A Dynamic Model for Smart Water Distribution Networks

While background leakages accounting for substantial water losses in supply networks remain undetectable, human reaction to even visible pipe bursts is insufficiently slow. We lose precious time in which not only water losses but severe damage to the surrounding infrastructure could be prevented.
These leakages can often be identified only during the minimum night flow and so repair work is delayed by days, causing intermittent water supply and outages. Increasing costs and the importance to ensure supply security require further measures.
Therefore, the application of holistic algorithms controlling proportional valves and pumps allows to act instantaneously on failures by isolating affected pipe sections and by reducing the pressure in that region. With the target to apply classical control theory and yet avoid too complex formulations, this paper presents a dynamic model using no more parameters than a typical, steady-state EPANET model. By means of a sophisticated network description, we modify the rigid water column theory in terms of pressure-driven demands. Other than traditional methods, this approach enables nodal consumptions to dynamically change inner system states such as pressure or flow values. Within this method, the nodal elevation undergoes proper treatment in the model equations and further ensures that pressure values will not become negative as one may have experienced in EPANET.