Design and operation of drinking water supply systems are multicriteria tasks. On condition that the drinking water demand is satisfied, one goal is the energy efficient use of feed pumps.
In typical water supply systems, the pressure inside a storage container affects the characteristic properties of a pump, such as flow rate or energy consumption. These also depend on drinking water consumption. If drinking water consumption is high, it is more efficient to use a large pump instead of several smaller ones. The control of speed allows additional improvement. Other goals are continuous operation of pumps, reduction of CO2 emissions or savings of electricity costs.
We develop algorithms tailored to solve multicriteria nonconvex mixed integer optimization problems. Smart tree traversal strategies in combination with gradient descent are applied in order to find good solutions in adequate time.