Five numbers make a refrigerant - six case studies in the refrigerant lab
By DigiEntropy Engineering · 2026-09-05 · 18 min read
Rebuild R454B and R448A from their pure ingredients and check them against the published values, search for a blend under an F-Gas GWP ceiling, screen molecules that have never been made, then run a real compressor on the result.
About this post — Authored by an AI assistant using DigiEntropy's polynomial corpus, the universal compressor predictor, and the project's chart-generation tooling. Charts are produced by Python scripts that read the same database the live site queries; tables and formulas are pulled from the same engineering modules. Findings reflect the project's current dataset and methodology — send corrections or deeper questions to admin@digientropy.com.
A refrigerant is not a brand. It is about five numbers — molar mass, critical temperature, critical pressure, the acentric factor and the molecular heat capacity — and corresponding-states physics builds most of the rest on top of them. That is the claim behind the lab at /god-refrigerants. This post tests it with six case studies instead of describing it.
Two of them check the lab against the real world: we rebuild two refrigerants the industry already uses from their pure ingredients and compare the answers with the published values. The other four put the lab to work: a natural-refrigerant trade, a search for a blend under an F-Gas GWP ceiling, a screen of molecules that have never been made, and a real catalogue compressor run on a fluid it was never rated for.
25 - pure molecules in the palette - plus 12 reference blends
The palette is 25 pure molecules — 11 HFCs, 8 naturals, 4 HFOs, one HCFO and one halogenated — plus 12 reference blends to compare against.
0.3 K - R454B critical-temperature gap - rebuilt from two molecules
Built from pure components, R454B's critical temperature comes out at 78.28 °C against the catalogue entry's 78.0 °C. The critical point is the real test, because it comes out of the mixture equation of state, not from a lookup.
A2L - what the arithmetic calls R448A - the burn-tested blend is A1
R448A contains flammable R32, R1234yf and R1234ze(E), all of them A2L, so the arithmetic says A2L. The real R448A, burn-tested as a finished mixture, is classified A1.
14 % - COP lost by a quarter-R32 blend - against R1234yf, and 15 % against R32
A quarter-R32 blend is 14 % less efficient than R1234yf and 15 % less efficient than R32, the two fluids it is made from.
15 of 117 - compositions met the F-Gas brief - GWP ceiling 300, class A2L, COP 2.5
117 compositions searched, 15 met the brief. Tighten the ceiling to 150 and 5 of the 117 survive.
0.69 % - evaporating-pressure error from five numbers - median over 18 fluids
Saturation pressure from five numbers, to within 0.7 % of the reference equation of state. That is the corresponding-states principle measured, not asserted.
…