Every liquefaction process ends identically: treated gas at around minus 160 degrees, let down to atmospheric pressure, running into a tank. What distinguishes them is how the heat is taken out on the way.
The underlying problem
Cooling gas from ambient to minus 162 is not one job but a range of jobs. Near ambient you are removing heat from a gas; near the bottom you are condensing it; in between the properties change continuously.
A refrigerant boiling at a single temperature is efficient only near that temperature. Use one refrigerant across the whole range and most of the cooling happens with a large temperature difference between the refrigerant and the gas, and every degree of unnecessary difference is wasted work.
There are two ways round this, and the commercial processes are variations on each.
Cascade: several pure refrigerants in series
Chill the gas with propane down to about minus 35, hand it to an ethylene loop down to about minus 95, hand it to a methane loop for the rest. Each loop operates near its own efficient range, and the staged handover keeps the temperature differences small.
The optimised cascade process, associated with ConocoPhillips, does exactly this with plate-fin exchangers and a characteristic machinery arrangement: two half-sized gas turbine and compressor sets per refrigeration service rather than one full-sized set. Losing a machine costs half a service rather than all of it, and availability is correspondingly high.
The cost is equipment count. Three complete refrigerant systems means three compressors, three inventories, three sets of everything.
Mixed refrigerant: one loop, many components
The alternative is to blend refrigerants — nitrogen, methane, ethane, propane, sometimes butane — into a single circulating mixture. A blend does not boil at one temperature; it boils across a range, and by choosing the composition the range can be matched to the cooling curve of the gas.
One loop then covers what a cascade needs three loops to do. Far less equipment, and in principle a very close match between refrigerant and gas temperature throughout.
The cost is control. The mixture’s composition is a live process variable that has to be maintained, and the main cryogenic heat exchanger doing the work is a single very large, very expensive item.
C3MR: the workhorse
Propane pre-cooled mixed refrigerant is the most widely used process, and it is a hybrid. Propane, in stages, does the warm end — where its single boiling point is a reasonable match. A mixed refrigerant in a coil-wound exchanger does everything below that.
It captures most of the benefit of both approaches: the simplicity of a pure refrigerant where a pure refrigerant works, and the range-matching of a mixture where it does not. Typical single-train capacity is around 5 Mtpa, and it is the process behind a very large share of world capacity.
AP-X: adding a third stage
C3MR’s ceiling is set by how much the mixed refrigerant circuit can handle. AP-X moves the ceiling by adding a nitrogen expander loop that takes over the final sub-cooling, relieving the mixed refrigerant of the coldest and most demanding part of the duty.
The result is a train of roughly 8 Mtpa rather than 5 — the step change that made Qatar’s build-out possible and drove unit costs down sharply.
DMR: for cold climates
Propane pre-cooling assumes the ambient is warm enough that propane has useful work to do. Where the air and sea are already cold, that assumption weakens.
Dual mixed refrigerant replaces the propane stage with a second mixed refrigerant loop whose composition can be tuned to the actual pre-cooling range. It suits Arctic and sub-Arctic projects, and it exists because the same process does not perform identically in the Barents Sea and the Persian Gulf.
SMR and nitrogen expanders: the simple ones
A single mixed refrigerant loop does the whole job in one circuit. It is meaningfully less efficient than a staged process and dramatically simpler, which is the right trade at small scale and offshore.
A nitrogen expander cycle is simpler still: compress nitrogen, expand it through a turbine so it becomes very cold, use it to chill the gas. Efficiency is poor by comparison with any hydrocarbon refrigerant, and the compensation is that nitrogen is inert — there is no flammable refrigerant inventory anywhere on the unit.
On a floating plant, where a hydrocarbon refrigerant leak has nowhere to disperse and every square metre is contested, that safety property is frequently worth more than the efficiency.
What the choice actually determines
Train size, first: whether a project needs two trains or three to reach its target capacity, which changes the whole site.
Efficiency, second, which sets how much of the feed gas is consumed rather than sold — typically 8 to 12 per cent, and the difference between a good and a poor match across that range is real money over decades.
Climate sensitivity, third. Every cycle rejects its heat to air or seawater, so all of them lose output on a hot day, but not equally.
And equipment count, which drives capital cost, plot area, maintenance burden and how much of the plant can be lost to a single failure.
None of that appears in a capacity figure. Two 5 Mtpa trains using different processes are different machines with different economics, and the tracker records the capacity rather than the process.