Optimization of alongwind and crosswind force coefficients on a tall building with horizontal limbs using surrogate modeling

Abstract

Summary For tall buildings, values of wind force coefficients can be obtained from wind tunnel tests or Computational Fluid Dynamics (CFD). This paper is concentrated to analyze a set of CFD data and propose parametric equations for determining force coefficients in the alongwind and crosswind direction (Cfx and Cfy) of tall buildings with horizontal limbs. Initially, a parametric study is performed with CFD analysis considering RANS k − ε turbulence models keeping a constant plan area 22,500 mm2. The length and velocity scales are taken as 1:300 and 1:5, respectively. The required design parameters are obtained and used for fitting parametric equations. The CFD data are further utilized for training artificial neural networks of Cfx and Cfy. The results of CFD, ANN, and parametric equations are compared. The parametric equations are validated by employing a wind tunnel study. Finally, three optimization studies are carried out using a genetic algorithm (GA), of which the first two aim to present the maximum and minimum force coefficients considering single objectives. The third optimization is a multi‐objective optimization problem, carried out to simultaneously minimize and maximize the two orthogonal force coefficients. Pareto‐optimal design results are presented.

Strength Optimization of Nanocomposite Cementitious Materials Using Nanoscale Modifications

“…Researchers considered the selection of algorithms, architecture networks, patterns, and neuron numbers taken as rational numbers for selecting the appropriate size and topography of the neural network. Not only is it used in the strength characteristic of cementitious material, but it is also used to obtain the force coefficient of wind load concerning tall structures [40]. Minor errors occur if the training pattern response is not suitable.…”

Improved Chaotic Algorithm-Based Optimization Technology of Architectural Engineering Drawing Parameters

“… Excitation function of the system includes instruction function and interference function. In order to facilitate the comparison of dynamic characteristics of the system, some typical signal functions are generally selected as instruction signals, such as step function, impulse function, and random signal [ 12 , 13 ]. For example, the initial time t 0 of the end and terminal conditions, the terminal time t f , the initial state x ( t 0 ), and the terminal state x ( t f ) are either given or arbitrarily valued according to the task conditions.…”

“…Excitation function of the system includes instruction function and interference function. In order to facilitate the comparison of dynamic characteristics of the system, some typical signal functions are generally selected as instruction signals, such as step function, impulse function, and random signal [ 12 , 13 ].…”

Reduced Frequency and Power Spectra of Wind Pressure of a Tall Building Having "Z" Shape in Plan

“…Off late, artificial neural networks are adopted to present the average pressure coefficients on all building facets of setback high-rise buildings (Bairagi and Dalui, 2020) and crossplan-shaped high-rise buildings (Paul and Dalui, 2020) [36,37]. Paul and Dalui (2021) further utilized artificial neural networks to compare the along and crosswind force coefficients obtained using CFD and parametric equations [38].…”