Designing High-Quality-Factor Electrospun Nanofiber Air Filters with Physics-Informed Machine Learning

High filtration efficiency ($\eta$) and low pressure drop ($\Delta P$) are desirable performance parameters for an air filter, and the quality factor \rev{(QF)} is the standard figure-of-merit that balances particle capture against breathability. In this work, we present a physics-informed machine-learning framework that predicts efficiency, pressure drop, and the derived quality factor for electrospun fibrous filters. Two physics-informed neural networks \rev{(PINNs), formulated as physics-constrained networks in which the filtration laws are imposed as hard constraints rather than as differential-equation residual penalties,} are constructed: a pressure-drop network constrained by Darcy's law and an efficiency network constrained by the classical exponential log-penetration relation. By embedding these filtration laws as differentiable constraints, the framework learns effective permeability and single-fiber efficiency, enabling physically consistent prediction and optimization of the quality factor. A dataset spanning multiple polymers, fiber diameters, packing fractions, membrane thicknesses, face velocities, and filtration particle sizes is compiled from published experimental studies. The pressure-drop PINN attains $R^2=0.982$, outperforming all five purely data-driven regressors, while the efficiency PINN attains $R^2=0.984$. Systematic variation of the input parameters identifies fiber diameter as the dominant lever for improving QF, shows that the optimal QF depends on particle size, and confirms the weak dependence of QF on membrane thickness. These results indicate that physics-guided learning offers a generalizable and physically consistent route to the predictive design of high-performance, energy-efficient electrospun filters.

Authors

Institutions

Publication Details

Journal
Nanotechnology
Published
2026-11-17
DOI
https://doi.org/10.1088/1361-6528/aec2fa
Primary Topic
Aerosol Filtration and Electrostatic Precipitation
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Designing High-Quality-Factor Electrospun Nanofiber Air Filters with Physics-Informed Machine Learning

Ashutosh Mahajan, Amrutha R Menon, Rohit Ramesh Nimje, Athira S et al.
3 links
Nanotechnology
Aerosol Filtration and Electrostatic Precipitation
article

Designing High-Quality-Factor Electrospun Nanofiber Air Filters with Physics-Informed Machine Learning

Ashutosh Mahajan, Amrutha R Menon, Rohit Ramesh Nimje, Athira S, Dharshini Sankaralingam, Vignesh S
article en

Abstract

High filtration efficiency ($\eta$) and low pressure drop ($\Delta P$) are desirable performance parameters for an air filter, and the quality factor \rev{(QF)} is the standard figure-of-merit that balances particle capture against breathability. In this work, we present a physics-informed machine-learning framework that predicts efficiency, pressure drop, and the derived quality factor for electrospun fibrous filters. Two physics-informed neural networks \rev{(PINNs), formulated as physics-constrained networks in which the filtration laws are imposed as hard constraints rather than as differential-equation residual penalties,} are constructed: a pressure-drop network constrained by Darcy's law and an efficiency network constrained by the classical exponential log-penetration relation. By embedding these filtration laws as differentiable constraints, the framework learns effective permeability and single-fiber efficiency, enabling physically consistent prediction and optimization of the quality factor. A dataset spanning multiple polymers, fiber diameters, packing fractions, membrane thicknesses, face velocities, and filtration particle sizes is compiled from published experimental studies. The pressure-drop PINN attains $R^2=0.982$, outperforming all five purely data-driven regressors, while the efficiency PINN attains $R^2=0.984$. Systematic variation of the input parameters identifies fiber diameter as the dominant lever for improving QF, shows that the optimal QF depends on particle size, and confirms the weak dependence of QF on membrane thickness. These results indicate that physics-guided learning offers a generalizable and physically consistent route to the predictive design of high-performance, energy-efficient electrospun filters.

Nanotechnology
Vellore Institute of Technology University (IN)
Aerosol Filtration and Electrostatic Precipitation
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon