The rapid electrification of transport is intensifying the demand for lithium-ion batteries that are high-performing, reliable, and safe. For electric vehicles and aerospace applications, Battery Management Systems (BMSs) and system design must rely on predictive, physics-aware models and experimentally validated algorithms to enable optimal battery sizing, robust online monitoring, and mission-level assurance. This doctoral dissertation, presented as a collection of peer-reviewed and draft articles, advances modelling, estimation, and model-based decision tools for Li-ion batteries targeted at high-performance and safety-critical applications. The thesis is organised along three complementary axes. The first axis develops models and experimental identification protocols for a broad range of battery dynamics: electrochemical, thermal and ageing. For the electrochemical work, non-invasive identification strategies are proposed for reduced Single Particle Models (SPMs) applied to different cells; these models outperform classical equivalent-circuit models (ECMs) in prediction and provide insight into otherwise unobservable internal states. The thermal work focuses on a prototypical high-power pouch cell for e-racing applications and spans models of increasing fidelity (from lumped to 2-D distributed), tailored identification protocols and tools for numerical stability analysis. Finally, ageing is experimentally characterised on a cell selected for an eVTOL application: an empirically calibrated cycle-ageing model is used to predict battery-pack lifetime under repeated mission profiles. The second axis focuses on real-time estimation. Two contributions are included: a new second-order sliding-mode algorithm applied to State of Charge (SoC) estimation, demonstrating strong robustness and accuracy; and a particle-filter algorithm for joint SoC/State of Health (SoH) estimation, calibrated on pre-aged cells and validated with good performance across varied operating conditions.\\ The third axis addresses applied design problems in powertrain and battery-pack sizing for helicopter applications. Two studies are presented at increasing levels of maturity: a methodology for preliminary conceptual sizing of a hybrid-electric powertrain for a light helicopter/UAV, and a pack-optimization method driven by experimental characterisation of promising cells and validated through mission-scaled testing, yielding practical insights for integration and thermal–durability trade-offs. \medskip These contributions combine experimental analysis, physics-based modelling and the development of estimation and decision algorithms to improve battery monitoring and the design of electric propulsion systems. The resulting model-based techniques can help create more efficient, higher-performing and safer systems, and support the electrification of vehicles.
La rapida elettrificazione dei trasporti sta aumentando la richiesta di batterie agli ioni di litio con elevate prestazioni, affidabilità e sicurezza. Per i veicoli elettrici e le applicazioni aerospaziali, i sistemi di gestione delle batterie (BMS) e il progetto dei sottosistemi devono basarsi su modelli fisici predittivi e su algoritmi convalidati sperimentalmente, in modo da permettere un corretto dimensionamento, un monitoraggio online affidabile e garanzie a livello di missione. Questa tesi, presentata come raccolta di articoli pubblicati e in bozza, propone strumenti per la modellazione, la stima e il supporto alle decisioni rivolti a batterie Li-ion per applicazioni ad alte prestazioni e con requisiti stringenti di sicurezza. La tesi è strutturata su tre assi complementari. Il primo asse sviluppa modelli e protocolli di identificazione per diverse dinamiche delle celle: elettrochimica, termica e invecchiamento. Per l’elettrochimica sono proposte strategie non invasive per l’identificazione di modelli ridotti di tipo Single Particle (SPM); questi modelli offrono previsioni migliori rispetto ai classici modelli a circuito equivalente (ECM) e consentono di inferire stati interni altrimenti non misurabili. Il lavoro termico riguarda una cella pouch ad alta potenza per applicazioni e-racing: si sviluppano modelli di fedeltà crescente (dal lumped ai modelli 2-D distribuiti), protocolli di identificazione dedicati e strumenti per l’analisi della stabilità numerica. Infine, l’invecchiamento è caratterizzato sperimentalmente su una cella scelta per applicazioni eVTOL: un modello empirico di degradazione da cicli viene calibrato per prevedere la vita utile del pacco sotto profili di missione ripetuti. Il secondo asse è incentrato sulla stima in tempo reale. Sono proposti due contributi principali: nuovo algoritmo sliding-mode di secondo ordine applicato alla stima dello State of Charge (SoC), che mostra elevata robustezza e accuratezza; e un particle filter per la stima congiunta di SoC e State of Health (SoH), calibrato su celle pre-invecchiate e validato in diverse condizioni operative. Il terzo asse affronta problemi applicativi di progettazione del powertrain e di dimensionamento del pacco batteria per impieghi elicotteristici. Il primo studio propone una metodologia per il dimensionamento concettuale di un powertrain ibrido-elettrico per un UAV/elicottero leggero; il secondo introduce un metodo di ottimizzazione del pacco basato su caratterizzazioni sperimentali di celle promettenti e test di missione scalati, fornendo indicazioni pratiche per l’integrazione e i compromessi tra gestione termica e durabilità. \medskip Nel complesso, questi lavori combinano analisi sperimentali, modelli fisici e algoritmi di stima e decisione per migliorare il monitoraggio e il progetto dei sistemi di propulsione elettrica. Le tecniche proposte possono supportare pacchi batteria più efficienti, performanti e sicuri e favorire la transizione verso la mobilità elettrica.
Advanced modeling and state estimation for Battery Management Systems
Trivella, Andrea
2025/2026
Abstract
The rapid electrification of transport is intensifying the demand for lithium-ion batteries that are high-performing, reliable, and safe. For electric vehicles and aerospace applications, Battery Management Systems (BMSs) and system design must rely on predictive, physics-aware models and experimentally validated algorithms to enable optimal battery sizing, robust online monitoring, and mission-level assurance. This doctoral dissertation, presented as a collection of peer-reviewed and draft articles, advances modelling, estimation, and model-based decision tools for Li-ion batteries targeted at high-performance and safety-critical applications. The thesis is organised along three complementary axes. The first axis develops models and experimental identification protocols for a broad range of battery dynamics: electrochemical, thermal and ageing. For the electrochemical work, non-invasive identification strategies are proposed for reduced Single Particle Models (SPMs) applied to different cells; these models outperform classical equivalent-circuit models (ECMs) in prediction and provide insight into otherwise unobservable internal states. The thermal work focuses on a prototypical high-power pouch cell for e-racing applications and spans models of increasing fidelity (from lumped to 2-D distributed), tailored identification protocols and tools for numerical stability analysis. Finally, ageing is experimentally characterised on a cell selected for an eVTOL application: an empirically calibrated cycle-ageing model is used to predict battery-pack lifetime under repeated mission profiles. The second axis focuses on real-time estimation. Two contributions are included: a new second-order sliding-mode algorithm applied to State of Charge (SoC) estimation, demonstrating strong robustness and accuracy; and a particle-filter algorithm for joint SoC/State of Health (SoH) estimation, calibrated on pre-aged cells and validated with good performance across varied operating conditions.\\ The third axis addresses applied design problems in powertrain and battery-pack sizing for helicopter applications. Two studies are presented at increasing levels of maturity: a methodology for preliminary conceptual sizing of a hybrid-electric powertrain for a light helicopter/UAV, and a pack-optimization method driven by experimental characterisation of promising cells and validated through mission-scaled testing, yielding practical insights for integration and thermal–durability trade-offs. \medskip These contributions combine experimental analysis, physics-based modelling and the development of estimation and decision algorithms to improve battery monitoring and the design of electric propulsion systems. The resulting model-based techniques can help create more efficient, higher-performing and safer systems, and support the electrification of vehicles.| File | Dimensione | Formato | |
|---|---|---|---|
|
Thesis.pdf
non accessibile
Descrizione: tesi di dottorato
Dimensione
45.6 MB
Formato
Adobe PDF
|
45.6 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/10589/255577