Small Mediterranean islands represent a paradigmatic example of remote, off-grid systems facing a large number of sustainability issues, mainly due to their distance from the mainland, the lack of accessible water sources, and the high seasonal variability of both water and electricity demand. Energy security is generally reliant on carbon intensive diesel generators, which are usually oversized to meet peak summer electricity demand driven by high touristic fluxes. Potable water is often produced by energy intensive desalination technologies, which strongly impact on the electricity system, increasing air pollution and greenhouse gas emissions. In order to improve the economic and environmental sustainability of small islands, the design of hybrid energy systems, combining traditional power generation with renewable energy sources and storage technologies, represents a viable and promising solution. However, traditional methods for designing such systems usually neglect important aspects and challenges. Major challenges include (i) the optimal control of the electricity system as well as its interconnection with other energy vectors (e.g., gas, heat) and domains (e.g., water system) for fully exploiting renewable power, (ii) the interdependency between system planning and its operation, (iii) the presence of multiple, potentially conflicting, objectives reflecting economic, environmental and other sustainability aspects, (iv) deep uncertainty in climate, technological and socio-economic conditions that may affect the system performance over a medium-to-long term horizon. Driven by these challenges, this thesis contributes novel methodologies for supporting energy systems transition towards decarbonization, helping decision makers to identify viable solutions at different temporal scales in light of plausible future conditions that might unfold. In particular, we develop a set of modelling and optimization tools for optimizing both the design and the operations of off-grid water-energy systems, also considering the uncertainty related to future changes in the main external drivers. The proposed methodologies allow us to (i) investigate the benefits of explicitly considering the interdependency between system design and operation with respect to multiple economical, environmental and efficiency objectives, (ii) assess the vulnerability of hybrid energy systems to the future uncertainty in the main external drivers, and (iii) design solutions that are robust with respect to this uncertainty. The first deliverable of this research is a novel multi-objective, dynamic approach for conjunctively optimizing design and operation of water-energy systems by focusing on the interconnection between electricity generation and water supply through the optimal control of desalination plant. Secondly, we propose a methodological framework to assess the vulnerability of hybrid energy systems with respect to changes in the main climate drivers (i.e., solar radiation, wind speed, temperature). More precisely, we evaluate how historical variability and future uncertainty in these climate variables affect the performance of highly renewable hybrid energy systems, designed under average historical conditions, in terms of different sustainability indicators. Finally, we focus on the challenge of directly including deep uncertainty in future climate drivers within the system design phase. Since the performance of hybrid energy systems in small Mediterranean islands strictly depends on multiple, deeply uncertain co-varying drivers, a very large number of future scenarios should be considered and, consequently, included within the optimization process for generating robust solutions, leading to very high, or even intractable, computational time for solving the problem. To address this issue, we develop ROSS (Robust Optimal Scenario Selection), a novel algorithm that uses active learning for adaptively selecting the smallest scenario subset to be included into a robust optimization process. We test our novel approaches on the real case study of the Italian Ustica island, which represents a paradigmatic example of off-grid Mediterranean island. Main thesis outcomes show that considering the interdependency between system design and operation by dynamically modelling the nexus between water production and electricity generation allows to significantly improve system performance by reducing the structural interventions, the investment costs and the environmental impacts. In addition, results suggest that wind speed represents the climate variable that mainly influences the performance of hybrid energy systems, which will likely degrade on a medium-to-long term horizon. Finally, our novel ROSS algorithm allows to obtain robust hybrid energy system designs reducing computational requirements between 23% and 84% compared with traditional robust optimization methods, depending on the complexity of the robustness metrics considered. Moreover, it is able to identify very small regions of the scenario space containing the most informative scenarios highlighting the main system vulnerabilities.
Le piccole isole del Mediterraneo rappresentano un esempio paradigmatico di sistemi remoti fuori rete che affrontano un gran numero di problemi di sostenibilità, principalmente a causa della loro distanza dalla terraferma, della mancanza di sorgenti d'acqua accessibili e dell'elevata variabilità stagionale della domanda di acqua ed elettricità. La sicurezza energetica dipende generalmente da generatori diesel, che di solito sono sovradimensionati per soddisfare il picco della domanda di elettricità estiva dovuta agli elevati flussi turistici. L'acqua potabile è spesso prodotta da tecnologie di dissalazione altamente energivore, che incidono fortemente sul sistema elettrico, aumentando l'inquinamento atmosferico e le emissioni di gas serra. Al fine di migliorare la sostenibilità economica e ambientale delle piccole isole, la progettazione di sistemi energetici ibridi, che combinano generazione elettrica da fonti tradizionali con fonti di energia rinnovabili e tecnologie di stoccaggio, rappresenta una soluzione praticabile e promettente. Tuttavia, i metodi tradizionali per la progettazione di tali sistemi di solito trascurano aspetti importanti. I principali aspetti includono (i) il controllo ottimale del sistema elettrico e la sua interconnessione con altri vettori energetici (ad es. gas, calore) e sistemi (ad es. sistema idrico) per sfruttare pienamente la risorsa rinnovabile, (ii) l'interdipendenza tra la pianificazione del sistema e la sua gestione, (iii) la presenza di molteplici obiettivi, potenzialmente conflittuali, che riflettono aspetti economici, ambientali e di sostenibilità, (iv) la profonda incertezza in condizioni climatiche, tecnologiche e socio-economiche che possono influire sulle prestazioni del sistema in un orizzonte di medio-lungo termine. Spinto da queste sfide, questa tesi propone nuove metodologie per supportare la transizione dei sistemi energetici verso la decarbonizzazione, aiutando i decisori a identificare soluzioni praticabili su diverse scale temporali, alla luce di possibili condizioni future che potrebbero verificarsi. In particolare, sviluppiamo una serie di strumenti di modellizzazione e ottimizzazione per ottimizzare sia la progettazione che le operazioni dei sistemi acqua-energia fuori rete, anche considerando l'incertezza relativa ai futuri cambiamenti nelle principali forzanti esterne. Le metodologie proposte ci consentono di (i) studiare i vantaggi di considerare esplicitamente l'interdipendenza tra la progettazione del sistema e il suo funzionamento rispetto a molteplici obiettivi economici, ambientali e di efficienza, (ii) valutare la vulnerabilità dei sistemi energetici ibridi rispetto all'incertezza futura nelle principali forzanti esterne, e (iii) identificare soluzioni che siano robuste rispetto a questa incertezza. Il primo risultato di questa ricerca è un nuovo approccio dinamico, multi-obiettivo per ottimizzare congiuntamente la progettazione e il funzionamento dei sistemi energetici ibridi, concentrandosi sull'interconnessione tra produzione di elettricità e fornitura di acqua attraverso il controllo ottimale dell'impianto di dissalazione. In secondo luogo, proponiamo un quadro metodologico per valutare la vulnerabilità dei sistemi energetici ibridi rispetto ai cambiamenti nei principali fattori climatici (ovvero radiazione solare, velocità del vento, temperatura). Più precisamente, valutiamo in che modo la variabilità storica e l'incertezza futura nelle variabili climatiche influenzano le prestazioni dei sistemi energetici ibridi progettati in condizioni storiche medie, in termini di diversi indicatori di sostenibilità. Infine, ci concentriamo sull'inclusione della profonda incertezza nelle forzanti climatiche nella fase di pianificazione del sistema. Siccome le prestazioni dei sistemi energetici ibridi nelle piccole isole del Mediterraneo dipendono fortemente da molteplici forzanti incerte e co-varianti, un numero molto alto di scenari dovrebbe essere considerato e, di conseguenza, incluso nel processo di ottimizzazione per generare soluzioni robuste, portando a tempi computazionali molto alti o addirittura intrattabili. Per affrontare questo problema, sviluppiamo ROSS (Robust Optimal Scenario Selection), un nuovo algoritmo che utilizza l'apprendimento attivo per selezionare in modo adattivo il sottoinsieme di scenari più piccolo da includere in un processo di ottimizzazione robusta. Testiamo i nostri nuovi approcci sul caso di studio reale dell'isola italiana di Ustica, che rappresenta un esempio paradigmatico di isola del Mediterraneo fuori rete. I principali risultati della tesi mostrano che considerare l'interdipendenza tra la progettazione e il funzionamento del sistema modellando dinamicamente il nesso tra produzione di acqua e generazione di elettricità consente di migliorare significativamente le prestazioni del sistema, riducendo gli interventi strutturali, i costi di investimento e gli impatti ambientali. Inoltre, i risultati suggeriscono che la velocità del vento rappresenta la variabile climatica che influenza maggiormente le prestazioni dei sistemi energetici ibridi, le quali molto probabilmente degraderanno in un orizzonte a medio-lungo termine. Infine, il nostro nuovo algoritmo ROSS consente di ottenere soluzioni di pianificazione robuste riducendo i requisiti computazionali tra il 23% e l'84% rispetto ai metodi tradizionali, a seconda della complessità delle metriche di robustezza considerate. Inoltre, ROSS è in grado di identificare regioni molto piccole dello spazio degli scenari contenenti gli scenari che evidenziano le principali vulnerabilità del sistema.
Optimal design of off-grid water-energy systems in small Mediterranean islands
GIUDICI, FEDERICO
Abstract
Small Mediterranean islands represent a paradigmatic example of remote, off-grid systems facing a large number of sustainability issues, mainly due to their distance from the mainland, the lack of accessible water sources, and the high seasonal variability of both water and electricity demand. Energy security is generally reliant on carbon intensive diesel generators, which are usually oversized to meet peak summer electricity demand driven by high touristic fluxes. Potable water is often produced by energy intensive desalination technologies, which strongly impact on the electricity system, increasing air pollution and greenhouse gas emissions. In order to improve the economic and environmental sustainability of small islands, the design of hybrid energy systems, combining traditional power generation with renewable energy sources and storage technologies, represents a viable and promising solution. However, traditional methods for designing such systems usually neglect important aspects and challenges. Major challenges include (i) the optimal control of the electricity system as well as its interconnection with other energy vectors (e.g., gas, heat) and domains (e.g., water system) for fully exploiting renewable power, (ii) the interdependency between system planning and its operation, (iii) the presence of multiple, potentially conflicting, objectives reflecting economic, environmental and other sustainability aspects, (iv) deep uncertainty in climate, technological and socio-economic conditions that may affect the system performance over a medium-to-long term horizon. Driven by these challenges, this thesis contributes novel methodologies for supporting energy systems transition towards decarbonization, helping decision makers to identify viable solutions at different temporal scales in light of plausible future conditions that might unfold. In particular, we develop a set of modelling and optimization tools for optimizing both the design and the operations of off-grid water-energy systems, also considering the uncertainty related to future changes in the main external drivers. The proposed methodologies allow us to (i) investigate the benefits of explicitly considering the interdependency between system design and operation with respect to multiple economical, environmental and efficiency objectives, (ii) assess the vulnerability of hybrid energy systems to the future uncertainty in the main external drivers, and (iii) design solutions that are robust with respect to this uncertainty. The first deliverable of this research is a novel multi-objective, dynamic approach for conjunctively optimizing design and operation of water-energy systems by focusing on the interconnection between electricity generation and water supply through the optimal control of desalination plant. Secondly, we propose a methodological framework to assess the vulnerability of hybrid energy systems with respect to changes in the main climate drivers (i.e., solar radiation, wind speed, temperature). More precisely, we evaluate how historical variability and future uncertainty in these climate variables affect the performance of highly renewable hybrid energy systems, designed under average historical conditions, in terms of different sustainability indicators. Finally, we focus on the challenge of directly including deep uncertainty in future climate drivers within the system design phase. Since the performance of hybrid energy systems in small Mediterranean islands strictly depends on multiple, deeply uncertain co-varying drivers, a very large number of future scenarios should be considered and, consequently, included within the optimization process for generating robust solutions, leading to very high, or even intractable, computational time for solving the problem. To address this issue, we develop ROSS (Robust Optimal Scenario Selection), a novel algorithm that uses active learning for adaptively selecting the smallest scenario subset to be included into a robust optimization process. We test our novel approaches on the real case study of the Italian Ustica island, which represents a paradigmatic example of off-grid Mediterranean island. Main thesis outcomes show that considering the interdependency between system design and operation by dynamically modelling the nexus between water production and electricity generation allows to significantly improve system performance by reducing the structural interventions, the investment costs and the environmental impacts. In addition, results suggest that wind speed represents the climate variable that mainly influences the performance of hybrid energy systems, which will likely degrade on a medium-to-long term horizon. Finally, our novel ROSS algorithm allows to obtain robust hybrid energy system designs reducing computational requirements between 23% and 84% compared with traditional robust optimization methods, depending on the complexity of the robustness metrics considered. Moreover, it is able to identify very small regions of the scenario space containing the most informative scenarios highlighting the main system vulnerabilities.File | Dimensione | Formato | |
---|---|---|---|
Giudici2019_phdDissertation.pdf
accessibile in internet per tutti
Descrizione: Thesis text
Dimensione
10.37 MB
Formato
Adobe PDF
|
10.37 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/151454