conversation worth having with your package supplier before the datasheet is frozen, not after.

Step 2: Match the Refrigerant to the Temperature Band

Once evaporating temperature is fixed, the refrigerant field narrows sharply. Each band has mature, well-proven options:

Evaporating temperature	Typical duty	Common refrigerants	System configuration
0 to −10 °C	Chilled water, process cooling	R134a, ammonia (R717)	Single stage
−15 to −25 °C	Brine / glycol secondary loops	Ammonia, R507A, propylene (R1270)	Single stage with economiser
−30 to −45 °C	Tank-farm direct cooling, low-temperature process	Propylene, two-stage ammonia	Two-stage, or single stage with economiser
Below −60 °C	Ethylene liquefaction	Ethylene cascade	Cascade system

The regulatory column you now have to add

Since March 2026, China's Ministry of Ecology and Environment has applied full-lifecycle registration to refrigerants — production, sale, use, recovery and destruction each logged, with direct venting prohibited and dedicated provisions for ammonia and CO₂ systems. The practical effect on new projects is that GWP has become a selection column alongside capacity and efficiency.

The spread is wide. R507A carries a GWP close to 4,000 and R134a around 1,430; both sit in the tightening band. Ammonia has a GWP of zero and propylene approximately 2. Ammonia still accounts for roughly 41 % of the global industrial refrigerant load, and petrochemical sites have a structural advantage with propylene: the plant already produces it, so make-up charge becomes an internal logistics question rather than a procurement one. Propylene's share of new-build process refrigeration duties has risen noticeably over the last few years for exactly this reason.

None of this makes fluorinated refrigerants unusable. Installed plant continues to run, and new installations are not banned outright. But specifying a high-GWP fluid on a twenty-year asset in 2026 is a decision to hand a retrofit project to your successor around 2031. For long-life process duties, the natural refrigerants deserve the first look.

Direct expansion or a secondary loop?

The refrigerant decision carries a system-architecture decision with it. Direct cooling is more efficient — you avoid a whole secondary circulation loop and its pumping power and temperature penalty — and suits duties where the cold users are concentrated. A secondary loop (glycol or brine) costs efficiency but confines the refrigerant charge to the compressor house, keeps it out of the process area entirely, and simplifies hazard management. On ammonia duties in congested plants, that containment argument frequently decides the layout on its own.

Step 3: Choose the Compressor Type

With temperature and refrigerant settled, compressor selection is largely constrained already. The three mainstream machine types divide up as follows:

Criterion	Screw	Reciprocating	Centrifugal
Capacity per machine	≈ 200–13,000 kW	Generally below 500 kW	Above 2,000 kW
Pressure-ratio capability	Strong; with economiser reaches deep-cooling duties	Strong	Weaker; surge risk at low load
Capacity control	Stepless slide valve, 10–100 %	Cylinder unloading, stepped	Inlet guide vanes, limited below ~40 %
Wear parts and maintenance	Few moving parts, long overhaul interval	Valve plates, piston rings replaced frequently	High precision, demanding maintenance regime
Typical position	Mainstream for petrochemical refrigeration	Small capacity, unusual pressure ratios	Large, stable base load

Condensed to three sentences: where capacity is small and the pressure ratio awkward, reciprocating machines still have a place; where capacity is very large and load genuinely flat, centrifugal efficiency wins; and the broad middle — a few hundred kW to five figures, fluctuating load, low evaporating temperature — belongs to the screw compressor.

That middle band describes most petrochemical process refrigeration, because petrochemical load is rarely flat. Start-up, grade changes, seasonal ambient swing and turndown all move the cold demand around, and this is precisely where stepless slide-valve control earns its keep. A centrifugal machine held at 35 % load is a different proposition from a screw package at 35 % load. The largest single screw package we have integrated was a twin-screw unit for a styrene plant at 10,863 kW — a duty that a decade ago would have been assumed to be centrifugal territory by default.

One point of transparency about scope: Lmart integrates screw compressor packages — compressor bare shafts are purchased from established compressor manufacturers. What we design and build in house is the code-compliant pressure envelope around them: pressure vessels, shell-and-tube heat exchangers, separators and receivers, oil systems, piping and the complete skid. The reciprocating and centrifugal comparison above is drawn from published industry data and from packages we have supplied alongside, and is offered as orientation rather than as a claim of manufacturing scope.

Two Worked Examples, Two Different Answers

Case 1 — −15 °C class brine package

A PMMA unit at a domestic chemical producer required low-temperature brine. The delivered configuration: three brine chiller packages, 980 kW each, R134a, glycol solution as the secondary fluid.

At this temperature band the pressure ratio is mild, so single-stage screw compression on R134a keeps the system simple. Three units in parallel give operational redundancy and let the plant follow load by taking machines offline rather than running everything at deep turndown. The secondary loop concentrates the refrigerant charge in the machine room, so no refrigerant reaches the process side — a straightforward safety case for the operator.

Case 2 — −38 °C ammonia direct-cooling package

A Rectisol (low-temperature methanol wash) unit at a domestic new-materials producer specified −38 °C evaporating temperature, 5,600 kW per unit, four units, ammonia.

At −38 °C the single-stage pressure ratio is already stretched, so the packages use economiser vapour injection to hold isentropic efficiency and control discharge temperature. Ammonia was chosen partly on efficiency in this band and partly on compliance: with GWP zero, the owner side-steps the phase-down question entirely. At 5,600 kW per machine only screw or centrifugal machines are candidates, and because the load swings, screw won.

Put the two side by side. The evaporating temperatures differ by just over 20 °C — and the refrigerant, the system architecture, the stage configuration and the whole safety design differ completely. That is the argument for the sequence in one picture.

Duties get more layered than this, too. One chemical plant's XDI unit needed 1,828 kW at −25 °C and 86 kW at −45 °C simultaneously. Rather than forcing a single dual-temperature system, the answer was to split the temperature levels and give the small deep-cooling duty its own dedicated machine — less power, less complexity, easier operation.

The Datasheet Checklist That Makes Bids Comparable

Selection discipline is worth little if the enquiry documents don't carry it through to the bidders. At minimum, a refrigeration package datasheet should state:

• Refrigeration capacity at the rated point, and the design margin basis
• Evaporating temperature — or the secondary fluid inlet/outlet temperatures and flow rate, so bidders derive the same number
• Condensing side conditions — cooling water supply temperature and available flow, or ambient design temperature for air-cooled duty
• Refrigerant, including any owner restriction on GWP or on ammonia in the process area
• Load range and expected operating profile — how many hours at what percentage load
• Shaft power at 50 %, 75 % and 100 % load as a required deliverable in the bid
• Electrical supply voltage and frequency, and the hazardous-area classification
• Design codes and third-party requirements for the pressure envelope — ASME BPVC Section VIII Div.1, PED 2014/68/EU, GB/T 150, and classification society involvement where the package goes offshore or on board
• Interface and footprint constraints — skid envelope, shipping limits, tie-in points, control system architecture

Miss any one of these and the bids you receive are not a like-for-like set.

Three Traps We See Repeatedly

Trap 1 — padding the evaporating temperature. The process needs −15 °C, the datasheet says −20 °C, and the owner pays 10–15 % extra on power for the life of the plant. Put margin on capacity instead.

Trap 2 — comparing prices without normalising the duty point. A RMB 200,000 gap between bids often turns out to be one supplier quoting −20 °C and another −25 °C. Normalise the datasheet before opening commercial envelopes, or the tender evaluation is measuring assumptions rather than suppliers.

Trap 3 — evaluating on full-load COP alone. If the plant spends 70 % of its hours at 60 % load, full-load COP is close to irrelevant. Ask for the part-load curve, and write the guaranteed part-load shaft power figures into the technical agreement rather than leaving them in the sales presentation.

Where an Equipment Supplier Fits In

Suzhou Lmart Energy Equipment has supplied more than 130 industrial refrigeration units since 2016, across R134a, R507A, ammonia, propylene and ethylene, with evaporating temperatures from 0 °C down to cascade duties below −60 °C. Our scope is screw compressor package integration and skid-mounted systems, built around pressure vessels and shell-and-tube heat exchangers manufactured in our own workshops, with design, fabrication, assembly and testing on one site.

We are the equipment supplier on these projects, not the process designer — the duty belongs to the licensor, the design institute and the owner. What we can usefully contribute at the specification stage is a reality check: whether the evaporating temperature and approach are consistent, whether the refrigerant suits the band and the site's compliance posture, and whether a comparable configuration already exists in our delivered project list. That conversation costs a week at the front end and routinely saves several rounds of technical clarification after bids are in.

If you have a refrigeration datasheet in preparation or out for tender, send it over and we will review the parameters against comparable delivered projects and come back with a technical proposal and budget pricing.