Combustion Modeling Using Regimes in Turbulent Non-Premixed Combustion
Institute for Combustion Technology, RWTH Aachen University
Local Project ID:
HPC Platform used:
JUWELS of JSC
In the scope of the last project period of JHPC09, a highly resolved, high turbulence intensity direct numerical simulation (DNS) of a temporally evolving non-premixed jet flame was conducted on the supercomputer JUWELS of the Jülich Supercomputing Centre. The starting conditions for the DNS were chosen to instigate both local extinction and re-ignition to provide a wide range of combustion phenomena for further investigation.
Turbulent combustion is omnipresent in technical energy applications, such as internal combustion engines and gas turbines. The most prominent approach to modeling both premixed and non-premixed turbulent combustion is called the flamelet concept. The idea of this concept is that a turbulent flame is composed of an ensemble of thin laminar flamelets. The flamelet structure itself is chemically well defined and is attached to an instantaneous flame surface, which itself is corrugated by turbulent fluctuations. However, as turbulence intensifies, the flamelet regime is departed and small scale mixing starts to affect the chemistry. For this regime, costly DNS with finite rate chemistry have to be deployed to resolve the combustion, as no reliable model exists for this regime. Since no models are employed for the chemistry or the turbulence in DNS, all involved physical scales must be completely resolved. The resulting system of equation features approximately 60 billion degrees of freedom, which has to be solved for each of the ten thousand time steps needed for a full transition of the flame into a fully turbulent state. The massive need for computational resources can only be provided by the most modern and powerful super computers.
The configuration of the DNS is illustrated by the iso-surface of the stoichiometric mixture fraction Zst in Fig. 1. This iso-surface marks the position of the optimal mixture of fuel and oxidizer and manifests itself as two highly corrugated flames on top of and below the centrally situated fuel slab. The local value of the heat release indicates the intensity of the combustion chemistry. Localized patches of low heat release are the result of turbulence induced extinction.
The chaotic and highly complex nature of turbulent combustion in the simulated conditions, as well as the huge size of data generated (1.2 terabyte in each time step), require very sophisticated methods of analysis. A promising and well tested method for measuring the local turbulent scales in large scale data sets is the Dissipation Element (DE) analysis. DEs are non-arbitrary, space-filling objects of monotonic regions in scalar fields. In this fashion, the flow can be subdivided into more easily understood sub-units, reducing the complexity of the entire flow domain demonstrated in Fig. 1 significantly and enabling a comprehensive analysis of the phenomena. Since the flamelet concept relies on the local smoothness of the chemical field, the attributes of the DEs encompassing the reaction zones will indicate whether the flamelet assumption is locally valid. Therefore, DEs can pinpoint regions which cannot be accurately modelled with the current combustion models and a systematic analysis of the mechanisms in these regions can be conducted.
Dr. Antonio Attili. Mathis Bode, Dominik Denker, Prof. Dr.-Ing. Heinz Pitsch (PI) – all: Institute for Combustion Technology, RWTH Aachen University (Germany)
D. Denker et al. “Dissipation Element Analysis of Non-premixed Jet Flames”. In: submitted to the Journal of Fluid Mechanics (2019)
D. Denker et al. “A New Modeling Approach for Mixture Fraction Statistics Based on Dissipation Elements”. In: Submitted to Proc. Comb. Inst (2019)
D. Denker et al. “Gradient Trajectory Analysis of the Burning Rate in Premixed Jet Flames”. In: Submitted to Proc. Comb. Inst (2020)
Fakultät für Maschinenwesen
Institut für Technische Verbrennung (ITV)
RWTH Aachen University
Templergraben 64, D-52062 Aachen (Germany)
e-mail: d.denker [@] itv.rwth-aachen.de
JSC project ID: cjhpc09