The increasing penetration of intermittent renewable power sources necessitates the development of robust long-term storage solutions to ensure grid stability and energy security. While lithium-ion batteries are highly effective for short-term storage, long-term storage requires different solutions. Hydrogen is one of the most promising among them, due to its high energy density and scalability. This thesis focuses on modeling a hydrogen storage system, specifically the fuel cell, and integrating it into a broader control architecture based on a hierarchical Model Predictive Control (MPC). First, an experimental campaign was conducted to characterize the fuel cell and refine thermal, voltage, and efficiency models. These revealed significant deviations from nominal manufacturer specifications, identifying a gross efficiency of 48% (excluding auxiliaries), and a peak net efficiency of 32 %. The campaign was also challenging as the fuel cell had hardware limitations, which led to frequent flooding events and caused many problems during startup. To mitigate these issues, the real-time MPC layer was modified to incorporate logic capable of handling this unpredictable behavior, using a battery buffer to compensate for the consequences of these faults. The algorithm was then tested experimentally, with necessary simplifications due to laboratory constraints. Subsequently, it was validated through a year-long simulation of the Ginostra microgrid, a small town on Stromboli island. Two scenarios were analyzed: one with a smaller tank and one that enabled fully renewable generation. The comparison between the two scenarios demonstrated the effectiveness of the algorithm and highlighted the challenges related to seasonality in the context of intermittent power generation, as capacity requirements increase significantly to achieve complete energy autonomy.
La crescente penetrazione di fonti rinnovabili intermittenti necessita di soluzioni robuste per l'accumulo di energia nel lungo periodo, al fine di garantire la stabilità della rete e la sicurezza energetica. Mentre le batterie agli ioni di litio sono altamente efficaci per l'accumulo a breve termine, quello a lungo termine richiede soluzioni diverse. L'idrogeno rappresenta uno dei candidati più promettenti, grazie alla sua elevata densità energetica e scalabilità. Questa tesi si concentra sulla modellazione di un sistema di accumulo ad idrogeno, in particolare la cella a combustibile, e sulla sua integrazione in una più ampia architettura di controllo, basata sul Model Predictive Control (MPC). Innanzitutto è stata condotta una campagna sperimentale, per caratterizzare la cella a combustione e affinare i modelli termici, della tensione e dell'efficienza. Questa ha rivelato una deviazione significativa dalle caratteristiche nominali indicate dal produttore, identificando un'efficienza lorda del 48% (esclusi gli ausiliari), ed un'efficienza netta massima del 32%. La campagna ha inoltre evidenziato limitazioni hardware, che spesso causavano allagamenti e instabilità durante l'avvio. Per mitigare questi problemi, il livello MPC real-time è stato modificato per incorporare una logica in grado di gestire questi comportamenti imprevedibili, utilizzando una batteria tampone per compensare le conseguenze di questi guasti. L'algoritmo è stato poi testato sperimentalmente, anche se con delle semplificazioni dovute a limitazioni nel laboratorio. Successivamente, è stato validato attraverso una simulazione di un anno della microrete di Ginostra (isola di Stromboli). Sono stati analizzati due scenari: uno con un serbatoio più piccolo, e uno con un serbatoio di capacità sufficiente da permettere una generazione completamente rinnovabile. Il confronto tra questi due scenari ha dimostrato l'efficacia dell'algoritmo, ed ha evidenziato le sfide legate alla stagionalità nel contesto della generazione intermittente, poiché i requisiti di accumulo crescono significativamente per raggiungere la completa autonomia energetica.
Fuel cell modeling for microgrid control applications
Berruti, Francesco
2024/2025
Abstract
The increasing penetration of intermittent renewable power sources necessitates the development of robust long-term storage solutions to ensure grid stability and energy security. While lithium-ion batteries are highly effective for short-term storage, long-term storage requires different solutions. Hydrogen is one of the most promising among them, due to its high energy density and scalability. This thesis focuses on modeling a hydrogen storage system, specifically the fuel cell, and integrating it into a broader control architecture based on a hierarchical Model Predictive Control (MPC). First, an experimental campaign was conducted to characterize the fuel cell and refine thermal, voltage, and efficiency models. These revealed significant deviations from nominal manufacturer specifications, identifying a gross efficiency of 48% (excluding auxiliaries), and a peak net efficiency of 32 %. The campaign was also challenging as the fuel cell had hardware limitations, which led to frequent flooding events and caused many problems during startup. To mitigate these issues, the real-time MPC layer was modified to incorporate logic capable of handling this unpredictable behavior, using a battery buffer to compensate for the consequences of these faults. The algorithm was then tested experimentally, with necessary simplifications due to laboratory constraints. Subsequently, it was validated through a year-long simulation of the Ginostra microgrid, a small town on Stromboli island. Two scenarios were analyzed: one with a smaller tank and one that enabled fully renewable generation. The comparison between the two scenarios demonstrated the effectiveness of the algorithm and highlighted the challenges related to seasonality in the context of intermittent power generation, as capacity requirements increase significantly to achieve complete energy autonomy.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Berruti_Ex_summary.pdf
accessibile in internet per tutti
Descrizione: Executive summary
Dimensione
1.42 MB
Formato
Adobe PDF
|
1.42 MB | Adobe PDF | Visualizza/Apri |
|
2025_12_Berruti_thesis.pdf
accessibile in internet per tutti
Descrizione: Tesi
Dimensione
5.71 MB
Formato
Adobe PDF
|
5.71 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/10589/247143