Optimised allocation of actuators for DWDS

• A new approach to booster station allocation within DWDS has been introduced.
• The allocation task is formulated as hierarchical (two layers) optimisation task where an upper layer is to handle integer decision variables by which fixed (for a single algorithm step) actuator structure is prescribed for lower layer optimising control algorithm.
• The approach uses open-loop MPC control, at the lower optimisation level, instead of offline scheduling during the allocation.
• Distributed GA with cluster-like computational environment has been utilized due to charactrisation of posed optimisation task.
• The two layered structure of the algorithm enables to utilise genetic information more efficiently than in previous works. Moreover two-layer structure connected with efficient computational environment leads to efficient computation and improves the quality of the results obtained in the meaning of the genetic information processed during the otimisation.

This paper addresses the problem of actuators’ allocation within networked structured system, namely allocation of disinfectant booster stations within Drinking Water Distribution System (DWDS), under receding horizon optimised control – Model Predictive Control (MPC) to be exact. The allocation task is kept within dynamic multiobjective optimisation framework. The MPC is defined as a single objective predictive operator. Two numerical examples illustrating the results of allocation procedure are given. The algorithm utilises multiple water quality and quantity scenarios to justify the allocation under normal and disturbed water quality operational conditions.