Speaker
Description
Filtration flows through nanoporous membranes play a crucial role in a range of cutting-edge technologies, including water purification, osmotic power generation, and targeted drug delivery. Molecular dynamics simulations are currently considered the state-of-the-art approach for modeling nanofiltration processes.
However, their high computational cost makes simulating large-scale filtration systems impractical, limiting the ability to conduct extensive parametric studies and optimize design strategies.
In the present contribution, we merge a molecular analysis of the nanofiltration problem with a homogenization technique [1], to upscale the filtration flow from the single nanopore description to the whole membrane scale phenomenon. The homogeneous model employed [1] allows replacing the detailed description of the flow through the whole membrane (figure 1a) with a simplified flow description, where the membrane is
a fictitious smooth interface (the red surface of figure 1c) between two macroscopic fluid regions. The model rigorously quantifies jumps in the macroscopic solvent velocity and stresses across the homogeneous membrane $\mathcal{C}$ via a set of tensorial quantities computed once and for all via characteristic problems at the pore-scale for given membrane properties. In [1], these quantities solve Stokes problems within the periodic microscopic domain of figure 1b.
We downscale the model by replacing the microscale Stokes problems with molecular dynamics simulations, enabling us to predictively quantify the membrane properties in the presence of nanoscopic pores. Such confined regions are indeed challenging for continuum mechanics: the usual concepts of density and viscosity, for example, are not well posed at these scales. The use of molecular dynamics is thus the only means to ensure the physical phenomena occurring at those scales are accurately reproduced. Finally, we validate the model by comparing it to molecular dynamics simulations of water flow through arrays of pores. Additional strategies for reducing total computational costs while preserving the predictive power of the molecular-homogeneous model are also discussed.
| References | [1] Zampogna et al., Journal of Fluid Mechanics 970, A39 (2023) |
|---|---|
| Country | Italy |
| Acceptance of the Terms & Conditions | Click here to agree |








