Evolution of the electric distribution network towards the Smart Grid

The traditional model of the electrical system, structured into the three stages of generation, transmission, and distribution, represents a passive network with a unidirectional flow of energy from power plants to the end user. However, the objectives of climate neutrality, coupled with the increasing electrification of final energy consumption, are driving significant evolutions within this system, making its centrality increasingly evident in achieving socio-economic development goals in modern society. The traditional model of the electrical system, particularly regarding distribution networks (which until recently played a predominantly passive role, distributing power from transmission grid interconnection points to the final user), is being replaced by a model in which the growing penetration of distributed generation technologies transforms the role of distribution networks into an active one. This new role requires the adoption of increasingly efficient technologies for the intelligent management of energy flows, while continuing to ensure the standards of service quality and continuity. In this changing and increasingly complex context, it becomes clear that distribution networks play a central role in the ecological transition, as they are required to fulfill a dual role: on the one hand, to integrate distributed generation, and on the other, to meet the growing electrical load. Thanks to the widespread use of information technologies, the potential for flexible network operation is emerging. When combined with "demand response" technologies, distributed storage, and advanced sensor systems, these capabilities support the development of networks in a "Smart Grid" framework, ultimately enhancing the quality of the electrical service. The main objectives that summarize the definition of "Smart Grid" include a network capable of delivering energy according to the desired standards of quality and continuity, facilitating the active participation of end users in demand response mechanisms, integrating new generation and distributed storage systems, being resilient to both physical and cyber threats, and managing the entire system in an optimized and efficient manner. Achieving these challenging goals requires substantial investments, which will inevitably impact system costs. Furthermore, it is important to consider that investments in distribution networks within highly urbanized contexts, such as the city of Rome, whose distribution network is managed by Areti S.p.A., are particularly complex to implement. These investments require careful planning, preliminary design, authorization, and execution phases due to the necessary regulatory constraints. Considerations include the massive presence of underground cable lines (which are difficult to inspect), protected natural areas and archaeological sites, limited availability of space for technical facilities, difficulty of access, and the inevitable inconvenience caused to citizens due to excavation works. On the other hand, it should be emphasized that distribution networks in highly urbanized contexts are particularly affected by the increase in load due to the electrification of consumption and the widespread adoption of electric mobility. This thesis aims to present the studies conducted with the following objectives: • To investigate the main challenges facing electricity distribution system operators (DSOs); • To propose tools and models to support the planning and operation of electricity distribution system operators. This thesis addresses these challenges by proposing innovative tools and methodologies to enhance the planning, operation, and resilience of distribution networks. The research is grounded in real-world applications, leveraging data and experiences from Areti S.p.A.'s operations within the city of Rome, a complex urban network characterized by extensive underground cabling and stringent regulatory constraints. The key contributions of this thesis include: • Development of artificial neural network models to predict medium voltage fault rates and perform sensitivity analyses, enabling better understanding of fault drivers. • Introduction of risk-based planning methodologies for low voltage networks, integrating fault rate and impact models to prioritize investments effectively. • Analysis of network expansion needs under load growth scenarios, presenting scalable approaches for asset adequacy evaluation. • Assessment of maximum load capacities under N-1 security conditions, proposing optimization-based solutions to minimize investment costs while ensuring system reliability. • Exploration of demand response mechanisms as a tool for cost-effective network planning, providing new insights into their potential benefits and implementation challenges. Current methodologies for network planning and fault prediction often lack the precision or scalability required to address these evolving needs, highlighting critical research gaps. By addressing these interconnected challenges, this work aims to provide actionable insights and tools for distribution system operators, contributing to the broader goal of a resilient and sustainable energy future. The thesis is structured to progressively address these topics, with each of the five chapters building on the preceding one to present a comprehensive view of the transition towards smart grids. The contents of each chapter are briefly summarized below: Chapter 2: Artificial Neural Network for Medium Voltage Fault Sensitivity Analysis One of the primary objectives of DSOs is to ensure the continuity of electricity supply to users connected to their networks. Service continuity indicators, such as the number and duration of interruptions (respectively SAIFI: System Average Interruption Frequency Index and SAIDI: System Average Interruption Duration Index), are closely linked to the fault rate of network components. This chapter presents a sensitivity analysis of faults occurring in medium voltage networks, with particular reference to the medium voltage network of Areti S.p.A. The study allows for an understanding of the influence of various endogenous and exogenous factors on the fault rate of components, starting from the modeling of a neural network trained on historical data. Chapter 3: Risk-Based Planning in Low Voltage Distribution Networks In order to ensure service continuity standards, the DSO must carry out accurate planning of interventions to maximize network benefits while minimizing costs. This chapter proposes a methodology for assessing the risk of power outages in distribution networks and for planning risk mitigation interventions. The study is applied to the low voltage network of Areti S.p.A. and consists of a risk model, divided into two components: fault rate and impact. This model identifies critical network assets from a power outage risk perspective, and a subsequent model is used to select risk mitigation interventions. Chapter 4: Assessing Distribution Network Expansion due to Load Growth In the context of the increasing electrification of consumption, the DSO's role is to plan and operate its network to meet the growing power demand of its users. This chapter presents a simple methodology that, based on asset adequacy evaluations, allows the DSO to make a macro estimation of the investments required to ensure the proper functioning of the network as the expected electrical load increases. The results obtained from applying the methodology to the entire Areti S.p.A. network are also presented. Chapter 5: Maximum Load for Medium Voltage Lines under N-1 Conditions As consumption electrification continues, the DSO is obliged to connect new users while minimizing the costs of the necessary investments, as these costs are ultimately borne by the users themselves. This chapter proposes a complex methodology, based on solving mixed-integer optimization problems, that allows the DSO to estimate the maximum additional load each medium voltage line can accommodate without violating the network's ability to operate under N-1 conditions. The proposed algorithm consists of an initial step to assess medium voltage lines under the current load to evaluate compliance with N-1 conditions, followed by a second step to determine the maximum load that can be connected to each line without violating these conditions. Chapter 6: Low Voltage Electric Distribution Network Planning with Demand Control This chapter presents a mixed-integer optimization model that allows the DSO to identify the set of investments that, at minimum cost, will enable the network to safely supply new loads. The electric distribution network expansion planning (EDNEP) problem is modeled here with much greater detail than in Chapter 4, and it also introduces the possibility for the DSO to leverage not only traditional investments in new network infrastructure but also flexibility in the form of demand response. The model is applied to two portions of the low voltage network of Areti S.p.A., yielding different results that allow general conclusions to be drawn regarding the potential of demand response mechanisms. The research presented in this thesis provides crucial insights and methodologies that support the transformation of distribution networks towards smart grids and a more sustainable and efficient future.