A systematic mixed-integer differential evolution approach for water network operational optimization.

Area of Science:

• Engineering
• Operations Research
• Environmental Science

Background:

• Potable water distribution network management faces challenges due to complex, nonlinear factors.
• Optimizing these networks is a mixed-integer, constrained decision-making problem involving discrete and continuous variables.

Purpose of the Study:

• To propose a systematic methodology for near real-time operation of water networks.
• To minimize water operational costs through a holistic, multidimensional approach.

Main Methods:

• A model-based predictive control scheme dynamically updates operational strategies with minimal human intervention.
• An adapted mixed-integer differential evolution (DE) algorithm handles discrete-continuous actuation variables.
• Hydraulic network models are integrated and simulated with various strategies.

Main Results:

• The proposed methodology and mixed-integer DE algorithm demonstrate effectiveness on a pilot water network.
• The mixed-integer DE algorithm outperforms genetic algorithms for this optimization task.
• Potential cost savings of 23.69% are identified with adaptive pricing for pumping stations.

Conclusions:

• The developed optimization methodology provides an effective approach for real-time water network operation.
• The novel mixed-integer DE algorithm is superior for optimizing networks with mixed-integer variables.
• Implementing adaptive pricing can lead to substantial cost reductions for water utilities.