EXPERIMENTAL VERIFICATION OF EXTENDED KALMAN FILTER APPROACH IN LEAK LOCALIZATION

Abstract

Water is the most important natural resource found everywhere in varying quantities and well appreciated in human history, and without it, life is non-existent. However, it is becoming scarce in the face of ever-increasing population and rapid economic development, and managing it properly is the timely demand. The main purpose of water distribution systems is to provide most economically the adequate quantity and satisfactory quality of water with applicable pressure. But, a large quantity of water ranging from 1% to 54% in different countries is lost in almost all water distribution systems via leakage at present times. This paper employs the extended Kalman filtering (EKF) approach (Lesyshen 2005) for detection and localization of leakage in pipe networks installed at the field laboratory of IIT Roorkee (India). This approach develops a transient model using momentum and continuity equations and these are solved using the method of characteristics. It simulates leakage magnitude and its position by dividing the pipeline into four equal parts having two pre-defined (hypothetical) leakages using a computer software developed in MATLAB. The pressure and discharge measurements at the pipe inlet and exit are the model inputs. The EKF approach is found to be efficacious in leak detection and localization for input pressure and leak discharge greater than 2.5 m and 0.25 lps, respectively. EKF accuracy is much more sensitive to changes in input pressure than those in leak magnitude, it increases with increasing input pressure and pipe size, and PVC pipes yield higher accuracy than do metallic Mild Steel and HDPE pipes.