IIR document

Accurate prediction of thermodynamic properties of R1234yf refrigerant using Gaussian process regression models.

Author(s) : YANG G., XU Q., LIU N., WU W., ZHOU M., WANG X.

Type of article: IJR article

Summary

In response to global phase-down efforts against high global warming potential (GWP) refrigerants under the Kigali Amendment, R1234yf is recognized as a promising low-GWP alternative for air conditioning applications. However, the accurate prediction of its thermophysical properties is hindered by scarce experimental data and the computational expense of traditional methods. This study develops efficient Gaussian process regression (GPR) models to predict six key properties of R1234yf. GPR is a Bayesian non-parametric method that uses a kernel function to model covariance between sample data points and provides predictions with inherent uncertainty. Based on a systematic evaluation, the optimal kernels for the six thermal properties are as follows: the
Mat´ern 5/2 kernel for viscosity, saturated vapor pressure, and density; the squared exponential (SE) kernel for specific heat capacity; and the Mat´ern 3/2 kernel for surface tension coefficient and thermal conductivity. For the present evaluation, the GPR model achieves R² values above 0.97, NRMSE below 4.6 %, and AARD less than 3 %, showing performance on par with or slightly better than the NIST REFPROP and the Peng–Robinson (PR) equation of state on this specific data collection. Compared with the data-driven deep feedforward neural network (DNN) model, GPR demonstrates comprehensive advantages. This work offers valuable tools for GPRbased property evaluation, thereby supporting the design of high-efficiency and sustainable refrigeration systems.

Available documents

Format PDF

Pages: 14 p.

Available

  • Public price

    20 €

  • Member price*

    Free

* Best rate depending on membership category (see the detailed benefits of individual and corporate memberships).

Details

  • Original title: Accurate prediction of thermodynamic properties of R1234yf refrigerant using Gaussian process regression models.
  • Record ID : 30034851
  • Languages: English
  • Subject: Technology
  • Source: International Journal of Refrigeration - Revue Internationale du Froid - vol. 186
  • Publication date: 2026/06
  • DOI: http://dx.doi.org/https://doi.org/10.1016/j.ijrefrig.2026.106907

Links


See other articles in this issue (27)
See the source

Indexing