Flow Boiling Characteristics of Water–Al ₂ O ₃ and R134a-Al ₂ O ₃ in Multi-layered Wavy Microchannels

This study presents a numerical investigation of flow boiling in multilayered wavy microchannels using nanofluids to enhance thermo-hydraulic performance. A coupled volume of fluid-Lee phase change method and Discrete Phase Model are employed to capture complex vapor-liquid interfaces and nanoparticle dynamics. Water–aluminum oxide and R134a-aluminum oxide are examined at various volumetric concentrations, mass fluxes, and heat fluxes. Unlike prior studies on straight channels, this work investigates wavy geometries to reveal heat transfer enhancements. An artificial neural network is trained to predict Nusselt number and friction factor ratios, and predictions are optimized using a non-dominated sorting genetic algorithm. Results indicate that double-layered wavy microchannels outperform single-layer configurations, achieving a maximum Nusselt number of 16 and a performance factor of 1.406. Water–aluminum oxide exhibits higher heat transfer coefficients, while the refrigerant mixture offers lower pressure drop. The machine learning model predicts results with minimal deviation from simulations, identifying an optimal solution with a performance factor of 1.422. This work offers new insights into the synergy between microchannel geometry, nanofluid properties, and data-driven optimization.