Skeletonization methods for water distribution network models to improve their optimization processes

Mathematical models of water distribution networks (WDNs) constitute a powerful tool for operation and management. Current date technologies and endless software options are capable of generating detailed models including include every element of a real network. However, sometime reduce models with lower computational costs are preferable (optimization problems, real-time management, reaction procedures to emergencies).
WDNs model skeletonization problem is a far from being fully solved. A large number of reduction methodologies have been developed for specific networks or apply only to certain cases, not giving a global solution. Some methods are based on rigid algorithms that restrict user's possibilities to adapt to particular needs. Additionally, in most cases reduction efficiency is only evaluated and at the end of the procedure, not allowing partial analysis. Finally, traditional these processes generally discard the information of the deleted elements, not permitting the user to relate node’s pressure values in both original and reduced network.
The objective of this work is the development of simple and independent simplification algorithms for mathematical models of WDNs. The development is approached in two directions. The first one are classical simplifications of individual elements (nodes or lines). In the second one a set elements are simplified at the same time using graph theory and optimization algorithms. In both approaches, the carry-over of the setpoint curve is considered as an added value for the use of the reduced models in operation actions. Validity, functionality and limitations of the proposed algorithms have been tested using different case studies.