As our energy infrastructure evolves, microgrids present a compelling vision for a more resilient and sustainable future. Unlocking their full economic potential, however, hinges on solving a critical challenge: how to optimally manage a diverse array of energy resources in real-time, especially when the microgrid is disconnected from the maingrid. This thesis tackles that challenge head-on, providing a rigorous performance analysis of several key optimization strategies applied to the economic load dispatch (ELD) problem within a benchmark islanded microgrid. To establish a baseline, this research first implements and compares three well-regarded metaheuristic algorithms, including Grey Wolf Optimizer (GWO), Particle Swarm Optimization (PSO), and Cuckoo Search Algorithm (CSA), on a static 24-hour dispatch problem. Building on this analysis, the work then investigates a dynamic approach using a simplified Model Predictive Control (MPC) framework to understand the operational trade-offs of a rolling-horizon strategy. A sensitivity analysis that quantified the financial effects of variations in fuel prices, load demand, and renewable energy production further confirmed the optimization framework's resilience. A key finding is the remarkable consistency of the metaheuristic methods; GWO, PSO, and CSA all independently discovered near-identical, cost-optimal dispatch schedules, confirming their reliability for this application. In contrast, the MPC strategy yielded a slightly higher total cost, a result that clearly illustrates the trade-off between the global knowledge of static optimization and the more limited, albeit adaptive, foresight of predictive control. Furthermore, the sensitivity analysis revealed that while the system’s operating cost is highly dependent on the availability of renewable sources, it is surprisingly resilient to sharp increases in diesel fuel prices. Ultimately, this thesis contributes a direct, replicable performance comparison on a published benchmark model, clarifying the practical strengths and trade-offs of these vital control strategies and offering a firmer foundation for future work in adaptive energy management.
La gestione economica ottimale delle microreti operanti in isola è una sfida cruciale per la transizione verso un'infrastruttura energetica resiliente e sostenibile. Questa tesi affronta tale sfida analizzando il problema del dispacciamento economico del carico (ELD) su una microrete di riferimento. A tal fine, vengono implementati e confrontati tre algoritmi metaeuristici, Grey Wolf Optimizer (GWO), Particle Swarm Optimization (PSO) e Cuckoo Search Algorithm (CSA), per l'ottimizzazione statica su 24 ore. Viene inoltre valutato un approccio dinamico tramite un controllo predittivo basato su modello (MPC) semplificato, e la robustezza del sistema è testata con un'analisi di sensibilità. I risultati evidenziano la notevole affidabilità degli algoritmi metaeuristici, che convergono verso soluzioni a costo ottimale quasi identiche. La strategia MPC, pur risultando leggermente più costosa, chiarisce il compromesso tra la pianificazione statica globale e il controllo predittivo ad orizzonte limitato. L'analisi di sensibilità dimostra inoltre come i costi operativi siano fortemente influenzati dalla disponibilità delle fonti rinnovabili, ma notevolmente resilienti alle variazioni del prezzo del gasolio. Il lavoro fornisce quindi un confronto diretto e replicabile delle prestazioni su un modello di riferimento pubblicato, offrendo una base solida per lo sviluppo futuro di strategie di gestione energetica adattiva.
Economic dispatch and sensitivity analysis of an islanded microgrid using metaheuristic optimization and model predictive control
ASADIAN, SINA
2024/2025
Abstract
As our energy infrastructure evolves, microgrids present a compelling vision for a more resilient and sustainable future. Unlocking their full economic potential, however, hinges on solving a critical challenge: how to optimally manage a diverse array of energy resources in real-time, especially when the microgrid is disconnected from the maingrid. This thesis tackles that challenge head-on, providing a rigorous performance analysis of several key optimization strategies applied to the economic load dispatch (ELD) problem within a benchmark islanded microgrid. To establish a baseline, this research first implements and compares three well-regarded metaheuristic algorithms, including Grey Wolf Optimizer (GWO), Particle Swarm Optimization (PSO), and Cuckoo Search Algorithm (CSA), on a static 24-hour dispatch problem. Building on this analysis, the work then investigates a dynamic approach using a simplified Model Predictive Control (MPC) framework to understand the operational trade-offs of a rolling-horizon strategy. A sensitivity analysis that quantified the financial effects of variations in fuel prices, load demand, and renewable energy production further confirmed the optimization framework's resilience. A key finding is the remarkable consistency of the metaheuristic methods; GWO, PSO, and CSA all independently discovered near-identical, cost-optimal dispatch schedules, confirming their reliability for this application. In contrast, the MPC strategy yielded a slightly higher total cost, a result that clearly illustrates the trade-off between the global knowledge of static optimization and the more limited, albeit adaptive, foresight of predictive control. Furthermore, the sensitivity analysis revealed that while the system’s operating cost is highly dependent on the availability of renewable sources, it is surprisingly resilient to sharp increases in diesel fuel prices. Ultimately, this thesis contributes a direct, replicable performance comparison on a published benchmark model, clarifying the practical strengths and trade-offs of these vital control strategies and offering a firmer foundation for future work in adaptive energy management.File | Dimensione | Formato | |
---|---|---|---|
2025_07_Asadian_Thesis.pdf
accessibile in internet per tutti
Dimensione
1.19 MB
Formato
Adobe PDF
|
1.19 MB | Adobe PDF | Visualizza/Apri |
2025_07_Asadian_Executive_Summary.pdf
accessibile in internet per tutti
Dimensione
845.15 kB
Formato
Adobe PDF
|
845.15 kB | 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/10589/240393