Multi-Variate Analysis of Shell & Tube Heat Exchanger using Principal Component Analysis

Abstract

Shell & Tube Heat Exchangers (STHEs) are a critical component for various industrial applications such as chemical, oil & gas, power, etc. Due to their complex design and high manufacturing cost, the efficient operation and optimum design are quite important for overall cost minimization. Multivariate Analysis (MVA) is a technique used for analysing data with more than one type of measurement. In this paper, MVA of STHEs is carried out using Principal Component Analysis (PCA). 12 variables which predicts the Thermo-Hydraulic Performance & the costs for STHEs are considered. In total, 100 data points are generated and analysed. Two Principal Components (PCs) are adopted and scores & loadings plots are plotted. It is concluded that the first principal component primarily measures design and flow characteristics of STHE while the second principal component has negative coefficients for the factors Res, Nt, Ret, L, do, di and At which signifies the inverse relationship between PC2 and these factors. The current work can be extended further with applications of Partial Least Square Regression and various Machine Learning algorithms.