A novel approach to design and fabricate foams with optimized fluid flow in porous media by combining the methods of computational fluid dynamics, machine learning and additive manufacturing

This study introduces an innovative approach integrating computational fluid dynamics (CFD), machine learning (ML), and additive manufacturing (AM) to design tailored foams for diverse industrial applications. To achieve this objective, CFD simulations were employed to analyze the impact of structural properties from 217 different foam types on fluid flow, utilizing air at varying velocities at a temperature of 300 K. The Voronoi tessellation method detailed foam structures, while pressure gradients related to fluid velocity were computed using flow coefficients. ML techniques were employed to predict these coefficients, lowering computational expenses significantly. Four algorithms were executed sequentially to identify the recommended foam types that meet the specified operational conditions, including pressure drop and fluid inlet velocity as defined by the user. The results of this approach demonstrate that the proposed method can effectively suggest suitable foams with specific structural parameters, including porosity (65–85%), pore diameter (0.2 to 1.2 mm), and strut diameter, based on the provided pressure drop and fluid inlet velocity. The method enables rapid prototyping and manufacturing of foams through AM, significantly reducing costs and maintaining high accuracy with an error margin of less than 6%, compared to traditional trial-and-error approaches. It also accelerates the design process, enhancing productivity and facilitating faster iterations to meet the evolving demands of the market. In conclusion, this approach enables customized foam designs for various industries, improving catalysts, heat exchangers, and filtration systems. Additionally, it fosters innovation and addresses specific challenges in energy, environmental engineering, and manufacturing.