State and Parameter Estimation in Water Distribution Systems

Abstract

Water Distribution Systems (WDS) are one of the most important infrastructures in a society, which has a direct impact on human lives. Water authorities around the world are expected to supply the required amount of water at adequate pressure and potable quality to the consumers at all times. For achieving this task, proper system and resource management need to be carried out. Optimal or near optimal management of the WDS is rendered challenging due to various issues the system is plagued with. WDS around the world have to deal with many issues such as leakage, water quality degradation, demand uncertainty, inequitable supply of water among the consumers etc. For better system management, monitoring, modelling and control of the WDS is required. Currently, many sensors are available commercially which are capable of measuring the water quality and quantity parameters of the system at very fine frequency. These sensors make real time monitoring of the water distribution system easier. In case of a water quality or quantity event in the WDS, this real time feed of data might be able to give a preliminary warning to the water authorities. For development of a decision support system, this monitored data need to be coupled with modelling and analysis of the WDS. Efficiency of the decision support system will be enhanced if modelling and analysis is carried out in real time or near real time, as the time required for event mitigation will be reduced. Automatic control of appurtenances in the WDS helps in appropriate operation and execution in the system as well. Data assimilation methods were used in many fields of science and engineering for parameter and state estimation of the system in real time. Different mathematical filters like Kalman Filter, Extended Kalman Filter, EnKF etc. and their variants are used in real time state estimation in the field of ocean sciences, remote sensing, ground water etc. Whereas, application of data assimilation techniques in WDS analysis is quite recent and limited. The non-linear nature of WDS, combined with dynamic operational conditions makes direct application of data assimilation techniques difficult in WDS The primary objective of the present study is to apply Ensemble Kalman Filter (EnKF) based data assimilation techniques for water quality and quantity state estimation in WDS under different parameter uncertainties. EnKF and its variants are used in this study, as EnKF is one of the most widely used mathematical filter, and it is easy to apply it to WDS, as it does not make any assumptions of system linearity. In this study, EnKF and its variants were formulated for hydraulic and water quality state estimation in WDS. EnKF and three different variants (Iterative or Dual- EnKF (I-EnKF), Non-Iterative Restart EnKF (NIR-EnKF) and Iterative Restart EnKF (IR-EnKF)) of it are used throughout the study. The main objectives of this thesis are as follows: i. Application of EnKF based data assimilation technique for hydraulic state estimation in WDS (Chapter 3). ii. Application of EnKF based data assimilation for water quality state estimation in presence of source uncertainty (Chapter 4) iii. Application of EnKF based data assimilation for water quality state estimation in presence of parameter uncertainty (Chapter 5) iv. Application of EnKF based data assimilation and Modified- Fault Sensitivity Matrix (FSM) for leak detection and localization in WDS(Chapter 6). v. Achieving equitable water supply for New Bangalore Inflow Network , using Dynamic Inversion-Proportional Integral Derivative (DI-PID) and Model Predictive Controller (MPC) technology (Chapter 7)