Artificial neural network-assisted prediction of Ti3C2-GO hybrid nanofluid flow over a curved surface

Purpose This study aims to investigate the hemodynamics characteristics of nanofluid and hybrid nanofluid over a curved surface. The analysis focuses on momentum and energy transport in the presence of nanoparticles and compares the numerical results with artificial neural network (ANN) results. Design/methodology/approach A Casson blood-based nanofluid (nf) and hybrid nanofluid (hnf) flow model over a curved region is considered. The nf consists of 4% MXene (Ti3C2) nanoparticles dispersed in blood, while the hnf contains 4% MXene (Ti3C2) and 4% graphene oxide (GO) nanoparticles dispersed in blood. The flow is influenced by magnetic forces and Darcy–Forchheimer porous effects, whereas the energy field is affected by radiant heat, ohmic heating and uneven heat-source/sink conditions. The mathematical model is solved numerically by the MATLAB bvp5c problem solver. In addition, an ANN model accomplished with the Levenberg–Marquardt backpropagation procedure is used. The random data split is 70% for training, 15% for validation and 15% for testing to predict the heat transfer rate, and the predictions are validated against numerical results. Findings Results show that hnf significantly improves the heat transfer rate compared with nf taken in this study. Also, the relative change in the magnitude of skin friction is approximately 10%–12% for nf and 22%–25% for hnf when compared with pure blood’s skin friction (with baseline ∼ −7.14). For a total sample size of 1,458 and 11 input parameters, the ANN has given the results with root mean squared error (RMSE) = 0.01303894 for nf and RMSE = 0.00066093 for hnf. It is observed that the heat transfer rates predicted by the ANN closely match the numerical results found using the MATLAB bvp5c solver. Practical implications The improved heat transfer rate and drag magnitude are important in many biomedical applications, particularly thermal therapies and drug delivery. The ANN-predicted outcomes validate the heat transfer values across various combinations of physical parameters. Originality/value This novel work investigates the flow of nf (Ti3C2) and hnf (Ti3C2-GO) over a curved-shape geometry under the combined effects. In addition, the ANN is used to guess the heat transmission rate in nf and hnf, and the predictions are verified using the bvp5c method.