Visualization of nanofluid flow field by adaptive-network-based fuzzy inference system (ANFIS) with cubic interpolation particle approach

Abstract

The interior of a square cavity containing nanofluid with copper (Cu) nanoparticles (NPs) was simulated by a computational fluid dynamics (CFD) method. The flow rate parameters were obtained with different directions of the computing elements and temperature of the fluid inside the cavity. The CFD outputs were studied using an adaptive-network-based fuzzy inference system (ANFIS) to create an artificial fluid flow domain. The CFD outputs were used as the input and output data in the ANFIS method. Subtractive clustering was applied used for data clustering to look at the impact of the cluster influence range on the performance of the ANFIS method. After the highest level of performance was reached, the cavity nodes that were not involved in the learning process were predicted. Very good accordance was observed between the ANFIS method prediction and the results of the CFD method. The ANFIS method reduced the calculation time dramatically compared to the CFD method and has the ability to predict far more nodes in a short period of time. The results show that the clustering framework can visualize the flow pattern in the square-shaped cavity in a short time. Additionally, the combination of CFD and smart modeling enables us to specifically analyze and visualize one part of a fluid structure without several complex CFD analyses.