National Technical University of Athens (NTUA)
Ανάπτυξη μεθοδολογίας ψευδοσυμπιεστότητας για την πρόλεξη μη μόνιμων τρισδιάστατων αεροδυναμικών ροών
Abstract
dc:descriptionThe aim of this work is the development of an artificial compressibility method for the prediction of 3D unsteady aerodynamic flows. Τhe Reynolds Averaged Equations (RANS) for incompressible flow are solved. The Reynolds stresses are modeled using the Boussinesq assumption for the eddy viscosity, which is computed from the Spalart-Allmaras, k-ω-TNT and k-ω-SST turbulence models. Equations are discretized using a conservative finite volume scheme with hybrid unstructured grids. A third order upwind interpolation scheme is used for the inviscid fluxes, while a second order central scheme is used for the viscous fluxes. A second order backward implicit time discretization scheme is used in physical time while in pseudotime a first order backward implicit Euler scheme is used. For the gradients computation the least-squares method and Green-Gauss theorem are used. The non linear system resulting from the discretization process is solved using Newton’s method. An acceleration method is also developed for the data structure construction needed by the serial flow solver. Parallel processing is also used for convergence acceleration. The parallel algorithm is optimized for data exchange, good scalability and deadlock avoidance. The solver is tested using known flow problems from literature and comparisons are made with corresponding numerical and experimental results. Τhe solver’s observed spatial and temporal order of accuracy is determined and mesh transparency is examined. The method proved to be mesh transparent with second order observed spatial and temporal accuracy. The numerical results of the present methodology were in good agreement with corresponding numerical and experimental results from literature. For the turbulent flows examined, the use of the least-squares method for data reconstruction proved to be more robust compared to Green-Gauss divergence theorem. The acceleration method for data structure creation gave significant speedup. The scaling of the parallel solver increases with increasing computational load and deadlocks are avoided. Speedup factors between 20÷24 were achieved using 32 processes.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2012
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Vrahliotis, Sokratis
- Βραχλιώτης, Σωκράτης
Subjects
dc:subject × 16- Ασυμπίεστη ροή
- Τεχνητή συμπιεστότητα
- Υβριδικό πλέγμα
- Παράλληλη επεξεργασία
- Μοντέλα τύρβης
- Υπολογιστική ρευστοδυναμική
- Incompressible flow
- Artificial compressibility
- Hybrid processing
- Parallel processing
- Turbulence models
- Computational fluid dynamics
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Μηχανολόγου Μηχανικού
- Engineering and Technology
- Mechanical Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/31786