| Resumo: | Modern cities have water distribution networks (WDN) that reliably meet the water demand of every individual household. These networks rely on pumps to move water from distant sources to the consumer. If inefficiently managed, this energy consuming process can become very costly. Using emerging computer power, simulation technologies and sensing devices, many techniques were developed that produce the most cost-efficient operational strategies. These methodologies rely on water consumption predictions to provide optimal pumping strategies. However, these predictions always contain errors, which may create control problems. Many efforts were made to develop progressively better predictions, some of which can change its parameters in real time. Nevertheless, these solutions still contain errors and cannot adapt appropriately to sudden changes in demand. To solve this problem, an adaptive controller that can efficiently update the pumping strategy based on monitored deviations of the predicted consumption is proposed. This methodology takes into account the tariffs of electricity over time and an optimal reference to continuously make the most cost-efficient updates in the pumping strategy. To validate the adaptive controller, two case studies were used: (i) a simple pump-reservoir network and (ii) the Richmond benchmark network. Both evaluations delivered positive results achieving the desired reliability while improving cost efficiency. The combination water consumption predictions and adaptive control methodologies provide a reliable and cost-efficient solution to operate a WDN automatically. This control model may become an essential feature in emerging water grids technology by closing the loop in the control system.
|
|---|