(301a) Experimentally Guided Time Series Models for Carbon Nanofibers Mixed Matrix Membranes for Enhanced Water/Salt Permselectivity

Processes involving liquid-vapor phase change mechanisms such as membrane distillation rely on solid-liquid interface with superior interfacial characteristics such as porosity, wettability, surface tension and adsorption. The talk will outline a systematic approach to designing efficient membranes for separation processes by using predictive models to forecast water vapor flux and salt rejection rate datasets (N= 434), thereby predicting long-term membrane stability and performance. The focus will be on the influence of carbon nanofibers mixed matrix membranes on membrane properties and desalination performance. The introduction of just 1 wt% Cu/CNF resulted in enhanced contact angle hysteresis and hierarchical morphologies, providing a pathway for water vapor escape and recording a 64% increase in water vapor flux compared to PVDF membranes, while maintaining a 99.9% salt rejection rate. The study highlights the superior accuracy of statistical time series models in predicting the performance of the membrane distillation process, while noting the limitations of deep learning algorithms in this context. The research suggests further exploration of alternative validation methods and additional models to enhance prediction accuracy. The findings could lead to more efficient membrane distillation processes and provide valuable insights for future research in developing precise models for predicting membrane performance.