**Soot Evolution in Turbulent Reacting Flows: A Multi-Fidelity Approach to a Multi-Scale, Multi-Physics Problem**

**Speaker:** Dr. Mike Mueller, Stanford University

**Series:** Other Events

**Location:**
J223 Equad

**Date/Time: **Thursday, February 23, 2012, 4:00 p.m.
- 5:00 p.m.

**Abstract:**

The next generation of efficient, low emission, fuel-flexible combustion systems will rely substantially on numerical simulations in addition to experimental testing of prototypes. Numerical simulations can provide details that cannot be obtained with experiments, and predictive numerical simulations can therefore shorten the design cycle and enable more aggressive, optimized designs. However, direct simulations of turbulent combustion, in general, and soot evolution, in particular, are currently intractable for realistic devices due to the large range of scales involved from soot particle interactions at the nanometer scale to chemical reaction zones and small-scale turbulence at the micron scale to large-scale turbulence at the meter scale. To gain the understanding needed to model such a complex multi-scale, multi-physics problem, a multi-fidelity approach is required. This understanding is then used to develop reliable engineering models for real combustion devices.

First, component models for soot and chemical kinetics are developed and validated in simple laminar flames. To model the evolution of the soot population, a statistical model is employed in which transport equations are solved for the moments of the soot Number Density Function (NDF). The moment source terms are closed with the Hybrid Method of Moments (HMOM), a robust model that is validated in a variety of laminar flames. Next, fully resolved Direct Numerical Simulations (DNS) of simple academic flows are used to understand the fundamental small-scale interactions between soot, turbulence, and chemistry. In particular, the interaction of soot precursor chemistry with turbulence and the unresolved structure of the soot fields are investigated in detail. Then, the component models are combined with the understanding gained from the DNS study to develop an integrated Large Eddy Simulation (LES) model for soot evolution in turbulent reacting flows. The LES approach leverages the soot model with models for the unresolved combustion and mixing phenomena that have been enhanced for soot evolution.

A posteriori validation of the integrated LES model against experimental measurements is performed in two laboratory-scale flames: a natural gas jet flame and an ethylene bluff body flame. Differences in the evolution of soot in the two configurations will be discussed. The integrated model is then applied to the simulation of a Pratt & Whitney aircraft combustor. Two global fuel-to-air ratios are simulated to investigate the effect on soot evolution.

Finally, in order to be used reliably in the design process, the uncertainties in LES must be quantified. One significant source of uncertainty is the chemical kinetic mechanism. An approach will be presented for propagating this uncertainty through turbulent combustion simulations.

LES results for the Pratt & Whitney PW6000 combustor: instantaneous temperature and droplet location (left) and soot volume fraction (right).