Anticipating Surface Mean Pressure Coefficient on Inner Surface of C-Shaped Irregular Buildings Using Artificial Intelligence Methodology

Abstract

Analysis of the effect of pressure on the unsymmetrical plan-shaped buildings due to impact of wind forces is more complicated than the symmetrical plan-shaped buildings. Toward this, the present paper is focuses on the study of the wind effects on the inner face of C-shaped unconventional buildings by considering the surface mean pressure coefficient (Cp¯) as the major influencing parameter. Experimental investigation was carried out to obtain the pressure coefficient (Cp) over the surfaces of the C-shaped building models by considering some important configurations like the side ratio, frontal ratio, height ratio, and angle of incidence in a subsonic wind tunnel. In the present study, model equations to predict the Cp¯ are developed by applying various artificial intelligence (AI) approaches such as the model tree (MT) and group method of data handling (GMDH) with neural network (GMDH-NN) as well as combinatorial (GMDH-C) algorithm to the experimental results. In the AI approach, the side width ratio (D/d), frontal ratio (B/d), depth ratio (H/d), relative width ratio (B/b), angle of incidence (θ), and face angle (ϕ) are considered as input to the algorithm to develop the model equation for predicting Cp¯. The performances of the model equations are tested through various statistical error analyses. The importance of input parameters is also analyzed through the sensitivity analysis technique. The results clearly indicate that the proposed GMDH-NN model is the best alternative approach to predict the surface mean pressure coefficient on C-shaped buildings with coefficient of determination (R2) of 0.96 and 0.92 during the training and testing phases respectively. To verify the model equation more accurately, the predicted results are also tested through uncertainty analysis which gave satisfactory results for GMDH-NN as compared to MT and GMDH-C, approaches.