4.7 Article

Experimental investigation and machine learning modelling of phase change material-based receiver tube for natural circulated solar parabolic trough system under various weather conditions

Journal

JOURNAL OF THERMAL ANALYSIS AND CALORIMETRY
Volume 148, Issue 14, Pages 7101-7124

Publisher

SPRINGER
DOI: 10.1007/s10973-023-12219-9

Keywords

Parabolic trough collector; Phase change material; Hot water generation; Thermal energy storage; Machine learning models

Ask authors/readers for more resources

This research evaluates the performance of a novel parabolic trough collector (PTC) with a naturally circulated open loop under different classes of solar radiation data. The use of phase change material (PCM) in the receiver tubes improves the overall thermal efficiency compared to bare tubes. Additionally, machine learning models, particularly the Gaussian process regression (GPR) model, can accurately predict the instantaneous thermal efficiency of the developed system under different solar radiation classes.
This research demonstrates the performance of a novel parabolic trough collector (PTC) that uses a naturally circulated open loop, and it is evaluated experimentally for several classes of solar radiation data. A concentric absorber tube packed with phase change material (PCM) is used in the development of an indoor PTC system. A comparative study has been performed for both bare and PCM-based receiver tubes for various classes of solar radiation data. According to the findings, the overall thermal efficiency of the naturally circulated PTC system in bare receiver tubes is 0.66, 0.57, 0.63, 0.64, 0.66, and 0.63, respectively, at classes 2, 3, 5, 6, 8, 9, and 10. Meanwhile, the overall thermal efficiency of PCM-based receiver tubes is 0.70, 0.62, 0.69, 0.65, 0.68, and 0.66, respectively. However, the instantaneous thermal efficiency of the developed system at various classes of solar radiation can be predicted by using different machine learning (ML) models. These models include linear regression (LR), polynomial regression (PR), decision tree (DT), random forest (RF), artificial neural network (ANN), KNN regression, and Gaussian process regression (GPR). Among all of these, the GPR model is showing higher prediction performance (i.e. RMSE = 0.0049, R-2 = 0.9977) than that of the remaining developed ML models.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available