ENGINEERING AND CFD

The direct numerical simulation performed in the course of this project – run on SuperMUC at LRZ – investigated a temporally evolving non-premixed syngas jet flame. Results of this simulation were used to validate a recently published set of extended model equations for the reaction zone dynamics in non-premixed combustion. Furthermore, the dataset was used to analyze the importance of curvature induced transport phenomena. Regions could be identified where curvature has a significant impact on the flame structure.

The increasing demand for power and environmental concerns motivates research activities in the field of alternative fuels. One example of these alternative fuels is synthesis gas (CO/H₂mixtures), which can e.g. be generated by partial oxidation/gasification of hydrocarbons such as biomass to obtain a hydrogen and carbon monoxide rich mixture. This mixture can be used for energetic and non-energetic applications. For example it can be used during the production process of methanol / fuel synthesis or it can be used directly as fuel in a gas turbine. Oxygen-enriched mixtures or even pure O₂ as oxidizer are employed in partial oxidation for syngas production.

During past decades computational fluid dynamics (CFD) have led to fundamental insights in the field of reactive flow systems which would not have been possible by experiments only. For example, the measurement of spatially (3D) and temporally resolved species concentrations in turbulent flames is almost impossible. On the other hand, combustion mostly occurs on the smallest scales where chemistry interacts with the smallest eddies, making numerical modeling a difficult task.

The rapid advancement of supercomputing capabilities helped to establish a powerful and promising tool in combustion science, which is able to overcome these difficulties. In the so-called direct numerical simulation (DNS) approach, the governing equations describing fluid motion, species and energy transport are solved without any modeling assumptions. Even if currently limited to simple geometries, spatially and temporally resolved datasets encourage the use of DNS for model development and validation.

The simulation performed in the course of this LRZ project investigated a temporally evolving non-premixed syngas jet flame and comprised nearly one billion grid points. The obtained results serve as a database for the validation of a recently published set of enhanced model equations for non-premixed combustion. The scientific question was how the curvature of the underlying fields affects molecular transport processes within the reaction zone. For reference, a snapshot of the simulation is shown in figure 1. The reaction zone is highlighted by the green iso-surface.

For the analysis it was necessary to extract several different flame structures, see figure 2, and to track their evolution while the simulation was running.

With this so called in-situ tracking the researchers were able to show for the first time that the extended set of model equations is actually applicable under turbulent conditions and that curvature-induced transport has an important contribution to combustion dynamics.¹

These findings are illustrated in figure 3. The plot shows the temporal evolution of temperature (at the point of stoichiometric mixture) of one flame structure. During early times of the flame structure significant deviations between the extended formulation and the classical formulation exist. These differences become negligible as soon as the underlying field flattens. Further details about the results can be found in Scholtissek et al.¹

These results in combination with the extended set of equations will be important for future modeling strategies when it comes to Large Eddy Simulations (LES). In contrast to DNS, LES resolves only large flow scales and requires a proper closure of the governing equations. However, with LES more realistic flow scenarios are feasible.