Numerical simulation and machine learning study on heat transfer enhancement of nanofluids in Taylor–Couette flow with an elliptical slit surface

Abstract

Energy-efficient and high-performance rotating machinery is essential to address the pressing global need for energy consumption saving and emission reduction. One critical design challenge for their thermal performance is managing the maximum hotspot temperature in annular gaps. To tackle this issue, nanofluids is used to enhance the heat transfer efficiency of Taylor-Couette flows. The flow and heat transfer characteristics of Al₂O₃/water nanofluid within annular gap is evaluated in present study. The Eulerian-Lagrangian method is employed to track the trajectories of the particles. In addition, machine learning is considered to predict the flow and heat transfer behavior of nanofluid. The findings indicate that the heat transfer performance of Taylor-Couette flow is positively correlated with volume fraction and negatively correlated with particle size, while the friction factor follows a similar trend. The maximum thermal performance factor is 1.064. The enhanced heat transfer performance of nanofluid is attributed to the migratory motion of particles from the inner to the outer cylinder and the microturbulence of particles within the boundary layer. Adaptive neuro-fuzzy inference system (ANFIS) serves as the most effective model in predicting Nu, while the Multigene genetic programming (MGGP) presents good results in estimating f. The high-precision predictive model for the convective heat transfer of nanofluid in annular gap is established with the assistance of machine learning.