ANN-based prediction of aerodynamic force coefficients and shape optimization using MOEO of a high-rise building with varying cross-section along height

The main goal of this study is to find the effect of various geometrical configurations with varying cross-sections along the height of a corner recessed square tall building, considering the wind flow around the buildings from 0 degrees to 360 degrees. The design variables considered in this study are the amount of corner recessing (S), the height of the square cross-section (h) and the wind incidence angle (& Oslash;). The design variables are generated using the simple random sampling technique in MATLAB. A series of numerical analyses have been performed for those randomly generated design variables using Computational Fluid Dynamics (CFD) in the ANSYS-CFX module to evaluate the along and across-wind force coefficients (Cfx and Cfy). The numerical simulations have been carried out using the k-epsilon turbulence model on a length scale of 1:300. The results of the CFD simulations are used to train the artificial neural network (ANN) of Cfx and Cfy. After simulating the networks of Cfx and Cfy, a shape optimization study has been carried out using Multi-Objective Equilibrium Optimization (MOEO) to find the optimal shapes of various building configurations considering different weights of the design function values (Cfx and Cfy). The results of the shape optimization study have been validated with separate CFD investigations and several wind tunnel experiments.