Artificial Intelligence – Machine Learning Based Framework For Design Optimization Of Preparation Conditions And Operating Parameters Of Thin Film Nanocomposite Membranes
The applications of artificial intelligence (AI) and machine learning (ML) are gaining popularityin the midst of industrial revolution 4.0 in several industries for rapid screening of possible combinations of operating inputs to achieve highest throughput. Among other fields, recentlyAI-ML based interventions have been used in the field of membrane development [1, 2] also. In this study, a novelAI-ML framework approach is proposed for optimization of the variable inputs for preparation of thin film nanocomposite membrane and operating parameters during operations for yielding the best performance in terms of product flux and salt rejection.Using published literature data, a detailed data mining exercise was carried out classifying the inputs into various operating as well as membrane synthesis parameters including operation pressure, feed salinity,feed flow rate, support polymer, solvents, temperature, concentration of aqueous & organic reagents for thin film formation, nanoparticle properties& concentration etc. Comparative analysis of these data was carried out using different AI-ML based models and also both single objective optimization (SOO) and multi-objective optimization (MOO) were implemented to find the optimal compositions and operating conditions. It was found that carbon nanotubes and metal organic frameworkswere the best suitednanoparticles with optimum nanoparticle loading to be around 0.0087 – 0.034 wt% and under these conditions, a thin film nanocomposite membrane can give water flux of > 85 l/m2h with > 99% salt rejection at feed salinity of 11,000 ppm andtemperature of 45 - 51oC.
Keywords - Artificial Intelligence; Machine Learning; Thin Film Nanocomposite; Data mining; Membrane.