Scale‐up of extraction processes: Dimensionless modeling and virtualization via lattice Boltzmann method

Abstract

AbstractWhile modeling and virtualization pose challenges and opportunities to food engineering, lattice Boltzmann method (LBM) is a recent computational technique. By relying on existing dynamic one‐dimensional models, in‐house LBM simulators have been developed toward extraction processes in agitated‐batch or continuous‐flow equipment. In view of supporting process scale‐up, this work aimed at casting the aforesaid extraction models (and related LBM simulators) in dimensionless form. Together with available experimental data for validation, pectin extraction from fruit peels in agitated reactor and essential oil extraction from gorse in fixed bed were taken as case studies. Primitive influencing parameters (e.g., intraparticle diffusivity, solid–fluid partition coefficient, fluid‐phase diffusivity, bed length, and interstitial fluid velocity) were fruitfully lumped into fewer dimensionless parameters (e.g., mass‐transfer Biot, Péclet and Damköhler numbers). In‐house LBM simulators remained operational in their dimensionless versions, with a reduced amount of dimensionless numbers indeed grouping the influence of different process parameters, which may help future sensitivity analyses and scale‐up procedures.Practical ApplicationsInitially used for academic purposes, computational modeling has experienced notable evolution and has become strategic to research, development, and innovation (RD&I) in food processing. Computational modeling can support novel food processes as different parameters and scenarios can be quickly simulated toward scale‐up and cost‐effectiveness. Accordingly, computational modeling may save valuable material, energy, water, and human resources in RD&I activities at industrial scale. The present work casts in‐house computer codes to numerically simulate extraction processes in either agitated‐batch or continuous‐flow equipment into dimensionless form. Related process parameters in primitive form were fruitfully lumped into fewer dimensionless numbers, which may ease scale‐up as well as sensitivity analysis.