The expansion in supersonic nozzles plays a fundamental role in determining rocket performances. Computational Fluid Dynamics is a useful tool to predict the flow field inside a nozzle, avoiding the drawbacks of the experiments. The computational time required by a numerical simulation represents one of the major limitations of this approach, leading in some cases to prohibitive costs. In the present work, OpenFOAM code is employed to perform reacting and non-reacting 2D-axisymmetric simulations by means of two solvers, rhoCentralFoam and rhoPimpleFoam, which implement density-based and pressure-based methods, respectively. The solver rhoCentralFoam is considered by the OpenFOAM community the best option to deal with supersonic flows, thanks to its density-based logics and shock capturing features. Nevertheless, its explicit scheme implies the use of small time steps making it a very time-consuming method. On the other side, rhoPimpleFoam allows the use of larger time steps, being based on the PIMPLE algorithm, but it is conceived to solve transonic flows at most. The goal of the present study is to establish if rhoPimpleFoam is able to deal with wall-bounded subsonic-to-supersonic expansion nozzle flows and to define the limits of its applicability. Two de Laval nozzle configurations are investigated: the first with a small throat radius of curvature operating at optimal conditions, the second with a large throat radius of curvature operating at over-expanded flow conditions. Both the test cases are simulated for inviscid and viscous flow conditions, and a mesh sensitivity analysis is performed. Given the interest of the aerospace community in studying the chemical processes in supersonic nozzles, a further test is conducted by considering a reacting flow inside the nozzle operating at optimal conditions. The pressure-based solver reactingFoam is employed and the new density-based solver rhoCentralReactingFoam is implemented. The results of the present study suggest that the pressure-based solvers are able to solve the flow field in the adapted nozzle with the same accuracy obtained by the density-based solvers and with a much lower computational cost. On the other hand, the pressure-based solvers result to be inadequate to deal with supersonic flows involving strong shock waves. The mesh sensitivity analysis on the over-expanded nozzle reveals that rhoPimpleFoam can actually capture the normal shock wave but it requires an extremely fine grid and, thus, its time-saving skills are completely lost. In conclusion, the results of the present work suggest the use of the OpenFOAM pressure-based solvers for the study of supersonic flows in de Laval nozzles whenever strong flow discontinuities are not involved, since it leads to a large decrease in the computational cost, without penalty on accuracy.
Il processo di espansione in ugelli supersonici ha importanti effetti sulle performance dei razzi. La fluidodinamica computazionale risulta uno strumento utile per predire il campo di moto all'interno di un ugello, evitando gli inconvenienti derivanti dagli studi sperimentali. Tuttavia, il tempo di calcolo richiesto da una simulazione numerica rappresenta uno dei maggiori limiti di tale approccio, portando spesso a costi proibitivi. Si utilizza OpenFOAM per effettuare simulazioni di flussi reagenti e non reagenti su geometrie 2D-assialsimmetriche per mezzo di due solver, rhoCentralFoam e rhoPimpleFoam, che implementano rispettivamente i metodi density-based e pressure-based. Il solver rhoCentralFoam è comunemente considerato come la migliore opzione per risolvere flussi supersonici, grazie al suo approccio density-based e alle sua capacità di shock capturing. Ciononostante, il suo schema esplicito richiede l'utilizzo di piccoli time step, rendendolo un solver decisamente costoso in termini di tempi computazionali. Dall'altro lato rhoPimpleFoam, basato sull'algoritmo PIMPLE, permette l'uso di time step più grandi, ma è concepito per risolvere, al massimo, flussi transonici. L'obiettivo del presente lavoro consiste nel determinare se rhoPimpleFoam sia in grado di trattare l'espansione del flusso subsonico-supersonico all'interno dell'ugello e stabilirne i limiti di applicabilità. Si analizzano due configurazioni di ugello di de Laval: la prima con piccolo raggio di curvatura della gola in condizioni di flusso adattato, la seconda con ampio raggio di curvatura della gola in condizioni sovraespanse. Entrambi i casi sono simulati in condizioni di flusso non viscoso e viscoso e si esegue un'analisi di sensitività sulla mesh. Dato l'interesse in campo aerospaziale nello studio dei processi chimici in ugelli supersonici, si effettua una simulazione di flusso reagente nell'ugello adattato. A tale scopo, si utilizza il solver pressure-based reactingFoam e si implementa rhoCentralReactingFoam, un nuovo solver density-based. I risultati del presente studio suggeriscono che i solver pressure-based sono in grado di predire il comportamento del flusso supersonico nell'ugello in condizioni di adattamento con la stessa accuratezza fornita dai solver density-based e ad un costo computazionale decisamente inferiore. Dall'altro lato, i solver pressure-based risultano inadeguati all'analisi di flussi supersonici che prevedono la presenza di forti onde d'urto. L'analisi di sensitività sulla mesh nel caso di ugello sovraespanso rivela che rhoPimpleFoam è in effetti capace di simulare l'onda d'urto, ma l'estrema raffinatezza richiesta alla mesh rende nulli i vantaggi di questo solver in termini di tempi di calcolo. In conclusione, i risultati del presente lavoro consigliano l'utilizzo dei solver pressure-based di OpenFOAM per la risoluzione di flussi supersonici in ugelli di de Laval ogniqualvolta non siano contemplate forti onde d'urto nel flusso, poichè porta a grandi vantaggi in termini di costo computazionale.
A comparison between pressure-based and density-based solvers in OpenFOAM applied to the expansion in supersonic nozzles
MESSA, TOMMASO
2018/2019
Abstract
The expansion in supersonic nozzles plays a fundamental role in determining rocket performances. Computational Fluid Dynamics is a useful tool to predict the flow field inside a nozzle, avoiding the drawbacks of the experiments. The computational time required by a numerical simulation represents one of the major limitations of this approach, leading in some cases to prohibitive costs. In the present work, OpenFOAM code is employed to perform reacting and non-reacting 2D-axisymmetric simulations by means of two solvers, rhoCentralFoam and rhoPimpleFoam, which implement density-based and pressure-based methods, respectively. The solver rhoCentralFoam is considered by the OpenFOAM community the best option to deal with supersonic flows, thanks to its density-based logics and shock capturing features. Nevertheless, its explicit scheme implies the use of small time steps making it a very time-consuming method. On the other side, rhoPimpleFoam allows the use of larger time steps, being based on the PIMPLE algorithm, but it is conceived to solve transonic flows at most. The goal of the present study is to establish if rhoPimpleFoam is able to deal with wall-bounded subsonic-to-supersonic expansion nozzle flows and to define the limits of its applicability. Two de Laval nozzle configurations are investigated: the first with a small throat radius of curvature operating at optimal conditions, the second with a large throat radius of curvature operating at over-expanded flow conditions. Both the test cases are simulated for inviscid and viscous flow conditions, and a mesh sensitivity analysis is performed. Given the interest of the aerospace community in studying the chemical processes in supersonic nozzles, a further test is conducted by considering a reacting flow inside the nozzle operating at optimal conditions. The pressure-based solver reactingFoam is employed and the new density-based solver rhoCentralReactingFoam is implemented. The results of the present study suggest that the pressure-based solvers are able to solve the flow field in the adapted nozzle with the same accuracy obtained by the density-based solvers and with a much lower computational cost. On the other hand, the pressure-based solvers result to be inadequate to deal with supersonic flows involving strong shock waves. The mesh sensitivity analysis on the over-expanded nozzle reveals that rhoPimpleFoam can actually capture the normal shock wave but it requires an extremely fine grid and, thus, its time-saving skills are completely lost. In conclusion, the results of the present work suggest the use of the OpenFOAM pressure-based solvers for the study of supersonic flows in de Laval nozzles whenever strong flow discontinuities are not involved, since it leads to a large decrease in the computational cost, without penalty on accuracy.File | Dimensione | Formato | |
---|---|---|---|
Messa Tommaso.pdf
non accessibile
Descrizione: Confronto tra pressure-based e density-based solver in OpenFOAM per lo studio di flussi supersonici in ugelli di de Laval
Dimensione
7.25 MB
Formato
Adobe PDF
|
7.25 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/10589/153769