Many end-use applications of hydrogen, such as fuel cells, operate at low pressure, typically below 4 bar(g). However, for storage and transportation, hydrogen is compressed to very high pressure, up to 700 bar(g). In most current systems, the pressure reduction from storage to end use is achieved through throttling valves, which dissipate the energy previously invested in compression. This work builds upon two previous researches investigating the recovery of part of this energy by converting it into electricity. The objective is to adapt an established technology, widely employed in other industrial sectors, namely multi-vane machines, and to evaluate its power output and efficiency when applied to hydrogen expansion from approximately 40 to 10 bar(g). To reduce cost and risk, similitude theory is applied, allowing experiments to be carried out with air instead of hydrogen. A dedicated test bench is designed and built for this purpose, enabling data acquisition at inlet pressures between 3 and 21.7 bar(g), focusing on expansion ratio of 4. These measurements are used to construct a dimensionless performance map. To improve the prediction accuracy, several artificial neural network architectures are developed and trained on the experimental dataset. A subsequent sensitivity analysis identifies the best-performing neural network, which achieves a mean relative error of 1.46% with respect to the test dataset, outperforming traditional fitting methods. Finally, a potential assembly configuration to minimize hydrogen leakages and improve generator cooling is designed, in accordance with the main European standards. The test condition closer to similitude with the target application corresponds to air operation at an inlet pressure of 18 bar(g) and an expansion ratio of 4. The expander achieves an efficiency of 35%, significantly higher than that obtained with the machine tested in previous work. However, challenges are encountered regarding the high mass flow rate required by the volumetric machine, exceeding the capacity of the test facility, and the limited mechanical strength of internal components, which are not originally designed for the higher operating pressures required by this application.
Molte applicazioni dell’idrogeno, come le celle a combustibile, operano a bassa pressione, tipicamente sotto di 4 bar(g). Tuttavia, per lo stoccaggio e il trasporto, l’idrogeno viene compresso a pressioni molto elevate, fino a 700 bar(g). Nella maggior parte dei sistemi attuali, la riduzione di pressione dallo stoccaggio all’utilizzo avviene tramite valvole di laminazione, che dissipano l’energia precedentemente spesa per la compressione. Questo lavoro prosegue le attività di due precedenti ricerche incentrate sul recupero di parte di tale energia sotto forma di elettricità. L’obiettivo è adattare una tecnologia consolidata in altri settori industriali, il motore a palette multiple, e valutarne potenza ed efficienza quando applicata all’espansione dell’idrogeno da circa 40 a 10 bar(g). Per ridurre costi e rischi è stata applicata la teoria della similitudine, consentendo di condurre prove in aria anziché idrogeno. A tal fine è stato progettato e realizzato un banco prova che ha permesso l’acquisizione di dati a pressioni di ingresso comprese tra 3 e 21.7 bar(g), con rapporto di espansione pari a 4. Le misure ottenute sono state utilizzate per costruire una mappa prestazionale adimensionale. Per migliorarne la precisione, sono state sviluppate e addestrate diverse architetture di reti neurali artificiali. Un’analisi di sensibilità ha individuato la rete con le migliori prestazioni, che ha raggiunto un errore relativo medio dell’1,46% rispetto ai dati sperimentali, superando i metodi di interpolazione tradizionali. Infine, è stata progettata una possibile configurazione dell’assieme, volta a ridurre le perdite di idrogeno e migliorare il raffreddamento del generatore, in conformità con i principali standard europei. La condizione di prova più prossima alla similitudine con l’applicazione finale corrispondeva a un funzionamento in aria con pressione di ingresso di 18 bar(g) e rapporto di espansione pari a 4. L’espansore ha raggiunto un’efficienza del 35%, significativamente superiore a quella ottenuta con la macchina testata in precedenti lavori. Tuttavia, sono emerse alcune criticità legate all’elevata portata massica richiesta dalla macchina volumetrica, superiore alla capacità dell’impianto, e alla limitata resistenza meccanica delle palette, non progettati per le pressioni di questa applicazione.
Rotary-vane expander for waste exergy recovery: experimental characterization of a full-scale machine and artificial neural network modeling
CONCA, GABRIELE;Artioli Bonati, Lorenzo
2024/2025
Abstract
Many end-use applications of hydrogen, such as fuel cells, operate at low pressure, typically below 4 bar(g). However, for storage and transportation, hydrogen is compressed to very high pressure, up to 700 bar(g). In most current systems, the pressure reduction from storage to end use is achieved through throttling valves, which dissipate the energy previously invested in compression. This work builds upon two previous researches investigating the recovery of part of this energy by converting it into electricity. The objective is to adapt an established technology, widely employed in other industrial sectors, namely multi-vane machines, and to evaluate its power output and efficiency when applied to hydrogen expansion from approximately 40 to 10 bar(g). To reduce cost and risk, similitude theory is applied, allowing experiments to be carried out with air instead of hydrogen. A dedicated test bench is designed and built for this purpose, enabling data acquisition at inlet pressures between 3 and 21.7 bar(g), focusing on expansion ratio of 4. These measurements are used to construct a dimensionless performance map. To improve the prediction accuracy, several artificial neural network architectures are developed and trained on the experimental dataset. A subsequent sensitivity analysis identifies the best-performing neural network, which achieves a mean relative error of 1.46% with respect to the test dataset, outperforming traditional fitting methods. Finally, a potential assembly configuration to minimize hydrogen leakages and improve generator cooling is designed, in accordance with the main European standards. The test condition closer to similitude with the target application corresponds to air operation at an inlet pressure of 18 bar(g) and an expansion ratio of 4. The expander achieves an efficiency of 35%, significantly higher than that obtained with the machine tested in previous work. However, challenges are encountered regarding the high mass flow rate required by the volumetric machine, exceeding the capacity of the test facility, and the limited mechanical strength of internal components, which are not originally designed for the higher operating pressures required by this application.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Conca_Artioli_Bonati_Tesi.pdf
accessibile in internet per tutti a partire dal 16/11/2028
Descrizione: Tesi
Dimensione
12.13 MB
Formato
Adobe PDF
|
12.13 MB | Adobe PDF | Visualizza/Apri |
|
2025_12_Conca_Artioli_Bonati_Executive_Summary.pdf
accessibile in internet per tutti a partire dal 16/11/2028
Descrizione: Executive Summary
Dimensione
1.81 MB
Formato
Adobe PDF
|
1.81 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/10589/246292