Design of metamodels of microcomponents to develop a microfluidic system and its optimization

Abstract

This study presents a metamodel-based optimization framework specifically designed for a microfluidic system comprising two valveless micropumps and a T-shaped micromixer. Comprehensive high-fidelity simulations were conducted employing Computational Fluid Dynamics (CFD) grounded in the Finite Element Method to meticulously analyze the behavior of individual components. A regression-based metamodel was developed for the micromixer, while a Response Surface Methodology (RSM) was employed for the micropump. These component-level metamodels were subsequently integrated into a Simulink environment to establish a system-level model, facilitating efficient predictions of critical performance parameters, including flow rate, pressure drop, and mixing efficiency. Optimization of the system was executed utilizing a Genetic Algorithm, wherein the frequency and pressure amplitude of the micropump diaphragm were systematically varied. A collection of Pareto-optimal solutions was identified, with the optimal configuration achieving a mixing efficiency of 91.94% and a pressure drop of 3036.6 Pa at an actuation frequency of 25.3 Hz and an amplitude of 2.52 kPa. The results obtained from the Simulink-based model were validated against CFD simulations and further corroborated through experimental testing of a microfluidic prototype fabricated via soft lithography. This study underscores the efficacy of metamodeling and evolutionary algorithms in optimizing microfluidic systems at the system level and emphasizes the significant interdependence between micropump actuation and micromixing performance.