Bulletin of the American Physical Society
72nd Annual Meeting of the APS Division of Fluid Dynamics
Volume 64, Number 13
Saturday–Tuesday, November 23–26, 2019; Seattle, Washington
Session P05: Turbulent Combustion Modeling and Methods |
Hide Abstracts |
Chair: James Brasseur, University of Colorado, Boulder Room: 204 |
Monday, November 25, 2019 5:16PM - 5:29PM |
P05.00001: LES/PDF of Sandia flame D using a coupled adaptive chemistry and tabulation approach Ashish Newale, Youwen Liang, Stephen Pope, Perrine Pepiot LES/PDF methods are known to provide accurate results for challenging turbulent combustion configurations. This higher level of fidelity, however, comes at the cost of added computational expense compared to other state-of-the-art methods. To reduce the magnitude of this cost differential, the majority of LES/PDF computations performed to date have used reduced mechanisms. We have recently proposed a coupled adaptive chemistry and tabulation approach to enable the use of the detailed mechanisms in LES/PDF computations. Specifically, we proposed a coupled pre-partitioned adaptive chemistry (PPAC) and in-situ adaptive tabulation (ISAT) method (Newale et al., CTM 2019). The proposed coupled method showed encouraging results in a partially stirred reactor configuration. In this work, we examine the performance of the coupled PPAC-ISAT method in a LES/PDF computation of Sandia flame D. We demonstrate that the coupled technique enables the use of detailed mechanisms at a significantly reduced computational cost, while retaining the level of fidelity attained by using the detailed mechanism without approximations. [Preview Abstract] |
Monday, November 25, 2019 5:29PM - 5:42PM |
P05.00002: In-Situ Adaptive Manifolds: Enabling simulations of complex turbulent reacting flows Cristian E. Lacey, Alex G. Novoselov, Michael E. Mueller Reduced-order manifold approaches to turbulent combustion modeling traditionally involve precomputation of manifold solutions and pretabulation of the thermochemical database versus a small number of manifold variables. However, additional manifold variables are required as the complexity of turbulent combustion processes increases, for example, multi-modal combustion or combustion featuring multiple and/or inhomogeneous inlets. This increase in the number of manifold variables comes with an increase in the computational cost of precomputing a greater number of manifold solutions, most of which are never actually utilized in a CFD calculation. The memory required to store the pretabulated high-dimensional thermochemical database also increases, limiting the complexity of reduced-order manifold combustion models. In this work, a new In-Situ Adaptive Manifolds (ISAM) approach is developed that overcomes this limitation by combining `on-the-fly' calculation of manifold solutions with In-Situ Adaptive Tabulation (ISAT), enabling the use of more complex manifold-based turbulent combustion models. The performance of ISAM is evaluated via LES calculations of canonical turbulent flames. [Preview Abstract] |
Monday, November 25, 2019 5:42PM - 5:55PM |
P05.00003: Fundamental Differences between Large-Eddy Simulation of Incompressible Turbulence vs Premixed Turbulent Combustion James Brasseur, Yash Shaw, Paulo Paes, Yuan Xuan In contrast with RANS where the modeled terms are of leading order, the LES framework requires that the modeled subfilter-scale (SFS) contributions be of lower order than the leading-order terms. This will be the case if the resolved-scale (RS) contributions to the triadic sum of advective nonlinearities in spectral space dominate the SFS triads, requiring an effective grid that resolves well Reynolds stress motions. Turbulent combustion deviates from the LES framework is several key ways, primarily in the existence of the chemical source terms that lead to the release of thermal energy at scales generally unresolved. In this study we quantify the dominant SFS contributions to the key nonlinearities that underlie RS evolution in LES of premixed turbulent combustion to isolate fundamental deviations from the LES framework. With a new method to remove spurious spectral content in inhomogeneous directions, we apply a concurrent physical-Fourier space methodology to compressible DNS of flame-turbulence interactions to isolate the triadic structure of advective nonlinearities and the quadratic structure of chemical nonlinearities in the Fourier representation from which the dynamically dominant contributions are determined. We find that when the RS fluctuations in momentum and energy are well resolved, the relative SFS contributions are very different depending on the class of nonlinearity and the relative RS vs. SFS contributions to the evolution of the individual species. \textit{Supported by AFOSR.} [Preview Abstract] |
Monday, November 25, 2019 5:55PM - 6:08PM |
P05.00004: Direct Numerical Simulation of the Temporally-Evolving Reacting Jet for Model Error Assessment of RANS-based Closures for Non-Premixed Turbulent Combustion Bryan Reuter, Todd Oliver, Robert Moser High-fidelity DNS data of the temporal reacting jet is generated using a laminar flamelet closure, the same chemistry model as is typically employed in RANS-based approaches. The simulations solve the low-Mach Navier-Stokes equations with a pseudospectral Fourier-Galerkin/B-Spline collocation approach and a novel second-order, explicit time marching scheme. The turbulence is fully resolved, but the chemistry is modeled by obtaining the density from a flamelet library which is a function of the local mixture fraction and scalar dissipation rate. This methodology allows for a clear assessment of the errors that arise from the three aspects of a model for combustion: turbulence closures, modeling the turbulence-chemistry interaction, and chemical kinetics closures. As there is no conflation of errors arising via the chemical kinetics, the data can be used to directly assess the performance of RANS closures in representing the turbulence and the turbulence-chemistry interaction. Conversely, by comparing our DNS with one closed with a higher fidelity chemistry model from Attili and co-authors, any discrepancies can be attributed to the different chemical kinetics closures. Results are shown for the $k-\varepsilon$ model with presumed-PDF approach for a non-premixed $n-$heptane flame. [Preview Abstract] |
Monday, November 25, 2019 6:08PM - 6:21PM |
P05.00005: Timescale Analysis Procedure Applied to Premixed and Non-Premixed Turbulent Combustion Salvador Badillo-Rios, Matthew Harvazinski, Venkateswaran Sankaran, Ann Karagozian Full detailed kinetics in turbulent combustion simulations can be computationally prohibitive for propulsion systems. While reduced kinetic models are often selected to reduce cost, there needs to be a clear understanding of their impact on the phenomena being modeled. The present study examines the effects of alternative kinetic models and flow parameters on turbulent combustion processes associated with non-premixed combustion in a shear-coaxial rocket injection configuration, as a means of determining the conditions under which turbulent reaction phenomena may be altered via the kinetics. A combination of 2D axisymmetric parametric studies and selected 3D simulations for a single element shear coaxial rocket injector are performed, incorporating GRI-Mech 3.0 and several alternative reduced kinetic models representing the combustion of gaseous methane and oxygen. Systematic timescale analysis procedures, e.g., the Chemical Explosive Mode Analysis (CEMA) procedure based on the Jacobian matrix of the chemical source term, are applied to explain differences in observed flame behaviors. This approach can serve as a quantitative method for the systematic detection of critical flame features, species, and reactions. [Preview Abstract] |
Monday, November 25, 2019 6:21PM - 6:34PM |
P05.00006: Large Eddy Simulations of turbulent flames using two-dimensional reduced-order manifold models Alex G. Novoselov, Cristian E. Lacey, Michael E. Mueller Recently, we have developed a set of two-dimensional manifold equations describing multi-modal combustion using the mixture fraction and a generalized progress variable. Information about the underlying mode of combustion is encoded in three scalar dissipation rates that appear as parameters in the two-dimensional equations. In this work, Large Eddy Simulations (LES) of turbulent hydrogen flames in both the asymptotic limits of combustion (i.e. nonpremixed and premixed) and multi-modal flames are performed using this new turbulent combustion model. These simulations are compared to LES performed using traditional one-dimensional manifold-based turbulent combustion models as well as experimental measurements. The new model is shown to describe the flames in the asymptotic limits just as well as the one-dimensional models but without any \emph{a priori} knowledge of the flame necessary to select the model. Additionally, the new model is shown to be able to describe the behavior of multi-modal turbulent flames where traditional one-dimensional models fail. [Preview Abstract] |
Monday, November 25, 2019 6:34PM - 6:47PM |
P05.00007: A Dispersion Model for Turbulent, Multi-Component Reacting Flows Omkar Shende, Ali Mani It is theoretically and computationally challenging to build reduced-order models for turbulent reacting flows as the underlying chemical and transport processes are individually complex. Furthermore, an understanding of the coupled effects of these phenomena remains elusive. However, deeper insight into turbulent transport effects on reaction dynamics is essential for the future design of efficient energy systems. Using theory developed for non-reactive dispersion of scalars and linear reactions, an algebraic Reynolds-averaged Navier-Stokes model for capturing unresolved interactions between multi-component scalar reactions in turbulent flows is developed. This work extends the modified gradient diffusion model by Corrsin (JFM, vol. 11, p.407-416) beyond single-component transport phenomena with linear reactions. Using two- and three-dimensional direct numerical simulations, it is shown that this model improves prediction of mean quantities compared to traditional results. [Preview Abstract] |
Monday, November 25, 2019 6:47PM - 7:00PM |
P05.00008: Conditional Reynolds stress modeling in turbulent premixed jet flames Jinyoung Lee, Michael Mueller Conventional turbulence modeling approaches based on traditional unconditional averaging implicitly assume that combustion heat release does not affect turbulence. However, in turbulent premixed flames at low Karlovitz number, combustion-induced dilatation and flame motion significantly modify turbulence. Instead of relying on unconditional averaging, simultaneously solving momentum and scalar transport equations conditionally averaged on a flame structure variable could provide a superior framework for modeling combustion-affected turbulence since the flame dynamics are embedded into the flame-conditioning. The primary challenge in this approach is developing closure models for conditional terms, which evolve in both physical and conditional spaces. In this work, a new model for conditional Reynolds stresses, which appear in the conditionally averaged momentum equations, has been developed. The new model consists of a conditional Boussinesq-like term and a new term representing turbulent momentum transport in conditional space. A theoretical scaling of the new model has been investigated, and the model performance has been evaluated in a priori analyses using DNS databases of turbulent premixed jet flames at low and high Karlovitz numbers. [Preview Abstract] |
Monday, November 25, 2019 7:00PM - 7:13PM |
P05.00009: DNS analysis of flame propagation for systematic variations in turbulence scales and intensity Shrey Trivedi, R. Stewart Cant Direct Numerical Simulation (DNS) of premixed hydrocarbon flames is performed using the DNS code Senga2 to investigate the role of varying integral length scale $l_{\mathrm{0}}$ as well as the turbulence intensity u` in the thin reaction zone regime. Different cases are studied by either systematically varying $l_{\mathrm{0\thinspace }}$at constant u` or by systematically varying u` at constant$ l_{\mathrm{0}}$. Several aspects of these flames are compared. The turbulent flame speed $s_{T}$ is found to decrease as $l_{\mathrm{0\thinspace }}$decreases or when u` increases. The ratio of $s_{T} / s_{L}$ is generally accounted for by the area change $A_{T} / A_{L}$ but a significant deviation is observed between these values at high u` cases. The size of the smallest scales of turbulence also decreases with $l_{\mathrm{0\thinspace }}$and the interaction of these small scales with the flame produces an increase in the mean curvature of the flame which eventually reaches a saturation point. There is also an increase in the number of flame-flame interactions with decreasing $l_{\mathrm{0}}$. Individual flame-flame interaction topologies are identified and analysed. All these findings serve to further our understanding of turbulent combustion and the role that the turbulence length scales play in flame propagation. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2020 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
1 Research Road, Ridge, NY 11961-2701
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700