Operational Optimization and Network Design of Large Water Supply Networks: Models and Algorithms

Production and distribution of high-quality drinking water at a low cost while guaranteeing supply security is gaining much interest lately. Nevertheless, companies in Flanders often still rely on the operators’ experience, which is trustworthy but ineffective when unexpected scenarios unfold. This can be combined with the use of simulation software, which is technically sound but can only be used for validation. To obtain an operational planning, both approaches fail in terms of energy-efficiency and cost minimization. In this study, we construct a decision support tool that minimizes the production and distribution costs while satisfying drinking water demand. This tool will be able to efficiently adapt to sudden changes which occur in the network, such as pipe fractures or unforeseen demand patterns.

Converting the components of the water supply network in a mathematical formulation leads to a very complicated model with many variables and constraints. To allow a solution to this model that is within reasonable computational limits, some simplifications are initially made. By developing new and more efficient algorithms, a more accurate model can be solved. This accuracy can be achieved by taking into account yearly constraints, e.g. yearly limits on the total groundwater extraction. In a next step, the model and algorithms are altered to allow support for network design problems. When pumps need replacement, the position and characteristics of the new pump can then be determined. As demand of drinking water is directly dependant on changing climate conditions, adding this stochastic variable to the model will lead to even more accurate solutions. A link with state-of-the-art simulation software to generate an exact operational planning, is the final step in this research.

Derek Verleye
  • El-Houssaine Aghezzaf
Aug. 9, 2016, 3:57 p.m.
Research group