Artificial neural network-based meshfree simulations for magneto-bioconvection flow of nano-enhanced phase change materials inside a partly porous circular cylinder

Magneto-bioconvection of nano-enhanced phase change materials with oxytactic micro-organisms is analyzed in a partially porous circular cylinder containing four rotating wavy inserts. An incompressible smoothed particle hydrodynamics (ISPH) solver provides flow–thermal–species fields, and a compact artificial neural network (ANN) predicts the mean Nusselt and Sherwood numbers for rapid parametric exploration. The study spans the Rayleigh number Ra=103–106 and the bioconvection Rayleigh number Rab=103−105, the Hartmann number Ha=0–100, the Darcy number Da=10−2–10−5, the Péclet number Pe=0–5, the Lewis number Le=1–5, the fusion temperature θf=0.1–0.4, and the insert radius RC=0.1–0.3. Increasing Ra and Rab intensifies recirculation, thins thermal layers, and elevates heat and mass transfer; for example, the peak velocity rises from 0.33 at Ra=103 to 30.96 at Ra=106. Stronger magnetic damping (higher Ha) and increased porous resistance (lower Da) suppress the velocity field (e.g., a 55.56% decrease in Vmax from Ha=0 to 100). Higher Pe reduces micro-organism intensity and lowers velocity (63.16% drop from Pe=0 to 5), while θf reshapes the apparent heat-capacity peaks and the mushy region. The Lewis number primarily modulates oxygen and micro-organism distributions. The ANN reproduces ISPH trends for Nu¯ and Sh¯ with high fidelity, enabling fast design-space sweeps. These results inform compact heat-exchange, electronics-cooling, and biomedical systems where porous structures, magnetic fields, and microbe-driven transport co-regulate phase-change performance.