Reactive and high-speed flows involve structures of varying spatial scales and can benefit from adaptive grid methods for reducing computational demands. Though very flexible, the traditional Adaptive Mesh Refinement (AMR) method lacks error control, and its refinement criteria are often adjusted iteratively, without guarantee of achieving a desired accuracy. Finite-volume wavelet methods aim to solve this issue by taking advantage of the multiresolution representation offered by Cartesian grids, combining it with wavelet multi-scale decomposition, compression and error control theory for adapting the grid, placing finer cells where the solution varies greatly. The performance of these methods is investigated in OpenFOAM over different shock problems, addressing the portability from their typically custom-tailored environment to widely available numerical packages and industrial applications who could benefit their accuracy. While differences in solution treatment may cause high-order wavelets to be overly sensitive, the Haar wavelet demonstrates reliable grid adaptation, avoiding over- or under-refinement issues from AMR and achieving similar accuracy with an order of magnitude less cells and significantly compressing uniform grid solutions. Time consuming threshold parametrization and set-up adjustments are nonetheless required, limiting its practical use.
I flussi reattivi e ad alta velocità coinvolgono strutture di diversa scala spaziale e possono trarre vantaggio dai metodi di griglia adattativa per ridurre i requisiti computazionali. Pur essendo molto flessibile, il tradizionale metodo AMR (Adaptive Mesh Refinement) manca di controllo degli errori e i suoi criteri di raffinamento sono spesso regolati iterativamente, senza garanzia di raggiungere l'accuratezza desiderata. I metodi wavelet a volume finito mirano a risolvere questo problema sfruttando la rappresentazione multirisoluzione offerta dalle griglie cartesiane, combinandola con la decomposizione wavelet multi-scala, la compressione e la teoria del controllo degli errori per adattare la griglia, posizionando celle più fini dove la soluzione varia notevolmente. Le prestazioni di questi metodi sono studiate in OpenFOAM su diversi problemi di shock, affrontando la portabilità dal loro ambiente tipicamente personalizzato a pacchetti numerici ampiamente disponibili e ad applicazioni industriali che potrebbero beneficiare della loro accuratezza. Sebbene le differenze nel trattamento delle soluzioni possano causare un'eccessiva sensibilità delle wavelet di ordine superiore, la wavelet di Haar dimostra un adattamento affidabile alla griglia, evitando i problemi di sovra o sotto-raffinamento dell'AMR e ottenendo un'accuratezza simile con un ordine di grandezza in meno di celle e comprimendo significativamente le soluzioni a griglia uniforme. Tuttavia, la parametrizzazione della soglia e le regolazioni di set-up richiedono molto tempo e limitano l'uso pratico di questo metodo.
On the implementation of adaptive finite-volume wavelet methods in OpenFOAM
EMBERGER, EVAN JOHANN
2023/2024
Abstract
Reactive and high-speed flows involve structures of varying spatial scales and can benefit from adaptive grid methods for reducing computational demands. Though very flexible, the traditional Adaptive Mesh Refinement (AMR) method lacks error control, and its refinement criteria are often adjusted iteratively, without guarantee of achieving a desired accuracy. Finite-volume wavelet methods aim to solve this issue by taking advantage of the multiresolution representation offered by Cartesian grids, combining it with wavelet multi-scale decomposition, compression and error control theory for adapting the grid, placing finer cells where the solution varies greatly. The performance of these methods is investigated in OpenFOAM over different shock problems, addressing the portability from their typically custom-tailored environment to widely available numerical packages and industrial applications who could benefit their accuracy. While differences in solution treatment may cause high-order wavelets to be overly sensitive, the Haar wavelet demonstrates reliable grid adaptation, avoiding over- or under-refinement issues from AMR and achieving similar accuracy with an order of magnitude less cells and significantly compressing uniform grid solutions. Time consuming threshold parametrization and set-up adjustments are nonetheless required, limiting its practical use.File | Dimensione | Formato | |
---|---|---|---|
2024_12_Emberger_Thesis_01.pdf
accessibile in internet per tutti
Descrizione: thesis main text
Dimensione
14.65 MB
Formato
Adobe PDF
|
14.65 MB | Adobe PDF | Visualizza/Apri |
2024_12_Emberger_Executive_Summary_02.pdf
accessibile in internet per tutti
Descrizione: executive summary text
Dimensione
2.47 MB
Formato
Adobe PDF
|
2.47 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/10589/230624