Descripción
|
|
---|---|
The simulation of the flow around Low Pressure Turbine airfoils at realistic Reynolds numbers is addressed using a Flux Reconstruction (FR) solver for the compressible Navier- Stokes that is executed on multiple Graphics Processing Units (GPUs). The solver exploits the massively parallel performances of GPUs very efficiently due to the strong data locality of the spatial discretization. The use of the GPU shared memory is very useful to maximize the computational throughput. When multiple GPUs are used in parallel, the low communications bandwidth between GPUs becomes a major bottleneck. It is alleviated overlapping the data communication and the solver execution and implementing a pipelining technique to increase the communication speed. The resulting parallel solver is used to study the influence of the Reynolds number on the aerodynamic performance of a low pressure turbine airfoil. The numerical results correctly reproduce the experimental measurements. | |
Internacional
|
Si |
Nombre congreso
|
AIAA Aviation 2019 Forum |
Tipo de participación
|
960 |
Lugar del congreso
|
Dallas, Texas |
Revisores
|
Si |
ISBN o ISSN
|
|
DOI
|
https://doi.org/10.2514/6.2019-2840 |
Fecha inicio congreso
|
17/06/2019 |
Fecha fin congreso
|
21/06/2019 |
Desde la página
|
2840 |
Hasta la página
|
2840 |
Título de las actas
|
AIAA Aviation 2019 Forum |