Large eddy simulation of turbulent combustion /

In the past decade, Large Eddy Simulation (LES) has been increasingly and successfully applied to both premixed and non-premixed reacting flows [1, 2, 3]. In application and testing, methods such as steady and unsteady flamelet modeling, the probability density function (PDF) assumption, and level set tracking have been shown to describe combustion in a LES context correctly. As capable as such methods are, however, researchers only very recently have begun to apply them to the detailed, multi-scale, and multi-physics flows that arise in the burners and engines found in modern, industrially relevant equipment. In the course of this transition of LES from a scientifically interesting method in its own right to a research platform useful for studying real world flows, the limits of current methods to be highlighted. More specifically, as the complexity of the flows being considered increases, so do the number of involved lengthscales, the details of the relevant chemistry, and the extent of the turbulence-chemistry interactions. In LES, where the large scales of the flow are resolved while the small scales are modeled, this increased complexity means that more accuracy will be demanded of sub-grid models. Sub-grid models for momentum have been studied rigorously, and computationally effcient methods exist. But sub-grid models for combustion-related phenomena have not yet demonstrated the ability to capture the physics needed to describe complex flows consistently and accurately. For example, quantities of interest such as flame lift-o heights and heat release near injection nozzles, often are predicted incorrectly using state-of-the-art codes. Furthermore, NOx and soot formation processes, which occur on timescales significantly longer than those of CO2 and H2O production, cannot be captured with the kind of steady flamelet models now being used [4]. Part I of this project was proposed in response to these challenges.Photocopy."March 15, 2006.""Final Performance Report : October 1, 2002 -- October 1, 2005."Includes bibliographic refereces (pages 28-30).In the past decade, Large Eddy Simulation (LES) has been increasingly and successfully applied to both premixed and non-premixed reacting flows [1, 2, 3]. In application and testing, methods such as steady and unsteady flamelet modeling, the probability density function (PDF) assumption, and level set tracking have been shown to describe combustion in a LES context correctly. As capable as such methods are, however, researchers only very recently have begun to apply them to the detailed, multi-scale, and multi-physics flows that arise in the burners and engines found in modern, industrially relevant equipment. In the course of this transition of LES from a scientifically interesting method in its own right to a research platform useful for studying real world flows, the limits of current methods to be highlighted. More specifically, as the complexity of the flows being considered increases, so do the number of involved lengthscales, the details of the relevant chemistry, and the extent of the turbulence-chemistry interactions. In LES, where the large scales of the flow are resolved while the small scales are modeled, this increased complexity means that more accuracy will be demanded of sub-grid models. Sub-grid models for momentum have been studied rigorously, and computationally effcient methods exist. But sub-grid models for combustion-related phenomena have not yet demonstrated the ability to capture the physics needed to describe complex flows consistently and accurately. For example, quantities of interest such as flame lift-o heights and heat release near injection nozzles, often are predicted incorrectly using state-of-the-art codes. Furthermore, NOx and soot formation processes, which occur on timescales significantly longer than those of CO2 and H2O production, cannot be captured with the kind of steady flamelet models now being used [4]. Part I of this project was proposed in response to these challenges.Report prepared by Leland Stanford Junior University, Stanford, California, under contract no.Mode of access: Internet