Regression Analysis on the Behaviour of Thin Spherical Shells with Various Parameters

Regression analysis is used to study the dependence of one variable with another explanatory variable. The need for the regression analysis is to establish a relationship, the effect, and the coefficient of dependency between the variables. This study focuses on the linear regression analysis on the numerical stress and displacement of thin spherical shells obtained from the previous research using ANSYS 18.1. The aim of this study is to establish a relation between the geometric characteristics of thin spherical shells. Thin spherical shells that have the ratio of curvature radius to thickness less than 20 are the structural roofing system that form an integral part of the structure. The shells are used in the construction for many decades because of its less consumption of materials and efficiency to covers larger spans. The behaviour of the spherical shells depends on the geometric characteristics that are span and thickness. The linear regression analysis using least square method is carried out for the varying geometric parameters of the spherical shell where stress and displacement are the dependent variables, and the span and thickness are the explanatory variables. The degree of dependency and coefficient of dependency of stress and displacement with an accuracy of 95% to span and thickness of the spherical shells are discussed.