A facile computational framework for predicting dynamic filtration performance of compressed coarse filter media using geometric compression and numerical simulation

The use of coarse, fibrous filter media as prefilters is an ideal approach to achieving high filtration efficiency, low pressure drop, and large dust-holding capacity in household Heating, Ventilation, and Air Conditioning (HVAC) systems, cleanrooms, and automotive/aircraft cabins. However, these coarse filter media are prone to compression under stress, and the compression-induced structural and performance changes have not been fully addressed. In this work, a facile computational framework was proposed to predict the performance of compressed fibrous media. Synchrotron radiation X-ray imaging was used to acquire the 3D filter structure. An image-processing-based algorithm was then applied, enabling the generation of compressed 3D structures within seconds. In contrast, the conventional mechanical simulations often require additional mechanical input parameters. A high-fidelity simulation of dynamic filtration performance under uncompressed conditions was achieved by determining the key parameters of the Hamaker collision model. Then, performance predictions were conducted utilizing these parameters and geometrically compressed 3D filter structures. The simulated results were in good agreement with experimental measurements. The generated digital twin structure was also validated for performance prediction. Furthermore, as the compression ratio increased from 0% to 50%, the D₅₀ pore size reduced from 73.53 μm to 53.62 μm, the initial filtration efficiency slightly reduced from 75.02% to 68.99%, the pressure drop increased from 6.7 Pa to 9.7 Pa, and the dust holding capacity reduced from 235.7 g/m² to 138.0 g/m². This reported facile computational framework could be widely employed to gather data for training machine learning models.