We designed the best semi-hermetic recip we could defend - here is the spec
By DigiEntropy Engineering · 2026-09-06 · 19 min read
Eleven physics numbers, 541 real machines, one duty profile - and a definition of best fixed before we started. The biggest single gain turned out to be the one number no datasheet argues about.
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.
We have been arguing that a compressor is about ten fitted numbers, eleven dials once the motor's rating is counted. This post does the obvious next thing: treat those numbers as design variables and build the best semi-hermetic reciprocating compressor we can defend.
Not the best-sounding one. The best one we can defend — which means the hard part of this post is not the physics, it is the definition.
8,315 - machines fitted in the God Theory lab - 6,147 of them trusted fits
The design space is not a blank sheet. It is the 8,315 machines already fitted in the God Theory lab, and the Atlas is where you look at all of them at once.
541 - trusted reciprocating fits, 300–900 cm³/rev - from nine brands
Narrowing to what we are designing — trusted reciprocating fits between 300 and 900 cm³/rev — leaves 541 machines from nine brands.
26 % - best vs worst published machine - same size class, same duty
The best published machine in this size class is 26 % more efficient than the worst, at the same duty.
82 % - motor efficiency pinned at 0.98 - an optimiser bound, not a measurement
Four of five are bounds. The motor number is the worst of them: 82 % of the population is pinned at exactly 0.98, because 0.98 is where the optimiser stops, not where the motors are.
+8.2 % - composite COP vs the 4NE-20Y - +14.2 % capacity, same displacement
Same displacement as the 4NE-20Y, same duty, the lab's physics on both sides: COP 2.859 against 2.643 (+8.2 %), cooling capacity 32.96 kW against 28.86 kW (+14.2 %).
+9.5 % - from the part-load de-rating alone - the largest single move, by a factor of four
Changing one number at a time from the real machine, the part-load de-rating (0.3039 to 0.0838) takes the COP to 2.895, +9.5 %. The single largest available gain, by a factor of four, is the part-load behaviour of the motor.
9,400 - EUR over 15 years, roughly - what the 8.1 % seasonal gain is worth
…