Low-cost bilayered structure for improving the performance of solar stills: Performance/cost analysis and water yield prediction using machine learning. (February 2022)

• Record Type: Journal Article
• Authors: Elsheikh, Ammar H.; Shanmugan, S.; Sathyamurthy, Ravishankar; Kumar Thakur, Amrit; Issa, Mohamed; Panchal, Hitesh; Muthuramalingam, T.; Kumar, Ravinder; Sharifpur, Mohsen
• Is Part Of: Sustainable energy technologies and assessments. Volume 49(2022)
• Publication Date: 2022-02
• Subjects: Low cost materials -- Water desalination -- Solar energy -- Solar still -- Heat localization -- Machine learning; Renewable energy sources -- Periodicals; Energy development -- Technological innovations -- Periodicals; Electric power production -- Periodicals; Energy storage -- Periodicals; 333.79
• DOI: 10.1016/j.seta.2021.101783
• Languages: English
• ISSNs: 2213-1388

Abstract

Highlights: A heat localization bilayered structure is utilized to enhance the water yield of a solar still. The cost of distilled water per liter is 0.015 $/l. The accumulated yield of the modified solar still reaches 4.99 L/m 2 /day. Three machine learning approaches were employed to predict the water yield. Support vector machine succeeded to predict the water yield with superior accuracy. Abstract: This paper aims to enhance the performance of conventional solar still (CSS) using a low cost heat localization bilayered structure (HLBS). The HLBS consists of a bottom supporting layer (SL) made of low thermal conductivity as well as low density material and a top absorbing layer (AL) made of a photo thermal material with a high sunlight absorptivity as well as an enhanced conversion efficiency. The developed HLBS helps in increasing the evaporation rate and minimize the heat losses in a modified solar still (MSS). Two similar SSs were designed and tested to evaluate SSs' performance without and with HLBS (CSS and MSS). Moreover, three machine learning (ML) methods were utilized as predictive tools to obtain the water yield of the SSs, namely artificial neural network (ANN), support vector machine (SVM), and adaptive neuro-fuzzy inference system (ANFIS). The prediction accuracy of the models was evaluated using different statistical measured. The obtained results showed that the daily freshwater yield, energy efficiency, and exergy efficiency of the MSS was enhanced by 34%, 34%, and 46% compared with that of CSS. The production cost per liter of the MSS is 0.015 $/L. Moreover, SVM outperformed other ML methods for both SSs based on different statistical measures.