CCWI2017: F3 'Exploring How Pipe-Level Energy Metrics Relate to Hydraulic Parameters Using Multivariate Analysis'
journal contributionposted on 01.09.2017 by Saeed Hashemi, Yves R. Filion, Vanessa Speight
Any type of content formally published in an academic journal, usually following a peer-review process.
PCA in this study builds upon a correlation matrix of 11 hydraulic parameters including pipe hydraulic data and pipe-level energy metrics by using Eigen values and Eigen vectors to compress the high- dimensional space of the data into a two-dimensional space. This makes the visualization of the data possible, including all hydraulic parameters simultaneously. The graphical presentation of the hydraulic parameters (mono-plots) and the data set (bi-plots) help which parameters can lead into inefficiency in pipes in the whole dataset. The results show that the metric energy lost to friction in a pipe along with average unit headloss, average flow rate and proximity to major components have a high influence in distinguishing poorly performing pipes from the others. Average pressure and the metric for energy needed by the user for each pipe tend to track closely, despite a lower statistical importance than previous parameters. Diameter and pipe roughness tend to stand alone with poor representations on the two principal component axes.