Pre-registration: Does a Water-Trained Dimensionless Critical Heat Flux Model Predict R-134a? A Fluid-to-Fluid Test Against the Scaled 2006 Look-Up Table

Pre-registration of a fluid-to-fluid test, frozen on 8 October 2026 before any test article was opened. Version 3 of a Buckingham π model for critical heat flux (doi:10.5281/zenodo.23225465), fitted to water alone, is applied unchanged to R-134a using R-134a properties and its own critical pressure. It is compared with the 2006 water look-up table carried to R-134a by the Ahmad (1973) and Katto (1978) scaling maps. The primary hypothesis (F1) is that version 3's relative RMS error is lower than that of the better of the two scaled tables, on primary points within version 3's training range. F1b is that it comes within 2.0 percentage points. The test articles are Rensch et al. (2026), Pioro et al. (2001), Kim and Chang (2005) and Chun et al. (2007). On the abstract ranges seen, no Kim and Chang point can be primary, so the test may be void with fewer than 10 primary points. That is stated in advance, and the domain will not be widened. Transcription, scoring and duplicate rules are fixed in the protocol and enforced by the code, which runs once against a hashed, publicly time-stamped transcription. The predictions of all three methods at fixed conditions are published here before any data are seen. Archive CHF_F2F_prereg_2026-10-08.zip, SHA-256 a75628d93e243011ce7fafeeba66dd12df7871323e30dd4bba51a380ece40535.