This study aims to investigate mixed-convection heat transfer of an Al2O3 – water nanofluid in a fixed straight elliptical cooling channel representative of an actuator-integrated cooling passage and develops an artificial neural network (ANN) surrogate for rapid prediction of thermal-hydraulic performance.
A three-dimensional two-phase mixture model is solved using the finite-volume method with SIMPLEC pressure–velocity coupling. Nanoparticle slip and drift are included under local thermal equilibrium. The effects of Reynolds number, Richardson number, nanoparticle volume fraction and particle diameter are examined. A 4–16–12–2 feedforward ANN is trained using the numerical database to predict the case-level Nusselt number and Fanning friction factor.
At the representative condition , axial inertia remains dominant and buoyancy-induced transverse circulation provides a secondary contribution. Increasing nanoparticle volume fraction reduces wall temperature and increases the local Nusselt number. Decreasing particle diameter also increases the predicted Nusselt number, particularly in the developing region, while friction-factor changes remain comparatively small. The representative 70 / 15/15 ANN realization achieved testing values of 0.99957 for Nusselt number and 0.99995 for Fanning friction factor.
This study integrates a two-phase mixture model for a fixed straight elliptical robotic-actuator cooling channel with simultaneous ANN prediction of Nusselt number and Fanning friction factor across variations in , , nanoparticle loading and particle size.
