In-flight icing is a complex, multi-physics phenomenon that occurs when an aircraft encounters a cloud of supercooled droplets. The ice modifies the aerodynamic shape of lifting surfaces, which may lead to a significant loss in aerodynamic efficiency through decreased lift, increased drag, a reduced stall angle, and degraded stability and controllability. Flight sensors and probes may freeze, delivering misleading information to pilots. Icing is a safety concern that must be addressed by aircraft designers early in the design loop, especially when designing new aircraft configurations. Regulators require aircraft manufacturers to comply with ever stricter requirements regarding in-flight icing. Most airworthy vehicles are usually certified for operations in icing hazard conditions. Several experimental methods may be used to certify the airworthiness of an aircraft under icing conditions, such as flight in natural icing, flight with simulated ice shapes, and wind tunnel tests. However, conducting planned certification tests in natural icing conditions is challenging because of the unpredictability of natural icing events and their high cost. On the other hand, icing wind tunnels can control the operating parameters (temperature, pressure, velocity, diameter, and concentration of the supercooled water droplets) but are still very expensive and have operational and model size limitations. Even well-established facilities may have up to $20\%$ uncertainty in cloud parameters, their limited test section allows only for sub-scale or component-level testing, and there is no consensus on the scaling laws of aircraft icing due to the complex multi-physics of the phenomenon, particularly when dealing with testing ice protection systems. Over the last decades, computational capacity and numerical tools have allowed for the development of ice prediction tools. Simple configurations can be studied along with a wide range of parameters compared to the limited scope of experimental research. In addition, these tools can predict and determine the most critical ice shapes more safely and cheaply compared to actual flight or wind tunnel testing. The numerical modeling of this phenomenon generally involves a sequential call to different modules, including mesh generation, aerodynamics, droplet trajectories, ice accretion, and geometry update. The automation of this process is critical because the solvers are embedded in a time loop, which is repeated several times (multi-step) to obtain an accurate ice shape prediction. Nowadays, research focuses on ice prediction on three-dimensional configurations that were previously limited because of the computational effort required and the complexity in performing automatic multi-step simulations. Indeed, three-dimensional icing models are not yet widely used due to their lack of efficiency and robustness, as the primary difficulty is often the manual remeshing and the geometry updating. This thesis presents a novel level-set-based methodology to address the issue of surface mesh entanglements and intersections that frequently arise when moving the ice front using standard mesh deformation techniques. Once the ice growth is computed over the aircraft surface, a new, good-quality, and body-fitted mesh is retrieved, both for 2D and 3D applications, necessary to assess the aerodynamic penalties after a single-step computation or to keep iterating the multi-step simulation. Combining wrapping techniques and the implicit definition of the new surface enables multi-step morphogenetic computations over arbitrary complex three-dimensional geometries. The mass conservation problem during the ice shape evolution is tackled, presenting the benefits introduced by multi-step computations and the morphogenetic approach. When transitioning to multi-step simulations, a challenging task is to assess the convergence of the simulated ice shape, which may depend on several factors, including the number of time steps employed and the surface discretization. An automatic and adaptive method for selecting the duration of each time step is proposed, eliminating the need for an a priori educated guess from the user. Multi-step simulations are then employed to assess the spatial and temporal convergence of ice shapes. This work also addresses the problem of ice shape convergence in morphogenetic methods, particularly when dealing with glaze ice, where the grid discretization can significantly impact the output of the random walk model. The analysis showed the benefits of multi-step computations, particularly when simulating mixed and glaze ice shapes. A very challenging task for ice accretion codes is to capture the location and thickness of glaze horns, or complex features such as rime feathers. Multi-step morphogenetic simulations are employed to investigate the numerical growth of such features, providing valuable insights into the physics of the ice accretion mechanisms. Finally, one of the ultimate goals of icing simulations is to generate ice shapes characterized by the same aerodynamic performance as those from wind tunnel and in-flight tests, but today it remains unclear what level of geometric detail is necessary for the scope. The final part of the work investigates, through mid- and high-fidelity simulations, the level of geometrical detail an ice shape requires to accurately predict the aerodynamic performance degradation and, therefore, be employed effectively for certification purposes.
L’accrescimento di ghiaccio in volo è un fenomeno complesso e multifisico che si verifica quando un velivolo attraversa una nube contenente goccioline d’acqua sopraffuse. Il ghiaccio che si forma modifica il profilo aerodinamico delle superfici portanti, causando una perdita significativa di efficienza aerodinamica, con riduzione della portanza, aumento della resistenza, diminuzione dell’angolo di stallo e peggioramento della stabilità e della controllabilità. Anche i sensori e le sonde di bordo possono ghiacciarsi, fornendo informazioni errate ai piloti. Il fenomeno rappresenta un serio problema di sicurezza che deve essere affrontato fin dalle fasi iniziali della progettazione, soprattutto nel caso di nuove configurazioni aeronautiche. Le autorità di regolamentazione richiedono ai costruttori di velivoli il rispetto di requisiti sempre più stringenti riguardo al volo in condizioni di ghiaccio. La maggior parte dei velivoli certificati è abilitata al volo in presenza di tali condizioni pericolose. Diversi metodi sperimentali possono essere utilizzati per certificare la navigabilità in presenza di ghiaccio, tra cui voli in condizioni naturali, test con forme di ghiaccio simulate e prove in galleria del vento. Tuttavia, pianificare test in ghiaccio naturale è difficile, sia per l’imprevedibilità del fenomeno, sia per gli alti costi. Le gallerie del vento per il ghiaccio permettono di controllare i parametri operativi (temperatura, pressione, velocità, diametro e concentrazione delle goccioline sopraffuse), ma restano comunque costose e soggette a limitazioni operative e dimensionali. Anche le strutture più avanzate possono avere un’incertezza fino al 20% nei parametri della nube, sezioni di prova limitate che consentono solo test in scala ridotta o su singoli componenti, e non esiste ancora un consenso sulle leggi di similitudine per l’ice accretion, a causa della complessità multifisica del fenomeno, in particolare nel caso dei sistemi anti-ghiaccio. Negli ultimi decenni, l’aumento della capacità computazionale e lo sviluppo di strumenti numerici hanno permesso la creazione di modelli predittivi del ghiaccio. Tali strumenti consentono di analizzare configurazioni semplificate con una gamma di parametri più ampia rispetto a quella accessibile tramite esperimenti. Inoltre, permettono di identificare in modo sicuro ed economico le forme di ghiaccio più critiche, rispetto a prove in volo o in galleria del vento. La simulazione numerica del fenomeno prevede tipicamente una sequenza di moduli — generazione della mesh, aerodinamica, traiettorie delle goccioline, accrescimento del ghiaccio e aggiornamento della geometria. L’automazione di questo processo è fondamentale, poiché i solutori sono inseriti in un ciclo temporale che deve essere ripetuto più volte (multi-step) per ottenere una previsione accurata della forma del ghiaccio. Oggi la ricerca si concentra sulla previsione del ghiaccio su configurazioni tridimensionali, in passato trascurate per l’elevato costo computazionale e la difficoltà di automatizzare le simulazioni multi-step. I modelli 3D non sono ancora ampiamente utilizzati per via della loro bassa efficienza e robustezza, dovute principalmente alla necessità di remeshing e aggiornamento geometrico manuale. Questa tesi propone una nuova metodologia basata su level-set per affrontare i problemi di intersezione e distorsione della mesh superficiale che si verificano frequentemente quando si fa avanzare il fronte di ghiaccio con tecniche standard di deformazione della mesh. Una volta calcolata la crescita del ghiaccio sulla superficie del velivolo, viene generata una nuova mesh conforme, di buona qualità, sia in 2D che in 3D, necessaria per valutare il degrado delle prestazioni aerodinamiche dopo un singolo step oppure per continuare la simulazione multi-step. Combinando tecniche di wrapping con una definizione implicita della nuova superficie, si rendono possibili simulazioni morfogenetiche multi-step su geometrie tridimensionali complesse. Il problema della conservazione della massa durante l’evoluzione della forma del ghiaccio viene affrontato, evidenziando i vantaggi introdotti dalle simulazioni multi-step e dall’approccio morfogenetico. Uno degli aspetti più difficili nel passaggio a simulazioni multi-step è valutare la convergenza della forma di ghiaccio simulata, che può dipendere da vari fattori come il numero di passi temporali e la discretizzazione superficiale. Si propone un metodo automatico e adattivo per selezionare la durata di ciascun passo temporale, eliminando la necessità di una stima a priori da parte dell’utente. Le simulazioni multi-step sono quindi impiegate per valutare la convergenza spaziale e temporale delle forme di ghiaccio. Il lavoro affronta inoltre il problema della convergenza delle forme di ghiaccio nei metodi morfogenetici, in particolare nel caso della galaverna (glaze ice), in cui la discretizzazione della griglia può influenzare sensibilmente l’uscita del modello a passeggiata aleatoria (random walk). L’analisi ha mostrato i vantaggi delle simulazioni multi-step, soprattutto nella simulazione di forme miste e di ghiaccio trasparente. Una delle sfide più complesse per i codici di accrescimento del ghiaccio è catturare con precisione la posizione e lo spessore dei corni di ghiaccio, o di strutture complesse come le "rime feathers". Le simulazioni morfogenetiche multi-step sono utilizzate per indagare la crescita numerica di tali caratteristiche, fornendo preziose indicazioni sulla fisica dei meccanismi di accrescimento. Infine, uno degli obiettivi principali delle simulazioni di formazione del ghiaccio è generare forme che riproducano le prestazioni aerodinamiche osservate nei test in galleria del vento e in volo. Tuttavia, resta ancora incerto il livello di dettaglio geometrico necessario a questo scopo. L’ultima parte del lavoro analizza, tramite simulazioni a media e alta fedeltà, il grado di dettaglio geometrico richiesto affinché una forma di ghiaccio consenta una previsione accurata della degradazione delle prestazioni aerodinamiche, e quindi possa essere efficacemente impiegata nei processi di certificazione.
Multi-step simulations of in-flight icing towards certification by analysis
Donizetti, Alessandro
2024/2025
Abstract
In-flight icing is a complex, multi-physics phenomenon that occurs when an aircraft encounters a cloud of supercooled droplets. The ice modifies the aerodynamic shape of lifting surfaces, which may lead to a significant loss in aerodynamic efficiency through decreased lift, increased drag, a reduced stall angle, and degraded stability and controllability. Flight sensors and probes may freeze, delivering misleading information to pilots. Icing is a safety concern that must be addressed by aircraft designers early in the design loop, especially when designing new aircraft configurations. Regulators require aircraft manufacturers to comply with ever stricter requirements regarding in-flight icing. Most airworthy vehicles are usually certified for operations in icing hazard conditions. Several experimental methods may be used to certify the airworthiness of an aircraft under icing conditions, such as flight in natural icing, flight with simulated ice shapes, and wind tunnel tests. However, conducting planned certification tests in natural icing conditions is challenging because of the unpredictability of natural icing events and their high cost. On the other hand, icing wind tunnels can control the operating parameters (temperature, pressure, velocity, diameter, and concentration of the supercooled water droplets) but are still very expensive and have operational and model size limitations. Even well-established facilities may have up to $20\%$ uncertainty in cloud parameters, their limited test section allows only for sub-scale or component-level testing, and there is no consensus on the scaling laws of aircraft icing due to the complex multi-physics of the phenomenon, particularly when dealing with testing ice protection systems. Over the last decades, computational capacity and numerical tools have allowed for the development of ice prediction tools. Simple configurations can be studied along with a wide range of parameters compared to the limited scope of experimental research. In addition, these tools can predict and determine the most critical ice shapes more safely and cheaply compared to actual flight or wind tunnel testing. The numerical modeling of this phenomenon generally involves a sequential call to different modules, including mesh generation, aerodynamics, droplet trajectories, ice accretion, and geometry update. The automation of this process is critical because the solvers are embedded in a time loop, which is repeated several times (multi-step) to obtain an accurate ice shape prediction. Nowadays, research focuses on ice prediction on three-dimensional configurations that were previously limited because of the computational effort required and the complexity in performing automatic multi-step simulations. Indeed, three-dimensional icing models are not yet widely used due to their lack of efficiency and robustness, as the primary difficulty is often the manual remeshing and the geometry updating. This thesis presents a novel level-set-based methodology to address the issue of surface mesh entanglements and intersections that frequently arise when moving the ice front using standard mesh deformation techniques. Once the ice growth is computed over the aircraft surface, a new, good-quality, and body-fitted mesh is retrieved, both for 2D and 3D applications, necessary to assess the aerodynamic penalties after a single-step computation or to keep iterating the multi-step simulation. Combining wrapping techniques and the implicit definition of the new surface enables multi-step morphogenetic computations over arbitrary complex three-dimensional geometries. The mass conservation problem during the ice shape evolution is tackled, presenting the benefits introduced by multi-step computations and the morphogenetic approach. When transitioning to multi-step simulations, a challenging task is to assess the convergence of the simulated ice shape, which may depend on several factors, including the number of time steps employed and the surface discretization. An automatic and adaptive method for selecting the duration of each time step is proposed, eliminating the need for an a priori educated guess from the user. Multi-step simulations are then employed to assess the spatial and temporal convergence of ice shapes. This work also addresses the problem of ice shape convergence in morphogenetic methods, particularly when dealing with glaze ice, where the grid discretization can significantly impact the output of the random walk model. The analysis showed the benefits of multi-step computations, particularly when simulating mixed and glaze ice shapes. A very challenging task for ice accretion codes is to capture the location and thickness of glaze horns, or complex features such as rime feathers. Multi-step morphogenetic simulations are employed to investigate the numerical growth of such features, providing valuable insights into the physics of the ice accretion mechanisms. Finally, one of the ultimate goals of icing simulations is to generate ice shapes characterized by the same aerodynamic performance as those from wind tunnel and in-flight tests, but today it remains unclear what level of geometric detail is necessary for the scope. The final part of the work investigates, through mid- and high-fidelity simulations, the level of geometrical detail an ice shape requires to accurately predict the aerodynamic performance degradation and, therefore, be employed effectively for certification purposes.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Tesi_Donizetti.pdf
non accessibile
Dimensione
123.8 MB
Formato
Adobe PDF
|
123.8 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/10589/241179