Artificial intelligence-assisted characterization and optimization of red mud-based nanofluids for high-efficiency direct solar thermal absorption

Abstract

The utilization of nanofluids (NFs) holds promise for enhancing the thermal efficiency of solar thermal collectors. Among the various NF solutions, red mud (RM) NFs have gained attention due to their effective absorption of solar thermal energy. RM comprises precious metal oxides, making it a proficient medium for direct solar heat absorption. This study aimed to formulate waterbased RM NFs with concentrations ranging from 0.1 to 0.75 vol%. Within the temperature range of 303–333 K, we assessed the specific heat (SH), viscosity (VST), and thermal conductivity (TC) of the NFs. To maintain stability, we employed polyvinylpyrrolidone (PVP) surfactant. The results indicated that the SH of RM NFs is lower than that of water. Additionally, as RM NF concentrations increased, there was a significant improvement in TC. The highest TC enhancement of 36.9 % is observed at 333 K for a concentration of 0.75 vol% compared to water. Based on the gathered data, unique equations were developed to estimate the properties of RM NFs within the studied range. Our findings suggest that RM NFs have the potential to effectively replace water in solar energy applications. Furthermore, we employed innovative ensemble-type machine learning (ML) techniques, namely Adaptive Boosting (AdaBoost) and random forest (RF), to address the problem. We also utilized these novel ML methods to construct metamodels for predicting the considered properties, offering accurate and efficient models for analyzing NF behavior. The incorporation of RM in solar thermal applications could contribute to resolving disposal challenges associated with this waste material, thereby aiding in its long-term management.