Neural network modeling of bioconvection flow in a curved-corner enclosure with magnetic potential


Gürbüz Çaldağ M., Pekmen B., Oztop H.

Engineering Analysis with Boundary Elements, cilt.187, 2026 (SCI-Expanded, Scopus) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 187
  • Basım Tarihi: 2026
  • Doi Numarası: 10.1016/j.enganabound.2026.106746
  • Dergi Adı: Engineering Analysis with Boundary Elements
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, MathSciNet, zbMATH, DIALNET
  • Anahtar Kelimeler: Curvy boundary, Fe3O4-water nanofluid, Induced magnetic field, Magnetotactic bacteria, Neural networks
  • TED Üniversitesi Adresli: Evet

Özet

This study investigates the magnetohydrodynamic bioconvection of magnetotactic bacteria inside Fe3O4-water nanofluid within a differentially heated cavity having a curved bottom-left corner. The mathematical model incorporates the coupled effects of thermal and solutal buoyancy, microorganism swimming behavior, magnetic field induction, and nanoparticle presence. The governing dimensionless equations are solved numerically using a radial basis function method. The dimensionless parameter investigations reveal that increasing the thermal Rayleigh number enhances heat transfer by 73.46%, while bioconvection effects provide additional flow enhancement of 15.49%. The Peclet number variation demonstrates that faster-swimming bacteria dramatically increase microorganism accumulation in iron-rich regions. Higher Lewis numbers create sharper concentration gradients, enhancing mass transfer by 228.59%. The Hartmann number increase suppresses convection by 44.49%, while magnetic Reynolds number effects remain modest at 2.47%. The buoyancy ratio significantly influences flow direction and intensity, with opposing buoyancy forces reducing heat transfer by 48.7%. Notably, the cavity geometry parameter demonstrates that sharper corners enhance transport phenomena by 31.16% compared to smoother geometries. A neural network (NN) model for important problem indicators, average Nusselt, Sherwood and bacteria density along the left heated wall, is also developed. Mean squared error metric results on test data is found less than 0.003 using trilayer NN.