Session G6 — Modeling G31 TOWARD NEXT-GENERATION MULTIPHYSICS SIMULATION SOFTWARE. James C. Sutherland, University of Utah Reacting flow simulation encompasses a very wide range of modeling approaches. Selection of models often influences the choice of numerical method (spatial and/or temporal discretization, algorithms, etc.) that one employs. Next-generation simulation codes will require the ability to implement a range of numerical methods that may be determined by the models that are selected. Furthermore, introduction of "smart algorithms" that may dynamically change modeling approaches as different physical regimes are realized will require wide flexibility in algorithm implementation. This talk addresses software development strategies that facilitate a wide range of algorithms, including switching from implicit to explicit time integration, mesh adaptivity, addition/removal of transport equations, modification of property evaluation schemes, etc. Aspects of such a design have been incorporated into SIERRA/Fuego, a massively parallel reacting flow code developed at Sandia National Laboratories to simulate turbulent reacting flow in regimes varying from laminar to turbulent using both RANS and LES approaches. G32 NOVEL SUBGRID MODELING OF THE LES EQUATIONS UNDER SUPERCRITICAL PRESSURE. Laurent Selle, Josette Bellan, Kenneth Harstad, Jet Propulsion Laboratory, California Institute of Technology, The Direct Numerical Simulation (DNS) conservation equations describing two fluids mixing are filtered to obtain the Large Eddy Simulation (LES) equations. The budget of the LES equations is calculated using a DNS database of 8 supercritical-pressure temporal mixing layer simulations. For strong departures from perfect gas and mixture ideality, the budget shows that in the momentum equation, the gradient of the pressure computed as a function of the filtered variables and the gradient of the difference between the filtered pressure and the pressure computed using the real-gas equation of state with the filtered variables, are of same order of magnitude as the resolved convective term, which is the leading term; this ordering of terms is independent of the filter size used to obtain the LES equations from those of DNS. This gradient difference is an important term that has so far been neglected in LES under supercritical pressure conditions. To model this additional term, a Taylor series expansion was used for the pressure. The model was implemented and tested on the database. At small filter size the model perform very well, but it deteriorates at larger filter size indicating the importance of LES grid resolution studies. G33 QUADRATURE METHOD OF MOMENTS FOR MODELING SOOT FORMATION IN TURBULENT NON-PREMIXED FLAMES. Qing Tang, Ying Huang, Mike Bockelie, Reaction Engineering International Rodney Fox, Dept. of Chemical and Biological Engineering, Iowa State University The computational fluid dynamics (CFD) modeling of soot formation in turbulent non-premixed flames requires explicit knowledge of the impacts of turbulent mixing on the soot particle size distribution (PSD). In this work, the Direct Quadrature Method of Moments (DQMOM) is applied to a univaritate population balance equation (PBE) where the number density function (NDF) depends on the size of the soot particles. DQMOM tracks the evolution of selected moments of soot particle NDF in space and time, overcoming the difficulty of discretizing the high- dimensional PBE. The soot model is coupled with a multi-environment probability density function (PDF) approach closed at the joint-scalar level to directly account for interactions between turbulence, and the solid and gas phase chemistry. The later is described by augmented reduced mechanism based on quasi steady state assumption. The interaction by exchange with the mean (IEM) model is used to close sub-grid mixing among fluid environments, each of which has its own NDF, and thus its own set of NDF moments represented by weights and abscissas. Given the initial distribution and the boundary conditions, the weights and abscissas of each environment and thus the Reynolds-averaged mean can be found directly. Radiation from soot and gas phase species is accounted for through the discrete ordinate method implemented in the context of the multi-environment PDF method. The overall modeling approach is validated by comparison with experimental data found in literature regarding soot formation in ethylene-air turbulent non-premixed flames, where turbulent jet of ethylene are burned in still air. G34 MODELING OF AN AUTOTHERMAL HEAT INTEGRATED WALL REACTOR FOR SIMULATION OF HYDROGEN PRODUCTION FOR FUEL CELLS. Cihan Banu Biçici, Hasan Bedir, Mechanical Engineering Department Boğaçi University An autothermal, heat integrated wall reactor for hydrogen production is numerically investigated. The two dimensional reactor model is composed of a hollow cylindrical tube enclosed in a bigger one, where different catalysts for exothermic partial oxidation and endothermic steam reforming reactions are deposited on the inside and outside surfaces of the tube. The reaction system is developed for methane reforming with one step chemical kinetics. Reactor operation at different feed ratios and with different catalyst configurations is analyzed under steady state conditions. Results show that hydrogen production is higher when the catalysts are placed closer as well as at low air-to-fuel and high steam-to-fuel ratios. G35 EFFECT OF OXYGEN ENHANCEMENT ON RADIATION CHARACTERISTICS OF NORMAL AND INVERSE DIFFUSION FLAMES. Manish Saini1, Sivakumar Krishnan2, Yuan Zheng1, Jay Gore1 1 Purdue University, 2Indiana University-Purdue University Indianapolis Radiative transfer in oxygen enhanced inverse flame configurations is an important area of study for spacecraft and terrestrial fire safety and for fundamental combustion research. Motivated by this, heat flux distributions, total radiative loss and the instantaneous spectral radiation intensity were investigated experimentally for oxygen enhanced normal and inverse laminar ethane diffusion flames with increasing heat release rates. The oxygen concentration in the oxidizer was varied as 21%, 30%, 50% and 100% with coflowing normal and inverse flame burners used to stabilize the flames. The inverse diffusion flames were essentially non-luminous while the normal diffusion flames with the same heat release rates were highly luminous. Oxygen enhancement was found to lead to reduced flame lengths increased luminosities and increased total radiative heat loss and spectral radiation intensity for both the normal and the inverse diffusion flames. Using flame length as the characteristic length parameter, the normalized radiative flux distributions for various flames approximately collapsed together, hence establishing the feasibility for the single point radiant output measurement technique. The radiation spectra were dominated by gas radiation from carbon dioxide and water vapor, with soot radiation becoming important for the high oxygen concentration flames. G37 NUMERICAL ANALYSIS OF COAL COMBUSTION IN A RACEWAY. Mingyan Gu1, N.K.C. Selvarasu1, Yongfu Zhao2, Dennis Lu3, Chenn Q. Zhou1 1 Department of Mechanical Engineering, Purdue University-Calumet 2United States Steel Corp., Research & Technology Center 3 United States Steel Corp., Gary Works Blast Furnace Engineering and Technology Blast furnace is widely used to produce molten iron in steel industry. Pulverized coal is used to provide a portion of heat for melting the raw materials. In order to lower the operational costs and reduce pollutant emissions, efforts have been made to increase the pulverized coal injection into a blast furnace to partially replace metallurgical coke. However, tuyere nose erosion has been found in some operating conditions. In this study, a 3-dimensional multiphase reacting computational fluid dynamics (CFD) model has been developed to simulate pulverized coal combustion inside the raceway. The velocity field, temperature distribution, and combustion characteristics have been analyzed in details to identify the causes of the tuyere nose erosion. The results can be used to provide guidance for selecting the correct tuyere size and for increasing the PCI rate and productivity. G37 MODELING MULTICOMPONENT SOLID-LIQUID PHASE EQUILIBRIUM OF SALTS AND SILICATES IN COAL COMBUSTION/GASIFICATION PROCESSES. Bing Liu, John L. Oscarson, Larry L. Baxter, University of Utah Reed M. Izatt, Department of Chemistry, Brigham Young University A knowledge of the solid-liquid phase equilibrium of salts and silicates is important in order to understand ash deposition and slagging problems in coal combustion/gasification processes. Computer codes simulating combustion/gasification processes being developed at BYU need a good thermodynamic subprogram in order to describe the effect of slagging on the process. In the present work, a modified quasichemical model has been used to correlate the equilibrium phase diagrams of binary and ternary systems. The calculated results have been compared with the measured data. The model will be discussed and a comparison of the model predictions and data will be given. G38 COMPUTATIONAL FLUID DYNAMICS MODELING OF THE OPERATION OF A FLAME IONIZATION SENSOR E. David Huckaby, Benjamin Chorpening, Jimmy Thornton,US Department of Energy, National Energy Technology Laboratory The sensors and controls research group at the United States Department of Energy (DOE) National Energy Technology Laboratory (NETL) is continuing to develop the Combustion Control and Diagnostics Sensor (CCADS) for gas turbine applications. CCADS uses the electrical conduction of the charged species generated during the combustion process to detect combustion instabilities and monitor equivalence ratio. As part of this effort, combustion models are being developed which include the interaction between the electric field and the transport of charged species. The primary combustion process is computed using a flame wrinkling model (Weller et. al. 1998) which is a component of the OpenFOAM toolkit (Jasak et. al. 2004). A sub-model for the transport of charged species is attached to this model. The formulation of the charged-species model similar that applied by Penderson and Brown (1993) for the simulation of laminar flames. The sub-model consists of an additional flux due to the electric field (drift flux) added to the equations for the charged species concentrations and the solution the electric potential from the resolved charge density. The subgrid interactions between the electric field and charged species transport have been neglected. Using the above procedure, numerical simulations are performed and the results compared with several recent CCADS experiments.