Towards high-fidelity modeling of turbulent reactive systems using highly-resolved and simplified simulation techniques
This work evaluates methods and models for Large-Eddy Simulation (LES) using Direct Numerical Simulation (DNS) data, with a specific focus on two aspects: grid resolution and subgrid modeling, and their influence on simulation quality. Recent model developments are assessed and applied to LES of turbulent reactive systems. The work is divided into two parts. The first part assesses the performance and quality of LES results for a canonical configuration and a numerically well-studied laboratory experiment. Different subgrid models, as well as DNS, are employed and compared. A workflow is proposed to assess the fidelity of the simulations performed, and the investigated models are recently developed subgrid models proposed in previous studies. The grid resolution is evaluated in a separate study, attempting to formulate an indicator to quantify resolution quality based solely on the performed LES, without the need for high-resolution reference data. The predictive ability of the indicator is validated using reference data from DNS, and a wide range of grid qualities are evaluated. The second part of this work applies high-fidelity models and numerical schemes to the investigation of reciprocating engines, representing as a turbulent reactive system. A back-tracing algorithm is proposed using Lagrangian tracers to trace the causes of physical events in technical systems in time and space. The method is shown to be applicable to both steady-state and transient engineering processes, providing insight into the convective transport behavior of fluid parcels involved in the physical events studied. Any desired data obtained from the simulation can be targeted and recorded in spatial and temporal protocols by the particles. Finally, the proposed tools are also used for the case of a piston engine exhibiting cyclic variability, demonstrating that high-order numerical models and methods can replace complicated supplementary models for ignition and early flame evolution.