Machine learning-based prediction of the coefficient of performance for low global warming potential refrigerants in a vapor compression system
DOI:
https://doi.org/10.15282/ijame.23.1.2026.19.1017Keywords:
Low-GWP refrigerant, Machine learning, Thermodynamic calculation, Refrigerant system , Vapor compression systemAbstract
This study addresses the energy significance of the coefficient of performance (COP) in vapor compression systems and the practical need to forecast COP quickly and reliably. Because COP directly reflects the amount of cooling delivered per unit of input power, accurate prediction supports energy savings, refrigerant selection, and early-stage design decisions, especially for low global warming potential (GWP) refrigerants. Authors develop data-driven models to estimate COP without full thermodynamic calculations. A synthetic dataset of 2,000 samples is generated in the Engineering Equation Solver (EES) for four refrigerants (R1234yf, R134a, R290, R600a) by using five inputs: refrigerant type, evaporation temperature, condensing temperature, subcooling, and superheat. Five supervised learning algorithms are trained and compared: linear regression, polynomial regression, random forests, decision trees, and support vector machines. The study evaluates model performance using the coefficient of determination (R²), root mean squared error (RMSE), and mean absolute error (MAE) based on an 80/20 train/test split. Results show Polynomial Regression (degree 3) delivers the highest accuracy (R² ≈ 0.9999; RMSE ≈ 0.0071; MAE ≈ 0.0053), with Random Forest as the next strongest baseline. The findings suggest that lightweight, well-tuned regressors can provide fast, precise COP predictions, reduce analysis time, and guide system design and parameter optimization. The approach offers an accessible tool for engineers seeking efficient, low-carbon refrigeration solutions.
References
[1] M. Mshragi, I. Petri, “FastML-GA: FPGA-accelerated machine learning for real-time energy HVAC optimization in buildings,” Neural Computing and Applications, vol. 38, no. 25, pp. 1–18, 2026.
[2] D. Zhou, Z. Zhu, “A digital twin model for cop prediction in refrigeration system using combined machine learning method, ”In IEEE Electrical Power and Energy Conference, Victoria, BC, Canada, pp. 412-417, 2022
[3] O. Pektezel, H.I. Acar, ”Experimental comparison of R290 and R600a and prediction of performance with machine learning algorithms, ” Science and Technology for the Built Environment, vol. 29, pp. 1-17, 2023
[4] H. Wang, D. Mai, Q. Li, Z. Ding, “Evaluating machine learning models for HVAC demand response: the impact of prediction accuracy on model predictive control performance,” Buildings, vol. 14, no. 7, pp. 1-17, 2024
[5] Q. Deng, Z. Chen, W. Zhu, Z. Li, Y. Yuan, W. Gui, “A performance prediction method for on-site chillers based on dynamic graph convolutional network enhanced by association rules,” Building Simulation, vol. 17, pp. 1213–1229, 2024.
[6] T.D. Nguyen, H.M Bui “Machine learning for fault detection in refrigeration systems using difluoromethane refrigerant,” Sustainable Chemistry One World, vol. 5, art. no. 100085, 2025.
[7] H.U. Cho, Y. Nam, E.J. Choi, Y.J. Choi, H. Kim, S. Bae, J.W. Moon, “Comparative analysis of the optimized ANN, SVM, and tree ensemble models using Bayesian optimization for predicting GSHP COP, ” Journal of Building Engineering, Vol. 44, p. 103411, 2021
[8] A. Senthilkumar, “Prediction and performance enhancement of R600a refrigerator adopting machine learning algorithms,” Research on Engineering Structures & Materials, vol. 11, pp. 3207-3223, 2025.
[9] F.W. Yu, W.T. Ho, C.F.J. Wong, “Improved energy management of chiller system with AI-based regression,” Applied Soft Computing, vol. 150, p. 111091, 2024
[10] M. Ghanbarpour, A. Mota-Babiloni, P. Makhnatch, B.E. Badran, J. Rogstam, R. Khodabandeh, “ANN modeling to analyze the R404A replacement with the low GWP alternative R449A in an indirect supermarket refrigeration system, ” Applied Sciences, vol. 11, no. 23, 2021
[11] R. Babu, R. Srikanth, “Design and analysis of different material to increase the performance of AC condenser,” International Journal of Research, vol. 3, no. 11, pp. 518–547, 2016.
[12] O. Altuntas, M.Z. Sogut, E. Yalcin, T.H. Karakoc, “Assessment of thermodynamic and environmental performances in subcooling process for different refrigerants, ”International Journal of Exergy, vol. 24, no. 2–4, pp. 216–234, 2017.
[13] P.W. Sunu, D.A. Simon, I.M. Rasta, Y.B. Lukiyanto, I.D.M.Susila, I.K. Suarsana, “Effect of various superheat condition on working condition of air-cooled chiller, ”International Conference on Applied Science and Technology, 2018.
[14] R. Dharmkar, J. Yeboah, I. Nti, “Smart temperature management in buildings using predictive analysis by machine learning algorithms,” International Conference on Computational Science and Computational Intelligence, pp. 284-290, 2023
[15] J. Marchante-Avellaneda, J.M. Corberan, E. Navarro-Peris, S.S. Shrestha, “A critical analysis of the AHRI polynomials for scroll compressor characterization,” Applied Thermal Engineering, vol. 219, p. 119432, 2023
[16] K. Nong, H. Zhang, Z. Liu, “Comparative study of different machine learning models for heat transfer performance prediction of evaporators in modular refrigerated display cabinets, ” Energies, vol. 17, no. 23, 2024
[17] A. Dhakar, B. Singh, P. Gupta, “Fault diagnosis of air compressor set-up using decision tree-based J48 classification algorithm,” Journal of Engineering Research, vol. 12, pp. 1011-1025, 2024.
[18] W. Tun, J.K. Wong, S. Ling, “Hybrid random forest and support vector machine modeling for HVAC fault detection and diagnosis, ” Sensors, vol. 21, no. 24, p. 8163, 2021
[19] A. Ivanov, S.M. Ailuro, “TMPNN: High-order polynomial regression based on Taylor map factorization, Proceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 11, pp. 12726-12734, 2024
[20] E. Halabaku, E. Bytyçi, E., “Overfitting in machine learning: a comparative analysis of decision trees and random forests, ” Intelligent Automation and Soft Computing, vol. 39, no. 6, pp. 987–1006, 2024.
[21] A. Al-Rashed, “Effect of evaporator temperature on vapor compression refrigeration system,” Alexandria Engineering Journal, vol. 50, pp. 283-290, 2011
[22] M.A. Ismael, S.G. Yahya, M. Azhar, I.M. Mahbubul, O. Ilori, “Improving the efficiency of a conventional vapor-compression refrigeration system used for geothermal and evaporative cooling techniques: case study in Iraq,” DJES, vol. 16, no. 1, pp. 15–26, 2023
[23] P. Singha, C. Das, M.S. Dasgupta, M.K. Sahu, “New approach to improve COP and heat recovery in transcritical CO2 refrigeration system for milk processing application,” Scientific Reports, vol. 15, art. no. 5502, 2025.
[24] M. Ma, O. Pektezel, V. Ballerini, P. Valdiserri, E.R. di Schio, “Performance predictions of solar-assisted heat pumps: methodological approach and comparison between various artificial intelligence methods,” Energies, vol. 17, no. 22, p. 5607. 2024.
[25] X. Ye, J. Lin, Y. Liu, C. Zhuo, S. Xiong, J. Zhao, et al., “Performance prediction of R290 refrigerant scroll compressors based on multitask Lasso-XGBoost algorithm,” Applied Thermal Engineering, Vol. 285, pp. 129285, 2026.
[26] S. Akyol, M. Das, B. Alatas, “Modeling the energy consumption of R600a gas in a refrigeration system with new explainable artificial intelligence methods based on hybrid optimization,” Biomimetics, vol. 8, pp. 1-20, 2023.
[27] A.M. Ashour, A.M. Al Jubori, A. Mota-Babiloni, H.M. Ali, H.M. Jaffar, A. Bouabidi, “Performance assessment of a hybrid vapor compression-thermoelectric refrigeration system using machine learning models,” Energy Conversion and Management, vol. 348, art. no. 120627, 2026.
[28] A. Senthilkumar, A., “Enhancing COP in R600a refrigerators by dispersing CuO/SiO2 Nanolubricants: A machine learning prediction,” Jordan Journal of Mechanical and Industrial Engineering, vol. 19, no. 3, pp. 547–556, 2025.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 The Author(s)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
