Numerical Study and POD-Based Prediction of Natural Convection in a Ferrofluids-Filled Triangular Cavity with Generalized Neural Networks

Selimefendigil F.; Öztop H.F.

Numerical Study and POD-Based Prediction of Natural Convection in a Ferrofluids-Filled Triangular Cavity with Generalized Neural Networks

Date

2015

Authors

Selimefendigil F.

Öztop H.F.

Publisher

Taylor and Francis Ltd.

Abstract

The effects of a magnetic dipole source on the natural convection of ferrofluids in a triangular cavity are studied. A partial heater is added to the left vertical wall of the cavity while the right vertical wall is kept at the constant temperature. A magnetic dipole source is placed outside the cavity close to the heater. The governing equations of a coupled multi-physics system are solved with a commercial solver using the finite element method. Computations are performed for different ranges of parameters: Rayleigh number (104 ≤ Ra ≤ 106), strength of the magnetic dipole (0 ≤ γ ≤ 8), horizontal and vertical location of the magnetic dipole (-2.5H ≤ a ≤ -0.5H, 0.2H ≤ b ≤ 0.8H). It is observed that the interaction between natural convection and ferrofluid convection under the influence of magnetic dipole affects the flow and thermal field in such the triangular enclosure. The external magnetic field acts in such a way to decrease local heat transfer in some locations and increase it in others for certain combinations of flow parameters and therefore it can be used as a control parameter for fluid flow and heat transfer. Furthermore, an interpolation method based on Proper Orthogonal Decomposition and Generalized Neural Networks is proposed to predict the thermal performance of the system. This approach gives satisfactory results in terms of local and averaged heat transfer values. © 2015 Taylor & Francis Group, LLC.