Bulletin of the American Physical Society
69th Annual Meeting of the APS Division of Fluid Dynamics
Volume 61, Number 20
Sunday–Tuesday, November 20–22, 2016; Portland, Oregon
Session E17: Reacting Flows: Numerical Methods |
Hide Abstracts |
Chair: D.C. Haworth, Pennsylvania State University Room: D131 |
Sunday, November 20, 2016 5:37PM - 5:50PM |
E17.00001: Improved Stiff ODE Solvers for Combustion CFD A. Imren, D.C. Haworth Increasingly large chemical mechanisms are needed to predict autoignition, heat release and pollutant emissions in computational fluid dynamics (CFD) simulations of in-cylinder processes in compression-ignition engines and other applications. Calculation of chemical source terms usually dominates the computational effort, and several strategies have been proposed to reduce the high computational cost associated with realistic chemistry in CFD. Central to most strategies is a stiff ordinary differential equation (ODE) solver to compute the change in composition due to chemical reactions over a computational time step. Most work to date on stiff ODE solvers for computational combustion has focused on backward differential formula (BDF) methods, and has not explicitly considered the implications of how the stiff ODE solver couples with the CFD algorithm. In this work, a fresh look at stiff ODE solvers is taken that includes how the solver is integrated into a turbulent combustion CFD code, and the advantages of extrapolation-based solvers in this regard are demonstrated. Benefits in CPU time and accuracy are demonstrated for homogeneous systems and compression-ignition engines, for chemical mechanisms that range in size from fewer than 50 to more than 7,000 species. [Preview Abstract] |
Sunday, November 20, 2016 5:50PM - 6:03PM |
E17.00002: Efficient partially implicit integration method for stiff chemistry in high-fidelity simulations of turbulent reacting flows Hao Wu, Matthias Ihme High-fidelity turbulent reactive flow simulations are typically associated with small time step sizes ($ h \le 10^{-8}$ sec) due to the CFL condition imposed by the fine gird. Although the step size is not sufficiently small to allow fully explicit time integration in the presence of stiff chemistry, it makes the use of classical implicit multi-step ODE solvers (e.g. VODE) an inefficient approach in combustion simulations due to the reduced number of internal iterations and excessive implicitness. In this study, an improved 4th-order Rosenbrock-Krylov (ROK4L) scheme is developed for the system of chemical reactions. This class of schemes replaces the Jacobian matrix by its low-rank Krylov approximation, thus introducing partial implicitness. The scheme is improved in both accuracy and efficiency by fulfilling additional order conditions and reducing the number of function evaluations. The ROK4L scheme is demonstrated to possess superior efficiency in comparison to CVODE due to the minimal degree of implicitness for small time-step sizes and the avoidance of other overhead associated with the start-up process of multi-step methods. [Preview Abstract] |
Sunday, November 20, 2016 6:03PM - 6:16PM |
E17.00003: A high-order immersed boundary method for high-fidelity turbulent combustion simulations Yuki Minamoto, Kozo Aoki, Kosuke Osawa, Tuo Shi, Alexandru Prodan, Mamoru Tanahashi Direct numerical simulations (DNS) have played important roles in the research of turbulent combustion. With the recent advancement in high-performance computing, DNS of slightly complicated configurations such as V-, various jet and swirl flames have been performed, and such DNS will further our understanding on the physics of turbulent combustion. Since these configurations include walls that do not necessarily conform with the preferred mesh coordinates for combustion DNS, most of these simulations use presumed profiles for inflow/near-wall flows as boundary conditions. A high-order immersed boundary method suited for parallel computation is one way to improve these simulations. The present research implements such a boundary technique in a combustion DNS code, and simulations are performed to confirm its accuracy and performance. [Preview Abstract] |
Sunday, November 20, 2016 6:16PM - 6:29PM |
E17.00004: ABSTRACT WITHDRAWN |
Sunday, November 20, 2016 6:29PM - 6:42PM |
E17.00005: Adjoint based sensitivity analysis of a reacting jet in crossflow Palash Sashittal, Taraneh Sayadi, Peter Schmid With current advances in computational resources, high fidelity simulations of reactive flows are increasingly being used as predictive tools in various industrial applications. In order to capture the combustion process accurately, detailed/reduced chemical mechanisms are employed, which in turn rely on various model parameters. Therefore, it would be of great interest to quantify the sensitivities of the predictions with respect to the introduced models. Due to the high dimensionality of the parameter space, methods such as finite differences which rely on multiple forward simulations prove to be very costly and adjoint based techniques are a suitable alternative. The complex nature of the governing equations, however, renders an efficient strategy in finding the adjoint equations a challenging task. In this study, we employ the modular approach of Fosas de Pando \textit{et al.} (2012), to build a discrete adjoint framework applied to a reacting jet in crossflow. The developed framework is then used to extract the sensitivity of the integrated heat release with respect to the existing combustion parameters. Analyzing the sensitivities in the three-dimensional domain provides insight towards the specific regions of the flow that are more susceptible to the choice of the model. [Preview Abstract] |
Sunday, November 20, 2016 6:42PM - 6:55PM |
E17.00006: Computational flow field in energy efficient engine (EEE) Kenji Miki, Jeff Moder, Meng-Sing Liou In this paper, preliminary results for the recently-updated Open National Combustor Code (Open NCC) as applied to the EEE are presented. The comparison between two different numerical schemes, the standard Jameson-Schmidt-Turkel (JST) scheme and the advection upstream splitting method (AUSM), is performed for the cold flow and the reacting flow calculations using the RANS. In the cold flow calculation, the AUSM scheme predicts a much stronger reverse flow in the central recirculation zone. In the reacting flow calculation, we test two cases: gaseous fuel injection and liquid spray injection. In the gaseous fuel injection case, the overall flame structures of the two schemes are similar to one another, in the sense that the flame is attached to the main nozzle, but is detached from the pilot nozzle. However, in the exit temperature profile, the AUSM scheme shows a more uniform profile than that of the JST scheme, which is close to the experimental data. In the liquid spray injection case, we expect different flame structures in this scenario. We will give a brief discussion on how two numerical schemes predict the flame structures inside the Eusing different ways to introduce the fuel injection. [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. |
© 2024 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
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700