Performance optimization of evacuated tube solar air heater with twisted tape by using various machine learning methods
The machine learning (ML) modelling techniques were evaluated for predicting the evacuated tube collector solar air heater with baffles and perforated twisted tape inserts (ETCSAHPTT) thermal performance. The experiments were run at flow of 100–500 kg/h, pitch ratios of 2, 2.5, and 3, and hole diameter ratios of 0.0714, 0.107, and 0.143. Output parameters such as temperature difference (TD), heat gain (HG) and thermal efficiency (TE) were taken into consideration. The five ML models (random forest (RF), multiple linear regression (MLP), support vector regression (SVR), multilayer perceptron (MLP) and gaussian process regression (GPR), ) were tested using 10-fold cross validation. The MLP model demonstrated the highest prediction accuracy, with R 2 values of 0.9996, 0.9867, and 0.9952 for TD, HG, and TE, respectively. The flow rate was the most important parameter identified using partial dependence analysis. An increase in flow rate from 100 kg/h to 500 kg/h resulted in a reduction in TD from about 45-48 °C to 15-17 °C; an increase in HG from about 1450 W to above 1800 W; and an increase in TE from about 41-42% to above 55%. Increasing the pitch ratio and hole diameter ratio resulted in lower predicted thermal performance, whereas the pumping power requirement increased nonlinearly with the operating conditions. These quantification results show that the MLP model can accurately predict the thermal performance of the studied ETCSAH in the experimental operating range.
Authors
- Tesfaye Olana Terefe (ORCID: https://orcid.org/0000-0001-6500-027X)
- Adaikalasamy Veerakumar (ORCID: https://orcid.org/0000-0001-9554-3988)
- T. Karthickmunisamy
- S. Prasath
- R. Sunilkumar
- C. Ramkumar
Institutions
- Adama Science and Technology University (ET)
Publication Details
- Journal
- Scientific Reports
- Published
- 2026-11-12
- DOI
- https://doi.org/10.1038/s41598-026-79596-7
- Primary Topic
- Heat Transfer Mechanisms
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Science and Engineering Research Board