Peak pressure events have long been observed during wind tunnel tests on the facades of scaled buildings. This thesis discusses challenges related to modeling of turbulent flows, in practical employment of Large-Eddie Simulation with Smagorinsky turbulence model, and provides new insights into the phenomena of peak pressure events, as the results are benchmarked against previous research work and validated through the experiments conducted at the wind tunnel of Politecnico di Milano. CFD analysis has been performed using OpenFoam open-source software running on CINECA HPC cluster. A precursor simulation method was adopted to generate the input turbulent field with characteristics similar to the ones during wind tunnel tests. Several models with various mesh refinements have been realized and the sensitivity analysis of the mesh size has been performed in order to evaluate the effect of subgrid scale filter and to provide quantitative measures for the quality of the numerical simulation. The data obtained in this study enabled to correlate the turbulent structures present in the flow field to the extreme pressure events, and provided a further understanding of their nature. Vortex identification was performed though implementation of Q-criterion as well as a computation of the velocity field curl. Moreover, the preliminary benchmark tests as well as simulation results confirmed the viability of a precursor method in generating the atmospheric boundary layer input for the CFD models, and its potential in further research. At the same time the limitations of the current numerical modeling techniques of turbulent flows are explored, such as the sensitivity to the boundary conditions, computational costs and difficulties associated to the definition of the effects of inadequate meshing and of implicit filtering in Large-Eddie Simulations.
Gli eventi di picco di pressione (Peak Pressure) sono stati a lungo osservati durante i test in galleria del vento sulle facciate degli edifici in scala. Questa tesi discute le sfide relative alla modellizzazione dei flussi turbolenti, nell'impiego pratico di Large Eddie Simulations (LES) con il modello di turbolenza Smagorinsky, e fornisce nuovi spunti sul fenomeno di Peak Pressure, in quanto i risultati sono confrontati con la riceraca precedente e convalidati attraverso gli esperimenti condotti presso la galleria del vento del Politecnico di Milano. L'analisi CFD è stata eseguita utilizzando il software open source OpenFoam sul HPC CINECA cluster. E' stato adottato un metodo di simulazione precursore per generare il campo turbolento di input con caratteristiche simili a quelle dei test in galleria del vento. Sono stati realizzati diversi modelli con vari perfezionamenti delle maglie e l'analisi della sensibilità delle dimensioni delle maglie è stata eseguita alla fine di valutare l'effetto del filtro di sottogriglia e di fornire misure quantitative per la qualità della simulazione numerica stessa. I dati ottenuti in questo studio hanno permesso di correlare le strutture turbolente presenti nel flusso agli eventi di estrema pressione e hanno fornito un'ulteriore comprensione della loro natura. L'identificazione dei vortici è stata eseguita attraverso l'implementazione del criterio Q ed atraverso un calcolo del rottore del campo di velocità. Inoltre, i test preliminari di valutazione ed i risultati delle simulazioni hanno confermato la validità del metodo precursore nella generazione dell'input dello strato limite atmosferico (ABL) per i modelli CFD ed il suo potenziale in ulteriori ricerche. Allo stesso tempo vengono esplorati le limitazioni delle attuali tecniche di modellazione numerica dei flussi turbolenti, come la sensibilità alle condizioni al contorno, i costi computazionali e le difficoltà associate alla definizione degli effetti di magliatura inadeguata e dei filtri impliciti nelle simulazioni LES.
CFD study of pressure peaks on building facades
MARYKOVSKIY, YURIY
2017/2018
Abstract
Peak pressure events have long been observed during wind tunnel tests on the facades of scaled buildings. This thesis discusses challenges related to modeling of turbulent flows, in practical employment of Large-Eddie Simulation with Smagorinsky turbulence model, and provides new insights into the phenomena of peak pressure events, as the results are benchmarked against previous research work and validated through the experiments conducted at the wind tunnel of Politecnico di Milano. CFD analysis has been performed using OpenFoam open-source software running on CINECA HPC cluster. A precursor simulation method was adopted to generate the input turbulent field with characteristics similar to the ones during wind tunnel tests. Several models with various mesh refinements have been realized and the sensitivity analysis of the mesh size has been performed in order to evaluate the effect of subgrid scale filter and to provide quantitative measures for the quality of the numerical simulation. The data obtained in this study enabled to correlate the turbulent structures present in the flow field to the extreme pressure events, and provided a further understanding of their nature. Vortex identification was performed though implementation of Q-criterion as well as a computation of the velocity field curl. Moreover, the preliminary benchmark tests as well as simulation results confirmed the viability of a precursor method in generating the atmospheric boundary layer input for the CFD models, and its potential in further research. At the same time the limitations of the current numerical modeling techniques of turbulent flows are explored, such as the sensitivity to the boundary conditions, computational costs and difficulties associated to the definition of the effects of inadequate meshing and of implicit filtering in Large-Eddie Simulations.File | Dimensione | Formato | |
---|---|---|---|
thesis.pdf
non accessibile
Descrizione: Thesis text
Dimensione
20.3 MB
Formato
Adobe PDF
|
20.3 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/10589/146478