Water supply networks represent key infrastructures to provide safe, reliable, drinking water with adequate pressure to communities, thus ensuring people’s health and well-being. These networks can be operated continuously or intermittently. Continuous water supply (CWS) is characterised by delivering permanently pressurised piped water to consumers with adequate pressure, meeting water quality standards and preventing potential contaminant intrusion. Intermittent water supply (IWS) also provides piped water, though only ensures delivery during limited periods of the day or the week, with interruptions from hours to days. This service is common in areas with limited water resources and with financial constraints. Despite the technological and management advancements in the water sector, most utilities with IWS have limited knowledge of the network performance due to unavailable or unreliable data or the lack of numerical models to better understand the systems’ operation. The development of numerical models to describe the phenomena in each IWS stage (filling, supplying and emptying) is important for design, diagnosis and management purposes. Most developed research focuses on the supply stage, using models with the assumption that the pipes are continuously pressurised. Since that is not the case, a model that allows simulating free-surface and pressurised flows is necessary to describe the other two IWS stages.

The thesis aims to develop and validate a new 1D model, based on the widely used SWMM solver, capable of describing the air-water interaction during pipe-filling events in IWS systems. The specific objectives of this research are: i) to identify, understand and characterise the most relevant air pocket related phenomena during pipe-filling events in single pipes and looped networks; ii) to learn how to incorporate the air pressurisation in SWMM solver as well as iii) the different mechanisms associated with the air pocket creation; iv) to understand the model's uncertainties related to these phenomena; and v) to test the developed model in a real-life network.

To accomplish objective i) and to contribute to objectives ii) - iv), an extensive experimental data collection program is developed to understand the phenomena related to the air pocket creation during the pipe-filling event. Collected data include time series of pressure and flow rate and video recordings of entrapped air pockets, for different pipe configurations and aeration conditions. Three pipe configurations are tested: a straight horizontal pipe, a single pipe with a high point and a single-loop pipe network. Three aeration conditions end are tested: no air release, restricted and unrestricted air release.

Several novel numerical developments are gradually implemented to fulfil key objectives ii) - iv). The first is the modification introduced in the existing SWMM hydraulic solver to incorporate the air phase. A conventional air accumulator model is implemented and coupled with SWMM flow calculations. Experimental data collected during the rapid filling of a single horizontal pipe for the three referred aeration conditions are used for model calibration and validation (fulfilling objective ii). Results show that the improved SWMM, AirSWMM(v1.0), describes better the effect of air behaviour during pipe-filling events than the original SWMM when using the EXTRAN surcharge method.

The AirSWMM(v1.0) model is improved to locate and quantify entrapped air pockets created during the pipe-filling events in single undulating pipe systems. Measurements are collected and video recordings are carried out to assess air pocket volumes for the three referred air release conditions. The stochastic nature of air pocket creation results in a range of air volumes predicted for the same aeration conditions. The new version of the model developed, AirSWMM(v2.0), is capable of simulating the air pocket creation, transport and entrainment (air and water mixing process). The stochastic nature of air pocket formation can be numerically simulated by conducting multiple runs of the new solver with different air entrainment ratios. The obtained numerical results show that AirSWMM(v2.0) can accurately locate and approximately quantify the entrapped air pocket volumes. These developments contribute to objective iii).

The AirSWMM(v2.0) model is further tested and validated using experimental data from a single-loop network laboratory setup. Experimental data consisting of pressure-head at multiple locations and video recordings of air entrapment for two high point locations and different nodal elevations, under three aeration conditions, are used. Experimental tests show that air entrapment occurs not only at the high point but along the pipe network, creating air pockets with elongated shapes and larger volumes than for single pipe systems. AirSWWM(v2.0) model results for the looped pipe network demonstrate that this model can correctly locate large air pockets with a tendency to underestimate their volumes. These developments contribute to objective iv).

The AirSWWM(v2.0) model is also tested using a case study of a real-life network published in the literature to assess the accuracy of predicted locations and volumes of the air pockets created during a pipe-filling event. For this purpose, pressure-driven analysis is implemented to better simulate the nodal demands, leading to AirSWMM(v2.1), since this feature was not originally included in SWMM. Results show that pressure-heads predicted by AirSWMM(v2.1) compare well with field data when constant spatial discretisation is used, provided the Courant number is close to 0.15. The recommendations from international guidelines for the location of air release devices (from the American Water Works Association and Deltares) are compared to the predicted air pocket locations. The locations of the estimated air pockets agree with those from the international guidelines for air valve installation. However, these locations only represent part of the air valves needed, those that are necessary for releasing entrapped air during the pipe-filling events, not accounting for other air valves important for pipe failure or conservative design purposes. These developments contribute to objective v).

Further research on AirSWMM should focus on assessing the spatial discretisation that corresponds to the best compromise between accuracy and computational effort to describe the air pocket dynamics in real-life networks. Additional numerical analyses should assess if the developed methodologies can be incorporated into the Preissmann slot pressurisation scheme. More experimental tests are needed to better quantify the air entrainment in piped flows and to analyse the effect of two-phase flows on leakage rate. Further field tests, collecting high-frequency pressure head data, should be carried out during pipe-filling events to validate the developed models.@en