Abstract: The most important factor in reducing leakage in water supply networks is pressure management. The use of pressure reducing valves (PRVs) causes excess pressure to be dissipated in the target nodes of the network, thus replacing equipment such as PAT (pump as turbine), reducing leakage as well as increasing excess pressure in the networks, inducing the generation of hydroelectric power, It is essential. In this article, a 25-node network is simulated using EPANET software; And the number and position of PATs are optimized by genetic algorithm. For optimization, a new objective function that simultaneously considers “pressure management in the water supply network, electricity generation from renewable sources and reduction of greenhouse gas effect; It is used to replace renewable energy instead of energy produced by non-renewable fuel power plants such as fossil fuel power plants. "
 
Introduction: The most important problem that always exists in the design of water supply networks is leakage from the networks. Pressure management is the first solution to solve this problem (Campisano and Creaco, 2020; Tabesh and Vaseti, 2006. Because, pressure reducing volves (PRVs) make dissipate excess pressure in consumer nodes, it is recommended to use PATs instead of PRVs (Latifi et al., 2021). Moreover, one of the reasons for global warming is the emission of greenhouse gases that have a direct relationship with countries' energy consumption (Kohansal and Bahraminasab., 2020). In most of the research, objective functions consider reducing leakage from networks and increasing electricity generation by renewable energy have been used to determine the optimal placement of PATs in a water supply network. Therefore, the need for environmental terms due to the effect of reducing greenhouse gases production by renewable energies in the optimization objective functions is completely felt.

In the current research, the hydraulic analysis of the 25-node water supply network is performed using the pressure-based analysis methodology with the assistance of EPANET software. The simulated water supply network is integrated with the powerful MATLAB software for applying the genetic algorithm as a powerful optimization approach. The objective function, optimizing PATs, considers three parameters of water pressure management, renewable energy, and environmental effects of greenhouse gases for finding the optimal location and number of equipment within the desired network.
 
Methodology: To check the performance of the optimization objective function of PATs, the hypothetic 25-node water supply network that has been used by several researchers such as Vairavamoorthy and Lumbers (1998), Fecarotta et al. (2014)، Giugni et al. (2014)، Garcia and Nabola (2020) and Nguyen et al. (2020) to research the optimization objective functions in water supply networks is evaluated. The hydraulic relations of the desired network are solved in Epanet software by the gradient method and pressure-based hydraulic analysis (Wagner's Eq). Also, this software uses the orifice equation (Germanopoulos (1985)) for the hydraulic analysis of leakage. After drawing and entering the basic info of the desired network in EPANET software and integrating the Epanet library into MATLAB, the new objective function will be optimized with the genetic algorithm's assistance. To define the new objective function and consider the parameters of pressure management and energy production from reverse pumps, the impact of reducing greenhouse gas production as a result of replacement energy made by fossil power plants with clean energy is also investigated. In the end, further to the equations of conservation of energy and conservation of mass, the minimum power made by reverse pumps and also the pressure limit within the nodes of consumers have been introduced as constraints of the problem.
 
Results and Discussion: For evaluation the new optimization objective function and aconducting a comparison between the induced outcomes and the reference results of Giugni et al. (2014), the cost parameters have been eliminated from Eq. 6. Then the production energy of PATs was annalysed according to the scenarios defined in Nguyen et al. (2020), including the network equipped with 1 PAT (minimum production power of 400 watts per day), 2 PATs (minimum production power of 1500 watts per day) and 3 PATs (minimum production power of 750 watts per day), ‎Fig. 5.
Optimization approach leads to the best place of installing PATs for the scenario1 at pipe number 20, scenario 2 at pipe numbers 20 and 18, and scenario 3 at pipe number 20 (PAT1), 1 (PAT2), and 18 (PAT3) and is completely in accordance with the results of Giugni et al. (2014). As shown in ‎Table 2 and ‎Fig. 5 and ‎Fig. 6, the production energy will increase with the increase in PATs from one to three. Also, the leakage in the network decreases, and thus the most optimal response of the desired network is to put in 3 PATs in the place of pipes No. 20, 18, and 1 is determined. As an example, in ‎Fig. 7, the amount of head drop and flow rate passing through PAT located at the situation of pipe number 18 is drawn. In this figure, it will be seen that during the peak consumption times (9:00-18:00), the flow through PAT increased, and the head drop decreased, which might be the result of the slow change of the water level within the reservoir even when there are different consumption patterns. After verifying the optimization function of the desired network with the assistance of the genetic algorithm, the optimization results of the new objective function also introduce the installation of three PATs within the same location of pipes 20, 1, and 18 as the best solution to the problem. ‎Fig. 8 and ‎Fig. 9 show the outturn and production power of every of the PATs, respectively.

Considering the importance of the return-on-investment factor in the final decision of the projects, using the results of the 25-node network of Garcia and Nabola (2020), and assuming the country of Iran, it was observed that the social cost of carbon dioxide gas is approximately equivalent to 15% of the energy tariff. Now, if this approximate number is added to the electricity production income, the investment return period of every PATs will show a decrease of 4. 7, 7. 8, and 32. 7 months, respectively. This reduction within the investment return period shows the importance of considering the environmental dimension of PAT energy.
 
Conclusion: The evaluated optimization method, using the new objective function, resulted in to placing three reverse pumps at the location of pipes 20, 18 and 1; as the best mode of optimization in the network. Conducted comparison showed, the number of PATs increases, the leakage significantly decreases but the energy production from these renewable sources increases. Furthermore, the impact of the social cost of greenhouse gases in the investment return period indicates a significant decrease in efficiency period for PATs. In general, adding the environmental term in optimization objective function emphasizes the necessity of using renewable energies (PAT) in water supply networks.