Intelligent RKF–BLMS-ANN prediction of unsteady bioconvective nanofluid flow containing swimming gyrotactic microorganisms under HOC reaction and SOVS effects
- 1Department of Mechanical Engineering, Babol Noshirvani University of Technology, Babol, Iran; Corresponding author.
- 2Department of Mechanical Engineering, Babol Noshirvani University of Technology, Babol, Iran
- 3Department of Mechanical Engineering, Babol Noshirvani University of Technology, Babol, Iran; Department of Computer Engineering, Biruni University, Istanbul 34010, Turkey; Corresponding author. Department of Mechanical Engineering, Babol Noshirvani University of Technology, Babol, Iran.
Abstract
This study presents an intelligent numerical framework for predicting the unsteady squeezing flow of a bioconvective nanofluid containing gyrotactic microorganisms between two parallel plates. The effects of second-order velocity slip, higher-order chemical reaction, Brownian motion, thermophoresis, and bioconvective transport are incorporated into the mathematical formulation. The governing nonlinear partial differential equations are transformed into a coupled system of ordinary differential equations using suitable similarity transformations and are then solved numerically through the Runge–Kutta–Fehlberg shooting method. The obtained numerical solutions are used to construct high-quality datasets for training, validating, and testing a Back-Propagated Levenberg–Marquardt Scheme Artificial Neural Network (BLMS-ANN). The physical results reveal that the squeezing parameter significantly redistributes the velocity field and enhances core-region momentum transport, while the slip parameters strongly modify the near-wall velocity gradient and shear stress. Increasing the chemical reaction parameter suppresses nanoparticle concentration, whereas thermophoresis enhances the temperature field but reduces concentration distribution. Moreover, the Peclet number plays a dominant role in controlling microorganism transport by strengthening convective effects relative to diffusion. The developed BLMS-ANN model demonstrates excellent predictive capability, with very low validation mean squared errors, correlation coefficients of R = 1, and residual errors tightly clustered around the zero-error line. The developed BLMS-ANN model reduced the computational time from 4.1 s required by the RKF45 solver to 1.1 s for prediction, corresponding to a speed-up factor of approximately 3.73 and a computational time reduction of 73.17%. These findings confirm that the proposed intelligent framework provides a highly accurate, stable, and computationally efficient surrogate model for predicting complex nonlinear transport phenomena in bioconvective nanofluid squeezing flows.
Related
- Unlocking the Synergy of Coupled Cold Plasma and Luminous Textile Photocatalysis for Indoor Air Purification: Simultaneous Elimination of Ethyl Acetate and Microorganisms
- Artificially intelligent modeling of macromolecular assemblies
- Valorization of Agro-Industrial By-Products in the Development of Bio-Polymer Films, Biodegradable Plasticulture, and Intelligent Food-Packaging Materials
- Lipid Matrix-Dependent Effects of Ceramide Enrichment on Cholesterol-Containing PSM and DLPC Membranes
- Essential Oil-Containing Mouthrinses for Plaque and Gingivitis Control: Formulation Chemistry, Human Salivary Biomarkers, Mechanistic Evidence, and Research Priorities
BibTeX
@article{mahboobtosi2026intelligent,
title = {{Intelligent RKF–BLMS-ANN prediction of unsteady bioconvective nanofluid flow containing swimming gyrotactic microorganisms under HOC reaction and SOVS effects}},
author = {Mehdi Mahboobtosi and Ali Gholami and Fateme Nadalinia Chari and Davood Domiri Ganji},
journal = {Results in Surfaces and Interfaces},
year = {2026},
month = oct,
doi = {10.1016/j.rsurfi.2026.101022},
url = {https://doi.org/10.1016/j.rsurfi.2026.101022},
note = {CC BY},
}
RIS
TY - JOUR
TI - Intelligent RKF–BLMS-ANN prediction of unsteady bioconvective nanofluid flow containing swimming gyrotactic microorganisms under HOC reaction and SOVS effects
AU - Mehdi Mahboobtosi
AU - Ali Gholami
AU - Fateme Nadalinia Chari
AU - Davood Domiri Ganji
PY - 2026
DA - 2026/10/30
JO - Results in Surfaces and Interfaces
DO - 10.1016/j.rsurfi.2026.101022
UR - https://doi.org/10.1016/j.rsurfi.2026.101022
AB - This study presents an intelligent numerical framework for predicting the unsteady squeezing flow of a bioconvective nanofluid containing gyrotactic microorganisms between two parallel plates. The effects of second-order velocity slip, higher-order chemical reaction, Brownian motion, thermophoresis, and bioconvective transport are incorporated into the mathematical formulation. The governing nonlinear partial differential equations are transformed into a coupled system of ordinary differential equations using suitable similarity transformations and are then solved numerically through the Runge–Kutta–Fehlberg shooting method. The obtained numerical solutions are used to construct high-quality datasets for training, validating, and testing a Back-Propagated Levenberg–Marquardt Scheme Artificial Neural Network (BLMS-ANN). The physical results reveal that the squeezing parameter significantly redistributes the velocity field and enhances core-region momentum transport, while the slip parameters strongly modify the near-wall velocity gradient and shear stress. Increasing the chemical reaction parameter suppresses nanoparticle concentration, whereas thermophoresis enhances the temperature field but reduces concentration distribution. Moreover, the Peclet number plays a dominant role in controlling microorganism transport by strengthening convective effects relative to diffusion. The developed BLMS-ANN model demonstrates excellent predictive capability, with very low validation mean squared errors, correlation coefficients of R = 1, and residual errors tightly clustered around the zero-error line. The developed BLMS-ANN model reduced the computational time from 4.1 s required by the RKF45 solver to 1.1 s for prediction, corresponding to a speed-up factor of approximately 3.73 and a computational time reduction of 73.17%. These findings confirm that the proposed intelligent framework provides a highly accurate, stable, and computationally efficient surrogate model for predicting complex nonlinear transport phenomena in bioconvective nanofluid squeezing flows.
SN - 2666-8459
ER -