Simulation of Turbulent Flows Interacting with Non-Spherical Particles

Principal Investigator:
Wolfgang Schröder

Institute of Aerodynamics, RWTH Aachen University (Germany)

Local Project ID:
GCS-SOPF (HLRS) and hac31 (JSC)

HPC Platform used:
Hazel Hen of HLRS and JUQUEEN of JSC

Date published:

Researchers of the Institute of Aerodynamics (AIA) at RWTH Aachen University conducted large-scale benchmark simulations on supercomputer Hazel Hen of the High-Performance Computing Center Stuttgart to analyze the interaction of non-spherical particles with turbulent flows. These simulations provide a unique data base for the development of simple models which can be applied to study complex engineering problems. Such models are required in a larger research framework to improve the efficiency of pulverized coal and biomass combustion to significantly reduce the CO2 emissions.

Although the renewable energy production experiences a continuous growth, medium-term goals as well as long-term goals of CO2 reduction are challenging, which applies even more for developing and emerging countries. To predict the required measures for the climate change goals, the International Energy Agency (IEA) develops scenarios to fulfill the goals of the Paris Agreements. In these scenarios, a combination of higher process efficiency, energy savings, substitution of carbon fuel by carbon neutral fuels, and in particular carbon capturing and storage (CCS) technologies are necessary to fulfill the climate change goals. The combustion of biomass may even serve as a CO2 sink in the overall budget. However, CCS technologies are still in a development phase. The development of such a CCS technology is the motivation for the framework "Oxyflame", where this project is embedded into. In this framework, the combustion of coal and biomass is decomposed into fundamental processes interacting on different scales, such that each process is studied in generic environments by specialized research groups, and the results have universal applicability.

In this project, large-scale benchmark simulations are performed which provide insights into interaction phenomena between non-spherical particles and turbulent flows. Such a particle-laden turbulent flow is encountered in combustion chambers where the non-spherical particles consist of needle-like biomass. Biomass particles have a complex shape and are approximately of the size of the smallest turbulent scales. The full resolution of all turbulent and all particle scales including the complex particle shape is a challenging task, which requires modern numerical methods and the access to high-performance computing systems.

Figure 1 shows a turbulent flow field laden with needle-like particles. Up to 60,000 particles are fully resolved to guarantee converged statistics. Additionally, the particles have a major impact on the turbulent flow field which can only be observed, if a sufficient number of particles are present.

Figure 2 shows a close-up view including the computational mesh which is used for the discretization. The mesh is locally refined in the vicinity of the particles to provide an accurate and efficient computation of the interaction between the particles and the flow field. Up to 8 billion Cartesian cells are required for the full resolution of the turbulent flow field and the particles.

The highly-resolved simulations performed in this project require the extensive usage of HPC systems. The simulations are performed on 48,000 processors on the HLRS HPC system Hazel Hen, and a single snap shot of the turbulent flow field requires more than 300 GB of disk space. Post-processing is therefore performed during the computation to avoid I/O overhead. These large-scale benchmark simulations are used to develop simplified particle models for industrial applications.

Project Team and Scientific Contact

M.Sc. Konstantin Fröhlich, Dr.-Ing. Matthias Meinke, Dr.-Ing. Lennart Schneiders, Prof. Dr.-Ing. Wolfgang Schröder (PI)

Institute of Aerodynamics, RWTH Aachen 
Wüllnerstraße 5a, D-52062 Aachen (Germany)
e-mail: office [@]


October 2019

Tags: CSE HLRS RWTH Aachen Large-Scale Project