Using a Genetic Algorithm with a Mathematical Programming Solver to Optimize a Real Water Distribution System

Figure 2. Pressures at the FRM network nodes [35]. 

In order to find a solution (water supply for all users), the existing hydraulic network, FRM, 

must be redesigned. For this specific network, changing the diameter of the pipes is not a feasible 

solution because it is too expensive. Additionally, it is important to mention that the objective of this 

work is to use the same network and make the smallest possible number of changes on it. The 

rebuilding process would involve excavating buried pipeline throughout the network, which 

probably means a new redesigned network, with economic and time costs that are higher than the 

resulting costs of just adding some elements. One solution, here proposed, is adding new storage 

tanks, and considering new pressure-reducing valves (PRVs), pipes, or pumps. To add the pumps, 

pipes, or PRVs to the existing hydraulic network, an optimum configuration must be obtained. This 

new configuration must be generated with the lowest possible cost. An optimization model is 

proposed, which takes into account the problem representation and the network. The objective of the 

model is to find the minimum number of necessary additional tanks, valves, and pipes to be added. 

In order to ensure efficient operation of the hydraulic network, an attempt is made to minimize the 

cost of the necessary modifications.