Climate change driven by fossil fuels consumption demands a quick energy transition to a carbon-neutral society. One of the most affected systems by this paradigm shift is, without a doubt, the electric power system. Once designed with few, typically high pollutant production nodes on the transmission side, production is also located in the distribution side, with a very high number of DERs, primarily renewable. This change, however, comes with a cost: many, intermittent small resources are hard to manage. They cannot always guarantee the meeting of demand, creating difficulties to the reliability, security, and power quality of the system. As a result, the demand side is becoming more and more active, as its contribution will be essential in addressing the new challenges. Moreover, diverse storage systems have been recently researched, prototyped, and commercialized that allow even more flexible management of the system, also creating other types of loads such as EVs. Such loads will modify load patterns in ways difficult to anticipate, which poses serious questions on how to reinforce the current distribution and transmission assets. Moreover, the role of the electricity markets, once liberalized to ensure low energy prices to all the consumers, are becoming a central tool for system operators. Indeed, markets are expected to procure all the flexibility needed to meet unexpected generation, load, and infrastructural contingencies. In this context, simulation, forecasting, and automated decision-making tools derived from business and data analytics are becoming crucial for modern power systems design, planning, and operation. During the PhD, the author has worked on research projects such as PREDICT and PODCAST, intending to develop data analytics tools better to manage distributed resources and all the related challenges. This thesis homogenizes the work done by framing it in the context of two great revolutions: one is the modern power system revolution, and the other is the advent of the analytics revolution, which has been possible thanks to the emergence of BD paradigm. The developed applications will be framed as particular cases of analytics subfields (descriptive, diagnostic, predictive, or prescriptive) after detailing the methodological aspects of the involved analytics branch that have much broader use than the single applications presented. In Chapter 1, this work is introduced by presenting the two current paradigm shifts of energy transition and data analytics and explaining how the two are interacting in the context of power systems. Some key definitions are laid out that will help to clarify the concepts explained in the following chapters. In Chapter 2, descriptive and diagnostics analytics and their possible applications to power systems are presented in the context of power systems. As an application, the proposed IGSC algorithm, devised during the PhD, is illustrated, which allows for flexible probabilistic modeling and simulation of quantities of interest via light online and adaptive clustering. In Chapter 3, predictive analytics, with a particular focus on forecasting, is presented, by following the CRISP-DM framework, together with some of its many applications present in the literature of power systems. Applications of distribution network load forecasting, building energy forecasting, and PV short-term forecasting are presented, all developed in the context of the projects PREDICT and PODCAST. In Chapter 4, the final step of analytics, that is, prescriptive analytics, is described together with some of the applications to the power system field. Battery profile optimization and optimal sizing and siting, developed as a functionality of a modern DMS, are the applications of this chapter and are drawn from the PODCAST project. Afterward, in the Conclusions, the key findings are summarized together with some final comments. Finally, the publications produced during the acs{phd}, and contributed projects, collaborations, and attended courses, are listed, while references cited in this thesis conclude the dissertation.

Simulation, forecasting, and control in power system analytics: methodological aspects and applications

MOSAICO, GABRIELE
2022-05-30

Abstract

Climate change driven by fossil fuels consumption demands a quick energy transition to a carbon-neutral society. One of the most affected systems by this paradigm shift is, without a doubt, the electric power system. Once designed with few, typically high pollutant production nodes on the transmission side, production is also located in the distribution side, with a very high number of DERs, primarily renewable. This change, however, comes with a cost: many, intermittent small resources are hard to manage. They cannot always guarantee the meeting of demand, creating difficulties to the reliability, security, and power quality of the system. As a result, the demand side is becoming more and more active, as its contribution will be essential in addressing the new challenges. Moreover, diverse storage systems have been recently researched, prototyped, and commercialized that allow even more flexible management of the system, also creating other types of loads such as EVs. Such loads will modify load patterns in ways difficult to anticipate, which poses serious questions on how to reinforce the current distribution and transmission assets. Moreover, the role of the electricity markets, once liberalized to ensure low energy prices to all the consumers, are becoming a central tool for system operators. Indeed, markets are expected to procure all the flexibility needed to meet unexpected generation, load, and infrastructural contingencies. In this context, simulation, forecasting, and automated decision-making tools derived from business and data analytics are becoming crucial for modern power systems design, planning, and operation. During the PhD, the author has worked on research projects such as PREDICT and PODCAST, intending to develop data analytics tools better to manage distributed resources and all the related challenges. This thesis homogenizes the work done by framing it in the context of two great revolutions: one is the modern power system revolution, and the other is the advent of the analytics revolution, which has been possible thanks to the emergence of BD paradigm. The developed applications will be framed as particular cases of analytics subfields (descriptive, diagnostic, predictive, or prescriptive) after detailing the methodological aspects of the involved analytics branch that have much broader use than the single applications presented. In Chapter 1, this work is introduced by presenting the two current paradigm shifts of energy transition and data analytics and explaining how the two are interacting in the context of power systems. Some key definitions are laid out that will help to clarify the concepts explained in the following chapters. In Chapter 2, descriptive and diagnostics analytics and their possible applications to power systems are presented in the context of power systems. As an application, the proposed IGSC algorithm, devised during the PhD, is illustrated, which allows for flexible probabilistic modeling and simulation of quantities of interest via light online and adaptive clustering. In Chapter 3, predictive analytics, with a particular focus on forecasting, is presented, by following the CRISP-DM framework, together with some of its many applications present in the literature of power systems. Applications of distribution network load forecasting, building energy forecasting, and PV short-term forecasting are presented, all developed in the context of the projects PREDICT and PODCAST. In Chapter 4, the final step of analytics, that is, prescriptive analytics, is described together with some of the applications to the power system field. Battery profile optimization and optimal sizing and siting, developed as a functionality of a modern DMS, are the applications of this chapter and are drawn from the PODCAST project. Afterward, in the Conclusions, the key findings are summarized together with some final comments. Finally, the publications produced during the acs{phd}, and contributed projects, collaborations, and attended courses, are listed, while references cited in this thesis conclude the dissertation.
30-mag-2022
Power Systems; Data Science; Artificial Intelligence; Data Analytics; Descriptive Analytics; Predictive Analytics; Prescriptive Analytics; Forecasting; Electricity Markets; Energy Transition
File in questo prodotto:
File Dimensione Formato  
phdunige_3800294.pdf

accesso aperto

Tipologia: Tesi di dottorato
Dimensione 12.5 MB
Formato Adobe PDF
12.5 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11567/1082325
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact