Integrating machine learning and experimental analysis for nanofluid-enhanced heat transfer in additively manufactured microchannel heat sinks

Abstract The rising thermal load in modern microelectronic devices necessitates compact cooling solutions with improved heat removal capability. This work examines the thermo-hydraulic performance of straight and secondary-wavy microchannels (SWMC), additively manufactured in AlSi10Mg using Direct Metal Laser Sintering (DMLS). Experiments were conducted under heat fluxes of 20–40 W/cm 2 , and the flow rates of 100–475 ml/min (Re = 25–123) using a water-ethylene glycol base fluid and Al 2 O 3 /CuO nanofluids at 0.02 and 0.05% concentrations. The nanofluids enhanced convective performance, yielding up to 12.35% improvement in straight channels and 16.97% in SWMC at 30 W/cm 2 and Re = 123, with the wavy geometry consistently offering superior heat transfer due to curvature-induced secondary flows. In parallel, five machine-learning models were developed to predict wall temperature and Nusselt number; among them, the Gradient Boosting model provided the closest agreement with experimental data. The findings highlight how additively manufactured microchannel geometries, nanoparticle-enhanced coolants, and data-driven predictive tools can be jointly leveraged to advance thermal management in high-power electronic applications.