Smart Network Control in Water Distribution Systems
Our work will advance smart network controls for water network to incorporate the identify energy recovery locations and optimise energy performance through the installation of MHP turbines with a dual pressure and energy production function. This work will also consider pump storage within water networks, including sizing and locating these storage systems.
Software development
From a control perspective, managing large scale water networks is particularly challenging. The complexity of network optimisation is due to several factors: (i) the high level of interconnectivity in a system, (ii) diverse performance criteria and stringent operational constraints, (iii) social and legislative factors, and (iv) consumer behaviour. Different approaches will be explored, for example model predictive control (MPC) has the ability of minimizing the cost associated with water treatment and pumping. Water demand models are important for forecasting water consumption patterns or trends for the network control system. The project has developed an innovative multi-scale control algorithm in the MPC framework for the efficient management of water networks containing multiple pump-as-turbines. This control framework enables the PATs to be simultaneously controlled to maximise energy production while also controlling pressure and leakage. The control frameworks also enable improved cost performance of water networks in Ireland and Wales due to higher energy yields and lower leakage.
Diagram of a smart water network
Flowchart of the MPC algorithm for the work presented at Exeter
Our research team has developed software solutions for the control of water distribution networks using MPC. The software has been applied to a small water network in Ireland and the developed control strategy using multi-objective model predictive control increased energy efficiency and costs.
Hardware solutions
One of the solutions investigated by the project for improvement of efficiency of water distribution networks (WDNs) is deployment of micro-hydropower (MHP) energy devices such pump as turbines (PATs) at locations of excess pressure as hardware to control pressure and also generate energy and savings. Different optimization techniques have been proposed in the literature for finding the optimal number and locations within the networks for deployment of these devices that would maximize the energy recovery and help to create smart water networks which control pressure, leakage and generate energy (Fernadez Garcia et al. 2019; Fecarotta and McNabola 2017; Morani et al. 2021; Nguyen et al. 2020).
Regardless of the optimization algorithms that were used, one of the problems that was reported by many researchers is the enormous computational time necessary for solving this complex problem, even for very small synthetic networks. One of the solutions investigated by Dwr Uisce team for reduction of the complexity of the problem is to generalize the optimal PAT design for sites within WDNs. The optimal performance of a PAT unit is defined with its best efficiency point (BEP). This point designates flow and hydraulic head (pressure) under which the unit attains the maximal mechanical efficiency, i.e., the maximal efficiency in transforming the energy of the water stream into the rotational energy of the machine’s shaft. By generalization of the PAT design, it is meant that the ratio between the theoretically optimal BEP and the average operating point of an examined site within the network is constant or at least varies within narrow bounds, regardless of its size and variation of the operating conditions.
To assess the above hypothesis (i.e., whether the optimal PAT design could be generalized), a large database with yearly recordings from 38 real-world pressure reduction valves (PRVs) sites has been compiled. To find the optimal theoretical solutions for each of 38 PRV sites, two optimization algorithms have been employed namely Nedler-Mead Simplex Direct Search and Particle Swarm Optimization.
As the PAT devices do not possess a flow control device as the conventional turbines, to control the flow passing through the machine, two control schemes have been considered. The first of two is classical hydraulic regulation (HR) scheme that considers installation of a PAT device rotating at constant rotational speed within a hydraulic circuit with two control valves (Fontana et al. 2016), see Fig. 1a. The second considered control scheme is so called hydraulic-electric (HER) scheme (Fontana et al. 2018a), see Fig. 1b. The addition of a variable speed drive (an inverter) in the HER scheme allows the tunning of the rotational speed of the PAT, so that the mechanical efficiency of the PAT can be optimized under variable operating conditions.
Considered PAT control schemes: a) Hydraulic regulation (HR); b) Hydraulic-electrical regulation (HER). Adapted from Fontana et al. (2021).
Preliminary results of this analysis will be presented at the 5th EWaS International Conference in Naples, in July 2022.
The future work will include a development of an optimization strategy for finding the optimal number and locations for PAT hardware within WDNs, that would use the above findings related to the generalized optimal design to reduce the solution space and thus reduce the convergence time. This should allow application of this algorithm to the large real-world networks. In the future we are also going to investigate the viability of MHP in combination with optimized water pumping and micro energy storage.