Towards a Digital Twin Model for the Management of the Laives Aqueduct ^†

Abstract

The digitalisation of water supply systems is essential to support crucial activities, ranging from the hydraulic modelling and calibration of distribution systems to the optimal planning of renewal measures. Digital Twins is the current challenge that aims to join accurate hydraulic models with Artificial Intelligence (AI) tools that employ a faithful digital copy of the original system and smart infrastructures. Thus, the aqueduct Digital Twin model is intended to be the core of future integrated water management systems to support efficient and sustainable management. In this paper, a process of implementing a digital model for a mountain aqueduct with its associated applications is proposed.

1. Introduction

Hydraulic simulations play a crucial role in the efficient and sustainable design and management of water supply systems [1,2]. To fulfil this function, such models have to be as close as possible to the real infrastructure behaviour and be able to reliably replicate the hydraulic components and control devices [3]. In recent decades, several improvements have been proposed in network modelling to enhance both the accuracy and the reliability of aqueduct models, involving, for instance, numerical solvers [4,5], demand schemes [6,7], leakage representation, and the embedding of control and smart devices [8,9].

The latest challenge in hydraulic simulations is the development of Digital Twins, which are beginning to be employed to improve the management of aqueducts, supporting operational planning and decision-making [10,11]. The need/opportunity for this new type of detailed hydraulic model has arisen due to the implementation of smart cities and smart grids that provide a large amount of digital details and real-time data streams. Some of the main benefits of using Digital Twins in an advanced integrated management system are optimal component design and network rehabilitation, leakages identification, efficient operational strategies, early warnings, water quality, and system anomalies detection [12,13,14,15,16].

In this contribution, the authors present the implementation of a Digital Twin model for the water supply system of Laives, which is a mountain town located in South Tyrol in northern Italy. This is one of the first approaches of utilising a Digital Twin to improve the management and planning of a fragmented mountain water distribution system, which is characterised by high elevation variability and several branches. Finally, the actual and potential benefits of implementing a Digital Twin within an integrated water management system are analysed, even in the case of small-scale water users. The rest of the paper is divided as follows: Section 2 presents the case study and the methodology for implementing a Digital Twin model, Section 3 shows the results, and the final remarks are proposed in Section 4.

2. Materials and Methods

2.1. Case Study

Laives is an Alpine town of approximately 180,000 inhabitants located in the Bozen-Bolzano Province. Similar to other large mountain urban centres, Laives lies between the final part of the mountain slopes and the valley floor of the Adige River. This configuration leads to a significant difference in altitude within the urban centre, consequently causing high pressure in the water distribution system, which consequently demands management. Therefore, the Laives aqueduct consists of more than 50 km of pipes, 8 water sources consisting of wells and springs, 4 tanks, and approximately 1700 connections. This system works in the same way as a typical mountain aqueduct, as water is supplied from mountain sources and is pumped from wells to water reservoirs in the upper part of the city, from where it is distributed by gravity to all users. The distribution system is characterised by three main interconnected districts called, from north to south, San Giacomo, Pineta, and Laives.

2.2. Hydraulic Modelling

For the design of the aqueduct of Laives, all the model elements resemble the actual water distribution system, meaning that the hydraulic model is fully detailed according to the principles of a Digital Twin. Indeed, the model is made of approximately 2100 pipes, 2800 nodes, and almost 700 hydraulic components, such as valves and gates. Figure 1 shows the simulation results regarding pressure and flow rate distributions.

This detailed model was developed with the water demand concentrated at the network nodes and simulated with a pressure-driven simulator. In addition, the leakages of the network have been implemented using the conventional formulation of the emitters concentrated in the network nodes. Furthermore, the network has been calibrated through an optimization procedure based on hydraulic simulations and genetic algorithms, as described in detail in [17,18]. This procedure is aimed at minimizing the differences between the simulated values by the model and the measurements collected with pressure and flow rate meters. The variables adjusted through the calibration are the roughness of the model pipes. This operation is accomplished by minimizing an objective function designed as the sum of the normalized difference between simulated and measured pressures in the monitored nodes. The resulting model is suitable for adoption as a Digital Twin, as described in the following Section.

3. Results and Discussion

The hydraulic model is embedded in an advanced integrated water management system consisting of a modern web-based application accessible via software-as-a-service. This interface allows one to consult the information stored in the database, including the characteristics of the components of the system, the chronological list of the maintenances and failures, and the metering data. As a result, the Digital Twin of the water supply system of Laives can fully exploit the simulations of the detailed hydraulic model and the metered data for optimal operational planning and proper decision-making. Figure 2 depicts the Digital Twin of the integrated water management system of Laives. Specifically, Figure 2 (a) shows the representation of the model with the main panel on the right, while (b) describes a detail of the information panel of an element valve. For instance, it is possible to simulate the behaviour of the network by changing the status of the valves in case of a need for network sectioning due to pipe bursts or pipe replacement.

4. Conclusions

This contribution presents the implementation of a Digital Twin of a mountain water supply system to support the intelligent management of similar distribution networks. A Digital Twin combined with suitable management software enables one to support the water companies to increase the quality of their service by employing hydraulic simulations with smart data. This management environment allows an advantageous development of several useful tools, such as water forecasting, metered districts design, pumping optimization, renewal planning, operational maintenance scheduling, and real-time anomaly detection [14,19,20].