Bulletin of the American Physical Society
71st Annual Meeting of the APS Division of Fluid Dynamics
Volume 63, Number 13
Sunday–Tuesday, November 18–20, 2018; Atlanta, Georgia
Session L25: Minisymposium: Prediction of Highly Turbulent Premixed Combustion in LES Framework 
Hide Abstracts 
Chair: James G. Brasseur, University of Colorado; Peter E. Hamlington, University of Colorado Room: Georgia World Congress Center B313 
Monday, November 19, 2018 4:05PM  4:31PM 
L25.00001: Interscale turbulencechemistry couplings in premixed turbulent combustion for largeeddy simulations Invited Speaker: Yuan Xuan With the everincreasing computing power and advances in computational technologies, LargeEddy Simulation (LES) has become a feasible tool for the design of aerospace and propulsion devices. In traditional LES, the filtered dynamical equations must be solved on a grid that resolves a large percentage of the variance associated with the turbulent fluctuations. As such, the evolution of turbulence motions and scalar quantities over these scales are primarily governed by the dynamical interactions explicitly resolved on the grid, with the effects of the unresolved, SubFilterScale (SFS) dynamics on the ResolvedScales (RS) of higher order and to be modeled. However, in turbulent combustion systems, chemical reactions leading to heat release typically occur over scales unresolvable by any practical LES grid, which violates the assumption underlying traditional LES as discussed above. In this talk, we explore new modeling elements that can be potentially embedded in existing LES frameworks to capture dynamicallyimportant interscale couplings between RS and SFS, necessary for more accurate predictions of the evolution of turbulent combustion events at RS. We first perform a scalebased decomposition of key primary variables, including momentum, energy, and chemical species. We then present the kinematic relationship between coherent structures in physical and scale spaces for these variables, characterized as SFS and RS contributions in the context of LES. Finally, we identify a subset of the SFS contributions that have substantial impact on the RS evolution through convective nonlinearities. These dynamicallydominant SFS modes are found to be localized near the SFSRS interface in scale space and near the thin flame heatrelease zone in physical space.

Monday, November 19, 2018 4:31PM  4:57PM 
L25.00002: Supergrid thermochemical closure of largeeddy simulation of turbulent combustion Invited Speaker: Alan Kerstein To illustrate turbulent combustion modeling challenges that influence the formulation of a largeeddysimulation (LES) closure, a strategy for thermochemical closure of LES of turbulent combustion that is currently under development is outlined, with emphasis on requirements for premixed combustion closure. The approach is intermediate between conventional closures, in which each LES cell contains an instantiation of the closure model, and the ‘representative interactive’ closure approach. The latter involves modeled time advancement of a single instantiation of a reference thermochemical field that is interrogated by each LES cell to obtain a cellspecific closure that takes into account limited state information residing at the LES level. This LESlevel information is generated by time advancing modeled transport equations for low moments of a progress variable (e.g., normalized local temperature) and, for cases that are not uniformly premixed, the mixture fraction (related to the local fuel mass fraction). In the intermediate approach, LES cells are grouped into clusters, e.g. using standard domain decomposition that is commonly used for parallelization. This forms a supergrid of LES cell clusters, where each cluster is associated with one instantiation of the timeevolving reference field. In order for each referencefield instance to correctly represent conditions in its associated cluster, the referencefield instances flux fluid to each other as specified by LES property fluxes through supergrid faces, including inlet, outlet, and boundary treatments. This fluxing is akin to fluxing in typical lineareddymodel (LEM) subgrid combustion closure of LES, but the representative interactive thermochemical closure method differs from the current use of LEM information for closure. The new approach will improve fidelity while reducing cost. Cost will be further reduced by using hierarchical parcel swapping (HiPS) instead of LEM to generate reference fields. 
Monday, November 19, 2018 4:57PM  5:23PM 
L25.00003: What is the potential of highorder methods for predicting turbulent combustion at highly turbulentconditions? Invited Speaker: Matthias Ihme Highorder methods have shown potential for improved predictions of aerodynamic flows. It is therefore expected that these advantages are translatable to the application of reacting flows at highly turbulent conditions. This talk examines the potential and reseach needs of highorder methods for predicting such flows. Of particular interest is the discontinuous Galerkin (DG) method. After discussing benefits of DGmethods for application to turbulent combustion, we will examine recent algorithmic developments to enable highorder predictions, which include solution stabilization, fluxdiscretization, and the treatment of reaction chemistry and complex transport. We will then proceed by reviewing progress on the development of combustion models and subfilterscale closures that are consistent with the highorder variational formulation. Recent developments of regularized deconvolution methods and variational multiscale methods are discussed as promissing strategies for subfilter terms, and potential benefits of solution adaptation and enrichment are illustrated. Results of turbulent combustion simulations and benchmark calculations against loworder methods are presented. We close by highlighting further opportunities of highorder methods for application to combustion problems at highly turbulent conditions. 
Monday, November 19, 2018 5:23PM  5:49PM 
L25.00004: Coutilization of experiments and large eddy simulations in turbulent combustion Invited Speaker: Adam Michael Steinberg Interactions between experiments and LES typically have taken similar forms to interactions between experiments and RANS. That is, experimental data may offer physical justification for a closure model, help specify boundary conditions for a simulation, and provide metrics – generally in the form of loworder moments – to compare against the simulation results. While this approach has led to steady improvements in LES fidelity, truly predictive simulations of complicated phenomena in realistic combustors remain an aspirational goal. On the experimental side, design of experiments and conversion of acquired signals to physical quantities of interest has progressed with limited input from simulations. This presentation will discuss richer methods of experiment/LES interaction. We first will consider the use of 3D (and 4D) experimental data to directly assess closure models. This has the potential to overcome some of the computational cost restrictions regarding a priori analysis of DNS, and to determine how well closure paradigms hold in realistic configurations. We then will discuss the assimilation of experimental data into LES, with an view towards improved state and parameterestimation in real combustors. Finally, we will consider improvements to experimental signal analysis that can be achieved through incorporation of prior information obtained from LES, particularly in the context of underdetermined signal conversion problems. 
Monday, November 19, 2018 5:49PM  6:15PM 
L25.00005: Stochastic Particle PDF Methods for Unresolved Fluctuations and InterScale Couplings in LES of Turbulent Premixed Combustion Invited Speaker: Daniel C Haworth Transported probability density function (PDF) methods have emerged as a powerful approach for dealing with the influences of unresolved turbulent fluctuations in LES of turbulent reacting flows. PDF methods provide an effective resolution to the closure problems that arise from spatial filtering of terms that correspond to nonlinear point processes: in particular, the important chemical reaction source terms. PDF methods invoke no inherent assumptions regarding the regime of combustion. Stochastic Lagrangian particle methods are the predominant approach for implementing PDF methods, and offer advantages in dealing with subfilterscale processes. In practice, hybrid Lagrangian particle/Eulerian mesh methods are used, where subfilterscale fluctuations are treated using a particlebased representation while resolvedscale dynamics are treated using conventional Eulerian CFD. Here the focus is on issues that are especially important in premixed systems. That includes discussions of the fundamental level of closure (e.g., velocitycomposition PDF versus composition PDF), the choice of thermochemical representation (e.g., progress variable/mixture fraction formulations versus reduced/skeletal/detailed chemical mechanisms), the modeling of molecular transport (the "mixing model"), and coupling of the subfilterscale fluctuations with the resolvedscale dynamics. Tight coupling between molecular transport and chemistry is important in premixed systems. That includes accommodating flamelet regimes of combustion, and differential diffusion. There are also consistency issues that arise from the dual LagrangianEulerian representation. Strategies that have been developed to address these issues, and to reduce the high computational cost of transported PDF methods, are discussed, including tradeoffs between accuracy and computational efficiency. The presentation concludes with a summary of key outstanding issues, and prospects for resolving those. 
Monday, November 19, 2018 6:15PM  6:41PM 
L25.00006: Large Eddy Simulation Subfilter Modeling of CombustionAffected Turbulence in Turbulent Premixed Combustion Invited Speaker: Michael E Mueller In turbulent premixed combustion, thermal expansion across the flame is a source of velocity fluctuations and turbulent kinetic energy. The influence of this thermal expansion effect on the turbulence depends on the ratio of the flame thickness to the Kolmogorov length scale, that is, the Karlovitz number. At high Karlovitz number, the thermal expansion effect is overwhelmed by dissipation and does not strongly influence the turbulence. However, at low Karlovitz number, the thermal expansion effect becomes the dominant source of turbulent kinetic energy. With Large Eddy Simulation, at low Karlovitz number, the flame will never be resolved, so the dominant thermal expansion production mechanism must be included in subfilter turbulence models. Direct Numerical Simulation databases of turbulent premixed planar jet flames at low and high Karlovitz number are utilized to assess a series of candidate subfilter turbulence models for both the subfilter stresses and the subfilter scalar fluxes. Three candidate models are considered: the Smagorinksy model, the Clark model, and an algebraic mixed model that explicitly accounts for thermal expansion effects presuming an infinitely thin flame. At high Karlovitz number, all models correctly capture the negligible influence of the flame on the turbulence. However, at low Karlovitz number, only the Clark model and the algebraic mixed model can capture the strong influence of the flame on the turbulence, manifested as generalized countergradient transport. This conclusion is independent of filter width, and countergradient transport is found to persist across all scales. In a posteriori tests, the algebraic mixed model is found to more accurately capture the influence of the flame on the turbulence since it explicitly includes information about the underlying flame structure. Looking forward, a new framework is proposed for subfilter turbulence modeling utilizing conditionally filtered velocities. 
Follow Us 
Engage
Become an APS Member 
My APS
Renew Membership 
Information for 
About APSThe American Physical Society (APS) is a nonprofit membership organization working to advance the knowledge of physics. 
© 2019 American Physical Society
 All rights reserved  Terms of Use
 Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 207403844
(301) 2093200
Editorial Office
1 Research Road, Ridge, NY 119612701
(631) 5914000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 200452001
(202) 6628700