The aim of this thesis is the comparison between two solvers, whose purpose is the optimization of the geometry of heat exchangers: the first solver can perform single region optimizations, whereas the second one performs multi-region optimizations. These solvers employ the continuous adjoint method and optimize the geometry with respect to the total pressure losses and the temperature extracted minimizing a properly defined objective function. The solvers are implemented in OpenFOAM-dev, the single region solver, called thermalAdjointShapeOptimizationFoam is an expanded version of the already present adjointShapeOptimizationFoam, whereas the multi-region solver adjointMultiregionFoam is the expansion of the single region solver for multiple regions problems, following the same structure of the already present chtMultiRegionFoam. In this document, after the presentation of the mathematical formalism that is behind the continuous adjoint optimization applied to a problem comprising fluid and solid regions, the structure of the multiple region solver itself is presented. After that, the geometry chosen for the comparison is presented, this latter is a heat exchanger devoted to cooling down the electric motor's inverter of the Formula SAE car of team Dynamis. The comparison of the results given by the two solvers highlights the best convergence speed of the single region solver, whereas the second solver produces geometry changes in the same zone as the first solver, but in general different, with a higher computational cost; given the differences in the results, an experimental verification would be beneficial to assess which geometry is the best.
L'obiettivo di questa tesi è comparare due solver il cui scopo è l'ottimizzazione della forma degli scambiatori di calore: il primo solver preso in esame ha la capacità di gestire solo problemi fluidi a singola regione, mentre il secondo riesce a risolvere casi con multiple regioni, siano esse solide o fluide. Questi solver impiegano il metodo dell'aggiunto continuo e ottimizzano la geometria sia da un punto di vista delle perdite di pressione totali che da un punto di vista della temperatura estratta minimizzando un'appositamente definita funzione obiettivo. I solver sono implementati in OpenFOAM-dev; il solver monoregione, chiamato thermalAdjointShapeOptimizationFoam è implementato sulla base del già presente adjointShapeOptimizationFoam, mentre il solver multiregione adjointMultiregionFoam è una espansione del solver monoregione, al quale viene aggiunta la possibilità di risolvere geometrie a più regioni seguendo la stessa struttura del già presente chtMultiRegionFoam. In questo documento, dopo la presentazione del formalismo matematico alla base del metodo dell'aggiunto continuo, viene presentata la struttura del codice di adjointMultiRegionFoam stesso. Dopodiché viene mostrata la geometria scelta per effettuare la comparazione, la quale rappresenta uno scambiatore di calore il cui scopo è quello di raffreddare l'inverter dei motori elettrici della vettura di Formula SAE realizzata dal team Dynamis. La comparazione dei risultati evidenzia la più rapida velocità di convergenza del solver monoregione, mentre il secondo solver genera un cambiamento di geometria nella stessa sfera d'influenza del primo solver ma in generale diversi, con un più alto costo computazionale; date le differenze nei risultati, una verifica sperimentale sarebbe necessaria per valutare quale tra le due geometrie sia la migliore.
A comparison between single and multi-region continuous adjoint optimization for couple thermal fluid problems
Bianchi, Fabio
2021/2022
Abstract
The aim of this thesis is the comparison between two solvers, whose purpose is the optimization of the geometry of heat exchangers: the first solver can perform single region optimizations, whereas the second one performs multi-region optimizations. These solvers employ the continuous adjoint method and optimize the geometry with respect to the total pressure losses and the temperature extracted minimizing a properly defined objective function. The solvers are implemented in OpenFOAM-dev, the single region solver, called thermalAdjointShapeOptimizationFoam is an expanded version of the already present adjointShapeOptimizationFoam, whereas the multi-region solver adjointMultiregionFoam is the expansion of the single region solver for multiple regions problems, following the same structure of the already present chtMultiRegionFoam. In this document, after the presentation of the mathematical formalism that is behind the continuous adjoint optimization applied to a problem comprising fluid and solid regions, the structure of the multiple region solver itself is presented. After that, the geometry chosen for the comparison is presented, this latter is a heat exchanger devoted to cooling down the electric motor's inverter of the Formula SAE car of team Dynamis. The comparison of the results given by the two solvers highlights the best convergence speed of the single region solver, whereas the second solver produces geometry changes in the same zone as the first solver, but in general different, with a higher computational cost; given the differences in the results, an experimental verification would be beneficial to assess which geometry is the best.File | Dimensione | Formato | |
---|---|---|---|
tesi_Fabio_Bianchi.pdf
Open Access dal 25/06/2023
Dimensione
1.52 MB
Formato
Adobe PDF
|
1.52 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/10589/190458