We designed the best refrigerant - and found out why it does not exist
By DigiEntropy Engineering · 2026-09-06 · 19 min read
Five constraints fixed in advance, 44 fluids and a 240-composition blend search screened against them, and zero survivors. Relax safety alone and the survivors appear - and the molecule that would win is already on the map.
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.
The companion to this post designs a compressor. This one designs the fluid — and it does not end the way we expected, because the honest answer to "what is the best refrigerant?" turns out to be "none of them, and here is exactly why."
That is a more useful answer than a winner, so we kept it.
0 of 44 - fluids meet all five constraints - every fluid with a computed cycle at this duty
Forty-four of them carry a computed cycle at this duty. Running all five constraints against all forty-four gives a single number: Zero fluids meet all five.
0 of 240 - blend compositions pass - across all 84 ingredient subsets
Zero of 240 compositions met all five constraints, across all 84 two- and three-ingredient subsets of the pool, at two duties each.
15 - near-misses - each meets exactly four of the five
Fifteen of the other 43 meet exactly four of the five constraints. Count the failure modes and the shape of the problem appears: capacity blocks 6 fluids, GWP blocks 4, safety blocks 4, efficiency blocks 1.
27 - compositions once flammables are in - relaxing GWP, glide or capacity finds none
Only safety moves it. Relax GWP, glide or capacity, one at a time, and the campaign still finds nothing. Let flammable fluids in, and it finds 27 compositions.
+12.8 % - R1270 COP over R404A - with +9.6 % capacity and GWP 2
Ideal-cycle COP: 3.087 for R404A, 3.481 for R1270, +12.8 %. More capacity and more efficiency and zero glide, at 0.05 % of the GWP, with pressures close to R404A's and a discharge temperature 6.6 K higher.
8,068 - property combinations qualify - 12 % of 66,912 clean points
4,056 are not subcritical cycles at this duty and 54,768 raise a correlation-validity flag, which leaves 66,912 clean points. 8,068 of them qualify, 12 %. The feasible region is wide, not a needle.
18 % - machine and fluid together - 8 % from each
8 % from the machine, 8 % from the fluid, 18 % together. Be careful what that proves: near-additive gains are what the model's structure delivers. It is not an independent test of anything.
What "best" means here, stated before we start
…