| dc.description.abstract | Intermittent water supply (IWS) systems are widely practiced around the world due to a lack of water resources, and economic and political factors. This practice leads to inequitable supply among consumers, water quality problems, water losses through pipe bursts, and meter malfunctioning alongside the cost of coping for consumers and utilities. Considering water’s essential role for human survival, it is crucial to address the problems associated with its supply. From a practical engineering perspective, one primary approach is to utilise modelling tools to understand and analyse the behaviour of IWS systems, a crucial step for effectively identifying and addressing issues, thereby enhancing water supply services. While a lot of efforts to develop modelling methods for IWS exist, there is still need for a rigorous modelling method to simulate IWS behaviour in a more realistic manner considering its unique characteristics. In addition, owing to the diverse practices and operations associated with IWS, significant uncertainty prevails in various aspects, including user water consumption, supply characteristics, and household tank sizes. Quantifying the uncertainty of model predictions is a critical task, yet this aspect has not been fully explored in the existing body of knowledge.
In this thesis, an improved EPA-SWMM-based IWS hydraulic analysis/simulation method is proposed, and experimental research is designed with a small-scale laboratory water distribution network, which replicates IWS system behaviour. The calibration of minor and major loss coefficients within the model of the network is accomplished using a Genetic Algorithm which demonstrates the method’s ability to achieve an improved replication of network behaviour. Further validation is provided by comparing the improved method with existing EPA-SWMM-based IWS modelling methods using a real case study from the literature. The obtained results illustrate the improved method’s advantages over current methods. Notable benefits include the ease of implementation, numerical stability, and accuracy in predicting network pressures and flow rates under extended period simulation.
Additionally, in this thesis, a novel uncertainty quantification framework is presented and demonstrated using the improved EPA-SWMM-based IWS modelling method, thereby opening a new field of research in IWS. The uncertainty quantification framework includes the characterisation and propagation of uncertain model input parameters, as well as sensitivity analysis. Uncertain input parameters are identified and characterised with probabilistic approaches while synthesising the IWS case studies found in the literature. Uncertainties are propagated using Monte Carlo (MC) simulations to quantify the uncertainty in IWS system performance. Following this, a global sensitivity analysis using Sobol’s method is conducted to identify the influential and non-influential uncertain input parameters. The results indicate that IWS system performance is predominantly determined by supply characteristics, in conjunction with household tank sizes and water usage rates from the household tanks. This underscores the importance of accounting for these uncertainties and emphasizes the need for a meticulous data collection. | en_GB |
