Andrea Beck, Claus-Dieter Munz
Institute of Aerodynamics and Gasdynamics, University of Stuttgart
Local Project ID:
HPC Platform used:
Hazel Hen of HLRS
Turbomachines are complex systems that are indispensable for energy generation and propulsion in a wide range of sectors. Due to their tightly integrated architecture and rotating parts, experimental investigations of the flow phenomena through those machines is at best difficult and costly and at worst currently not feasible yet. Large scale numerical simulation can help here: It allows researchers and engineers a look at the intricate flow patterns inside these devices and helps to understand, predict and influence them. This knowledge can then be harnessed by the designers and developers to improve efficiency, reliability and safety and to reduce the ecological footprint of these machines.
Simulating the flow through turbomachines or their components is a challenging task. As the fluid passes through the system, it encounters a range of drastic changes in a very short time: It is accelerated and decelerated rapidly in the rotating components, it is compressed, heated, mixed with fuel and combustion takes places. Naturally, this leads to very complex flow patterns with a high degree of irregularity and unsteadiness. This requires numerical methods with excellent scale-resolving capabilities and high efficiency on parallel systems. From a computational perspective, the rotating components pose an additional challenge: The communication pattern between the rotating and stationary parts of the geometry also become dynamic.
In the current project, researchers of the Institute of Aerodynamics and Gas Dynamics at the University of Stuttgart develop numerical methods to conduct Large Eddy Simulations of flows in turbomachinery components, in particular with efficient coupling of stator and rotor domains. The complexities of the flow field require a very high resolution of the important scales in space and time, which needs a large computational effort and can only be achieved on massively parallel systems.
An example of such a computation is shown in Fig. 1. Here, the flow through a stator – rotor - stator system of a turbine is shown. The inflow Mach number is about 0.1, and the Reynolds number based on the exit velocity and second stator chord is approx. 800000. The scientists use the FLEXI framework based on a Discontinuous Galerkin spectral element discretization. It is based on a high order polynomial approximation of the solution within each grid element. A newly developed sliding mesh interface allows for a consistent inclusion of arbitrary rotating domains, while keeping the excellent scaling properties of the original implementation.
For the setup shown above, a 6th order ansatz was chosen, leading to 208 million degrees of freedom on a grid with close to 1 million grid cells. All computations were run on the Cray HPC System Hazel Hen at HLRS. The simulation costs per one rotor period (so called blade passing frequency) were about 27,000 CPUh, with the number of cores ranging from 4000 to 20000. Similar simulations with the same setup are currently already running on HAWK, the next supercomputer at HLRS.
Based on these simulation results, the complex transition and separation behaviour of the flow in turbomachines will be examined in this ongoing project. In the future, this knowledge will help to improve the design of these components and – further down the road – of the full system.
A video showing the vortical structures and acoustic waves derived from this project can be found at https://www.youtube.com/watch?v=7qwaQTwJS9M
a) Andrea Beck, Philip Ortwein, Patrick Kopper, Nico Krais, Daniel Kempf, Christian Koch (2019) Towards high-fidelity erosion prediction: On time-accurate particle tracking in turbomachinery, Manuscript submitted for publication in International Journal of Heat and Fluid Flow
b) Beck, Bolemann, Flad, Frank, Krais, Kukuschkin, Sonntag, Munz (2018) Application and Development of the High Order Discontinuous Galerkin Spectral Element Method for Compressible Multiscale Flows in High Performance Computing in Science and Engineering ’ 17, pp. 387–407, Springer, Cham
c) Krais, Nico, et al. "FLEXI: A high order discontinuous Galerkin framework for hyperbolic-parabolic conservation laws." arXiv preprint arXiv:1910.02858 (2019).
Research Team & Scientific Contact:
Dr.-Ing. Andrea Beck, Prof. Dr. Claus-Dieter Munz
Institute of Aerodynamics and Gas Dynamics, University of Stuttgart
Pfaffenwaldring 21, D-70569 Stuttgart (Germany)
HLRS project ID: HPCDG