Water-energy issues

Director: Sophie Demassey

The optimization of water supply networks is a flourishing research domain that goes hand in hand with several concomitant factors, such as increased water consumption, which entails creating new networks; gradual deterioration of installations, which require renewal; rationalization of installation costs and energy consumption, which involves detecting leaks, automation and more efficient use of pumps; and changes in the electricity supply, with the integration of intermittent energy sources and fluctuating prices, which calls for careful management of pumping.

Mathematical programming is the preferred approach and applies to all time scales. However, two problems stand out in published studies: at a strategic level, the sizing of gravity-fed networks and, at operational level, the planning of pumping station operations. 

Due to their algorithmic complexity, the two problems are studied separately. They are however intrinsically linked in practice: the conveyance of drinking water is more likely to be backstreamed than gravity-fed, and the daily programming of a pumping station depends on its size.

The thesis that Gratien BONVIN commenced in December 2014 aims to develop an integrated, optimization approach to water supply systems from short to long term, going from the real-time management of a network up to its design.

The first achievement, published in Applied Energy [Bonvin et al., A convex mathematical program for pump scheduling in a class of branched drinking water networks, Applied Energy (2016)], is a new, quadratically constrained convex programming model for “day-ahead” planning to operate the type of pumping station typically used in French rural areas. In comparison with current manual management, the approach has multiple benefits, i.e. operational, automated pumping plans; substantial energy and financial savings (on average 15% of the electricity bill); a pressure drop in the network and thus fewer leaks in the pipeline.

Following the success of this new convex programming approach, we are currently extending it to a more general context, including urban and irrigation networks that are larger in size or made up of components that are more complex to model.

The second part of the thesis will center on integrating this short-term optimization model into resolving mid-range decision-making problems (e.g. choice of electricity supply contract) and long-term ones (resizing the network).

Centre de Mathmatiques Appliqus

Agence Web Spinat