A probabilistic approach for estimating water permeability in pressure-driven membranes

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Cellulose triacetate (CTA) membranes are widely used in water purification.
• Accurate estimation of water permeability is critical for optimizing membrane performance in reverse osmosis.
• Molecular-level understanding of water transport through membranes remains a challenge.

Purpose of the Study:

• To develop and validate a probabilistic approach for estimating water permeability in CTA membranes.
• To utilize non-equilibrium molecular dynamics (MD) simulations and hierarchical Bayesian modeling for permeability assessment.
• To predict full-scale water flux based on molecular-scale simulations.

Main Methods:

• Non-equilibrium molecular dynamics (MD) simulations were employed to model water transport across different CTA membrane configurations.
• Annealing MD simulations were used to generate diverse membrane structures.
• A hierarchical Bayesian model was applied to analyze MD simulation results and estimate membrane permeability.
• Process-level Monte Carlo simulations were used to predict full-scale water flux.

Main Results:

• Model parameters converged within a 5-ns simulation time, indicating efficient model calibration.
• The study highlights the importance of using unique structural configurations in MD simulations for realistic molecular-scale membrane properties.
• Predicted full-scale water flux closely matched bench-scale experimental measurements (order of magnitude).

Conclusions:

• The proposed probabilistic approach effectively estimates water permeability in CTA membranes.
• Molecular dynamics simulations combined with Bayesian analysis provide valuable insights into membrane transport mechanisms.
• This methodology offers a reliable pathway for predicting membrane performance and guiding material design.