Enhancing thermal efficiency in rotating third-grade fluids: A local nonsimilarity framework with ANN-based entropy analysis

The third-grade viscoelastic model, formulated within the Rivlin–Ericksen constitutive framework, offers a distinct advantage over other models by simultaneously capturing normal stress differences and shear-thinning behavior in nonlinear fluids. This work applies third-grade fluid model to investigate viscoelastic rotating flow with heat dissipation effects over a stretchable surface, highlighting the combined impact of rotation, viscous dissipation and entropy generation, which has not been addressed in prior studies. The involvement of third-grade fluid’s stresses generates nonsimilar terms in both momentum and heat transfer equations. Compared with the prior studies conducted for nonrotating frame, this work reports a locally nonsimilar analysis wherein the derivatives along the streamwise direction are retained. Irreversibility effects in the model are further scrutinized by evaluating entropy generation rate and its dependence on fluid’s rheological properties. A neural network framework trained with the Levenberg–Marquardt (LM) algorithm is also applied to predict entropy rates and Bejan number. Multiple validation metrics including regression plot (RP), mean squared error (MSE) and histograms are used to demonstrate the reliability of artificial neural network (ANN) based forecasts. The results obtained from the ANN closely align with those from the bvp4c method, exhibiting an exceptionally low mean absolute error. The flow fields and associated fluid dynamic characteristics are analyzed under varying rotation rates, elasticity parameters and shear-thinning effects. This study reveals that entropy generation intensifies near the wall when the temperature difference increases, primarily due to stronger thermal gradients. As viscous dissipation intensifies, more mechanical energy transforms into heat and the associated entropy generation rate becomes progressively higher.