Numerical simulations of turbulent mixing in complex flows

by Elizaveta Ivanova

Institution: University of Stuttgart
Department: Fakultät Luft- und Raumfahrttechnik und Geodäsie
Degree: PhD
Year: 2012
Record ID: 1105857
Full text PDF: http://elib.uni-stuttgart.de/opus/volltexte/2013/7813/


For an accurate gas turbine combustion simulation the quality of the mean and fluctuating velocity, temperature, and species concentration field predictions is of critical importance. A typical flow pattern of a gas turbine combustion chamber comprises several complex jet configurations intricately interacting. Numerical representation of turbulence and mixing in such flows is a challenging task. Different aspects of modeling the turbulence and turbulent scalar mixing in complex jets are considered in this dissertation. Three complex jet test cases are selected for a systematic investigation in the present work: jet in crossflow, confined coaxial swirling jets, and confined coaxial jets without swirl. A comprehensive literature overview on the previous numerical investigations of these flows is given in chapter 1. Three main goals of the present work are formulated. The first one is the critical assessment of the widespread turbulent viscosity and turbulent scalar diffusivity Reynolds-Averaged Navier-Stokes (RANS) approaches in application to the flow and mixing modeling in the selected test cases. The second aim is a thorough evaluation of the statistical data obtained in Large Eddy Simulations (LES) of three main test cases. This goal is set in order to provide a more comprehensive database for validating RANS models and to obtain variables of interest which cannot be determined easily in experiments due to measurement instrumentation limitations. The third aim of the present work is to understand the potential of the Scale-Adaptive Simulation(SAS) method and of the Unsteady RANS (URANS) approach for the mixing modeling in the considered flows. The most important question here is the clarification of the ability of different methods to correctly predict the flame propagation processes in the practically relevant case of autoignition in a jet in crossflow. Chapter 2 develops the theoretical framework of this dissertation. RANS and LES concepts are introduced and common modeling closure approaches for both methods are discussed. Chapter 3 is devoted to the study of the accuracy of RANS in three main test cases and to LES data evaluation. The results of RANS mean and fluctuating velocity and scalar field simulations are validated against experimental data and the outcome of accompanying LES calculations. The LES results, for the most part, agree better with the respective experimental data. Many of the tested RANS models show a clear trend of the underestimation of turbulence and mixing in the considered complex jet configurations. The influence of different equation terms and corrections on the RANS model accuracy is investigated. For a more thorough study, the resolved data fields from LES are used for the evaluation of the Reynolds-averaged turbulent viscosities, turbulent scalar diffusivities, and the main budget terms of the turbulent kinetic energy and the turbulent scalar variance transport equations. This data is also used to extract practically important information on the turbulent Schmidt numbers in the…