Here - Fifteenth International Conference on Numerical Combustion
Transcription
Here - Fifteenth International Conference on Numerical Combustion
th 15 International Conference on Numerical Combustion Palais des Papes, Avignon, France April 19-22, 2015 www.nc15.ecp.fr Organized by Laboratoire EM2C CNRS and CentraleSupélec 2 Organizing committee Organizing Committee Denis Veynante (chair) Matthieu Boileau Nasser Darabiha Sébastien Ducruix Benoît Fiorina Olivier Gicquel Frédérique Laurent-Nègre Brigitte Llobel Franck Richecoeur Nathalie Rodrigues Philippe Scouflaire Thomas Schmitt Ronan Vicquelin Laboratoire EM2C UPR 288 CNRS and CentraleSupélec Grande Voie des Vignes 92295 Châtenay-Malabry Cedex France http://www.em2c.ecp.fr 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. Conference sponsors 3 CONFERENCE SPONSORS sponsors Bronze sponsor FLUIDYN France 7 boulevard de la Libération 93200 Saint-Denis - France Contact: Dr. Amita Tripathi email: [email protected] Tél: +33 (0)1 42 43 16 66 Fax: +33 (0)1 42 43 50 33 Silver sponsors NUMECA International Ch. de la Hulpe 189 Terhulpse Steenweg 1170 Brussels - Belgium Contact: Dr.-Ing. Jan E. Anker Head - Combustion Modeling Group Tel: +32 (0)2 642-2824 Fax: +32 (0)2 647 93 98 SGI Corporate Headquarters 900 North McCarthy Blvd. Milpitas, CA 95035 - USA Sales Contact: 1-800-800-7441 http://www.sgi.com/sales/askarep.html 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. Monday April 20th, 2015 4 Program summary: Monday April 20th, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 5 Tuesday April 21st, 2015 Program summary: Tuesday April 21st, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 6 Program summary: Wednesday April 22nd, 2015 Wednesday April 22nd, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. Convention Center map 7 Palais des Papes Convention Center map Rooms Level 0 (Convention Center entrance): Salle des Gardes Level 1: Cellier Benoît XII Level 2: Salle de la Paneterie, access to Cloître Benoît XII Level 3: Salle du Conclave, Chambre du Trésorier, Cubiculaire, Grand Promenoir Registration desk: Salle des Gardes (level 0) Plenary lectures: Salle du Conclave (level 3) Coffee breaks, lunches: Salle des Gardes (level 0), Salle de la Paneterie, Cloître Benoît XII (level 2) Gala dinner: Salle Jeanne Laurent (5 mn walk from Palais des Papes) 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 12 h 00 - 13 h 30 10 h 00 - 12 h 00 9 h 30 - 10 h 00 8 h 30 - 9 h 30 8 h 15 - 8 h 30 Session 4 (Paneterie) Plenary lecture: Prof. Luc VERVISCH (CORIA, France) Welcome Session 3 (Trésorier) ! RANS-based modeling and uncertainty quantification of soot formation in flames, H. Koo, M. Mueller, V. Raman, B. Dally Session 5 (Promenoir) !Theoretical, numerical, and modeling issues in LES, V. Raman Org.: V. Raman and M. Mueller Mini-symposium MS1 (1/2) Large eddy simulation: Challenges and opportunities !Simple and fast integration method for stiff chemical kinetic ODEs, Y. Morii, H. Terashima, M. Koshi, T. Shimizu, E.Shima Chair: V. Giovangigli Contributed presentations CP2 (1/2) Numerical methods Org.: F. Zhang, H. Bockhorn, C.O. Paschereit, W. Schröder, J. Janicka Mini-symposium MS2 Combustion Noise Coffee Break Lunch ! Modelling of flame lift-off evolution: numerical implementation of densitypressure coupling, Z. Chen, S. Ruan, N. Swaminathan ! G-equation based modelling of partially premixed compressible reactive flows, L. Gastaldo, J.C. Latché Org.: H. Schmidt Session 6 (Cubiculaire) ! ! Trivariate chemistry model for the simulation of high Karlovitz premixed turbulent flames, B. Savard, G. Blanquart ! Flamelet/progress variable CFD modeling of high-pressure partial oxidation flame, M. Vascellari, H. Xu, S. Hartl, C. Hasse ! Introduction of unsteadiness and small-scale resolution into steadystate reacting-flow solvers using LEM3D, S. Sannan, A.R. Kerstein ! Comparison between ODT and DNS for ethylene/air auto-ignition, A. Abdelsamie, D. Thévenin !The representative interactive linear-eddy model (RILEM) for regime-independent turbulent nonpremixed combustion modeling combustion, T. Lackman, A.R. Kerstein, M. Oevermann !Reduction of chemical mechanisms for aviation fuels with RCCE, W. Jones, P. Koniavitis, S. Rigopoulos ! Investigation of corrugated premixed methane-air flame structure using REDIM, A. Neagos, V. Bykov, U. Maas !On the benefits of using LEM- and ODT-based approaches in numerical combustion, H. Schmidt !Extraction of Chemical kinetics insights with special CSP data, S. Lam Mini-symposium Contributed presentations MS3 CP3 (1/3) Advances in linear-eddy and Kinetics reduction and one-dimensional-turbulence tabulation modeling Chair: C. Westbrook Simulation of turbulent flames: from high-performance computing to low-order stochastic models for process control Session 2 (Cellier) !Direct combustion noise of premixed flames: experiments and simulation using compressible LES and DNS, F. Zhang, H. Nawroth, ! Using LES of ignition, quenching P. Habisreuther, J. Moeck, and instabilities to design future gas C.O. Paschereit, H. Bockhorn turbines, T. Poinsot, B. Cuenot, ! Physics-based preconditioning and ! Lewis number effects in turbulent L. Gicquel, G. Staffelbach, dual timestepping for stiff nonpremixed sooting flames, A. Dauptain, F. Duchaine combustion problems with detailed ! A hybrid approach to predict A. Attili, F. Bisetti, M.E. Mueller, chemical mechanisms, M. Hansen, combustion noise of premixed H. Pitsch J. Sutherland flames, A. Hosseinzadeh, G. Geiser, ! Assessment of LES combustion J. Janicka, W. Schröder models, H. Pitsch, P. Trisjono ! Large eddy simulation of soot !Adaptation in time and space for formation in an oxy-coal combustor, ! multi-scale combustion fronts with !LES based simulations of D. Lignell, A. Josephson, B. Isaac, error control based on operator combustion noise in a helicopter T. Fletcher splitting and multiresolution, engine, T. Livebardon, L. Gicquel, M. Duarte, S. Descombes, T. Poinsot, E. Bouty M. Massot, C. Tenaud, S. Candel ! Sensitivity of soot production to gaseous kinetic models in LES of ! Estimating indirect combustion aero-engine combustors, ! A new approach to secure scalar noise in nozzle flows with a 2D B. Franzelli, E. Riber, B. Cuenot, boundedness in flame simulation analytical model, J. Zheng, M. Huet, M. Ihme with high-order spectral difference A. Giauque, F. Cléro, S. Ducruix methods on unstructured meshes, E. Bossenec, G. Lodato, ! PDF Modelling of Soot Emissions, P. Domingo, L. Vervisch M.A. Schiener, R.P. Lindstedt !Accounting for strain-rate effects in soot modeling of turbulent flames, B. Franzelli, A. Cuocci, A. Stagni, C. Saggese, A. Frassoldati, T. Faravelli, M. Ihme Chair: S. Dworkin Contributed Presentations CP1 (1/2) Soot Session 1 (Conclave) Monday April 20th, 2015 8 Detailed program: Monday April 20th, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 15 h 30 - 16 h 00 13 h 30 - 15 h 30 ! The challenge of pollutant emission predictions in realistic burners, V. Moureau, G. Lartigue !Validation of multi-physics LES against sparse data, M. Mueller ! ! An efficient operator splitting Monte Carlo for aerosol dynamics simulation with application to soot formation, A. Abdelgadir, K. Zhou, F. Bisetti ! Development of a unique function for soot surface reactivity while oxidation and surface growth in laminar diffusion flames, A. Khosousi, S. Dworkin ! Modeling soot formation and oxidation in laminar premixed flames using the method of moments and a continuous reconstruction of the soot number density function, A. Wick, T.T. Nguyen, F. LaurentNègre, M. Massot, R.O. Fox, H. Pitsch ! Design optimization with LES, Q. Wang ! Modeling the soot particle size distribution in flames using an extended quadrature method of moments approach, S. Salenbauch, ! Data-driven modeling for LES, A. Cuocci, T. Faravelli, C. Hasse K. Duraisamy !Heterogeneous PAH dimerization as an Important contributor to soot nucleation, N. Eaves, S. Dworkin, M. Thomson Org.: V. Raman and M. Mueller Mini-symposium MS1 (2/2) Large eddy simulation: Challenges and opportunities Contributed Presentations CP1 (2/2) Soot Chair: C. Shaddix Session 2 (Cellier) Session 1 (Conclave) Session 4 (Paneterie) !Assessment of double conditioning of reactive scalar transport equations for HCCI, F. Salehi, M. Talei, E.R. Hawkes, A. Bhagatwala, J.H. Chen, S. Kook ! Using large eddy simulation (LES) for pollutant prediction in turbulent flames, T. Jaravel, E. Riber, B. Cuenot !Application of a regularization method to the choice of the strain parameter in LES of a turbulent premixed jet burner, K. Kleinheinz, E. Knudsen, E.R. Hawkes, H. Pitsch !Compressible combustion and nonlinear dynamics for thermoacoustic analysis, C.T. Lee, S. Cant ! Effect of compressibility in the flashback of premixed flames in confined channels, M. Hassanaly, H. Koo, V. Raman ! Simulation of high-pressure turbulent mixing and reacting real gas flows, M. Pfitzner, H. Müller ! Reduced schemes in combustion, A. Felden, B. Cuenot, E. Riber !Using genetic algorithms for optimizing reduced chemicalscheme for turbulent flames simulation, N. Jaouen, P. Domingo, L. Vervisch !An implicit, density-based algorithm for coupled simulations of arbitrary gas mixtures, F. di Mare, L. di Mare Org.: F. di Mare Mini-symposium MS5 Numerical methods for compressible combustion Session 6 (Cubiculaire) ! A reduced and optimized chemical mechanism for n-dodecane oxidation, L. Cai, L.C. Kröger, H. Pitsch ! An adaptive strategy for the ! Towards an integrated approach to efficient implementation of complex realistic engine modeling with large- chemistry in turbulent flame scale computing, S. Aithal simulations, Y. Liang, S. Pope, P. Pepiot ! Coffee Break ! Uncertainties in theoretically predicted elementary rate coefficients , S. Klippenstein ! A DNS study of ignition characteristics of a lean H2/air mixture under HCCI conditions within an enclosed geometry including wall heat transfer, M. Bolla, M. Schmitt, E.R. Hawkes, ! Probabilistic inference of reaction rate parameters based on summary K. Boulouchos statistics, M. Khalil, H. Najm ! Investigations on pressure wave generation and chemical kinetics ! Uncertainty quantification in the during end-gas autoignition on oxidation of vinyl radical, knocking combustion, H. Terashima, F. Goldsmith M. Koshi, ! Correlated uncertainty ! Prediction of super-knock in a quantification: application to downsized spark-ignition engine complex chemical kinetics using Large-Eddy Simulation, mechanisms, K. Truffin, A. Robert, O. Colin, J. Sutton, W. Guo, M. Katsoulakis, C. Angelberger D. Vlachos ! Combining theory and experiment to reduce uncertainties in the derivation of phenomenological rate coefficients for combustion systems, A. Tomlin, R. Shannon, P. Seakins, M. Pilling, M. Blitz, S. Roberson A. Tomlin Chair: D. Goussis Session 5 (Promenoir) Mini-symposium Contributed presentations Contributed presentations MS4 (1/4) CP3 (2/3) CP4 Uncertainty quantification in Kinetics reduction and Internal combustion engines computational combustion tabulation Chair: O. Colin Org.: H. Najm, M. Frenklach, Session 3 (Trésorier) Monday April 20th, 2015 Detailed program: Monday April 20th, 2015 9 18 h 00 - 18 h 30 16 h 00 - 18 h 00 ! Direct numerical simulation of low Mach number combustion beyond petascale, C. Frouzakis, A. Tomboulides, S. Kerkemeier, P. Fisher ! Legacy codes and the jump to exascale : Fluid dynamics simulation with AVBP and HPC, G. Staffelbach, O. Vermorel, S. Jaure, C. Fournier, I. D'Ast, E. Oseret !Combustion simulation on nextgeneration architectures, J. Bell ! Toward large-eddy simulation of complex burners with exascale super-computers: a few challenges and solutions, V. Moureau !Legion as a programming model for combustion simulation at the exascale, J. Chen, M. Bauer, S. Treichler, A. Aiken, A. Bhagatwala Org.: J. Chen and R. Sankaran ! Statistical calibration of simplified chemical mechanisms for Diesel engine combustion, L. Hakim, G. Lacaze, M. Khalil, H. Najm, J. Oefelein ! Uncertainty quantification of the rate parameters of the syngas combustion system, T. Varga, C. Olm, I.G. Zsély, T. Nagy, E. Valkó R. Palvolgyi, H.J. Curran, T. Turányi ! Combining theoretical and experimental data in uncertainty quantification across multiple scales, M. Burke ! Comparison of probabilistic and deterministic frameworks of uncertainty quantification, M. Frenklach, A. Packard, J. Sacks, R. Pablo, G. Garcia-Donato A. Tomlin ! Flow curvature method applied to dynamical systems analysis, J.M. Ginoux ! Optimal dimension of the combustion mechanism in the ignition problem, V. Bykov, V. Gol'dshtein, U. Maas !Analytical prediction of syngas induction times, P. Boivin, A. Sanchez, F. Williams !The developing explosive dynamics in the vicinity of the third explosion limit of H2/air, E.A. Tingas, T. Turanyi, D. Goussis ! Uncertainties in predicting the ignition properties of biofuels: what constraints can measurements provide?, A. Tomlin, E. Agbro !How details of fuel molecular structure can accelerate or retard autoignition, C. Westbrook Org.: M. Valorani Session 4 (Paneterie) Session 5 (Promenoir) !Three-dimensional structure of hypergolic ignition process for hydrazine/nitrogen dioxide un-like doublet impinging gas jets, Y. Daimon, H. Tani, H. Terashima, M. Koshi ! Computational modeling of boron particle fueled ducted rocket combustion chambers, B. Kalpakli, ! A consistent approach for coupling E. Acar the devolatilization of coal particles with tabulated chemistry, !Numerical simulation of turbulent R. Knappstein, A. Ketelheun, combustion between a hydrogen jet G. Künne, J. Köser, A. Dreizler, and a supersonic vitiated air A. Sadiki, J. Janicka crossflow, A. Techer, G. Lehnasch, A. Mura ! Macropore growth in char particles under different gasification !Numerical investigation of conditions , K. Wittig, A. Richter, transverse hydrogen jet into M. Kestel, P. Nikrityuk, B. Meyer supersonic crossflow using detached eddy simulation, L. Zhaoyang, W. Zhenguo, !How does turbulent fluid motion S. Mingbo, W. Hongbo influence heterogeneous combustion?, J. Krüger, N. Haugen, T. Løvås ! Numerical study of flame dynamic characteristics in a supersonic combustor with dual cavity, Y. Yixin, ! Application of a skeletal kinetic mechanism on LES of a pulverized W. Hongbo, W. Zhenguo, S. Mingbo coal jet flame, S. Ahn, H. Watanabe, K. Tanno, N. Hashimoto ! Numerical simulation on detonation initiation and propagation in supersonic combustible mixtures with nonuniform velocities and species, X. Cai, J. Liang, Z. Lin, R. Deiterding, Y. Che ! A modified two-step model for devolatilization, A. Richards, T.H. Fletcher !Adaptive kinetic model for coal devolatilization in oxy-coal combustion conditions, S. Iavarone, C. Galletti, F. Contino, L. Tognotti, A. Parente and P.J. Smith Chair: C. Fureby Session 6 (Cubiculaire) Mini-symposium Mini-symposium Contributed presentations Contributed presentations MS4 (2/4) MS7 (1/2) CP7 CP6 Uncertainty quantification in Diagnostics for auto-ignition Rocket engines, ramjets, Coal combustion computational combustion processes scramjets Chair C. Hasse Org.: H. Najm, M. Frenklach, Session 3 (Trésorier) ! Bayesian inference of chemical ! Large eddy simulation of turbulent kinetic models from proposed combustion with a dynamic second- ! Physical and algorithmic reactions, N. Galagali, Y. Marzouk order moment closure model, approaches to numerical K. Luo, J. Yang, Y. Bai, J. Fan combustion at the exascale, S. Cant !Region of influence-based sensitivity analysis in turbulent ! A dynamically adaptive combustion combustion, V. Carey, R. Moser framework for the general description of complex flame configurations, H. Wu, Y.C. See, Q. Wang, M. Ihme ! A dynamic sub-grid model for turbulent combustion based on conditional averaging, G. R. Hendra, M.M Salehi, W.K. Bushe ! Analysis of a dynamic flame wrinkling factor model for large eddy simulations of turbulent premixed combustion, P. Stefanin-Volpiani, T. Schmitt, D. Veynante ! Modelling of lean premixed flames based on a new artificial thickened flame framework, K. Vogiatzaki, S. Navarro-Martinez, S. Hong, S. Shanbhogue, A. Ghoniem !Modelling of the subgrid scale wrinkling factor for large-eddy simulation of turbulent premixed combustion, F. Thiesset, G. Maurice, F. Halter, N. Mazellier, C. Chauveau, I. Gökalp Chair: O. Gicquel Mini-symposium MS6 (1/2) Towards exascale simulation of turbulent combustion Session 2 (Cellier) Contributed presentations CP5 (1/6) Turbulent combustion Session 1 (Conclave) Monday April 20th, 2015 10 Detailed program: Monday April 20th, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 12 h 00 - 13 h 30 10 h 00 - 12 h 00 ! Using GPUs to accelerating nonstiff and stiff chemical kinetics in combustion simulations, K. Niemeyer ! The framework of multiple mapping ! GPU based chemistry acceleration conditioning, its applications and for reacting flow simulations , J. Lee, prospects, A. Klimenko, M. Cleary, V. Sankaran B. Sundaram, L. Dialameh ! Accelerating combustion modeling !Radiation effects on a turbulent with the Intel Xeon Phi, C. Stone premixed syngas flame, S. Karimkashi, M. Bolla, H. Wang, E.R. Hawkes, H.G. Im, P.G. Arias, M. Talei !Analysis of approximate deconvolution strategies and explicit filtering for LES of turbulent flames, ! Efficient abstractions for exascale P. Domingo, L. Vervisch software design, J. Sutherland ! Chemical source term modeling in LES of reacting flows using deconvolution method, Q. Wang, M. Ihme Org.: J. Chen and R. Sankaran Mini-symposium MS6 (2/2) Towards exascale simulation of turbulent combustion !Reacting flow simulation framework for heterogeneous and many-core architectures, R. Sankaran, B.J. Lee, H. Im ! Radiative properties of molecular combustion products and of some hydrocarbons: state of the art, P. Rivière, A. Soufiani Session 4 (Paneterie) !Numerical study of alkane-air spray combustion, C. Nicoli, P. Haldenwang, B. Denet Chair: M. Arienti Lunch ! Analysis of multi-phase material model for Al/AlOx droplet, K. Lee, D.S. Stewart ! Spray modeling using a method of moment approach with droplet size ! LES analysis of the influence of the distribution reconstruction, turbulence intensity on the critical M. Pollack, S. Salenbauch, ignition energy of a lean C. Hasse methane/air flame, S. Mouriaux, O. Colin, B. Renou, D. Veynante ! Counter-flow spray pulsated flames: from experimental ! The flame consumption speed – measurements to numerical comparison of theory and numerical simulation , J.C. Brändle de Motta, simulations, G. Giannakopoulos, M. Boileau, L. Zimmer, B. Robbes, C. Frouzakis, M. Matalon, F. Laurent-Nègre, M. Massot A. Tomboulides ! Large eddy simulations of ! A statistical model to predict polydisperse turbulent n-heptane ignition probability, L. Esclapez, spray flames with a dynamic ATF E. Riber, B. Cuenot procedure coupled with the FGM chemistry reduction method, F.L. Sacomano Filho, M. Chrigui, ! A method for predicting sensitive rate coefficients with high accuracy A. Sadiki, J. Janicka tested for the H2/CO/O2-system, T. Methling, M. Braun-Unkhoff, ! Toward the fully realizable large U. Riedel eddy simulation of disperse phase flows: Eulerian kinetic modelling and associated numerical methods, ! M. Sabat, A. Vié, A. Larat, M. Massot ! Direct numerical simulations of kernel growth of turbulent premixed syngas/air flames, G. Giannakopoulos, C. Frouzakis, M. Matalon, A. Tomboulides Chair A. Trouvé Contributed presentations Contributed presentations CP8 (1/2) CP9 (1/2) Ignition / quenching Two-phase flows Coffee Break Session 5 (Promenoir) Org.: A. Kapila and D. Schwendeman Mini-symposium MS8 Modeling and computation of detonations in highenergy explosives !A discrete model of gas combustion in porous media, F. Sirotkin, R. Fursenko, S. Minaev Chair: A. Parente Contributed presentations CP10 (1/2) New technologies Session 6 (Cubiculaire) !Large-eddy simulation of a MILD combustion burner using conditional source-term estimation , J. Labahn, C. Devaud ! Direct numerical simulation of MILD combustion, U. Göktolga, J. van Oijen, P. de Goey ! The effect of the three! Coupled simulations of condensed- dimensionality of the flame front phase explosives and elastic-plastic inside a radial-flow porous burner – solids, L. Michael, N. Nikiforakis a detailed DPLS numerical study, I. Dinkov, P. Habisreuther, H. Bockhorn ! Propagation of detonations in curved geometries of compliantly confined solid explosives , ! Lean hydrogen flames in a narrow E. Ioannou, L. Michael, adiabatic channel, C. Jiménez, N. Nikiforakis D. Fernández-Galisteo, V. Kurdyumov ! Application of two-dimensional shock-fitting to detonation ! A numerical investigation of propagation in high explosives, heterogeneous/homogeneous T. Aslam, C. Romick combustion of n-dodecane over Pt for mcrocombustion, E. Tolmachoff, A. Booth, I. Lee !Sensitivity of run-to-detonation distance in practical explosives, J Gambino, D. Schwendeman, A. Kapila The hydrodynamic theory of premixed flames: laminar tot turbulent propagation !Spectral model of premixed turbulent flame, S. Rashkovskiy Chair: B. Fiorina Contributed presentations CP5 (2/6) Turbulent combustion Session 3 (Trésorier) Plenary lecture: Prof. Moshe MATALON (University of Illinois, Urbana-Champaign, USA) Session 2 (Cellier) 9 h 30 - 10 h 00 8 h 30 - 9 h 30 Session 1 (Conclave) Tuesday April 21st, 2015 Detailed program: Tuesday April 21st, 2015 11 15 h 30 - 16 h 00 13 h 30 - 15 h 30 Org.: F. Duchaine and L. Gicquel Mini-symposium MS9 (1/2) High-fidelity coupled simulation of reacting systems Session 2 (Cellier) ! Modal analysis of reacting jet in cross flow using direct numerical simulations, T. Sayadi, S. Lyra, C. Hamman, H. Kolla, P. Schmid, J.H. Chen ! Direct numerical simulation of a turbulent lifted DME jet flame in a heated coflow at elevated pressure, Y. Minamoto, J.H. Chen ! Direct numerical simulation of spontaneous flame propagation under reactivity controlled compression ignition conditions, A. Bhagatwala, R. Sankaran, S. Kokjohn, J.H. Chen !Analysis of large-eddy simulations of laboratory-scale fire, M. Rochoux, B. Cuenot, F. Duchaine, E. Riber, D. Veynante, N. Darabiha !An a priori DNS study of molecular !Large eddy simulation and the mixing models in a planar stationary filtered probability density function premixed flame, X. Zhao, H. Kolla, method, W. Jones J.H. Chen ! Heterogeneous multi-scale LES ! 3D DNS of methane-air turbulent using skeletal reaction mechanisms, premixed planar flame in thin C. Fureby, N. Zettervall, reaction zones, B. Yenerdag, K. Nordin-Bates Y. Naka, M. Shimura, M. Tanahashi ! Coupling approach to account for !A peta-scale direct numerical heat losses in a practical simulation study on pollutant combustor, D. Mira, M. Zavala-Ake, formation in turbulent premixed S. Gövert, M. Vazquez, flames, P. Trisjono, H. Pitsch G. Houzeaux Chair: S. Menon Contributed presentations CP5 (3/6) Turbulent combustion Session 1 (Conclave) ! Numerical simulation of shock wave dynamics internal ballistics of gun, M. Chirame, D. Pradhan, S. Naik ! Flame acceleration in channels with cold walls, C. Dion, B. Demirgok, V. Akkerman, D. Valley, V. Bychkov Chair: A. Matsuo Session 5 (Promenoir) ! Modeling and simulation of a dense evaporating spray, F. Doisneau, M. Arienti, J. Oefelein ! Detonation shock propagation !Compressibility effects in the initial through nitromethane embedded transient of high-pressure Diesel metal foam, B. Lieberthal, injection, M. Arienti, M. Sussman D.S. Stewart S. Rashkovskiy, A. Dolgoborodov ! Numerical simulations of kerosene spray atomization with various ! Simulations of compressible gashybrid breakup models, T. Liu, dust flows, E. Oran, R. Houim L. Hu, Y. Wu, D. Zhao, G. Wang ! A monotonic mixing-describing composition-space variable for the flamelet formulation of spray flames, B. Franzelli, A. Vié, M. Ihme ! Symmetric Baer and Nunziato model, R. Saurel !Numerical investigation of sheardriven primary breakup of liquid fuel !3D simulation of discrete flows, C. Bilger, T. Pringuey, combustion waves in mechanically R.S. Cant activated powder mixtures, !Development of a turbulent liquid flux model for Eulerian-Eulerian multiphase flow simulations, S. Puggelli, A. Andreini, C. Blanchini, B. Facchini, F.X. Demoulin Coffee Break ! Adjoint-based sensitivity for understanding chemistry modeling, V. Raman, K. Braman, T. Oliver !Effect of uncertainties in detailed chemistry on combustion model parameters, N. Dumont, R. Vicquelin, O. Gicquel ! Combustion properties of H2/CO mixtures: consistent chemical mechanism from collaborative data processing, N. Slavinskaia, U. Riedel, J.H. Starcke ! Facilitating uncertainty quantification for large, detailed reaction mechanisms, R. West, S. S. Goldsborough, A. Fridlyand ! Handling model error in the calibration of physical models, H. Najm, K. Sargsyan, R. Ghaneim Org.: H. Najm, M. Frenklach, A. Tomlin Chair: P. Haldenwang Session 4 (Paneterie) ! ! Computational modeling of plasma assisted combustion in gas turbines for electric power generation, A. Ehn, J. Zhu, P. Petersson, Z. Li, M. Alden, C. Fureby, T. Hurtig, N. Zettervall, A. Larsson, J. Larfeldt ! Simulation of turbulent premixed combustion with a self-consistent plasma model for initiation, H. Sitaraman, R. Grout ! Application of a two-fluid plasma model to the simulation of premixed flames under electric fields, T. Casey, P. Arias, J. Han, M. Belhi, F. Bisetti, H. Im, J.Y. Chen ! Steady laminar one-dimensional flame solvers for modelling ions in flames, B.J. Lee, M.S. Cha, H. Im, S.H. Chung !Numerical aspects of the shockless explosion combustion, P. Berndt, R. Klein, C.O. Paschereit Chair: P. Domingo Session 6 (Cubiculaire) Mini-symposium Contributed presentations Contributed presentations Contributed presentations MS4 (3/4) CP9 (2/2) CP11 (1/3) CP10 (2/2) Uncertainty quantification in Two-phase flows Detonation New technologies computational combustion Session 3 (Trésorier) Tuesday April 21st, 2015 12 Detailed program: Tuesday April 21st, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 19 h 30 - 23 h 00 18 h 00 - 18 h 30 16 h 00 - 18 h 00 !A 3D coupled approach for the thermal design of aero-engine combustor liners, A. Andreini, B. Facchini, L. Mazzei ! LES flamelet-progress variable modeling of a turbulent piloted dimethyl ether flame, S. Popp, F. Hunger, S. Hartl, D. Messig, B. Coriton, J. Franck, F. Fuest, C. Hasse ! DNS and experiments of H2/He jets in a vitiated turbulent crossflow, ! Towards coupled simulation of S. Lyra, B. Wilde, H. Kolla, aero-engines HP system using T.C. Lieuwen, J.H. Chen industrial CFD solvers, P. Legrenzi ! Numerical simulation of NOx formation in syngas combustion, H. Watanabe, T. Kawai, S.Y. Ahn !Turbulent diffusion flame modelling ! Automatic determination of the in Large Eddy Simulation (LES), coupling period in unsteady F. Shum-Kivan, B. Cuenot, E. Riber conjugate heat transfer. Application to large eddy simulation of flamewall interaction, C. Koren, ! LES combustion modelling of the R. Vicquelin, O. Gicquel dynamics of turbulent pulsed premixed flames, A. Chatelier, R. Mercier, N. Bertier, B. Fiorina ! Aerothermal prediction of an aeronautical combustion chamber based on the coupling of large eddy ! DNS of high turbulence intensity simulation, solid conduction and premixed methane-air flames in a radiation numerical solvers, 3D inflow-outflow configuration, S. Berger, F. Duchaine, L. Gicquel, G. Nivarti, S. Cant S. Richard Org.: F. Duchaine and L. Gicquel Mini-symposium MS9 (2/2) High-fidelity coupled simulation of reacting systems Contributed presentations CP5 (4/6) Turbulent combustion Chair: W. Jones Session 2 (Cellier) Session 1 (Conclave) ! ! Reducing chemistry table sizes by using polynomial functions, W. Leudesdorff, A. Ketelheun, J. Janicka Chair: P. Pepiot Session 5 (Promenoir) ! Investigation of the effect of correlated uncertain rate parameters on a model of hydrogen combustion using a generalized HDMR method, E. Valkó, A. Tomlin, T. Varga, T. Turányi ! Numerical investigation of ignition ! From fully premixed to highly in laminar methane/LOx flamelet at stratified combustion using hybrid supercritical pressures, transported tabulated chemistry, P.E. Lapenna, P.P. Ciottoli, F. Creta, B. Duboc, G. Ribert, P. Domingo M. Valorani !Comparative analysis of kinetic ! Impact of pilot injections on ignition mechanism reduction methods behaviour at low in-cylinder within the tabulation of dynamic temperatures from extreme miller adaptive chemistry framework, valve timing in a heavy-Duty Diesel F. Contino, T. Lucchini, S. Backaert, engine, S. Pandurangi, Y.M. Wright, N. Bourgeois, A. Parente, C. Brückner, P. Kyrtatos, H. Jeanmart K. Boulouchos ! Principal component transport in ! Regimes of auto-ignition in turbulent combustion, T. Echekki, homogeneous reactant mixtures H. Mirgolbabaei with temperature fluctuations, H. Im, P. Pal, M. Wooldridge, ! Principal component analysis for A. Mansfield modelling turbulent premixed flames, A. Coussement, O. Gicquel, ! Auto-ignition diagnostics using the B. Isaac, A. Parente tangential stretching rate concept , M. Valorani, S. Paolucci ! !The developing explosive dynamics during the autoignition of a DME/air mixture, E.A. Tingas, D. Kyritsis, D. Goussis Banquet ! Industrial & applied V/UQ for a 15MW coal-fired boiler, P. Smith, S. Smith ! Optimal experimental design of furan shock tube kinetic experiments, D. Kim, F. Bisetti, Q. Long, R. Tempone, A. Farooq, A. Knio ! Uncertainty quantification of rate rules for normal alkanes using Bayesian analysis, L. Cai, L.C. Kröger, V. Raman, H. Pitsch !Needs and progress in assessing the accuracy of LES, J. Oefelein, G. Lacaze A. Tomlin Org.: M. Valorani Session 4 (Paneterie) Session 6 (Cubiculaire) ! ! Development of an unstructured CFD solver for the modeling of industrial furnaces, L. Romagnosi, J.E. Anker, K. Claramunt, C. Hirsch ! A laboratory-scale downsizing procedure for highly resolved LES of large-scale combustion system, B. Farcy, D. Midou, L. Vervisch, P. Domingo ! Large eddy simulation of coal and biomass co-firing, M. Rabacal, M. Costa, A. Kempf ! Avoiding ring formation in cement kilns, D. Lahaye !Effect of particle shrinkage on the biomass pyrolysis behavior in a hightemperature entrained-flow reactor, X. Ku, T. Li, T. Løvås Mini-symposium Mini-symposium Contributed presentations Contributed presentations MS4 (4/4) MS7 (2/2) CP3 (3/3) CP12 Uncertainty quantification in Diagnostics for auto-ignition Kinetics reduction and Industrial furnaces computational combustion processes tabulation Chair: A. Kempf Org.: H. Najm, M. Frenklach, Session 3 (Trésorier) Tuesday April 21st, 2015 Detailed program: Tuesday April 21st, 2015 13 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 12 h 00 - 13 h 30 10 h 00 - 12 h 00 9 h 30 - 10 h 00 8 h 30 - 9 h 30 Session 4 (Paneterie) Plenary lecture: Dr. Joseph OEFELEIN (Sandia, USA) Session 3 (Trésorier) Chair: A. Tomboulides !Direct numerical simulation of an igniting, turbulent jet with global nheptane chemistry at enginerelevant conditions, A. Krisman, E.R. Hawkes, A. Bhagatwala, ! DNS investigation on thermoacoustic oscillation characteristics of J. Chen turbulent swirling premixed flame in a cuboid combustor, K. Aoki, ! Direct numerical simulation of M. Shimura, Y. Naka, M. Tanahashi investigation of autoignition with split fuel injection, D.H. Shin, E. Richardson ! Gas expansion and shear layer effects in turbulent premixed flames, S. Schlimpert, A. Feldhusen, ! J. H. Grimmen, B. Roidl, M. Meinke, W. Schröder ! Effects of hydrogen added on the ignition of iso-octane in a diffusion ! Large eddy simulation of layer, Z. Li, Z. Chen transverse modes in a swirled kerosene/air combustor, A. Ghani, ! Predicting extinction in condensed T. Poinsot, L. Gicquel phase combustion in a counterflow geometry, S. Koundinyan, ! Acoustically induced vortex core D.S. Stewart, M. Matalon, J. Bdzil flashback in a staged swirlstabilized combustor, C. Lapeyre, ! Large eddy simulation of extinction M. Mazur, P. Scouflaire, limits in two-dimensional plane F. Richecoeur, S. Ducruix, T. Poinsot buoyant turbulent diffusion flames, S. Vilfayeau, J. White, A. Trouvé ! Dynamical modeling of combustion instabilities in liquid rocket engines, M. Gonzalez-Flesca, T. Schmitt, S. Ducruix, S. Candel !Nonlinear stability study of a timedelayed thermoacoustic system, X. Yang, A. Turan, S. Lei Chair: T. Shimizu Session 5 (Promenoir) Coffee Break !A modeling study of the lowtemperature oxidation of n-hexane, 2-methyl-pentane and 3-methylpentane, Z. Serinyel, O. Herbinet, R. Fournet, V. Warth, F. Battin-Leclerc ! Transition state theory based semiautomatic generation of surface reaction mechanisms, P. Kraus, P. Lindstedt ! Predictive fire simulations with the software ISIS, C. Lapuerta, S. Suard, F. Babik, G. Boyer ! PoKiTT: an efficient, platform agnostic package for thermodynamics, kinetics, and transport properties in reactive flow simulations, N. Yonkee, J. Sutherland ! Lunch ! Tabulation strategies of partially premixed dimethyl-ether flames, S. Hartl, A. Zschutschke, D. Messig, C. Hasse Scalable parallel and computationally-conservative implementations of the conditional moment closure model in large eddy ! Kinetic modeling of the combustion simulations, H. Zhang, of 2-methyl-2-butene, C. A. Garmory, E. Mastorakos Westbrook, P.A. Glaude, F. BattinLeclerc, O. Herbinet, O. Mathieu, ! A methodology for the integration E. Petersen of stiff chemical kinetics on GPUs and application to the LES-PDF ! Comparison of the performance of simulation of a turbulent nonseveral recent hydrogen and syngas premixed flame, F. Sewerin, combustion mechanisms, C. Olm, S. Rigopoulos I.G. Zsély, R. Pálvölgyi, T. Varga, T. Nagy, H.J. Curran, T. Turányi ! Load balancing, dynamic repartitioning, and data migration in ! Kinetic study of the combustion of turbulent reactors' simulation, phenolic tars from biomass, P. Pisciuneri, E. Meneses, A. Zheng, M. Nowakowska, O. Herbinet, A. Labrinidis, P. Chrysanthis, P. Givi A. Dufour, P.A. Glaude Chair: N. Darabiha Chair: G. Staffelbach ! Finite volume methods for evolving surfaces, K. Mikula, P. Frolkovic, M. Remesikova ! Regularized flux-based gradient for level set methods, J. Hahn !A flame surface tracking method for calculating the premixed combustion in engines, P. Priesching Org.: J. Hahn and P. Priesching Mini-symposium MS11 Surface evolution methods in gasoline direct injection combustion engines Session 6 (Cubiculaire) ! Simulation of turbulent premixed combustion in turbocharged direct injection gasoline engines using a ! Transported probability density level set based flamelet model, function modeling of pulverized coal D. Linse combustion, X. Zhao ! Oxy-coal power boiler simulation and validation through extreme computing, P. Smith, J. Thornock, Y. Wu, S. Smith, B. Isaac ! LES and RANS of pulverized coal oxy-combustion in swirl burners, A. Bogusławski, P. Warzecha !Developing and validating an oxyfuel CFD modelling methodolody, R. Knappstein, A. Kethelheun, G. Kuenne, S. Farazi, M. Baroncelli, A. Sadiki, H. Pitsch, J. Janicka Org.: A. Sadiki and H. Pitsch Mini-symposium MS10 LES modeling of oxy-coal combustion Modeling and Simulation of Real-Fluid Thermodynamics and Transport in Advanced Combustion Systems Session 2 (Cellier) Contributed presentations Contributed presentations Contributed presentations Contributed presentations CP13 CP8 (2/2) CP14 CP15 Instabilities Ignition / quenching HPC / Software engineering Kinetics Session 1 (Conclave) Wednesday April 22nd, 2015 14 Detailed program: Wednesday April 22nd, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 15 h 30 - 16 h 00 13 h 30 - 15 h 30 ! Challenging modeling strategies for LES of non-adiabatic turbulent stratified combustion, B. Fiorina, R. Mercier, G. Kuenne, A. Ketelheun, A. Advic, J. Janicka, D. Geyer, A. Dreizler, E. Alenius, C. Duwig, P. Trisjono, K. Kleinheinz, S. Kang, H. Pitsch, F. Proch, F. Cavallo-Marincola, A. Kempf ! Large-eddy simulation/probability density function modelling of Cambridge stratified flame, H. Turkeri, M. Muradoglu ! Resolved flame LES of the Cambridge stratified burner using PFGM, F. Proch, A.M. Kempf ! Preferential diffusion effects in LES of mild combustion with flamelet generated manifolds , S.E. Abtahizadeh, R. Bastiaans, J. van Oijen, P. de Goey ! Analysis of ignition regimes in rapid compression machines, M. Ihme, K. Grogan, S. Goldsborough ! Direct numerical simulation of the auto-ignition of turbulent ethylene/air mixtures, A. Abdelsamie, D. Thévenin ! LES simulation of reactive flow with effective PDF models at a minimal cost, Y. Ge, F. Ferraro, M. Pfitzner ! DNS of plasma-assisted ignition in quiescent and turbulent flow conditions , M. Castela, B. Fiorina, O. Gicquel, A. Coussement, C. Laux, N. Darabiha ! Chair: M. Matalon ! Propagation of premixed flames in long narrow channels from a closed end: transition from constant speed to rapid acceleration, V. Kurdyumov, M. Matalon ! Direct numerical simulation of circular expanding premixed flames in a lean quiescent hydrogen-air mixture: Phenomenology and detailed flame front analysis, C. Altantzis, C. Frouzakis, A. Tomboulides, K. Boulouchos ! Numerical study of interaction between Darrieus-Landau instability and spatially periodic shear flow, D. Valiev, A. Gruber, C.K. Law, J.H. Chen ! Diffusive-thermal instabilities in premixed flames: stepwise ignitiontemperature kinetics, L. Kagan, I. Brailovsky, P. Gordon, G. Sivashinsky ! The analysis diffusional-thermal stability of rich hydrogen-air combustion waves in model with detailed kinetic mechanisms, V. Gubernov, A. Korsakova, A. Kolobov, V. Bykov, U. Maas !Transient response of a laminar premixed flame to a radially diverging/converging flow, M. Sahafzadeh, L. Kostiuk, S. Dworkin Coffee Break !Compressible mixing of liquid fuels, A. Hernandez, S. Stekovic, D.S. Stewart, A. Wass ! Development, validation and application of a DNS approach for heterogeneous reactive flows, A. Chabane, K. Truffin, A. Nicolle, F. Nicoud, C. Angelberger !Large Eddy Simulation using adaptive mesh refinement with a multi-level subgrid scaled closure for turbulent reacting flows, B. Muralidharan, S. Menon !High-Order CENO Finite Volume Scheme for Large-Eddy Simulation of Turbulent Reactive Flows, L. Tobaldini Neto, C. Groth !Ignition in a premixed propane/air gas by turbulent jets of its exhaust products, A. Ghorbani, S. Fischer, G. Steinhilber, D. Markus, U. Maas Chair: J. Sutherland !Numerical and experimental analysis of laminar and turbulent oxy-fuel jet flames using a direct comparison of the Rayleigh signal, F. Hunger, M.F. Zulkifli, B.A.O. Williams, F. Beyrau, C. Hasse ! Fuel effects on bluff-body stabilized turbulent premixed flames, V. Katta, W. Roquemore Session 4 (Paneterie) Session 5 (Promenoir) Session 6 (Cubiculaire) ! LES of transcritical LOx/GH2 injection in a rocket engine, L. Matuszewki, P. Gaillard, V. Giovangigli ! The absence of a dense potential core in transcritical injection, D. Banuti, K. Hannemann !Diffuse interface methods for diffusion flame from subcritical to supercritical pressure, P. Gaillard, V. Giovangigli, L. Matuszewki Chair: J. Oefelein ! Two-phase simulations of TNT/Aluminium afterburning during explosions, E. Fedina, C. Fureby A. Kasimov, B. Emolaev ! Large Eddy Simulation of transverse acoustic activity in a 42injector rocket engine, A. Urbano, L. Selle, G. Staffelbach, B. Cuenot, T. Schmitt, S. Ducruix, S. Candel ! Oscillatory combustion in a subscale liquid rocket chamber, T. Shimizu, Y. Morii, Y. Mizobuchi, Y. Numone, T. Tomita, H. H. Kawashima, M. Hishida ! A compressible LEM-LES strategy (CLEM-LES) for modeling turbulent detonation propagation, B. Maxwell, ! Large eddy simulations of a S. Falle, G. Sharpe, M. Radulescu naturally unstable transcritical coaxial injector, A. Coussement, ! Set-valued solutions for T. Schmitt, S. Ducruix, S. Candel, detonations with losses, Y. Nunome R. Semenko, L. Faria, ! ! Weakly nonlinear detonations: theory and numerics , L. Faria, A. Kasimov, R. Rosales !Three-dimensional simulations of shock-induced combustion around a spherical projectile with various diameters, Y. Sakuragi, A. Matsuo Chair: G. Lodato Contributed presentations Contributed presentations Contributed presentations Contributed presentations CP2 (2/2) CP16 CP11 (2/3) CP17 Numerical methods Laminar flames Detonation Real gases effects Session 3 (Trésorier) Org.: D. Markus, U. Maas, D. Thévenin Mini-symposium MS12 (1/2) Safety related ignition process Session 2 (Cellier) Chair: E. Hawkes Contributed presentations CP5 (5/6) Turbulent combustion Session 1 (Conclave) Wednesday April 22nd, 2015 Detailed program: Wednesday April 22nd, 2015 15 16 h 00 - 18 h 00 !Ignition at sub-millimeter sized hot particles: a numerical and experimental study, D. Roth, T. Häber, H. Bockhorn !Numerical investigation of ignition over heated spheres, J. Melguizo-Gavilanes, J. Shepherd ! ! Sensitivity of internal flow dynamics and boundary conditions on a turbulent opposed jet flame configuration, A. Ruiz, G. Lacaze, B. Coriton, J. Franck, J. Oefelein ! Analysis of natural gas fired furnaces with different oxygen enrichments, R. Prieler, M. Demuth, C. Hochenauer ! Methanol ignition by a line energy source embedded in a wall, ! Implicit large-eddy simulation of a M. Sanchez-Sanz, Bunsen-like burner with multi-scale E. Fernandez-Tarrazo, A. Sanchez turbulent forcing, S. Zhao, N. Lardjane, I. Fedioun ! Quasi-spectral element method for numerical integration of stiff ! LES/FDF of a lean premixed unsteady combustion systems, methane counterflow flame, A. Koksharov, V. Bykov, U. Maas M. Rieth, J.Y. Chen, F. Proch, P. Lindstedt, A. Kempf !Effects of preheat temperatures and turbulence intensities on the disruption of the reaction zone in high Karlovitz premixed flames, S. Lapointe, G. Blanquart Org.: D. Markus, U. Maas, D. Thévenin ! Large eddy simulation of a twostage liquid fueled swirl burner: influence of injection strategies, B. Cheneau, A. Vié, S. Ducruix Session 6 (Cubiculaire) ! On the influence of instabilities on the direct blast initiation of twodimensional detonations, H.D. Ng, C.B. Kiyanda, G.H. Morgan, N. Nikiforakis ! Interactions of turbulence and scalar structures in shear-driven premixed turbulent flames using DNS, H. Wang, E. Hawkes, H. Kolla, J. Chen ! ! Numerical investigation on detonation initiation and propagation in supersonic reactive flows, J. Li, Q. Xiao, W. Fan ! Detonation in a radial supersonic outflow with simplified and realistic chemistry, S. Korneev, A. Kasimov ! Numerical investigation on cellular H2-O2-Ar detonation evolution in mixtures with different hydrogen concentrations, Q. Xiao, W. Fan, H. Li, K. Wang, Q.Y. He ! Is is possible to achieve DDT behind a continuously growing Mach stem?, Y. Lv, M. Ihme Chair: M. Ihme ! Two- and three-dimensional numerical simulations of wildland fires, A. Lamorlette, D. Morvan ! Evaluation of a sensor-driven wildland fire spread modeling strategy using the FireFlux experiment, C. Zhang, M. Durand, W. Tang, M. Goliner, A. Trouvé, M. Rochoux, S. Ricci, B. Cuenot, J.B. Filippi, C. Clements ! LES modelling of burning wildland fuels, M. El Houssami, A. Lamorlette, D. Morvan, A. Simeoni ! How simulating the behaviour of wildland fires?, D. Morvan ! Mathematical modelling of the interaction between crosswind and aviation-fuel fire engulfing a full scale aircraft, G. Wang, H.Y. Wang !CFD simulations of fire-induced doorway flows in a small scale enclosure, S. Suard, A. Koched, H. Petrel, L. Audouin Chair: B: Cuenot Contributed presentations Contributed presentations CP11 (3/3) CP19 Detonation Fires Session 5 (Promenoir) ! Extinction and reignition Dynamics in turbulent dimethyl ether jet flames, A. Bhagatwala, E. Hawkes, P.T. Bremer, A. Gyulassy, J.Y. Chen !Topology of turbulent premixed flame interaction, R. Griffiths, H. Kolla, W. Kollman, J. Chen, S. Cant Org.: S. Cant Mini-symposium MS13 Flame topology Session 4 (Paneterie) ! Combustor effusion cooling design parameters analysis and modeling, ! Effective normal strain rate and N. Savary, R. Hervé, G. Cottin scalar gradient enhancement, L. Cifuentes, C. Dopazo, J. Martin, P. Domingo, L. Vervisch, C. Jimenez ! Direct numerical simulation of premixed flames flashback in turbulent channel flows with fuel stratification, A. Gruber, J.H. Chen ! Modelling chemistry in coupled combustor and turbine CFD simulations, R. Eggels ! Detailed analysis of light-round in an annular multiple-injector combustor, M. Philip, R. Vicquelin, M. Boileau, T. Schmitt, S. Candel !Numerical computation methods of hot gases radiation in a turbofan combustion chamber: performance comparison for a design process, R. Daviot, F. Loureiro Chair: T. Poinsot Contributed presentations CP18 Gas turbine combustion Mini-symposium MS12 (2/2) Safety related ignition process Contributed presentations CP5 (6/6) Turbulent combustion Chair: L. Vervisch Session 3 (Trésorier) Session 2 (Cellier) Session 1 (Conclave) Wednesday April 22nd, 2015 16 Detailed program: Wednesday April 22nd, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. Detailed program: Wednesday April 22nd, 2015 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 17 18 Invited plenary lectures INVITED PLENARY LECTURES PL1: Monday April 20th (8h30 – 9h30) Simulation of turbulent flames: from high-performance computing to low-order stochastic models for process control Luc Vervisch CORIA, CNRS and INSA de Rouen, France. The objectives and challenges of turbulent combustion modeling are summarized in introduction, with an overview of strategies and methods to simulate flames and practical burners. Then, most recent sub-grid scale modeling approaches for Large Eddy Simulation (LES) of turbulent premixed and non-premixed flames are discussed and some applications are presented. The time history of turbulent mixing and how it impacts on the requirements for sub-grid scale modeling is specifically addressed with strategies to perform well-resolved and multi-physics simulations of industrial systems. Finally, it is demonstrated how low-order stochastic models for process control, may be derived using results from LES. Contact: [email protected] PL2: Tuesday April 21st (8h30 – 9h30) The hydrodynamic theory of premixed flames: Laminar to turbulent propagation Moshe Matalon Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, IL. USA. Significant advances in understanding the dynamics of premixed flames under laminar and turbulent conditions were recently achieved using an asymptotic model that exploits the disparity between the length scales associated with the fluid dynamic field, the diffusion processes and the highly temperature-sensitive reaction rates. In this hydrodynamic model the flame, represented as a surface separating burned from unburned gas, propagates relative the fresh mixture at a speed that depends on the local stretch rate modulated by a Markstein length that mimics the influences of diffusion and chemical reaction occurring inside the flame zone. The propagation is therefore affected by the local mixture composition and flow conditions and the flow field is modified, in turn, by the gas expansion resulting from the increase in temperature caused by the heat released during combustion. The hydrodynamic model has been a valuable framework for understanding flame-flow interactions under laminar conditions; the onset of flame instabilities and the nonlinear development of flames subjected to the Darrieus-Landau instability; and, more recently, the structure and propagation speed of turbulent flames. The present talk will focus on recent results within this context. Because of the omnipresence of the Darrieus-Landau instability stemming from the gas expansion, laminar flames are seldom flat. In sufficiently large domains, small cells developing on the surface of a planar flame tend to grow and merge into larger cells that develop eventually into a single peak conformation with a pointed crest intruding into the burned gas and wide rounded troughs towards the fresh mixture. These structures are stable and propagate steadily at a constant speed significantly larger than the laminar flame speed. The bifurcation properties of planar flames, the nonlinear development beyond the instability threshold and the structure and propagation speed of the stable cusped-like flames have been well-established within the context of the hydrodynamic theory and validated by DNS. Turbulent flames have similar characteristics. Flame brushes that are statistically planar, i.e., of zero mean curvature, can be only observed when the Markstein number (scaled by the lateral size of the domain) exceeds a critical value. For Markstein numbers below criticality, the turbulent flame brushes are much thicker, exhibit frequent sharp intrusions into the burned gas reminiscent of the Darrieus-Landau instability, and remain partially resilient to the turbulence, at least at low intensities. The influence of the Darrieus-Landau instability progressively decreases as the turbulence level increases with the flame 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. Invited plenary lectures 19 becoming increasingly controlled by the turbulence. Of greatest importance is the determination of the turbulent flame speed and its dependence on turbulence intensity, which is shown to transition from a quadratic law at low intensities to a sub-linear scaling at higher turbulence levels, in agreement with the experimental record. We show that the increase in speed with increasing turbulence intensity is primarily due to the increase in the flame surface area as envisioned by the pioneering work of Damköhler, while the leveling in the rate of increase of the turbulent flame speed with turbulence intensity (the so-called bending effect) is partially due to frequent flame folding and detachment of pockets of unburned gas from the main flame surface area, and partially due to flame stretching (for positive Markstein length). The proposed scaling laws for the turbulent flame speed deduced from the simulations are free of any turbulence modeling assumptions and/or adjustable parameters, and exhibit an explicit dependence on physically measurable quantities. These include flow characteristics, such as turbulence intensity and integral scale and overall stretch rate (accounting for both flame front curvature and hydrodynamic strain), and combustion characteristics, such as the effective Markstein length controlling the fuel type, mixture composition and system pressure, and the thermal expansion coefficient determined by the ambient state conditions and the total heat released. Contact: [email protected] PL3: Wednesday April 22nd (8h30 – 9h30) Modeling and simulation of real-fluid thermodynamics and transport in advanced combustion systems Joseph C. Oefelein Sandia National Laboratories, Combustion Research Facility, Livermore, CA, USA. Real-fluid thermodynamics and transport are prevalent in a wide variety of advanced combustion systems such as liquid rockets, Diesel engines, and gas turbines. For example, imaging has long shown that under some high-pressure conditions, the presence of discrete two-phase flow processes becomes diminished. Under such conditions, liquid injection processes transition from classical atomization and spray dynamics, to densefluid diffusion dominated mixing with no drops present. When and how this transition occurs, however, is not completely understood. When it does, the classical view of liquid atomization and spray dynamics as an appropriate model is questionable. Instead, non-ideal, real-fluid behavior must be taken into account using a multicomponent formulation that applies to mixtures at high-pressure, supercritical conditions. This presentation will focus on the progression of issues related to these high-pressure phenomena and treatment of the extreme real-fluid, multicomponent, gas-liquid processes that occur as a consequence. First, a summary of recent progress in understanding the processes responsible for the transition from spray dynamics to densefluid mixing will be presented to give perspective on the modeling challenges. The theoretical and numerical treatment of these flows will then be analyzed within the formalism of Large Eddy Simulation (LES). Using the filtered conservation equations as a foundation, closure issues using a model framework based on a generalized treatment of the equation of state, thermodynamics, and transport processes will be described. The related numerical issues in treating the extremely large gradients in thermo-physical properties that arise will then be addressed. Throughout the presentation, representative results will be used to illustrate various aspects of the problem, with emphasis on recent studies that focus on experiments of liquid hydrocarbon injection processes being conducted in the high-pressure combustion vessel at Sandia National Laboratories. In the latter, LES calculations reveal the structural characteristics of the inherent turbulent scalar mixing processes at supercritical conditions directly relevant to advanced Diesel engines. The presentation will conclude with a summary of key observations and suggestions for future research. Contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 20 MS1: Large eddy simulation: challenges and opportunities MINI-SYMPOSIA MS1: Large eddy simulation: challenges and opportunities V. Raman1 and M. Mueller2 1 2 University of Michigan, USA. Princeton University, USA. Large Eddy Simulation (LES) has become a vital part of combustion science, and has been the focus of computational modeling for the past two decades. While the growth of computing aided its rise, significant theoretical and computational challenges remain unaddressed. The goal of this mini-symposium is to invite discussion on these challenges and future opportunities. Two sessions are planned. The first will have three senior researchers discuss their perspectives on LES modeling. The second will have promising young researchers provide their vision for the future. MS1 (1/2) Monday April 20th (10h00 – 12h00) Theoretical, numerical, and modeling issues in LES V. Raman University of Michigan, USA. Large eddy simulation (LES) has become the preeminent tool for simulating complex reacting flows. Yet, several fundamental theoretical and numerical challenges remain unresolved. In this talk, a summary of the issues and emerging consensus on LES of turbulent combustion is discussed. In particular, the limitations of LES in terms of the range of applications will be presented. This talk will set the stage for the two sessions that provide insight from leading researchers in this field. contact: [email protected] Using LES of ignition, quenching and instabilities to design future gas turbines T. Poinsot1, B. Cuenot2, L. Gicquel2, G. Staffelbach2, A. Dauptain2 and F. Duchaine2 1 2 IMF Toulouse, CNRS, France. CERFACS, France. LES has replaced RANS in most laboratories and the same evolution takes place now for companies. This requires significant evolutions of LES codes to adapt to the specificities of real combustion chambers: multiphysics becomes an issue and LES codes have to be coupled now to heat transfer codes to predict wall temperatures, to soot and radiation models but also to more precise inlet and outlet conditions to provide realistic boundary conditions at inflow and outflow. The interfaces between users and softwares also become a difficulty. Even if parallel machines allow to perform LES in reasonable human times, the CPU cost of LES for companies is exploding, calling for drastic reductions of costs for future LES through code optimization but also mesh adaptation and smart interfaces to minimize CPU waste. This talk will discuss these issues in the specific case of gas turbine engines. contact: [email protected] Assessment of LES combustion models H. Pitsch and P. Trisjono Institute for Combustion Technology, RWTH Aachen University, Germany. Development and validation of combustion models for large-eddy simulations is not straightforward. Often, because of resolution requirements for the fluid flow, combustion models are not really tested for the typical laboratory flame validation cases. Furthermore, uncertainties in specifying boundary condition make validation difficult. On the other hand, DNS data can be constructed in a way that circumvents these difficulties, but because of restrictions in computational resources, such data sets are even more removed from practical combustion systems. In this talk, a joint validation approach using experiments and DNS is proposed. Systematic model development strategies will be discussed. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS1: Large eddy simulation: challenges and opportunities 21 MS1 (2/2) Monday April 20th (13h30 – 15h30) Data-driven modeling for LES K. Duraisamy Validation of multi-physics LES against sparse data M. Mueller Princeton University, USA. As multi-physics LES models continue to become more complex, validation of complete, integrated models becomes an increasingly significant challenge. The difficulty of the validation process is compounded by the fact that experimental data is rarely available in a single configuration to validate all components of a complex model. In this talk, a pair of case studies is discussed in utilizing novel error propagation and uncertainty quantification algorithms to enhance the model validation process for complex multi-physics models with sparse experimental data. contact: [email protected] University of Michigan, Ann Arbor, USA. Opportunities for data-driven techniques to improve predictive modeling of near-wall turbulence will be discussed. High Reynolds number large eddy simulations of turbulence require near-wall modeling, a situation that is unlikely to change — at least not over the next few decades. It is emphasized that neither DNS nor experimental data exists in a form that is directly useful to improve predictive models. Adjoint-driven inverse problems are used to transform data into information that is relevant to - and in the context of the model. This presentation will address the issue of how to identify and formulate properly-posed datadriven problems and how machine learning techniques can be effectively used to transform inferred information into modeling knowledge. contact: [email protected] The challenge of pollutant emission predictions in realistic burners Design optimization with LES V. Moureau and G. Lartigue CORIA, CNRS UMR6614, Normandie Université, France. The aeronautical industry faces the challenge of designing novel engines with better specific fuel consumption while decreasing pollutant emissions such as nitrogen oxide (NOx) and carbon monoxide (CO). These pollutants have chemical time scales that are different from those of the flame advancement and require dedicatedlarge-eddy simulation models. This presentation will focus on the prediction of NOx and CO emissions in complex burners using tabulated and finite-rate chemistry. This latter modeling approach is very promising despite its high CPU cost but novel strategies enable to benefit from many-cores supercomputers. Q. Wang MIT, USA. How can large eddy simulation be used in the engineering design process? How can they leverage recent progress in extreme scale computing? We identify key challenges specific to the engineering design process, and some of the critical outstanding problems and promising research directions. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 22 MS2: Combustion noise MS2: Combustion noise Monday April 20th (10h00 – 12h00) F. Zhang1, H. Bockhorn1, C.O. Paschereit2, W. Schröder3 and J. Janicka4 1 Karlsruhe Institute of Technology, Germany. TU Berlin, Germany. 3 RWTH Aachen, Germany. 4 TU Darmstadt, Germany 2 Unsteady combustion is regarded as the major source of noise generated by industrial combustion devices like gas turbines. Therewith, a freely burning flame constitutes an acoustic monopole source due to the volumetric expansion of the burning gases, which is related to the direct combustion noise. In case of confined systems, the combustion noise may be augmented by indirect sources that arise during acceleration of unevenly heated gases through the system exhaust. The indirect combustion noise is manifested as an acoustic dipole source involving interactions between temperature inhomogenities and mean flow variation. This mini-symposium presents experiments and different numerical studies using LES/DNS/CAA methods to analyse both direct and indirect combustion noise generated by turbulent premixed flames. Direct combustion noise of premixed flames: experiments and simulation using compressible LES and DNS F. Zhang1, H. Nawroth2, P. Habisreuther1, J. Moeck2, C. Paschereit2 and H. Bockhorn1 1 2 KIT Karlsruhe, Germany. TU Berlin, Germany. The direct combustion noise generated by an unconfined, premixed flame operated by a generic burner has been experimentally and numerically investigated. Comparison of the sound pressure levels (SPL) obtained from the microphone measurements and the direct numerical simulation (DNS) shows a very good agreement, whereas it is slightly underestimated by the large eddy simulation (LES). With help of the DNS, the strong correlation between the OH* concentration and the heat release rate have been proven. In addition, the joint probability density function (PDF) of the heat release rate and OH* mass fraction has been derived to give quantitative findings of both parameters. contact: [email protected] A hybrid approach to predict combustion noise of premixed flames A. Hosseinzadeh1, G. Geiser2, J. Janicka1 and W. Schröder2 1 2 TU Darmstadt, Germany. RWTH Aachen, Germany. In the current work a hybrid approach (LES/CAA) is proposed to analyze the combustion noise of premixed flames. The low-Mach combustion LES uses the artificially thickened flame approach coupled with FGM tabulated chemistry. The CAA uses acoustic perturbation equations (APE). Diverse configurations of a generic premixed burner are numerically investigated using the proposed hybrid approach. The predicted sound pressure levels (SPL) using the hybrid approach and the measurements are in a very good agreement. Besides direct combustion noise mechanisms induced by heat release fluctuations, indirect mechanisms by acceleration of entropy inhomogeneities and nonisentropic mixing processes are also found to be the major noise sources. contact: [email protected] LES based simulations of combustion noise in a helicopter engine T. Livebardon1, L. Gicquel1, T. Poinsot2 and E. Bouty3 1 CERFACS, France. Institut de Mécanique des Fluides de Toulouse, France. 3 Turbomeca S.A., France. 2 Reducing helicopter noise disturbances has become a major issue because of the growth of air traffic at the vicinity of populated areas and ACARE 2020 objectives. At low and medium frequencies, helicopter engines are known to be major contributors in the radiated noise, especially during take-off. Indeed, an emitted broadband noise at the exhaust, called corenoise, is composed of turbomachinery and combustion noise in the absence of jet noise. In this scope, a method to predict combustion noise in real aero-engines using large eddy simulations of the combustion chamber coupled with an analytical approach to model the acoustic transmission of acoustic and entropy noise through the turbine stages is described. The proposed strategy is tested by comparing predictions of the computed noise with experimental results obtained for a full helicopter engine with pressure sensors located in the chamber and in all turbine stages. The waves leaving the combustion chamber are extracted from these simulations at the outlet of the chamber, and an analytical method based on actuator disk theory gives 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS2: Combustion Noise 23 noise levels at the various turbine stages using the waves amplitudes at the chamber outlet. LES of the combustion chamber for two representative operating points are achieved and discussed. contact: [email protected] Estimating indirect combustion noise in nozzle flows with a 2D analytical model J. Zheng1, M. Huet1, A. Giauque2, F. Cléro1 and S. Ducruix3 1 2 3 ONERA, France. Laboratoire de Mécanique des Fluides et d’Acoustique, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. combustion noise use quasi-1D flow assumptions. A new 2D model estimating indirect combustion noise generated by the acceleration of non-uniform temperature regions through a subcritical nozzle is presented. This model extends the 1D models proposed by Giauque et al. (JEGTP, 2012) and Duran and Moreau (JFM, 2013) by taking into account the deformation of the fluctuating entropy fronts. An original approach is proposed that treats the acoustic fluctuations as 1D along the nozzle while entropy fluctuations are modeled as 2D with longitudinal and radial dependencies. contact: [email protected] To achieve reduced computational times, most of the semi-analytical methods for the estimation of MS3: Advances in linear-eddy and one-dimensional-turbulence modeling Monday April 20th (10h00 – 12h00) H. Schmidt BTU Cottbus-Senftenberg, Germany. DNS is currently the numerical tool to investigate fundamental problems in combustion, since it solves the governing physical conservation equations without assumptions. Unfortunately due to the range of spatial and temporal scales emerging in reactive flows currently it is limited to mostly fundamental research. Interesting alternatives are stochastic approaches like the linear-eddy model (LEM) and the one-dimensional-turbulence (ODT) model and multi-dimensional Eulerian and Lagrangian approaches that incorporate them. LEM and ODT resolve molecular effects (as DNS) on small scales, but use a one-dimensional map-based stochastic process compatible with conservation laws to reduce the degrees of freedom. The models have been successfully applied to many flows and are under continuous development and extension. During the minisymposium we present (i) counterflow flames results, including comparison to DNS, (ii) a new LEM-based regime- independent modeling concept, (iii) ODT results on the auto-ignition of ethylene/air flames, and (iv) a 3D extension of LEM. The representative interactive linear-eddy model (RILEM) for regime-independent turbulent nonpremixed combustion modeling On the benefits of using LEM- and ODT-based approaches in numerical combustion H. Schmidt T. Lackmann1, A. Kerstein2 and M. Oevermann1 BTU Cottbus-Senftenberg, Germany. 1 We summarize the progress in analyzing, applying, and improving stochastic turbulence models in reactive fluid mechanics that are based on either the linear-eddy model (LEM) or on one-dimensional turbulence (ODT). Compared to direct numerical simulation (DNS), these models span a wider range of scales. Compared to Reynolds averaged Navier Stokes (RANS) and large eddy simulation (LES), (i) the molecular effects are retained and (ii) no assumption of scale separation is made. After a general overview we will show and discuss recent results on counterflow flames and compare them to DNS. contact: [email protected] 2 Chalmers University of Technology, Sweden. Consultant, USA. Future engines with efficient and environmentally friendly combustion of fuel are likely to operate under low temperature combustion (LTC) conditions. However, most combustion models used in todays CFD codes are based on the assumptions of a specific combustion mode and regime. Their predictive capabilities might be limited under LTC. Here we present a new regime independent modeling concept for non-premixed turbulent combustion based on the lineareddy model. The model in its current development state features a single linear-eddy line representative of the combustion process. Local coupling with the flow is achieved via a presumed PDF for the mixture fraction. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 24 MS3: Advances in linear-‐eddy and one-‐dimensional-‐turbulence modeling Comparison between ODT and DNS for ethylene/air auto-ignition Introduction of unsteadiness and small-scale resolution into steady-state reacting-flow solvers using LEM3D A. Abdelsamie and D. Thévenin S. Sannan1 and A. Kerstein2 University of Magdeburg, Germany. 1 The auto-ignition of a turbulent ethylene/air mixture is a very complex physical process. Resolving all turbulence and flame scales in a DNS leads to extremely expensive computations. Therefore, ODT would be - if suitable - an excellent alternative. Qualitative and quantitative comparisons between ODT and DNS are conducted concerning ethylene/air autoignition in this work. First result shows that ODT is able to predict correctly auto-ignition or misfire of an atmospheric, stoichiometric mixture in a temperature range of [1000-1400] K. If this observation can be generalized, the probability of auto-ignition could be efficiently deduced from ODT simulations. contact: [email protected] 2 SINTEF Energy Research, Norway. Consultant, USA. A 3D subgrid closure approach, LEM3D, has been developed. The approach is based on the Linear Eddy Model (LEM), and structurally involves three orthogonally intersecting arrays of LEM domains that are coupled so as to capture the 3D character of fluid trajectories. LEM3D thus provides small-scale resolution in all three spatial directions of the flow field. The conceptual basis for LEM3D is presented, together with model performance evaluation for multi-stream mixing and differential diffusion in turbulent round jets. Progress toward implementation of two-way coupling between LEM3D and RANS for combustion applications is described. contact: [email protected] MS4: Uncertainty quantification in computational combustion H. Nahjm1, M. Frenklach2 and A. Tomlin3 1 2 3 Sandia National Laboratories, USA. University of California at Berkeley, USA. University of Leeds, UK. Predictive combustion computations rely on calibrated and validated models. These models rely on parameters that are estimated either from experimental measurements or from theory/computations, and are, to some extent, uncertain. Accordingly, it is important that computational predictions based on these models explore the impact of input uncertainties, and provide reliable quantification of uncertainties in their predictions. This minisymposium focuses on challenges and recent advances in uncertainty quantification (UQ) in chemical models and combustion computations. It brings together experts from a wide range of backgrounds working in computational combustion and interested in UQ, parameter estimation, and model assessment. MS4 (1/4) th Monday April 20 (13h30 – 15h30) Combining theory and experiment to reduce uncertainties in the derivation of phenomenological rate coefficients for combustion systems A. Tomlin, R. Shannon, P. Seakins, M. Pilling, M. Blitz and S. Robertson University of Leeds, UK. Statistical rate theory calculations, in particular formulations of the chemical master equation (ME), are widely used in order to calculate rate coefficients that are used within kinetic mechanisms describing the combustion of many different fuels. Such calculations can be applied over wide ranges of temperatures and pressures and hence in this sense are more general than individual experimental kinetic studies. However despite the increasing accuracy of ab initio data, small uncertainties in the input parameters for ME calculations can lead to relatively large uncertainties in the calculated rate coefficients. ME input parameters may be constrained further using experimental data over more limited temperature and pressure conditions. The relationship between experiment and theory will be discussed in this paper, and in particular, whether experiments performed at lower temperatures and pressures can help to constrain reaction rates at combustion relevant conditions. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS4: Uncertainty quantification in computational combustion Uncertainties in theoretically predicted elementary rate coefficients S. Klippenstein Argonne National Laboratory, USA. A survey will be provided of the uncertainty in various components of ab initio transition state theory master equation based theoretical predictions for thermal rate constants. Uncertainties in the electronic energies, the vibrational frequencies, anharmonicity treatments, tunneling estimates, and energy transfer parameters will each be discussed at length. The role of variational optimizations will also be discussed. A number of sample studies covering a range of reaction types will be used as illustrations. contact: [email protected] previous estimates. C2H3 + O2 will be predominantly chain branching above 1700 K. Uncertainty analysis is presented for the two most important transition states. The uncertainties in these two barrier heights result in a significant uncertainty in the temperature at which CH2CHO + O overtakes all other product channels. Compared to previous mechanisms, the present work is expected to accelerate fuel consumption in combustion models. The impact of these mechanistic changes on combustion properties will be presented. contact: [email protected] Correlated uncertainty quantification: application to complex chemical kinetics mechanisms J. Sutton1, W. Guo1, M. Katsoulakis2 and D. Vlachos1 1 2 Probabilistic Inference of reaction rate parameters based on summary statistics M. Khalil and H. Najm Sandia National Laboratories, USA. We present the results of an application of Maximum Entropy and Approximate Bayesian Computation methods to the estimation of the joint probability density on the Arrhenius parameters of the H + O2 => OH + O reaction. Available published results in the form of summary statistics, being nominal values and error bars of the rate coefficient at a number of temperatures, are used to generate consistent OH concentration profiles, and associated joint posterior densities on the Arrhenius parameters, using a nested Markov Chain Monte Carlo procedure. A consensus joint posterior on the parameters is obtained by pooling the aforementioned densities. University of Delaware, USA. University of Massachusetts, USA. We demonstrate that surface kinetics and in general reaction mechanisms have certain correlations. To address this problem, we formulate a correlated parameter sensitivity analysis method and apply to catalytic kinetics. We then perform a number of density functional theory calculations (DFT) and estimate the posterior by using Bayesian statistics and by injecting the DFT data into a prior. We show that uncertainty is significant for surface kinetics and can rationalize most but not all of the deviations from experimental data. contact: [email protected] MS4 (2/4) Monday April 20th (16h00 – 18h00) Comparison of probabilistic and deterministic frameworks of uncertainty quantification contact: [email protected] Uncertainty quantification in the oxidation of vinyl radical 25 M. Frenklach1, A. Packard1, J. Sacks2, R. Pablo3 and G. Garcia-Donato4 1 University of California at Berkeley, USA. National Institute of Statistical Sciences, USA. ISEG Technical University of Lisbon, Portugal. 4 Universidad de Castilla-La Mancha, Spain. 2 F. Goldsmith Brown University, USA. State-of-the-art calculations of the C2H3O2 potential energy surface are presented. A new method is described for computing the interaction potential for R + O2 reactions. The method, which combines accurate determination of the potential on the quartet surface with multi-reference calculations of the doublet/quartet splitting, decreases the uncertainty in the potential and thence the rate constants by more than a factor of two. The temperature and pressure dependent rate coefficients are computed using variable reaction coordinate transition state theory, variational transition state theory, and conventional transition state theory, as implemented in a new RRKM/ME code. The product distributions are considerably more complex than 3 Numerical modeling of physical phenomena must accommodate sources of uncertainty rooted in the formulation of underlying physical models, their mathematical realization and numerical implementation, accounting for uncertain model parameters and experimental calibration data. How to use the experimental data and the numerical realization to enable prediction of new settings and estimate unknowns has had wide attention in recent years. Even when a physical model and its numerical realization is an accurate representation of the phenomenon, methods and capacity to utilize experimental data for calibration and prediction is not yet standardized. Such a setting is found, for instance, in combustion chemistry where the physical model is a complex network of many chemical 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 26 MS4: Uncertainty quantification in computational combustion species related through hundreds and thousands of chemical reactions involving many uncertain model parameters. Moreover, the limited experimental data are heterogeneous with sketchy assessments of measurement error. Here we compare a deterministic approach to the problem, namely the Bound-to-Bound Data Collaboration, with a Bayesian approach that embeds the analysis into a probabilistic framework. The comparison is performed using methane combustion as an illustrative example. Our findings show that, in general, the predictions obtained match but, in terms of estimation of unknown parameters, subtle differences arise that add insights about limitations of each of the methodologies. contact: [email protected] data, consisting of ignition measurements in shock tubes and rapid compression machines, and flame velocity measurements covering wide ranges of temperature, pressure, equivalence ratio and H2/CO ratio. Arrhenius parameters A, n, and E of 18 elementary reaction steps were optimized, and also 9 third body collision efficiency coefficients of four pressure dependent reactions. The joint covariance matrix of the optimized parameters was calculated, which characterizes the temperature dependent uncertainty of the optimized reactions rate coefficients and also their correlation. The obtained posterior uncertainties were compared with the prior uncertainties, determined on the basis of literature sources for the important rate parameters. contact: [email protected] Combining theoretical and experimental data in uncertainty quantification across multiple scales M. Burke Columbia University, USA. Recent developments in and results from a multi-scale informatics approach to kinetic model uncertainty quantification will be discussed. The approach simultaneously integrates information from a wide variety of sources and scales: ab initio electronic structure calculations of molecular properties, rate measurements of isolated reactions, and global measurements of multi-reaction systems. The resulting model representation consists of a set of theoretical kinetics parameters (with constrained uncertainties) that are related through elementary kinetics models to rate constants (with propagated uncertainties) that in turn are related through physical models to global behavior (with propagated uncertainties). Comparisons of multi-scale informed model predictions and recent data will be shown that reveal a high level of predictive accuracy for both small and large scales (molecular properties and global observables). Additionally, the utility of the approach for design of both experiments and theoretical calculations, in a manner accounting for existing theoretical and experimental data as well as accounting for relevant parametric and structural uncertainties in interpreting potential new data, will be demonstrated. contact: [email protected] Uncertainty quantification of the rate parameters of the syngas combustion system T. Varga1, C. Olm1, I. Zsély1, T. Nagy2, É. Valkó1, R. Pálvölgyi1, H. Curran3 and T. Turányi1 1 Institute of Chemistry, Eötvös University (ELTE), Hungary. MTA Research Centre for Natural Sciences, Hungary. 3 Combustion Chemistry Centre, NUI Galway, Ireland. 2 Statistical calibration of simplified chemical mechanisms for Diesel engine combustion L. Hakim, G. Lacaze, M. Khalil, H. Najm and J. Oefelein Sandia National Laboratories, USA. In this work, a reduced chemical mechanism for large eddy simulation (LES) is proposed to predict autoignition and flame propagation in Diesel engines, two features that directly influence efficiency, reliability and emissions. The mechanism is calibrated using Bayesian inference, providing optimal parameter estimates and respective uncertainties. The latter can then be propagated through the LES to quantify the uncertainties in the simulations. As an example, the technique is applied to the Engine Combustion Network (www.sandia.gov/ECN) Spray-A case where a twostage auto-ignition of n-dodecane in air has been observed. contact: [email protected] Bayesian inference of chemical kinetic models from proposed reactions N. Galagali and Y. Marzouk Massachusetts Institute of Technology, USA. We present a new framework for tractable Bayesian inference of chemical kinetic models in case of a large number of model hypotheses generated from a set of proposed reactions. The approach involves imposing point-mass mixture priors over rate constants and exploring the resulting posterior distribution using an adaptive Markov chain Monte Carlo method. We show that further gains in sampling efficiency can be realized by analyzing the chemical network structure in order to reduce the space of MCMC exploration. contact: [email protected] An optimized syngas combustion mechanism has been developed using a large set of indirect experimental 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS4: Uncertainty quantification in computational combustion Region of influence-based sensitivity analysis in turbulent combustion V. Carey and R. Moser University of Texas at Austin, USA. Adjoint-based sensitivity analysis is a powerful tool for UQ in many-parameter systems. Time-dependent turbulent systems present several obstacles for this methodology, including solution storage, analytic jacobians, and adjoint blowup over long time. We address these challenges when applied to turbulent combustion with a combination of checkpointing, automatic differentiation software, and space-time domain reduction to a "Region of Influence" for combustion-relevant quantities of interest. contact: [email protected] MS4 (3/4) Tuesday April 21st (13h30 – 15h30) Handling model error in the calibration of physical models H. Najm1, K. Sargsyan1 and R. Ghanem2 1 2 Sandia National Laboratories, USA. University of Southern California, USA. We present a statistical model calibration strategy that focuses on model error, and is tailored for physical models. We require that the mean prediction is centred on the data, and that the predictive uncertainty is consistent with the range of discrepancy between the means of the prediction and the data. We embed statistical model error terms in select model components. We then formulate the problem as a density estimation problem and solve it using approximate Bayesian computation methods. We demonstrate this construction for calibration of a global chemical model for methane-air against a detailed kinetic mechanism in an ignition problem. contact: [email protected] Facilitating uncertainty quantification for large, detailed reaction mechanisms R. West1, S. Goldsborough2 and A. Fridlyand1 1 2 Northeastern University, USA. Argonne National Laboratory, USA. Large, detailed reaction mechanisms describe the ignition and combustion behavior of real, transportation-relevant fuels like gasoline and diesel, represented by surrogate blends, e.g., n-heptane/isooctane/toluene. We describe an approach to quantify the uncertainty for such mechanisms at realistic combustion conditions, where global sensitivity analysis can be employed. Realistic uncertainty bounds are applied for 27 each reaction, with the base chemistry, i.e., C0-C3, treated in detail and fuel-specific reactions treated by reaction class or family. Automated means are developed to generate the UF file for an existing mechanism, by first identifying the reacting species, then categorizing the reactions. contact: [email protected] Combustion properties of H2/CO mixtures: consistent chemical mechanism from collaborative data processing N. Slavinskaia, U. Riedel and J. Starcke DLR, Germany. The revision of the state of the H2/CO DLR kinetic mechanism and related experimental data has been performed. This revision was performed on the base of numerical algorithms and tools developed in the DataCollaboration module of the automated datacentric infrastructure, Process Informatics Model (PrIMe). The uncertainty-quantification framework of Data Collaboration can establish consistency or inconsistency of a data-and-model system, when the kinetic parameters of a reaction mechanism and experimental observations used for model validation are known within its uncertainties. On this way for the studied H2/CO DLR kinetic mechanism, the influence of parameter and experiment uncertainties on the optimal solution have been examined, and experimental targets that are most difficult to match as well as model parameter values that are likely to be questionable have been identified. The 100 selected experimental observations (targets) for ignition delays and laminar flame speeds of H2/CO systems with boundary uncertainties for measured observations followed from references were included in the Data Set of DataCollaboration module. Performed analysis enhanced the quality of experimental data adopted for a model parameter optimization over the feasible region of the parameter space. Adding experimental uncertainties to parameter uncertainties the rate coefficients of the 12 most important reactions have been then optimized. The applied new approach to optimizing combustion models by constraining the optimization not only to parameter uncertainties but also to the uncertainties in experimental data assures the best-fit solution. The predictions, calculated with so obtained model, deviate the least from the experimental targets while remaining within their experimental uncertainties. The applied methods allow producing the best-fit solution predictive model with evaluated uncertainty level. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 28 MS4: Uncertainty quantification in computational combustion Effect of uncertainties in detailed chemistry on combustion model parameters N. Dumont, R. Vicquelin and O. Gicquel Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Parameters such as auto-ignition delay, laminar flame speed, quenching strain rate and so on are extensively used in turbulent combustion models. Their evaluations rely on numerical simulations using detailed chemistry mechanisms in canonical flame configurations: Homogeneous reactors, premixed and non-premixed flamelets. However, these model constants are uncertain because of the inherent uncertainty in the kinetic parameters of detailed mechanisms. In this study, the uncertainties in these quantities of interest are evaluated with a Monte-Carlo method due to the high number of uncertain parameters in detailed mechanism. This is done for several types of fuel (hydrogen, methane and heavier hydrocarbon fuels) in different flame configurations in a high-performance computing framework. Several operating conditions are investigated to identify the most sensitive conditions. contact: [email protected] Adjoint-based sensitivity for understanding chemistry modeling V. Raman1, K. Braman2 and T. Oliver2 1 2 University of Michigan, USA. University of Texas at Austin, USA. Chemistry models are calibrated mainly using zero or one dimensional flame experiments. In this work, adjoint-based sensitivity is used as the basis for evaluating multi-dimensional flames with the goal of calibrating and understanding model performance. Continuous adjoint equations are derived for low-Mach number flows, and applied to laminar and turbulent flame simulations. It is shown that adjoints provide a unique measure for determining the adequacy of chemistry models through a metric named field sensitivity contact: [email protected] MS4 (4/4) Tuesday April 21st (16h00 – 18h00) Needs and progress in assessing the accuracy of LES J. Oefelein and G. Lacaze Sandia National Laboratories, USA. complicated by the interdependence of different subfilter models, competition between model and numerical errors, model variability, and numerical implementation. This presentation will provide an overview of needs and progress to date in the development of predictive LES and application of UQ as a tool to validate its accuracy. contact: [email protected] Uncertainty quantification of rate rules for normal alkanes using Bayesian analysis L. Cai1, L. Kröger1, V. Raman2 and H. Pitsch1 1 2 RWTH Aachen University, Germany. Universitiy of Michigan, Ann Arbor, USA. The uncertainties of the rate rules for the mechanism of C5-C12 normal alkanes, expressed as probability densities, are estimated using the stochastic Bayesian approach. Sampling from these estimated probability densities, the uncertainties of the model prediction for global combustion targets can be determined. Joint probability densities of rate rules and model responses are evaluated to reveal the effect of various rate rules on fuel reaction pathways and on combustion targets. The results are also compared to those determined by the uncertainty quantification method based on the conventional optimization technique with polynomial chaos expansions. contact: [email protected] Optimal experimental design of furan shock tube kinetic experiments D. Kim, F. Bisetti, Q. Long, R. Tempone, A. Farooq and O. Knio KAUST, Saudi Arabia. A Bayesian optimal experimental design methodology has been developed and applied to refine the rate coefficients of elementary reactions in Furan combustion. We focus on the Arrhenius rates of Furan + OH = Furyl-2 + H2O, and Furan + OH = Furyl-3 + H2O, and rely on the OH consumption rate as experimental observable. A polynomial chaos surrogate is first constructed using an adaptive pseudospectral projection algorithm. The PC surrogate is then exploited in conjunction with a fast estimation of the expected information gain in order to determine the optimal design in the space of initial temperatures and OH concentrations. contact: [email protected] The Large Eddy Simulation (LES) technique is widely viewed as an engineering tool of the future that is conceptually capable of representing a wide variety of turbulent reacting flow processes. The potential benefits, however, are shadowed by the fact that LES is 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS4: Uncertainty quantification in computational combustion Industrial & applied V/UQ for a 15MW coal-fired boiler P. Smith and S. Smith The University of Utah, USA. Computational and experimental results from industrial combustion applications are expensive to obtain, sparse in number and high in uncertainty. However, there is high value in extracting as much information as possible from the combined set of experimental and computational data. This study targeted a validation with uncertainty quantification using LES simulations 29 of a 15MW ox-coal power boiler operated by Alstom Power and using 50 different measurements of the heat flux, temperature and O2 concentration as the specific quantities of interest as a function of wall thermal conductivity, fuel feed rate, and coal reactivity. Our validation approach identified the region of simultaneous consistency between experimental and simulation data, according to a consistency constraint, where the defect between the quantity of interest as predicted by the model and as measured by the experiment is bounded by the measurement uncertainty. contact: [email protected] MS5: Numerical methods for compressible combustion Monday April 20th (13h30 – 15h30) F. di Mare TU Darmstadt / DLR, Germany. The solution methods for the simulation of combustion system developed in the last decades have relied on the assumption that the reacting mixtures behaved ideally and that dynamic effects were largely negligible (low-Mach approximation). However, as the integrated modelling and simulation of propulsion systems in their entirety will become necessary for their global design and optimisation, novel, comprehensive solution methods accounting for complex thermodynamic effects must be developed. The contributions presented during this mini-symposium will describe advances in efficient solution approaches for fully-compressible solvers where non-ideal effects are taken in consideration as well as compressibility effects in complex combustion phenomena, such as flashback and thermoacoustic instabilities. An implicit, density-based algorithm for coupled simulations of arbitrary gas mixtures F. di Mare1 and L. di Mare2 1 2 TU-Darmstadt/DLR, Germany. Imperial College London, UK. The development of new design concepts for cleaner and efficient propulsion and energy transformation devices (gas turbines, combined gas-steam generation groups) requires the global optimization of all machine components. In the present work a novel numerical approach is illustrated, which allows to model accurately and efficiently departures from an ideal behaviour of the working fluid in density-based solvers. The concept of residual internal energy permits a ready extension of classic flux-difference splitting or flux-splitting solution algorithms to (reacting) mixtures of thermally and calorically imperfect gases of variable composition. A fully implicit formulation is then derived in terms of mass fractions. Simulation of high-pressure turbulent mixing and reacting real gas flows M. Pfitzner and H. Müller Bundeswehr University Munich, Germany. Many technically relevant combustion systems operate at elevated pressures, where the use of real gas equations of state might be important to accurately simulate the turbulent mixing and combustion phenomena. Examples are rocket combustors, gas turbine combustors and the combustion in diesel engines. In the presentation, we will show how to use pressure correction numerical schemes to simulate lowMach number flows without and with combustion. The method is implemented in OpenFOAM (RANS and LES). The resulting real gas CFD code using modern cubic equations of state is validated with high pressure mixing and combusting flows. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 30 MS5: Numerical methods for compressible combustion Effect of compressibility in the flashback of premixed flames in confined channels Compressible combustion and nonlinear dynamics for thermoacoustic analysis M. Hassanaly, H. Koo and V. Raman C. Lee and S. Cant The University of Michigan, USA. University of Cambridge, UK. Predictive models that could capture the onset and the propagation of a flashback event would be indispensable for high-hydrogen content gas turbines combustors. In this work, we use large eddy simulation (LES) for describing the boundary layer flashback. A low-Mach number assumption has been previously used to study the accuracy of LES chemistry models. However, Direct Numerical Simulations (DNS) have shown that the flame interface was also inducing a steep pressure gradient, whose impact was neglected with the low-Mach number assumption. The purpose of this work is to describe and quantify the role of compressibility on the flashback propagation. Acoustically-coupled combustion instabilities are difficult to predict and hard to eradicate in practical systems. Many of the existing design tools are based on assumptions of low Mach numbers and linear perturbations. In the present work, a more general approach has been developed using fully-compressible CFD and a set of nonlinear analysis methods. The approach is described, and results are presented for the classic Volvo afterburner test case, for which experimental data is available. Comparisons between the present results and the experimental data serve to validate the approach, and conclusions are drawn regarding the mechanisms behind the observed combustion instabilities. contact: [email protected] contact: [email protected] MS6: Towards exascale simulation of turbulent combustion J. Chen1 and R. Sankaran2 1 2 Sandia National Laboratories, USA. Oak Ridge National Laboratory, USA. Exascale computing will enable combustion simulations in parameter regimes relevant to next-generation combustors burning alternative fuels. High-fidelity simulations are needed to provide the underlying science base required to develop vastly more accurate predictive combustion models used ultimately to design fuel efficient, clean burning vehicles, planes, and power plants for electricity generation. However, making the transition to exascale poses a number of algorithmic, software and technological challenges. As Moore’s Law and Dennard scaling come to an end exascale computing will be achieved only through massive concurrency. Addressing issues of data movement, power consumption, memory capacity, interconnection bandwidth, programmability, and scaling are critical to its success. In this minisymposium numerical algorithms, programming models and domain specific languages needed to support high fidelity numerical simulations of turbulent combustion are presented. MS6 (1/2) Monday April 20th (16h00 – 18h00) Legion as a programming model for combustion simulation at the exascale J. Chen1, M. Bauer2, S. Treichler2, A. Aiken2 and A. Bhagatwala1 1 2 Sandia National Laboratories, USA. Stanford University, USA. Legion is a programming model designed for exascale machines with a number of features designed for heterogeneous, hierarchical machines that are expected at the exascale. Specifically, Legion supports hierarchical decomposition of both tasks and data to match the structure of the machine and provides a deferred execution model for hiding long and variable latencies inherent in large, distributed memory architectures. Legion is designed to provide a level of abstraction above the concrete hardware with an explicit mapping from the computation to a specific machine. Mapping is done dynamically, and so decisions about where tasks will run and where data will be placed can take advantage of runtime information. Legion promotes code portability as porting to a new machine involves changing only the mapping tuned for the new machine. Lessons learned from a recent port of a petascale DNS combustion code, S3D, to Legion on Titan at ORNL will be discussed. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS6: Towards exascale simulation of turbulent combustion Toward large-eddy simulation of complex burners with exascale super-computers: a few challenges and solutions V. Moureau 31 focus on accelerators, massively many core and highly vectorised architectures and programming models along with a few recent examples. contact: [email protected] CORIA, CNRS & INSA de Rouen, France. Large-Eddy Simulation (LES) is a powerful approach for the simulation of complex burners as it resolves the most energetic scales of turbulence and their interactions with the reactive fronts. However, LES is very CPU intensive as it requires fine meshes and long accumulation times. Future exascale machines present a number of challenges for combustion LES, which need to be addressed: i) efficient solving of the low-Mach number Navier-Stokes equations on a large number of cores, ii) load balancing of the stiff integration of the chemical source terms arising in operator splitting approaches, iii) data mining of billion cells LES. contact: [email protected] Combustion simulation on next-generation architectures J. Bell Lawrence Berkeley Laboratory, USA. Computer architectures are shifting to many-core, possibly heterogeneous nodes with reduced memory per core. Block-structure adaptive mesh refinement (AMR) is well-suited to this type of architecture. AMR reduces the number of degrees of freedom needed to represent the solution and provides a hierarchical model for computation ideally suited to many-core systems. However, developing AMR methodology for next generation architectures raises a number of challenges related to the increased importance of data movement and load balancing. In this talk, we discuss some of these issues in the context of both compressible and low Mach number formulations. contact: [email protected] Legacy codes and the jump to exascale : fluid dynamics simulation with AVBP and HPC G. Staffelbach1, O. Vermorel1, S. Jaure1, C. Fournier1, I. D'Ast1 and E. Oseret2 1 2 CERFACS, France. Exascale Computing Research, France. Combustion is one of the fields where HPC is the most needed. Leadership class systems enable large advances for research on more reliable and pollution compliant systems using unsteady combustion simulations. Adapting current programs to benefit from the everchanging improvements of architecture is the key to a sustainable effort. This talk will present recent developments for the AVBP code used for LES of combustion in gas turbines, engines and rockets. It will Direct numerical simulation of low Mach number combustion beyond petascale C. Frouzakis1, A. Tomboulides2, S. Kerkemeier1 and P. Fischer3 1 2 3 ETH Zurich, Switzerland. University of Western Macedonia, Greece. Argonne National Laboratory, USA. The extension of the open source incompressible flow solver Nek5000 for the direct numerical simulations of low Mach number combustion employs the spectral element method (SEM) for spatial discretization, where the solution is approximated by tensor product polynomials of order N (typically 8-12) on each of E elements. In combination with implicit or semi-implicit temporal discretization schemes, small dispersion errors can be efficiently obtained in complex geometries at a cost per gridpoint that is comparable to low-order methods. The analysis of the performance and scalability of the domain decomposition approaches indicate that the solver can perform well at exascale for a granularity of one element per (current-day) core. Recent applications and current development efforts will be presented. contact: [email protected] Physical and algorithmic approaches to numerical combustion at the exascale S. Cant Cambridge University, UK. Exascale computing offers great opportunities and also some potentially serious challenges. The fundamental physical scaling laws for turbulent combustion ensure that there is continuing demand for ever-greater computational resources. On the other hand, it is not obvious that currently-successful explicit algorithms will continue to scale well on hardware involving millions of compute cores, possibly equipped with numerical co-processors and with increasingly nonuniform access to distributed memory. The present work explores the options for progress in spatial discretisation methods and time stepping schemes, with emphasis on examples from turbulent premixed combustion, as well as two-phase flow and spray combustion. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 32 MS6: Towards exascale simulation of turbulent combustion MS6 (2/2) st Tuesday April 21 (10h00 – 12h00) Reacting flow simulation framework for heterogeneous and many-core architectures R. Sankaran1, B. Lee2 and H. Im2 1 2 ORNL, USA. KAUST, Saudi Arabia. With the advent of exascale computing, the computer architectures are expected to shift more heavily towards accelerators and many-core processors. These new architectures require legacy simulation codes to be rewritten and engineered to take full advantage of the hardware. This talk describes a new reacting flow simulation framework that is being developed to transition to heterogeneous architectures. The multiphysics models and kernels required for combustion simulations are expressed as function objects that are executed on the target processors using parallel frameworks such as Kokkos. Performance results on GPU and MIC accelerators are compared for representative benchmark problems. contact: [email protected] Efficient abstractions for exascale software design J. Sutherland University of Utah, USA. Exascale computing will require software abstractions that expose and exploit hierarchical parallelism while maintaining programmability. This is particularly challenging given the complexity of the physics models that we seek to deploy on exascale systems as well as uncertainty in what future computer architectures will be. This talk explores strategies for effectively dealing with both task and data parallelism in complex PDE solvers, with particular emphasis on turbulent combustion simulation. Specifically, directed acyclic graph approaches for exposing task-level parallelism, coupled with a domain-specific language that provides matlab-like syntax and supports deployment on multicore or GPU systems are discussed. Together, these abstractions increase programmer productivity while enabling more complexity in terms of physics and hardware targets. contact: [email protected] Using GPUs to accelerating nonstiff and stiff chemical kinetics in combustion simulations K. Niemeyer Oregon State University, USA. Combustion simulations with detailed chemical kinetics require the integration of a large number of ordinary differential equations (ODEs), with at least one ODE system per spatial location solved every time step. This task is well-suited to the massively parallel processing capabilities of graphics processing units (GPUs), where individual GPU threads concurrently integrate independent ODE systems for different spatial locations. However, the typical high-order implicit algorithms used in combustion modeling applications (e.g., VODE, LSODE) to handle stiffness involve complex logical flow that causes severe thread divergence when implemented on GPUs, thus limiting the performance. Alternate algorithms are therefore needed. This talk will discuss strategies and results using integration algorithms for nonstiff and stiff chemical kinetics on GPUs. contact: [email protected] GPU based chemistry acceleration for reacting flow simulations J. Lee and V. Sankaran United Technologies Research Center, USA. Advances in GPU and their programmability make them attractive for Computational Fluid Dynamics (CFD). Here we benchmarked a hybrid CPU-GPU paradigm by simulating an idealized, but fully resolved combustion problem. This problem retained the complexity of three dimensional turbulent reactive flow physics including detailed chemistry, while neglecting geometrical complexities. We achieved 2-5X overall speed-up compared to CPU-only simulations. Further details of the CFD problem, hybrid methodology, performance metrics definition and benchmarking results will be presented. This promising technology, if exploited properly, could quickly enable accurate predictions of finite rate chemistry effects, such as pollutant emissions from combustors. contact: [email protected] Accelerating combustion modeling with the Intel Xeon Phi C. Stone Computational Science & Engineering, USA. Promising speed-up of combustion simulations with GPUs, particularly in regards to kinetics, has been demonstrated over recent years. Here, we will examine the speed-up potential of the Intel Xeon Phi coprocessor for combustion applications and compare to GPU performance. Application benchmarks will be presented using the Xeon Phi in isolation as well as in tandem with the host CPU cores to demonstrate the scalability in the heterogeneous environment. A novel single-instruction multiple-data vectorization approach will be presented that improves both Xeon Phi and host CPU performance by a factor of between two to three times compared to traditional implementation. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS7: Diagnostics for autoignition process 33 MS7: Diagnostics for autoignition processes M. Valorani Sapienza University of Rome, Italy. The efficient design of combustion devices requires a detailed understanding on the dependence of basic autoignition characteristics (e.g., ignition delay) to various operative conditions, such as type of fuel fuel, initial mixture fraction, pressure and temperature, uncertain parameters, etc. In the past, researchers were resorting to analytic (mostly asymptotic) methodologies, so that it was possible to analyze only the cases were a simple reduced model was available. The development of alternative algorithmic methodologies has recently allowed the analysis of physical problems modeled by complex kinetic mechanisms. This mini-symposium focuses on recent advances in the development of algorithmic tools for the study of basic autoignition characteristic. It brings together experts from a wide range of backgrounds working in computational combustion. MS7 (1/2) Monday April 20th (16h00 – 18h00) How details of fuel molecular structure can accelerate or retard autoignition C. Westbrook This presentation will investigate the inherent uncertainties in predicting ignition delays for a variety of bio-fuels where many of the rate constants have been estimated rather than the subject of detailed kinetic studies. The constraints that ignition delay data may provide on key parameters within these schemes will be addressed. contact: [email protected] LLNL, USA. Autoignition of hydrocarbon and other fuels is responsible for many important combustion properties that affect fuel performance in practical systems and fuel safety, and fuel molecular structure can have important influences on autoignition rates. Kinetic modeling can show how the presence of different C-H bonds, C=C double bonds, the specific location of O atoms in the fuel, the length of carbon atom chains, the degree of branching in the fuel molecule, and other factors can either accelerate or retard autoignition in fuel molecules, and this presentation will discuss those many factors. contact: [email protected] Uncertainties in predicting the ignition properties of biofuels: what constraints can measurements provide? A. Tomlin and E. Agbro The developing explosive dynamics in the vicinity of the third explosion limit of H2/air E. Tingas1, T. Turanyi2 and D. Goussis3 1 National Technical University of Athens, Greece. Eötvös University (ELTE), Hungary. 3 University of Leeds, UK. 2 While the discussion on the first and second explosion limits of H2/air mixtures seems to have come to an end, the prevailing chemistry for the development of the third limit is still a subject of research. This topic is further investigated here by resorting to the CSP algorithmic tools, in order to identify the reactions and species that relate the most to the explosive modes, in the case of a homogeneous constant volume mixture. Two different sets of initial conditions are considered; one below the third limit and one above (T0=800 K; P0=1 atm and P0=5 atm, respectively). contact: [email protected] University of Leeds, UK. Correctly predicting the ignition properties of fuels and fuel mixtures is an essential component of chemical mechanisms if they are to be used within models of practical combustion devices. Ignition delay measurements from a variety of apparatus including low and high pressure shock tubes and rapid compression machines is therefore commonly used as a method of evaluating combustion mechanisms. For newly developed mechanisms (e.g. for alternative/biofuels), such data has a role to play in providing realistic parameterisations of key model inputs. Nevertheless, such models still contain a high level of uncertainty. Analytical prediction of syngas induction times P. Boivin1, A. Sanchez2 and F. Williams3 1 IUSTI, CNRS and Aix-Marseille Université, France. Universidad Carlos III de Madrid, Spain. 3 UCSD, USA. 2 Computations indicate that, under all possible conditions of practical interest, including temperatures both above and below the crossover temperature at which the rates of the H2 + O2 branching and termination steps are equal, eleven irreversible elementary steps suffice to provide accurate values of induction times in autoignition processes of fuels 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 34 MS7: Diagnostics for autoignition processes consisting of mixtures of H2, CO, and inerts. At high temperatures, this time is controlled by the time required for HO2 to reach a steady state, with heat release being negligible during that time, while below the crossover temperature, the time to reach this steady state becomes shorter than the induction time, and the heat release becomes non-negligible once HO2 has reached a steady state, resulting in the induction time progressively approaches that of a thermal explosion, which includes an effectively autocatalytic production of H2O2. Analytical approximations are introduced here that enable induction times to be calculated accurately under all conditions. contact: [email protected] Optimal dimension of the combustion mechanism in the ignition problem approximation (up to a suitable order) of the slow invariant manifold of n-dimensional slow fast autonomous dynamical systems. Then the Flow Curvature Method (FCM) will be applied for determining an approximation of the slow invariant manifold of the Van der Pol system in dimension 2, of the Chua’s models in dimension 3, 4 and 5. Then, in order to compare the FCM with the so-called Computational Singular Perturbation Method (CSPM), the slow invariant manifold of the “Davis–Skodje problem” in dimension 2, of “a 3-species kinetics problem” in dimension 3, of the Lorenz-Krishnamurthy model in dimension 5 and of a model of a network of coupled enzymatic reactions in dimension 6 will be also analytically computed. contact: [email protected] MS7 (2/2) V. Bykov1, V. Gol'dshtein2 and U. Maas1 1 2 Karlsruher Institut für Technologie (KIT), Germany. Ben-Gurion University of the Negev, Israel. In order to model and analyse ignition process a very detailed description of the chemical kinetics is required. Therefore, typical combustion mechanisms have become extremely large during the last few decades. The question of an optimal dimension (as a number of degrees of freedom required) to model and analyze the ignition process accurately enough has become very important but it is not answered yet in a general context. In the proposed talk a combination of local (ILDM based) and global (GQL based) approaches for mechanism analysis is suggested that answers this question. The methodology and results for its application are illustrated and benchmarked by the methane/air combustion system. contact: [email protected] Flow curvature method applied to dynamical systems analysis J. Ginoux Université de Toulon, France. The Flow Curvature Method is based on the idea that the trajectory curve, integral of any n-dimensional dynamical system may be considered as curve in Euclidean n-space having local metrics properties of curvatures. Thus, the location of the points where the curvature of the trajectory curve, integral of any ndimensional dynamical system, vanishes defines a manifold called: flow curvature manifold. This manifold (a curve in dimension, a surface in dimension 3, a hypersurface in higher dimension) enables to find again the main features of such n-dimensional dynamical system (fixed points and their stability, center manifolds, local bifurcation of codimension 1,…). More particularly, it has been established that the flow curvature manifold directly provides an Tuesday April 21st (16h00 – 18h00) The developing explosive dynamics during the autoignition of a DME/air mixture E. Tingas1, D. Kyritsis2 and D. Goussis1 1 National Technical University of Athens, Greece. Department of Mechanical Engineering, Khalifa University, Abu Dhabi, UAE 2 Bio-DME is currently being considered as a potential second-generation biofuel that can be produced from ligno-cellulosic biomass. Using a detailed chemical kinetics mechanism (1542 reactions and 253 species), the reactions and species related to the generated explosive modes are identified in the case of a homogeneous constant volume DME/air mixture, where phi=1, P0=50 atm and T0=1,100 K. The analysis is based on various CSP algorithmic tools. The role of the developing explosive modes is examined and conclusions are drawn regarding the chemical-thermal coupling that control the process. The influence of various additives (i.e., CH4, CH2O etc) on ignition delay is examined. contact: [email protected] Numerical investigation of ignition in laminar methane/LOx flamelet at supercritical pressures P. Lapenna, P. Ciottoli, F. Creta and M. Valorani University of Rome "La Sapienza" Dept. of Mechanical and Aerospace Engineering, Italy. High-pressure combustion devices, such as liquid rocket engines, are characterized by transcritical and supercritical conditions which influence ignition processes. In the present study, real fluid effects on both forced and auto-ignition phenomena are investigated in the framework of unsteady laminar flamelet equations, which provide a well established representation and diagnostic tool for non premixed combustion transient 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS7: Diagnostics for autoignition processes 35 phenomena. Real fluid properties are taken into account using a computationally efficient cubic equation of state, written in a comprehensive three-parameter fashion. High pressure conditions for unsteady flame structure calculations are representative of a methane/liquid-oxygen rocket engine operating conditions. hot spots leading to ignition front propagation, and is considered to be the main cause for the discrepancies in the prediction of the ignition delay times. The present study reports 1D and 2D parametric studies of computational simulations to represent the distinct ignition characteristics. Further scaling consideration leads to a diagram to map out parametric conditions in which different ignition regimes are identified. contact: [email protected] contact: [email protected] Impact of pilot injections on ignition behaviour at low in-cylinder temperatures from extreme miller valve timing in a heavy-duty Diesel engine S. Pandurangi, Y. Wright, C. Brückner, P. Kyrtatos and K. Boulouchos ETH Zürich, Aerothermochemistry and Combustion Systems Laboratory, Dept. of Mechanical Engineering Simulations of a heavy-duty single-cylinder research Diesel engine operating with pilot injections at extreme Miller valve timing leading to end-of-compression temperatures between 710–790 K have been performed with a 3D CFD solver, coupled with the conditional moment closure (CMC) combustion model. The influence of adding pilot injections is studied, concerning their impact on auto-ignition, heat release as well as soot and NOx formation. An extended CMC model, capable of incorporating an arbitrary number of injection events, is employed. An embedded twoequation soot model is included while NOx is calculated using the Zeldovich mechanism. Predicted apparent heat-release rates and their response to pilot injections are in very good agreement with experiments. Numerical results are subsequently employed to gain further insight into the impact of pilot injection on sootspecific processes such as formation and oxidation, as well as on NOx emissions. The auto-ignition behaviour is further elucidated by discussion of evolutions of conditional quantities as predicted by CMC. The findings suggest that the extended CMC model shows considerable promise for the simulation of Diesel engines operating with multiple injections under extreme-Miller conditions. contact: [email protected] Regimes of auto-ignition in homogeneous reactant mixtures with temperature fluctuations H. Im1, P. Pal2, M. Wooldridge2 and A. Mansfield2 1 2 King Abdullah University of Science and Technology, Saudi Arabia. University of Michigan, USA. “Mild” and “strong” ignition phenomena have been experimentally observed in shock tubes and rapid compression machines in which the mixture condition is expected to be nearly homogeneous. The “mild” ignition behavior is attributed to early auto-ignition of Auto-ignition diagnostics using the tangential stretching rate concept M. Valorani1 and S. Paolucci2 1 2 Sapienza University of Rome, Italy. University of Notre Dame, USA. We analyze ignition phenomena by resorting to a Tangential Stretching Rate (TSR) parameter which combines the concepts of stretching rate in dynamical systems with the decomposition of the local tangent space in eigenmodes. The main feature of TSR is its ability to identify unambiguously the most energetic scale at a given space location and time instant. The TSR of a state function can be readily computed during the post processing of reactive flow fields data. We verified the properties of the TSR with reference to different simple model problems and for the autoignition of hydrocarbon fuels. contact: [email protected] Investigation of the effect of correlated uncertain rate parameters on a model of hydrogen combustion using a generalized HDMR method É. Valkó1, A. Tomlin2, T. Varga1 and T. Turányi1 1 2 Institute of Chemistry, Eötvös University (ELTE), Hungary. Energy Research Institute, University of Leeds, UK. The High Dimensional Model Representation (HDMR) method has previously been applied to obtain global sensitivity indices of uncorrelated model parameters for several combustion systems. Here a generalized HDMR method was obtained by using the Rosenblatttransformation on a correlated model parameter sample to obtain a sample of independent normally distributed variables. The sensitivity indices were calculated for these transformed variables using the HDMR method. This generalized HDMR method was applied for the determination of sensitivity indices of a hydrogen combustion model, knowing the covariance matrix of rate parameters. The effect of the correlation of the rate parameters was investigated on the calculated uncertainty of ignition delay times of hydrogen combustion at various initial conditions. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 36 MS8: Modeling and computation of detonations in high-‐energy explosives MS8: Modeling and computation of detonations in high-energy explosives Tuesday April 21st (10h00 – 12h00) A. Kapila and D. Schwendeman Rensselaer Polytechnic Institute, USA. Detonations in condensed-phase, high-energy explosives present special challenges not only in experiment but also in modeling and computation. This mini-symposium gathers four presentations that focus on the modeling and computation of detonation initiation and propagation in practical explosives, addressing different but related aspects of the problem. The issues addressed include sensitivity of detonation initiation to initial porosity and initiating stimulus, hot-spot generation due to binary cavity collapse, role of confinement in transition to detonation, and development of an accurate numerical framework that applies shock-capturing to detonation propagation. Sensitivity of run-to-detonation distance in practical explosives Propagation of detonations in curved geometries of compliantly confined solid explosives J. Gambino, D. Schwendeman and A. Kapila E. Ioannou, L. Michael and N. Nikiforakis Rensselaer Polytechnic Institute, USA. University of Cambridge, UK. In condensed-phase explosives, run-to-detonation distance subsequent to the application of an initiating stimulus is an important measure of both performance and safety. This study, which continues our work on the evaluation of multi-phase DDT models, considers practical HMX-based granular explosives, models them as two-phase materials, employs realistic constitutive input including equations of state and chemical kinetics, and examines the sensitivity of the run-to-detonation distance to system parameters including the initial porosity and the initiating stimulus. Results are compared with recently-obtained experimental data. Challenges encountered in the computational process are also discussed. This study deals with the propagation of detonation waves in a solid explosive which is compliantly confined within a duct with straight and curved sections. The focus is on the process of transition to steady state after the detonation wave has entered the curved region. Previous studies have suggested that the transition is characterized by an exponential decay of the outer wall detonation velocity. We show that this is only part of the process, as local effects also play a role in the early stages and can be prominent or not, depending on the specific parameters of the curved geometry. contact: [email protected] contact: [email protected] Application of two-dimensional shock-fitting to detonation propagation in high explosives Coupled simulations of condensed-phase explosives and elastic-plastic solids T. Aslam1 and C. Romick2 1 L. Michael and N. Nikiforakis University of Cambridge, UK. This research is concerned with the simulation of combustion and transition to detonation of condensedphase explosives, which are confined by (or in general interacting with one or more) compliant inert materials. We couple the chemically active model for the explosive to an elastic-plastic model in the case of solid confiner and a separate fluid model for fluid inert materials. The ghost-fluid method is used to evolve the explosive/inert material interface and the coupling is achieved by means of an approximate mixed Riemann solver. Examples of such simulations are presented. contact: [email protected] 2 Los Alamos National Laboratory, USA. University of Notre Dame, USA. A highly accurate numerical shock-fitting scheme composed of fifth order spatial and third order temporal discretizations is applied to the two-dimensional reactive Euler equations. The steady detonation phase speed that results from this scheme was compared with the equivalent from shock-capturing. It was found that, unlike a shock-capturing scheme, the shock-fitting scheme allowed for the steady detonation phase speed to converge at higher than first order while the errors at the coarser solutions were significantly reduced in comparison with shock-capturing. Several examples are given, demonstrating efficiency gains of O(100010000) for shock-fitting versus shock-capturing. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS9: High-‐fidelity coupled simulation of reacting systems 37 MS9: High-fidelity coupled simulation of reacting systems F. Duchaine and L. Gicquel CERFACS, France. Multiphysics, multi-scale and multi-component simulations will play an increasing role for better understanding of reacting flows or in the design of new technologies. Indeed, although complex fluid motions are of prime importance for first order flow predictions, higher quality numerical predictions will require advanced Computational Fluid Dynamics (CFD) coupled to more physics. In many fields (combustion, rotating machines, aerodynamic, acoustic...) unsteady flow methods such as Large Eddy Simulation (LES) or Direct Numerical Simulation (DNS) give already very accurate results. The step forward is hence to couple these solvers with other physics to compose multiphysics, multi-component frameworks addressing coupled real flows. The symposium on " High fidelity coupled simulations of reacting systems " aims at bringing together international researchers and engineers involved in the development and/or use of coupled solutions. Specific but not limited topics include aerothermal, aeromechanics and aeroacoustic with interests in applications, models, HPC, methodologies, framework... MS9 (1/2) Tuesday April 21st (13h30 – 15h30) Large eddy simulation and the filtered probability density function method and detailed reaction mechanisms and experimental combustor data will be used for validation of the global and skeletal reaction mechanisms and the LES combustion model, respectively. contact: [email protected] Coupling approach to account for heat losses in a practical combustor W. Jones Imperial College London, UK. In large eddy simulation of combustion flows a method is needed to account for the influence of unresolved sub-grid-scale motion on chemical reaction rates. The evolution equation for the filtered fine grained or subgrid joint probability density function of the scalars quantities needed to describe reaction represents a basis on which such a model can be constructed. Because of the large dimensionality of the equations stochastic solution techniques are required. For this the Eulerian stochastic field solution method is adopted. The capabilities of the approach will be illustrated by comparisons of LES results with measurements in a range of configurations. contact: [email protected] Heterogeneous multi-scale LES using skeletal reaction mechanisms C. Fureby, N. Zettervall and K. Nordin-Bates The Swedish Defence Research Agency - FOI, Sweden. In the paper to be presented we plan to examine the use of skeletal reaction mechanisms in heterogeneous multiscale Large Eddy Simulations (LES) of turbulent combustion. Physical, chemical as well as numerical aspects will be discussed, and LES results, using global and skeletal reaction mechanisms, will be presented and discussed for an engineering application involving a premixed turbulent bluff-body stabilized flame. Comparison with laminar flame data from experiments D. Mira1, M. Zavala-Ake1, S. Gövert2, J. Kok2, M. Vazquez1 and G. Houzeaux1 1 2 BSC-CNS, Spain. University of Twente, The Netherlands. Heat transfer is a key issue to evaluate the overall performance and life duration of practical systems. The heat losses carry a reduction in efficiency that eventually might lead to an economical cost and potential failure. The understanding of the distribution of heat in thermal devices is not only beneficial to improve particular operating conditions, but also to enhance aerodynamic, thermal design, and selection of appropriate materials for a given application. The proposed work addresses the modelling of heat losses using a conjugate-heat transfer approach in the context of large-eddy simulation (LES). The emphasis is given to the analysis of thermal stresses on the solid domain with its corresponding boundary conditions for a given operating condition of the combustor. The exchange of information between fluid and solid domains is performed using a parallel coupler and the application case is a combustor subjected to strong heat losses. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 38 MS9: High-‐fidelity coupled simulation of reacting systems Analysis of large-eddy simulations of laboratory-scale fire M. Rochoux1, B. Cuenot2, F. Duchaine2, E. Riber2, D. Veynante3 and N. Darabiha3 Aerothermal prediction of an aeronautical combustion chamber based on the coupling of largeeddy simulation, solid conduction and radiation numerical solvers S. Berger1, F. Duchaine2, L. Gicquel2 and S. Richard3 1 Meteorological Research Division - Environment Canada, Canada. CERFACS, France. 3 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. 2 1 2 3 A large-eddy simulation (LES) approach, coupling combustion with radiation heat transfer and biomass fuel pyrolysis, is proposed to solve for the buoyant flame structure specific to natural fire propagation. These multi-physics coupled simulations of fire propagation are performed at laboratory scale and are compared to measurements to provide a comprehensive understanding of the mechanisms as well as of the characteristic time-/length-scales underlying fire propagation. This approach is found promising to examine the assumptions used to estimate the rate of spread as well as the emission factors, the modeling amount of species released in the atmosphere. contact: [email protected] MS9 (2/2) CERFACS / TURBOMECA, France CERFACS, France. TURBOMECA, France. Wall temperature evaluation is a multiphysics problem that can be solved numerically jointly through the use of several dedicated numerical and algorithmic approaches. In this paper, the partitioned coupling methodology used relies on a high fidelity Large Eddy Simulation (LES) reacting flow solver coupled to a conduction and radiation solvers. A efficient parallel coupling of the three solvers is obtained thanks to a code coupler that manages the parallel execution of the parallel codes as well as the exchange of data between each physics. Issues linked to the performance and the accuracy of such high-end coupled simulations are addressed and the effect of each physical component of the coupled simulation is discussed based on a real industrial burner application. contact: [email protected] st Tuesday April 21 (16h00 – 18h00) Automatic determination of the coupling period in unsteady conjugate heat transfer. Application to large-eddy simulation of flame-wall interaction 1 2 C. Koren , R. Vicquelin and O. Gicquel A 3D coupled approach for the thermal design of aero-engine combustor liners A. Andreini, B. Facchini and L. Mazzei Department of Industrial Engineering - University of Florence, Italy. 2 1 Air Liquide - Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. 2 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Combustors’ increase in size and power leads to a need of better estimation of wall heat fluxes. To reach this goal, coupled numerical simulations are carried out using state of the art numerical tools for each physical phenomenon. One of the main issues for such an approach is the determination of an accurate and efficient coupling period between the different solvers. We present an approach where this coupling period is automatically determined. For conjugate heat transfer, a hybrid cells layer that overlaps on both domains enables to derive an ordinary differential equation (ODE) for the wall temperature. Mathematical framework for time step adaptation in ODEs is then used to estimate the coupling frequency for a given error criterion. After being validated in one-dimensional tests, this approach is carried out in a 3D case of flame-wall interaction. The implementation of lean burn concept in modern aero-engine combustors involves a significant reduction in the available coolant amount. In order to limit the computational effort due to the presence of effusion cooling, the proposed 3D coupled approach for metal temperature estimation relies on the accurate prediction of hot side heat loads through CFD, modelling coolant injection with local mass sources, whereas an external code provides mass flow rate and heat loads on the cold side, where the flow field can be approximated reliably through a correlative approach. The application of the procedure to a representative test case is presented in this work. contact: [email protected] Towards coupled simulation of aero-engines HP system using industrial CFD solvers P. Legrenzi Loughborough University, UK. contact: [email protected] This work aims to develop a reliable and robust methodology to couple two existing industrial CFD solvers for the simulation of compressor-combustor and combustor-turbine interactions in an aero-engine gas turbine. A low-Mach number pressure-based code specialised for combustion systems and a compressible 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS9: High-‐fidelity coupled simulation of reacting systems density-based code specialised for turbomachinery applications are computed simultaneously, each in its own domain, exploiting a zonal coupling type of approach. Two communication methodologies, file based and memory based, have been used for steady 39 and unsteady RANS simulations respectively. Testing is carried out using simplified geometries. The approach is applied to a real combustor-turbine interaction problem. contact: [email protected] MS10: LES modeling of oxy-coal combustion Wednesday April 22nd (10h00 – 12h00) A. Sadiki1 and H. Pitsch2 1 2 Technische Universitaet Darmstadt, Dept. of Mechanical and Processing Engineering, Germany. RWTH Aachen University, Germany. A major part of the carbon dioxide emissions due to human activities come from fossil fuels. Three main approaches are currently used to capture such undesirable carbon dioxide, namely, pre-combustion capture, post-combustion capture, and oxy-fuel combustion. Oxy-coal combustion is energetically favorable, but features complex unsteady fluid flow, mass and heat transfer interaction processes along with heterogeneous combustion and radiation in an atmosphere that is not usually studied. In order to foster the understanding of these interacting processes and to support the exploration of design concepts for oxy-fuel burners, boilers and combustors while minimizing trial-and-error iterations, reliable CFD tools are essential. During this mini-symposium, some efforts related to developing CFD for oxy-coal combustion will be reported and discussed. Especially CFD tools based on Large-Eddy Simulation related techniques are considered. Four well-known research groups from Germany (TU-Darmstadt, RWTH-Aachen), Poland (TU- Częstochowa), USA (Utah University) and USA (Sandia-Lab.) will address this topic. Developing and validating an oxy-fuel CFD modelling methodolody R. Knappstein1, A. Kethelheun1, G. Kuenne1, S. Farazi2, M. Baroncelli2, A. Sadiki1, H. Pitsch2 and J. Janicka1 1 2 Technische Universitaet Darmstadt, Germany. RWTH Aachen, Germany. Oxy-fuel combustion is recognized as one of the advanced clean fossil fuel technologies to allow for capturing CO2. This contribution reports on current efforts achieved in developing and validating a CFD modelling methodology for oxy-fuel combustion. Relying on Large Eddy Simulation techniques, this modelling approach will enable to provide early information and analysis of intrinsically unsteady fluid flow, mixing and reaction characteristics of complex oxy-fuel combustion systems. An advanced flamelet based tabulated strategy with a detailed chemistry mechanism will be used. The modelling endeavors are closely supported by both DNS and experimental data for validation. LES and RANS of pulverized coal oxy-combustion in swirl burners A. Bogusławski and P. Warzecha Czestochowa University of Technology, Poland. The contribution presents the results of numerical simulations of pulverized coal combustion process in swirl burners using RANS and LES methods for turbulent flow in two test facilities with swirl burners for combustion in air and in O2/CO2 atmospheres. Most of the numerical simulations of pulverized coal oxy-combustion process have been done using RANS method. Therefore authors took an attempt to analyze the influence of turbulence model on the results of pulverized coal oxy-combustion technology. The test cases analyzed show clearly that proper modelling of underlying turbulent flow is crucial for quality of combustion process predictions. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 40 MS10: LES modeling of oxy-‐coal combustion Oxy-coal power boiler simulation and validation through extreme computing Transported probability density function modeling of pulverized coal combustion P. Smith, J. Thornock, Y. Wu, S. Smith and B. Isaac X. Zhao University of Utah, USA. Sandia-Lab, USA. We have studied the operation of a 15 MW boiler under oxy-coal firing with flue gas recirculation for carbon capture. This study was performed using highperformance-computing (HPC) at an extreme scale and dynamic large eddy simulation. These techniques allowed for 1-2 cm resolution of turbulence structures within the boiler and its temporal resolution at microsecond time scale. All of the scales of the particle transport and reaction are fully resolved for the entire particle size distribution. Particle size segregation and clustering are spatially and temporal resolved. Experimental measurements collected in an ALSTOM Boiler Simulator Facility are used for validation. A transported probability density function method has been constructed to model the process of pulverized coal combustion. A Lagrangian particle/Eulerian mesh (LPEM) algorithm has been employed for the gas phase, and a particle-source-in-cell method is applied to the solid phase. Gas-phase chemistry is accounted for by detailed chemical mechanisms. Temperature, gasphase concentrations, and ignition properties are calculated. Parametric studies are carried out, including the variations in turbulence models, turbulent combustion models, heat transfer models, and the phase-coupling models. contact: [email protected] contact: [email protected] MS11: Surface evolution methods in gasoline direct injection combustion engines Wednesday April 22nd (10h00 – 12h00) J. Hahn and P. Priesching AVL List GmbH, Austria. In this minisymposium, we focus on extension of advanced numerical methods on surface evolution into a general unstructured mesh and their practical usages of gasoline direct injection combustion engines. In surface evolution, level set (Eulerian) methods and direct (Lagrangian) methods are reviewed and novel high order explicit and semi-implicit finite volume level set methods are proposed on general computational meshes. An efficient and fast parallel computing of proposed methods is discussed on cell-based overlapping domain prevalent in many commercial numerical CFD softwares. Even if each domain only allows to contain the first neighboring buffer cells, a high experimental order of convergence is obtained. In the context of application examples, contemporary standard and innovated numerical methods in combustion engine simulation from BMW, Germany and AVL, Austria are presented. Moreover, future challenges in gasoline engines are discussed in the view of practical necessity in real application. A flame surface tracking method for calculating the premixed combustion in engines calculations are based on measured data from a variety of application examples. contact: [email protected] P. Priesching AVL List GmbH, Austria. A new modeling approach for the calculation of premixed combustion is presented. This model is able to track the flame surface in a quite accurate manner, based on the combination of a Lagrangian and Eulerian method. Comparisons to other modeling approaches are discussed. In the context of application examples it is outlined, which problems generally dominate in developing spark ignited combustion engines today and how CFD based methods can help to better understand combustion and emission phenomena. Results of the Regularized flux-based gradient for level set methods J. Hahn AVL List GmbH, Austria. We propose a regularized flux-based gradient method to solve surface evolution equations in level set literature. Unlike to using a standard structured mesh, the proposed method extends well-known advantages in level set method for surface evolution into a general unstructured mesh. A high experimental order of 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS11: Surface evolution methods in gasoline direct injection combustion engines convergence is obtained by a regularized gradient computation with compatible discrete operators. An efficient and fast parallel computing is also achieved on cell-based overlapping domains which only allow to have the first neighboring cells. Standard examples in surface evolution and combustion models are numerically investigated to show the advantage of regularized flux-based gradient method. contact: [email protected] evolved by nonlinear advection-diffusion equation, and we present novel high order explicit and semi-implicit finite volume level set methods suitable for computations on general Eulerian computational grids. contact: [email protected] Simulation of turbulent premixed combustion in turbocharged direct injection gasoline engines using a level set based flamelet model D. Linse Finite volume methods for evolving surfaces K. Mikula, P. Frolkovic and M. Remesikova Department of Mathematics, Slovak University of Technology, Slovakia. We present and discuss new finite volume numerical schemes for evolving surfaces both in level set (Eulerian) and direct (Lagrangian) formulations. The methods are capable to solve precisely and in a stable way the complex surface evolutions advected by external vector fields and mean curvature as arising in flame front propagation in numerical combustion. In the Lagrangian formulation, a triangular surface representation is evolved and crucial point for stable and robust computations is usage of a proper redistribution of discrete nodes along the surface. In the level set formulation, the so-called level set function is 41 BMW Group, Germany. We have implemented the G-equation, a level-set based flamelet model, into the 3D-CFD Solver ANSYS CFX. It has proven in recent years to be successful in predicting the turbulent flame front propagation for turbulent premixed combustion. In this talk we discuss the applicability of level set based approaches for the simulation of combustion in turbocharged direct injection spark ignition engines. Special emphasis is put on the characterization of the interaction of turbulence and flame structure. Finally, current and future challenges for the simulation of combustion in gasoline engines are highlighted. contact: [email protected] MS12: Safety related ignition process D. Markus1, U. Maas2 and D. Thévenin3 1 Physikalisch-Technische Bundesanstalt, Germany. Karlsruhe Institute of Technology, Germany. 3 University of Magdeburg, Germany. 2 The interaction of the various transport processes with chemical reactions during an ignition process is a multi scale problem which comprises several time and length scales and different physical and chemical phenomena. Therefore, models to describe ignition processes have to consider a large number of parameters. Especially in the case of safety relevant ignition processes the challenge remains to cover turbulence, plasma processes, heterogeneous and homogeneous combustion etc. This mini‐symposium covers different ignition sources near the minimum energy necessary for ignition. Additionally, the impact of turbulence and mixing on a successful ignition event are investigated. MS12 (1/2) Wednesday April 22nd (13h30 – 15h30) Ignition in a premixed propane/air gas by turbulent jets of its exhaust products A. Ghorbani1, S. Fischer2, G. Steinhilber2, D. Markus1 and U. Maas2 1 2 Physikalisch-Technische Bundesanstalt (PTB), Germany. Karlsruhe Institute of Technology, Germany. The ignition of a combustible environment by hot jets is a safety concern in many industries and understanding the ignition of combustible mixtures by its exhaust products plays an important role in explosion protection. This work addresses numerical aspects of the investigation of such configuration. Calculations are preformed using a probability density function (PDF) method in conjunction with a reaction-diffusion manifold (REDIM) which is used to reduce the number of dependent variables in the description of the thermochemical state. The test case considered in the current work are combustion of premixed hydrogen/air and propane/air gases which are among the exemplary fuels in the investigation of maximum experimental safe gaps (MESG). contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 42 MS12: Safety related ignition process DNS of plasma-assisted ignition in quiescent and turbulent flow conditions M. Castela1, B. Fiorina1, O. Gicquel1, A. Coussement2, C. Laux1 and N. Darabiha1 1 2 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Université Libre de Bruxelles, FSA, Belgium. This numerical study focuses on the ignition of CH4-air mixtures by nanosecond repetitively pulsed electric discharges. A sequence of pulses is computed by means of Direct Numerical Simulations where different initial conditions of turbulence and pulse frequencies are considered. This work examines the impact of the oxygen dissociation, temperature and production of vibrational states of molecules induced by the plasma discharges on the flame kernel formation and propagation. The results show a minor contribution of the discharge energy stored as gas vibrational energy on the initial stage of ignition. In turn, the early production of O atoms favors the initial branching reactions increasing the concentration of radicals in the vicinity of the discharge zone. The initial condition of turbulence inside the discharge zone impacts on the number of pulses needed to ignite the mixture. contact: [email protected] Direct numerical simulation of the auto-ignition of turbulent ethylene/air mixtures A. Abdelsamie and D. Thévenin University of Magdeburg "Otto von Guericke", Germany. The probability of auto-ignition of turbulent ethylene/air mixtures has been examined by means of Direct Numerical Simulations. Various configurations (isotropic homogenous turbulence and jet, in 2D and in 3D) are considered. The impact of different parameters (initial temperature, pressure, equivalence ratio and shear) has been quantified. The UCSD mechanism (64 species, 235 elementary reactions) is used based on Cantera 1.8 to obtain kinetic, thermodynamic and transport properties in the DNS. First results show that 2D DNS are sufficient to predict auto-ignition probability, but further comparisons are needed to generalize this observation. contact: [email protected] Analysis of ignition regimes in rapid compression machines M. Ihme1, K. Grogan2 and S. Goldsborough3 1 Stanford University, USA. no affiliation 3 Argonne National Laboratory, USA. 2 Rapid compression machines (RCMs) are used to study ignition properties of different fuels over a wide range of pressure and temperature conditions. This results in accessing different ignition regimes, which are currently not fully understood. This presentation gives an overview about relevant system-describing quantities. Essential parameters associated with turbulence, ignition chemistry, and heat-transfer are identified, and a regime diagram is constructed to delineate different ignition phenomena. Comparison with measurement are performed to validate the theoretically developed regime diagram. contact: [email protected] MS12 (2/2) Wednesday April 22nd (16h00 – 18h00) Numerical investigation of ignition over heated spheres J. Melguizo-Gavilanes and J. Shepherd California Institute of Technology, USA. Ignition of combustible atmospheres by hot particles is a common issue in industrial safety. Results show the importance of flow separation in creating zones conducive to ignition. Far from the ignition threshold, reaction starts upon contact with reactive mixture, and ignition times are comparable to or less than the transit time of a parcel of fluid from front stagnation point to the separation. Closer to the threshold however, the volume of gas confined in the separation region ignites homogeneously after a longer induction time. Competition between convective losses and heat addition due to chemical reaction is dynamic in this regime, with quenching and re-ignition events observed. contact: [email protected] Ignition at sub-millimeter sized hot particles: a numerical and experimental study D. Roth, T. Häber and H. Bockhorn Karlsruhe Institute of Technology (KIT), Engler-Bunte-Institute (Combustion Division), Germany. Mechanical sparks are small hot particles formed in mechanical shaping processes like grinding and hammering. If these glowing, hot particles come in contact with an explosive gas mixture they may cause an ignition event. We investigate numerically and experimentally the required conditions and governing influences to achieve ignition of combustible-air mixtures by sub-millimeter sized particles. In favor of experimental and computational accessibility the particle generation and consecutive movement will be neglected, thus we cover ignition processes caused by spherical particles in quiescent atmospheres. The surface temperature needed to achieve ignition correlates well with the minimum ignition energy of various combustibles. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. MS12: Safety related ignition process 43 Methanol ignition by a line energy source embedded in a wall Quasi-spectral element method for numerical integration of stiff unsteady combustion systems M. Sanchez-Sanz, E. Fernandez-Tarrazo and A. Sanchez A. Koksharov, V. Bykov and U. Maas Karlsruhe Institute of Technology, Germany. Universidad Carlos III de Madrid, Spain. We consider the transient near-isobaric ignition process of a semi-infinite methanol-air mixture following the sudden application of a constant heat flux from a line energy source placed along a bounding flat wall. In contrast with the case of an infinite atmosphere, when the resulting problem is axisymmetric and the momentum equation becomes secondary for the computation of the temperature and velocity fields, the solution involves in this case the integration of the Navier-Stokes equations, supplemented by the energy and species conservation equations. The description of the ignition history provides information of interest in connection with the problem of micro-scale combustion in rotary engines. Although a typical combustion system is both stiff in time and in space, within an appropriate scale the solution remains sufficiently smooth. This property is used to approximate the solution with a set of polynomials over a proper space-time grid, which consists of a minimal set of irregular elements in space and time providing the solution approximation with the defined tolerance. Starting with a polynomial family and a provided tolerance our method dynamically optimises the distribution and the number of elements during the system integration. This method is benchmarked by 1D premixed hydrogen ignition problems using different polynomials. contact: [email protected] contact: [email protected] MS13: Flame topology Wednesday April 22nd (16h00 – 18h00) S. Cant University of Cambridge, UK. The dynamics of turbulent flame propagation and extinction depend strongly on the local geometry of the flame surface. Statistical analysis of quantities such as the flame surface curvature has already yielded a great deal of useful information. More advanced methods based on ideas from topology are now emerging which allow for a more detailed understanding of the local flame behaviour and its dependence on interactions with the surrounding turbulent flow field. The minisymposium brings together four distinctively different approaches to the common theme of flame topology. These novel topological analysis methods are based on the extensive use of DNS data, including some of the largest available datasets. New results have been obtained for the topology of flame-flame interaction, the tracking of flame extinction holes, and the influence of flame geometry and flame alignment on propagation. New insight has been gained which will assist in the future development of models. Topology of turbulent premixed flame interaction R. Griffiths1, H. Kolla2, W. Kollmann3, J. Chen2 and S. Cant1 1 University of Cambridge, UK. Sandia National Laboratories, USA. 3 University of California Davis, USA. occurrence of each type of interaction topology for two different Damkohler numbers cases. Inferences are drawn concerning the underlying physical processes behind each of the observed interaction types. contact: [email protected] 2 Flame-flame interaction is an important contributor to flame surface area production and destruction in turbulent premixed flames. The present work makes use of a general method for finding the critical points of line, surface and volume-type geometrical features. A large DNS database for turbulent interacting hydrogenair flames is analysed and a complete set of flame interaction topologies is identified using surface shape factors. Statistical data is obtained on the frequency of 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 44 MS13: Flame topology Extinction and reignition dynamics in turbulent dimethyl ether jet flames Effective normal strain rate and scalar gradient enhancement A. Bhagatwala1, E. Hawkes2, P. Bremer3, A. Gyulassy4 and J. Chen1 L. Cifuentes1, C. Dopazo1, J. Martin1, P. Domingo2, L. Vervisch2 and C. Jimenez3 1 Sandia National Laboratories, USA. University of New South Wales, Australia. 3 Lawrence Livermore National Labs, USA. 4 University of Utah, USA. 2 Recent direct numerical simulations of turbulent dimethyl ether temporal slot jet flame at a jet Reynolds number of 13,050 and low Damkohler number are used to understand the mechanisms of local extinction and re-ignition. Topological segmentation methods are used to understand the spatio-temporal dynamics associated with the creation and healing of extinction holes and their dependence on turbulent strain, stretch, and mixing. Conditional statistics following topological structures undergoing extinction and re-ignition will be presented. contact: [email protected] Interactions of turbulence and scalar structures in shear-driven premixed turbulent flames using DNS H. Wang1, E. Hawkes1, H. Kolla2 and J. Chen2 1 School of Mechanical and Manufacturing Engineering, UNSW, Australia. 2 Combustion Research Facility, Sandia National Laboratories, USA. 1 LIFTEC, CSIC-University of Zaragoza, Spain. CORIA - CNRS, Normandie Université, INSA de Rouen, France. 3 Department of Energy, CIEMAT, Spain. 2 Two DNS datasets for turbulent premixed flames (Flame-A: inflow-outflow configuration; Flame-B: Jet in a co-flow) are used to investigate local flow topologies and scalar iso-surface geometries [1]. The scalar-gradient modulus, |∇ Y|, evolution depends on the sign of (a_N+ ∂V^Y /∂ x_N), the ‘effective’ strain rate normal to the iso-scalar surfaces [2]. This study aims at characterizing the ‘effective’ normal strain rate for different scalar field geometries and flow topologies for Flame-A [1] and Flame-B [3]. References [1] L. Cifuentes, C. Dopazo, J. Martin, and C. Jimenez. Local flow topologies and scalar structures in a turbulent premixed flame. Phys. Fluids, 26 (6):065108, 2014. [2] C. Dopazo, L. Cifuentes, J. Martin, and C. Jimenez. Strain rates normal to approaching iso-scalar surfaces in a turbulent premixed flame (under review). Combustion and Flame, 2014. [3] L. Cifuentes, C. Dopazo, J. Martin, P. Domingo, and L. Vervisch. Local volumetric dilatation rate and scalar geometries in a premixed methane-air turbulent jet flame. Proc. Combust. Inst., 35, 2014. contact: [email protected] Direct numerical simulations of lean H2/air flames in a temporally evolving premixed slot-jet configuration are employed to investigate the alignments of vorticity, strain rate, and reacting scalar gradient. It is found that at early times, dilatation from heat release dominates, and the flame normal n is aligned preferentially with the most extensive strain e1. As the flame front interacts with the shear, local strain rate dominants and the alignments are close to those in non-reacting flows. Local flow patterns are identified using the invariants of the deformation rate tensor, and they are shown to be connected with the alignment characteristics. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP1: Soot 45 CONTRIBUTED PRESENTATIONS CP1: Soot CP1 (1/2) Large eddy simulation of soot formation in an oxy-coal combustor Monday April 20th (10h00 – 12h00) Accounting for strain-rate effects in soot modeling of turbulent flames B. Franzelli1, A. Cuoci2, A. Stagni2, C. Saggese2, A. Frassoldati2, T. Faravelli2 and M. Ihme3 1 Laboratoire EM2C (CNRS - Ecole Centrale Paris, France) - CTR (Stanford University, USA). 2 Politecnico di Milano, Italy. 3 CTR, Stanford University, USA. The objective of this work is to identify the impact of reduced descriptions for the modeling of the soot production in turbulent flames. The performance of a reduced soot model is here compared against a detailed description. In a first approach, the role of turbulence on the soot production is examined in the context of a onedimensional counterflow diffusion flame. To introduce curvature and transient effects, a two-dimensional configuration of a flame/vortex interaction is then investigated. Results show that strain rate and curvature have a strong effect on the soot production, which has to be accounted for by the reduced description. D. Lignell1, A. Josephson1, B. Isaac2 and T. Fletcher1 1 2 Brigham Young University, USA. University of Utah, USA. Large eddy simulation (LES) of a 100 kW coal-fired oxy-fuel combustor is presented. Oxy-coal combustion is a promising technology for carbon capture in coalfired power plants. The combustor is a 1.7 m long, 0.6 m diameter, down fired unit. We compare simulation results to experiments including heat flux, temperature, and exit compositions. Soot formation in coal combustion is important for accurate capture of radiative emission and flame temperatures. Soot formation is modeled using a method of moments. We present results of a coal soot model, and uncertainty quantification of key soot rate parameters and representation of the particle size distribution. contact: [email protected] Sensitivity of soot production to gaseous kinetic models in LES of aero-engine combustors B. Franzelli1, E. Riber2, B. Cuenot2 and M. Ihme3 contact: [email protected] 1 Lewis number effects in turbulent nonpremixed sooting flames A. Attili1, F. Bisetti1, M. Mueller2 and H. Pitsch3 1 King Abdullah University of Science and Technology, Saudi Arabia. Princeton University, USA. 3 Aachen University, Germany. 2 Two direct numerical simulations of planar nheptane/air turbulent jets are compared to assess the effect of the gas-phase species diffusion model on soot formation and growth. The statistics of temperature and major species obtained with a mixture average formulation are very similar to those in the unity Lewis number case. The total mass of PAH precursors decreases by 10 to 20% with the Le=1 approximation, but their distribution is more homogeneous in space and time. Due to the non-linearity of the soot growth rate with respect to the precursors’ concentration, the soot mass yield decreases by a factor of two. Laboratoire EM2C (CNRS - Ecole Centrale Paris, France) - CTR (Stanford University, USA). 2 CERFACS, France. 3 CTR, Stanford University, USA. The objective of this study is to evaluate the LES approach for the prediction of soot production in an experimental swirl-stabilized non-premixed ethylene/air aero-engine combustor, for which soot and flame data are available. Three simulations are performed using a two-equation soot model to compare the performance of a skeletal kinetic, a hybrid chemical description (reduced chemistry for the flame structure/tabulated chemistry for soot precursor chemistry) and a classical tabulation method. Discrepancies for soot concentrations between the three LES calculations will be analyzed and the sensitivity to chemical models will be investigated. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 46 CP1: Soot PDF modelling of soot emissions CP1 (2/2) M. Schiener and R. Lindstedt Monday April 20th (13h30 – 15h30) Imperial College London, UK. The present work investigates the prediction of soot in the Delft/Adelaide turbulent natural gas diffusion flame. A transported pdf approach, closed at the joint-scalar level, is used in conjunction with a second moment solution for the velocity field and coupled with the method of moments for modelling soot coagulation and agglomeration. Soot oxidation via reactions with O, OH and O2 is taken into account, and the inclusion of surface reactions based on a second ring PAH analogy is investigated. A parametric study of the surface chemistry growth model is conducted and results are compared to previous experimental and numerical data. contact: [email protected] RANS-based modeling and uncertainty quantification of soot formation in flames H. Koo1, M. Mueller2, V. Raman3 and B. Dally4 1 University of Texas, USA. Princeton University, USA. University of Michigan, USA. 4 University of Adelaide, Australia. 2 Heterogeneous PAH dimerization as an important contributor to soot nucleation N. Eaves1, S. Dworkin2 and M. Thomson1 1 2 University of Toronto, Canada. Ryerson University, Canada. Given the upcoming EURO 6 regulations, which include limits on particle number density, soot models must be capable of accurately predicting soot particle sizes. In recent work [N. Eaves et al., Proc. Comb. Inst. 2015], a fundamental model for both the nucleation and condensation processes that directly accounted for their reversibility was developed. The model has been modified to allow nucleation and condensation to occur from PAHs as small as naphthalene and as large as benzo[a]pyrene. Preliminary results demonstrate heterogeneous dimers containing a small and a large PAH represent the largest contribution to net soot nucleation rates. contact: [email protected] 3 The simulation of turbulent sooting flames poses a tremendous modeling challenge due to the complex interaction between soot chemistry, particle dynamics, and the background turbulent flame. Recently, highfidelity experimental data of canonical sooting flames has been made available by the University of Adelaide. In this work, Reynolds-averaged Navier Stokes (RANS) simulation of this flow configuration with detailed chemistry and soot models are presented. In any turbulent flow simulation, a number of model coefficients are used in the sub-models to describe the participating physics. These model coefficients are determined from other canonical flows and are subject to large uncertainties. In order to better judge the simulation capabilities, the RANS computations are combined with a sampling-based uncertainty quantification (UQ) approach. Variations in model coefficients that determine turbulence properties, inflow conditions, and soot and combustion models are considered. The results are then compared with experimental data to better understand the performance and failings of the model. Overall, soot volume fraction is predicted lower compared to experiments, and the shape of the jet spread (controlled by turbulence model parameters) has an important role in determining soot evolution. Modeling the soot particle size distribution in flames using an extended quadrature method of moments approach S. Salenbauch1, A. Cuoci2, T. Faravelli2 and C. Hasse1 1 2 TU Bergakademie Freiberg, Germany. Politecnico di Milano, Italy. Moment methods are an efficient and accurate approach to model the soot evolution in flames. However, as a drawback, the fully resolved particle size distribution is usually not available. In this study, the recently developed 'Extended Quadrature Method of Moments [1]', which allows the reconstruction of the size distribution out of a few moments, is applied to a detailed, phenomenological soot model [2]. Retaining the numerical efficiency of a moment approach, the method enables a very accurate closure of the moment source terms. Finally, the model is extended to a bivariate description, which accounts for primary particles and aggregates. [1] C. Yuan, F. Laurent, R. Fox, Journal of Aerosol Science 51 (2012) 1-23. [2] J. Appel, H. Bockhorn, M. Frenklach, Combustion and Flame 121 (2000) 122 - 136 contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP1: Soot 47 Modeling soot formation and oxidation in laminar premixed flames using the method of moments and a continuous reconstruction of the soot number density function 1 2 2 3 A. Wick , T. Nguyen , F. Laurent-Nègre , M. Massot , R. Fox4 and H. Pitsch5 1 Institute for Combustion Technology, RWTH Aachen University, Germany. 2 CNRS, UPR 288, Laboratoire EM2C, France; Ecole Centrale Paris, France; CNRS, FR 3487, Fédération de Mathématiques de l'Ecole Centrale Paris, France. 3 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France; CNRS, Fédération de Mathématiques de l'Ecole Centrale Paris, France. 4 Department of Chemical and Biological Engineering, Iowa State University, USA; CNRS, UPR 288, Laboratoire EM2C, France; Ecole Centrale Paris, France; CNRS, FR 3487, Fédération de Mathématiques de l'Ecole Centrale Paris, France. 5 Institute for Combustion Technology, RWTH Aachen University, Germany. Quadrature-based moment methods have shown to give a high accuracy for the solution of the Population Balance Equation for several processes of engineering interest including the formation of soot. Formally, the quadrature is equivalent to a representation of the Number Density Function by a series of delta functions. In case of particle disappearance due to oxidation, the flux of particles at minimum size needs to be known, which requires the reconstruction of the Number Density Function by a continuous function or a superposition of continuous kernel functions. Such methods have recently been developed, and are now for the first time applied and evaluated in the context of soot formation and oxidation. An efficient operator splitting Monte Carlo for aerosol dynamics simulation with application to soot formation A. Abdelgadir1, K. Zhou2 and F. Bisetti1 1 Clean Combustion Research Center, King Abdullah University of Science and Technology, Saudi Arabia. 2 Wuhan University of Science and Technology, Wuhan, China. A highly efficient Monte Carlo scheme is developed for aerosol dynamics simulation. The method accounts for nucleation, growth and coagulation and has applications in aerosol dynamics and combustion (soot formation). It adopts an operator splitting technique to separate coagulation from other processes. A careful study of the errors and convergence properties is conducted using forms of the collision kernel for which the analytical solutions of the extended Smoluchowski equation are available. The method is applied to the simulation of soot in a laboratory flame and the results are in good agreement with the results obtained using method of moment. contact: [email protected] contact: [email protected] Development of a unique function for soot surface reactivity while oxidation and surface growth in laminar diffusion flames A. Khosousi and S. Dworkin Ryerson University, Canada. Numerical modelling of laminar coflow diffusion ethylene/air flames is performed using complex chemistry, detailed transport and sectional soot models. A newly developed surface character model accounts for soot surface reactivity in oxidation and surface growth by considering soot aging and its effects on the particle surface. It has been shown that soot particles become less reactive as they age. This new development eliminates one parameter used in soot numerical simulations, α, and is implemented using temperature and residence time. The model has proven reasonably capable of predicting soot concentrations in a wide range of laminar flames. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 48 CP2: Numerical methods CP2: Numerical methods CP2 (1/2) Monday April 20th (10h00 – 12h00) Adaptation in time and space for multi-scale combustion fronts with error control based on operator splitting and multiresolution Simple and fast integration method for stiff chemical kinetic ODEs 1 2 3 Y. Morii , H. Terashima , M. Koshi , T. Shimizu and E. Shima1 M. Duarte1, S. Descombes2, M. Massot3, C. Tenaud4 and S. Candel5 1 1 Japan Aerospace Exploration Agency, Japan. 2 University of Tokyo, Japan. 3 Yokohama National University, Japan. The simple and fast explicit time integration method for stiff chemical kinetic ordinary differential equations is proposed. Numerical simulations of ignition problems using the proposed method are performed, and the speed and accuracy of the proposed method are investigated for comparison with earlier explicit time integration method, MTS and CHEMEQ2. As a result, the proposed method shows the superior performance to the other explicit methods. contact: [email protected] Physics-based preconditioning and dual timestepping for stiff combustion problems with detailed chemical mechanisms 1 CCSE, Lawrence Berkeley National Laboratory, USA Laboratoire J.A. Dieudonné, CNRS and Université de Nice - Sophia Antipolis, France. 3 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France; CNRS, Fédération de Mathématiques de l'Ecole Centrale Paris, France. 4 LIMSI, CNRS, France. 5 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. 2 We introduce a new resolution strategy for multi-scale reaction waves based on time operator splitting and space adaptive multiresolution. Based on theoretical numerical analysis studies, this strategy leads to a splitting time step which is not restricted neither by fast scales in the source term nor by restrictive stability limits of diffusive or convective steps, but only by the physics of the phenomenon and can be dynamically adapted using a posteriori error estimates. The main goal is then to perform computationally efficient as well as accurate in time and space simulations of the complete dynamics of multi-scale phenomena. contact: [email protected] M. Hansen and J. Sutherland University of Utah, USA. Detailed simulation of turbulent, low-Mach combustion is difficult due to the breadth of length and time scales involved. A principal difficulty of combustion is stiffness due to strong coupling between 'slow' processes such as convection and mixing and 'fast' processes such as ignition and equilibration. The inherent stiffness and nonlinearity of detailed chemical mechanisms severely constrain traditional integration methods. Matrix preconditioning is a versatile acceleration technique which has seen little development for combustion. We present results that demonstrate the efficiency of dual time-stepping and a novel preconditioning approach in resolving lowfrequency physics of stiff, 0-D reactors with detailed chemical mechanisms. contact: [email protected] A new approach to secure scalar boundedness in flame simulation with high-order spectral difference methods on unstructured meshes E. Bossennec, G. Lodato, P. Domingo and L. Vervisch CORIA, CNRS and INSA de Rouen, France. In flame simulation, it is essential to preserve scalar boundedness. Various approaches exist to do so, but none is actually fully satisfactory (clipping, additional diffusivity, etc.). In the proposed approach, a scalar discontinuity sensor is obtained from the modal decomposition of the signal within every cell, in the context of spectral difference (SD) methods. The objective of SD is to secure a given degree of accuracy, whatever the mesh topology and stretching. From the modal decomposition of the scalar signal, an autocalibration of an additional dissipative term is designed within every cell element and various tests cases are discussed. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP2: Numerical methods G-equation based modelling of partially premixed compressible reactive flows L. Gastaldo and J. Latché IRSN, France. We present a model and a numerical scheme for the simulation of partially premixed compressible reactive flows, implemented in the IRSN software P2REMICS. The flame brush location is determined by a level-setlike equation, the so-called G-equation. To cope with partially premixed situations, the species mass balances are however solved, but with a reaction rate depending on the level-set function. This system is solved by a fractional-step algorithm, which ensures the essential stability properties, like keeping the species mass fractions or thermodynamic variables (pressure, internal energy) within their physical bounds. The simulation of a free hydrogen jet explosion is presented. contact: [email protected] Modelling of flame lift-off evolution: numerical implementation of density-pressure coupling Z. Chen, S. Ruan and N. Swaminathan University of Cambridge, UK. Three-dimensional unsteady simulations of turbulent lifted methane/air jet flames are performed to investigate transient evolution of flame lift-off process. Premixed flamelets based partially premixed combustion modelling framework is implemented using the OpenFOAM CFD platform. The implementation effects of using different density-pressure coupling approaches on the temporal and spatial evolution of the most upstream point of flame leading edge are studied. It is found that the final flame lift-off height is influenced only marginally by these effects, whereas a highly coupled density-pressure solving approach is required to capture the transient propagation process of flame lift-off evolution. contact: [email protected] CP2 (2/2) Wednesday April 22nd (13h30 – 15h30) High-order CENO finite volume scheme for largeeddy simulation of turbulent reactive flows L. Tobaldini Neto and C. Groth University of Toronto, Institute for Aerospace Studies, Canada. A novel, parallel, high-order, central essentially nonoscillatory (CENO), finite-volume scheme is proposed for large-eddy simulation (LES) of turbulent reactive flows. The high-order CENO finite-volume scheme is applied to the solution of the Favre-filtered NavierStokes equations governing a fully-compressible 49 reactive mixture on multi-block, body-fitted, computational mesh consisting of hexahedral elements. The CENO method uses a hybrid reconstruction approach based on a fixed central stencil that avoids the complexities of other ENO schemes and offers several computational advantages. Numerical results are discussed for laboratory-scale turbulent premixed methane-air flames and the performance of the highorder approach relative to standard lower-order schemes is assessed. contact: [email protected] LES simulation of reactive flow with effective PDF models at a minimal cost Y. Ge, F. Ferraro and M. Pfitzner der Universität der Bundeswehr München, Germany. LES is gradually becoming a reliable design tool in computational fluid dynamics. However, in combustion CFD, chemical reactions and their interaction with turbulence are also important. Advanced PDF methods have been proved to work well in non-premixed combustion. In a Lagrangian implementation, the computational cost is linked to the number of particles involved. Two methods are presented to incorporate the PDF model into the LES simulation at a minimal cost. The sparse-Lagrangian simulation requires less Pope particles than the LES grid cells while a hybrid LES/RANS-PDF simulation performs the PDF calculation within RANS context using an averaged LES flow field. contact: [email protected] Large eddy simulation using adaptive mesh refinement with a multi-level subgrid scaled closure for turbulent reacting flows B. Muralidharan and S. Menon Georgia Institute of Technology, USA. We employ a new strategy to use adaptive mesh refinement (AMR) within a LES to couple the largescale features resolved on AMR grid with a subgrid turbulence-chemistry linear-eddy mixing (LEM) model. Unlike conventional past implementation of LEM within a fixed grid LES, in this hierarchical scheme the subgrid closure for turbulence and for turbulencechemistry occur on independent grids that exploit the strengths of each approach. Application to turbulent Hydrogen/Air reacting jet in cross flow (JICF) show that features such as lifted flame structure and triple flame formation are captured in good agreement with low Re DNS. More comparisons with high-Re experiments of reacting H2/AIR JICF will be shown in the final presentation. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 50 CP2: Numerical methods Development, validation and application of a DNS approach for heterogeneous reactive flows 1 2 2 3 A. Chabane , K. Truffin , A. Nicolle , F. Nicoud and C. Angelberger2 1 IFPEN/ECP, France. IFPEN, France. 3 Université Montpellier 2, France. Compressible mixing of liquid fuels A. Hernandez1, S. Stekovic1, D. Stewart2 and A. Wass1 1 Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, USA. 2 no affiliation 2 Appropriate boundary conditions accounting for mass, momentum and heat transfers near a reactive wall are derived for the Direct Numerical Simulation of heterogeneous reactive flows. A specifically developed implicit solver for the resolution of detailed surface and gas-phase chemistry is coupled with the explicit LES/DNS code AVBP. The numerical developments are validated using analytical solutions and academic configurations for which experimental data are available. The impact of heterogeneous reactions and obstacles on transfers near a catalytic surface, on a catalytically-stabilized reaction zone and on pollutant conversion kinetics are studied for the case of a channel-flow monolithic catalyst. We present a multidimensional combustion model for high pressure, compressible mixing of liquid fuels that vaporize and mix with entrained air to burn. A robust multi-component, fully parallel, finite difference code was developed to solve these three-dimensional complex reactive flow problems. The spatial component of the governing equations is solved by a positive preserving WENO interpolation solver and the temporal component uses a TVD Runge-Kutta scheme. For reflective boundary conditions we developed a simplified novel point-wise level set scheme that is local and can efficiently run on parallel machines. Computed results show good agreement between numerical and experiment data. contact: [email protected] contact: [email protected] CP3: Kinetics reduction and tabulation CP3 (1/3) Monday April 20th (10h00 – 12h00) Extraction of Chemical kinetics insights with special CSP data S. Lam Princeton University, USA. In order to claim insights on a massively complex chemical reaction system, one needs answers to the following questions: (1) What are the current time scales of the reactions? (2) Which reactions are currently rate-controlling? (3) Which chemical species and which chemical reactions included in the full chemistry model are currently unimportant? (4) What simplified-chemistry models could honor a userspecified species-by-species error tolerance? (5) What happens to the above answers if some of the parameters were changed? This paper shows how to computationally obtain insightful and definitive answers to such questions by exploiting reaction-specific special CSP data (S. Lam, Combustion and Flame, 2014). Reduction of chemical mechanisms for aviation fuels with RCCE W. Jones, P. Koniavitis and S. Rigopoulos Department of Mechanical Engineering, Imperial College London, UK. Detailed mechanisms for aviation fuels contain hundreds of species and thousands of reactions, indicating a necessity for reduced mechanisms. RateControlled Constrained Equilibrium (RCCE) provides a physical and mathematical framework for deriving reduced mechanisms, where the kinetically controlled (slow) species are determined from the kinetics of the detailed mechanism, while the equilibrated (fast or steady-state) species are calculated by minimising the Gibbs free energy of the system. In this work, we derive reduced chemical mechanisms with RCCE for kerosene. Several sets of constraints and surrogate fuels are employed, and a considerable degree of reduction is obtained. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP3: Kinetics reduction and tabulation 51 Investigation of corrugated premixed methane-air flame structure using REDIM Trivariate chemistry model for the simulation of high Karlovitz premixed turbulent flames A. Neagos, V. Bykov and U. Maas B. Savard and G. Blanquart Karlsruhe Institute of Technology, Germany. California Institute of Technology, USA. The correct description of transient turbulent processes within thickened and corrugated flames represents a challenging problem for model reduction. In the present study the performance of the REDIM model reduction in such flame scenarios is investigated. Therefore an examination of the relaxation process of perturbed 1D premixed methane-air flames is carried out. Perturbations are implemented via triplet maps leading to corrugated flame structures. Among other investigation a comparison of detailed and reduced calculations regarding relaxation time scales dependent on size and magnitude of the triplet perturbation is performed. It is shown that the reduced models (1D and 2D) can capture such critical transient flame behavior for different equivalence ratios. A trivariate chemistry model is proposed to capture accurately the chemical source term fluctuations of thermodiffusively-stable premixed turbulent flames in the thin/broken reaction zones regimes. The model is derived mathematically from a local coordinate transformation. A progress variable, its dissipation rate, and the curvature of its isosurfaces are used to represent the chemical state through a flamelet-generated manifold. This manifold is first validated using direct numerical simulations of turbulent flames with detailed chemistry. The model, which requires a single scalar transport equation, is then used to simulate these flames. The model can predict most of the flames' characteristics at a significantly lower computational cost. contact: [email protected] contact: [email protected] Flamelet/Progress Variable CFD Modeling of HighPressure Partial Oxidation Flame CP3 (2/3) Monday April 20th (13h30 – 15h30) M. Vascellari, H. Xu, S. Hartl and C. Hasse TU Bergakademie Freiberg, Germany. In partial oxidation processes two significant reactions regimes can be observed. In the flame zone, the fast oxidation reactions are mainly governed by the mixing of the fuel with the reactants, similarly to conventional lean diffusion flames. In the post-flame zone, the reforming reactions takes place, slowly approaching to the chemical equilibrium conditions. In order to describe both regimes, the flamelet/progress variable (FPV) approach was successfully applied to model a pilot-scale High-Pressure Partial OXidation (HP-POX) reactor fired with CH4 and O2/H2O, respectively. The chemistry was pre-computed using laminar one-dimensional premixed freely propagating flames in physical space for different mixture fractions and stored in a look-up table. Turbulent fluctuations were accounted by means of the PDF-averaging of the laminar solution in flamelet and progress variable spaces. The laminar solutions were converted from physical space to a progress variable (PV) space, which was defined combining different chemical species in order to have a monotonically growing function. Additionally, the PV should be accurately defined for resolving the slow reactions in the post-flame zone. Finally, the PDF look-up table generated was coupled to the CFD solution of the turbulent flow in the HP-POX reactor. The numerical results shows that the FPV allows to describe both the fast oxidation and the slow reforming reactions, giving a reasonable agreement with the experiments. A reduced and optimized chemical mechanism for ndodecane oxidation L. Cai, L. Kröger and H. Pitsch Institute for Combustion Technology, RWTH Aachen University, Germany. A strongly reduced kinetic mechanism for n-dodecane oxidation will be presented. The mechanism consists of around 60 species and is quite compact in comparison with mechanisms in the published literature. For improved model performance, the reduced n-dodecane mechanism is further automatically optimized against a large number of experimental measurements over a variety of conditions. The resulting mechanism is compared with detailed and other reduced mechanisms in the literature to assess model performance in predictions of combustion properties and engine simulations. The reduced mechanism presented here is shown to be quite accurate and besides other quantities also reproduces burning velocities very accurately. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 52 CP3: Kinetics reduction and tabulation Reduced schemes in combustion 1 1 1 A. Felden , B. Cuenot , E. Riber and P. Pepiot 2 1 CERFACS, Toulouse, France. Sibley School of mechanical and Aerospace Engineering, Cornell University, USA. 2 Recent progress in CFD applied to complex combustion systems, coupled to the increase in available computational power, now allows for the use of a more accurate chemistry description through reduced kinetic schemes. Derived in a fast and efficient way, such schemes greatly improve flame predictions at a reasonable cost. The methodology and available tools for their derivation will be presented first, followed by a demonstration of their performances in terms of accuracy and CPU cost for the combustion of various hydrocarbons. Finally, a LES of a realistic combustion chamber using the obtained reduced kinetic schemes is performed and analyzed. In particular, the potential for increased prediction capability of such simulations is discussed. contact: [email protected] Using genetic algorithms for optimizing reduced chemical-scheme for turbulent flames simulation N. Jaouen, P. Domingo and L. Vervisch CORIA, CNRS and INSA de Rouen, France. quality of it depends heavily on the choice of the mass fraction. A regularization method is applied to select strain parameters, which are optimal with respect to different sets of constraints. The resulting parameterizations are tested in large-eddy simulations of a turbulent premixed jet burner and the sensitivities of the results are assessed. contact: [email protected] Using large eddy simulation (LES) for pollutant prediction in turbulent flames. T. Jaravel, E. Riber and B. Cuenot CERFACS, France. The accurate prediction of pollutants emitted by combustion systems remains a challenge due to the complexity of the involved chemistry and flow. To describe pollutant chemistry, an analytically reduced mechanism for methane-air oxidation, including carbon monoxide (CO) and nitrogen oxide (NOx) is proposed. This mechanism is first evaluated on laminar flames, showing a good prediction capability for both CO and NOx. It is then combined to LES to compute the Sandia D turbulent diffusion flame. Results are found in good agreement with measurements in terms of flame properties and pollutant formation. Effects of grid resolution and reduced chemical scheme are also discussed. contact: [email protected] A strategy to optimize reduced chemical schemes based on a genetic algorithm is discussed. The targets of the optimization procedure are the chemical source profiles of a reduced number of species. These reference chemical sources are reconstructed from a detailed chemical scheme, so that the balance equations of the reduced number of species are closed when imposing the profiles obtained with detailed chemistry and for given transport properties. The optimization is first performed for H2/Air combustion, to then add the necessary reactions to simulate CH4/Air premixed flames at various equivalence ratios. contact: [email protected] Application of a regularization method to the choice of the strain parameter in LES of a turbulent premixed jet burner K. Kleinheinz1, E. Knudsen2, E. Hawkes3 and H. Pitsch1 1 Institut for Combustion Technology, RWTH Aachen University, Germany. 2 Dept. of Mechanical Engineering, Stanford University, USA. 3 Faculty of Engineering, The University of New South Wales, Australia. An adaptive strategy for the efficient implementation of complex chemistry in turbulent flame simulations Y. Liang, S. Pope and P. Pepiot Cornell University, USA. The integration of detailed chemical kinetics for practical fuels in turbulent combustion simulations is in most cases prohibitive in terms of computational cost. We will describe an adaptive strategy tailored for LES/particle PDF simulations of turbulent flames that relies on an a priori partition of the composition space into a specified number of regions, over which suitable reduced chemical mechanisms are identified prior to the simulation. The method is combined in LES/PDF with In-Situ Adaptive Tabulation (ISAT) and reduced computational particle representations to show significant gain, both in terms of memory and CPU usage. contact: [email protected] A strained flamelet model has recently been proposed, which accounts for turbulence’s perturbations of the flame structure. Its novelty is the strain parameterization via a species mass fraction. The 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP3: Kinetics reduction and tabulation CP3 (3/3) th Tuesday April 20 (16h00 – 18h00) Reducing chemistry table sizes by using polynomial functions W. Leudesdorff, A. Ketelheun and J. Janicka Energy and Power Plant Technology, TU Darmstadt, Germany. Flamelet generated manifolds are well-established in large eddy simulations of combustion processes. However, the tables tend to get very large due to the required resolution and to ensure an efficient searchfree access during the simulation and thus occupy much memory in parallel computations. With an increasing number of control variables, this problem gets worse. Within this work, a method was developed to avoid the fine equidistant tabulation by using polynomial functions to represent the variables’ dependencies especially on the mixture fraction. The results show a significant reduction in required memory by maintaining the accuracy of the chemistry representation. contact: [email protected] From fully premixed to highly stratified combustion using hybrid transported tabulated chemistry B. Duboc, G. Ribert and P. Domingo CORIA-Normandy University, France. A novel way to downsize detailed chemistry, HTTC, has been recently proposed by Ribert et al. (FTaC 2014). In that method, the major species are transported while the intermediate species are tabulated, making use of their self-similarity properties. In the present work, the capacities of HTTC to handle combustion regimes from fully premixed to highly stratified flames, up to the limit of diffusion flames, are explored. The performances of HTTC will be compared toward fully detailed chemistry and classic tabulated chemistry approaches. The numerical set-up consists in a V-flame, fed by a flow featuring various gradients of equivalence ratio. contact: [email protected] Comparative analysis of kinetic mechanism reduction methods within the tabulation of dynamic adaptive chemistry framework F. Contino1, T. Lucchini2, S. Backaert3, N. Bourgeois3, A. Parente4 and H. Jeanmart3 1 Vrije Universiteit Brussel, Belgium. Politecnico di Milano, Italy. 3 Université catholique de Louvain, Belgium. 4 Université Libre de Bruxelles, Belgium. 2 53 simulations, many methods reduce the size of these mechanisms to the set of species and reactions needed for the specific thermal-chemical conditions of the case being investigated. In the context of on-the-fly reduction coupled to a dynamic tabulation technique, the size of the mechanism changes during the simulation which has a great impact on the coupling. We compare the efficiency of five different reduction methods in the framework of the TDAC method. This paper presents the most effective reduction techniques for two test cases: an inhomogeneous auto-ignition case and an unsteady flame in a closed vessel. contact: [email protected] Principal component transport in turbulent combustion T. Echekki and H. Mirgolbabaei North Carolina State University, USA. The solution of turbulent combustion problems based on the transport of principal components (PCs) is demonstrated. The PCs are derived from a priori principal component analysis (PCA) of computational data. This analysis is used to construct and tabulate the PCs’ chemical source terms and diffusion coefficients in terms of the PCs using artificial neural networks (ANN). The a posteriori validation is implemented on different combustion problems and yields a very good reconstruction of the original thermo-chemical scalars profiles with a significant reduction in the number of transported variables. contact: [email protected] Principal component analysis for modelling turbulent premixed flames A. Coussement1, O. Gicquel2, B. Isaac3 and A. Parente1 1 Université Libre de Bruxelles, ATM Department, Belgium. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. 3 University of Utah, Department of Chemical Engineering, USA. 2 In the last years the potential of principal component analysis for the development of reduced order turbulent combustion models has been investigated. Recently, a few methods have been able to perform a posteriori computations, but no comparative analysis has yet been performed. This work aims at filling this gap by presenting a comprehensive comparison between the MG-PCA and Score-PCA technique for laminar and turbulent premixed flames. Computations are done using direct numerical simulations to consider only the quality of the combustion modelling. Finally, special care is also applied to the treatment of differential diffusion including further developments of Score-PCA. contact: [email protected] To mitigate the impact of large chemical kinetic mechanism on the computational time of CFD 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 54 CP4: Internal combustion engines CP4: Internal combustion engines Monday April 20th (13h30 – 15h30) Assessment of double conditioning of reactive scalar transport equations for HCCI 1 2 1 Investigations on pressure wave generation and chemical kinetics during end-gas autoignition on knocking combustion 3 F. Salehi , M. Talei , E. Hawkes , A. Bhagatwala , J. Chen3 and S. Kook1 H. Terashima1 and M. Koshi2 1 The University of New South Wales, Australia. 2 The University of Melbourne, Australia. 3 Sandia National Laboratories, USA. 1 This work investigates doubly conditional moment closure (CMC) equations under HCCI conditions. For this purpose, a set of direct numerical simulation (DNS) data is used. In the DNS cases, ignition of a lean ethanol/air mixture with thermal inhomogeneities is simulated. Two possible variables including scalar dissipation and progress variable are introduced as a second conditioning variable into the first-order CMC model. A priori test using DNS data shows that the choice of progress variable significantly improved results compared to the choice of scalar dissipation however the former leads to more difficulties in modelling the joint PDF and the cross dissipation rate. A knocking combustion modeled using a onedimensional constant volume reactor is simulated in a manner of direct numerical simulations, in which large detailed chemical kinetic mechanisms for two premised gases, n-butane (113 species) and n-heptane (373 species), are directly and efficiently used with the compressible Navier-Stokes equations. Detailed mechanisms of pressure wave generation and chemical kinetics during end-gas autoignition are discussed, which would lead to an explanation of super-knocking combustion. The study would also discuss the effects of wall temperature conditions on the strength of knocking combustion. contact: [email protected] contact: [email protected] A DNS study of ignition characteristics of a lean H2/air mixture under HCCI conditions within an enclosed geometry including wall heat transfer M. Bolla1, M. Schmitt2, E. Hawkes1 and K. Boulouchos2 2 University of Tokyo, Japan Yokohama National University, Japan. Prediction of super-knock in a downsized sparkignition engine using Large-Eddy Simulation K. Truffin, A. Robert, O. Colin and C. Angelberger IFPEN, France. 1 The University of New South Wales, Australia. 2 Swiss Federal Institute of Technology, Switzerland. This work presents results from a two-dimensional (2D) direct numerical simulation (DNS) of a lean H2/air mixture under homogeneous charge compression ignition conditions within an enclosed geometry including wall heat transfer. Initial and boundary conditions have been taken from a previous nonreacting DNS [Schmitt et al., Proc. Comb. Inst. 35 (2015) 3069-3077] in an engine-like geometry. The domain consists of a rectangular 2D slice (75mm x 7.5mm). The initial H2-air mixture is compositionally homogeneous and is thermally stratified according to the compression stroke history. The DNS are analysed to understand the effects of the wall on the ignition behaviour. Downsizing is a major pathway to improve the fuel efficiency and reduce CO2 emissions from Spark Ignition Engines (SIE). Its principle is to increase the specific load at which the engine is operated. The resulting thermodynamic conditions inside the cylinder however tend to favour the appearance of uncontrolled auto-ignition and can result in the onset of detonation. They are therefore very complex to study experimentally due to the risks of engine damage. The fundamental novelty of the proposed research is to develop and validate LES approaches for studying such soliton cycles and quantifying the probability for their occurrence. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP4: Internal combustion engines Towards an integrated approach to realistic engine modeling with large-scale computing S. Aithal Argonne National Laboratory, USA. Predicting the performance and emissions of automotive engines over a wide-range of operating conditions is a daunting task. High-fidelity simulations can provide details of the flow physics but the required computational effort precludes their use for parametric 55 studies. Reduced-order models can be used for largescale parametric sweeps but require tuned parameters for realistic results. An integrated approach to realistic engine modeling combines the unique features of both the high-fidelity simulations with reduced-order simulations. This work describes the development of a framework to conduct large-scale simulations using such an integrated approach. contact: [email protected] CP5: Turbulent combustion CP5 (1/6) Monday April 20th (16h00 – 18h00) Modelling of the subgrid scale wrinkling factor for large-eddy simulation of turbulent premixed combustion F. Thiesset1, G. Maurice1, F. Halter1, N. Mazellier2, C. Chauveau1 and I. Gökalp1 1 2 ICARE, CNRS Orléans, France. PRISME, Univ. Orléans, France. We propose a model for assessing the unresolved wrinkling factor in LES of turbulent premixed combustion. It relies on a power-law dependence of the wrinkling factor to the filter size and an original expression for the ’active’ corrugating strain rate. The latter is written as a product of a new efficiency function by a recent expression for the turbulent strain intensity. Yields functional expressions for the fractal dimension and the inner cut-off length scale, the latter being (i) filter-size independent and (ii) consistent with the Damköhler asymptotic behaviours at both large and small Karlovitz numbers. A new expression for the wrinkling factor which incorporates finite Reynolds numbers effects is also proposed. contact: [email protected] sub-grid wrinkling. However, if the TFM transformation is applied to the complete Navier-Stokes equations, the effect of sub-grid wrinkling can be directly captured. Two sets of simulations will be presented and compared with experimental data. The effects of TFM on the scalar equation and in the full Navier-Stokes equations will be investigated. contact: [email protected] Analysis of a dynamic flame wrinkling factor model for large eddy simulations of turbulent premixed combustion P. Stefanin-Volpiani, T. Schmitt and D. Veynante Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Dynamic models where model parameters are automatically adjusted from known resolved fields are a very attractive formulation for large eddy simulations. Now widely used for unresolved momentum transport, this approach remains rather marginal to describe filtered reaction rates despite of very promising results. The influence of physical (flame wrinkling inner cut-off length scale) and numerical (test filter width, averaging procedure, filtering frequency, …) characteristics of a flame wrinkling factor dynamic model for turbulent premixed combustion is investigated. Numerical results are discussed in terms of mean flow fields as well as dynamical behaviors. contact: [email protected] Modelling of lean premixed flames based on a new artificial thickened flame framework K. Vogiatzaki1, S. Navarro-Martinez2, S. Hong3, S. Shanbhogue3 and A. Ghoniem3 A dynamic sub-grid model for turbulent combustion based on conditional averaging G. Hendra, M. Salehi and W. Bushe 1 City University, London, UK. 2 Imperial college London, UK. 3 MIT, USA. In this work we present a numerical investigation of lean premixed flames behaviour at a backward-facing step combustor with a modified version of the Thickened Flame Model (TFM). In conventional TFM approach, an efficiency function is used to represent the University of British Columbia, Canada. The Germano dynamic sub-grid model in Large Eddy Simulation requires averaging of turbulence parameters across ensembles of statistical homogeneity. We propose that binning by mixture fraction (a proxy for instantaneous position within the mixing layer) could provide more accurate results than the traditional 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 56 CP5: Turbulent combustion approach of binning by radius (a proxy for average position). Properly accounting for sub-grid variation in mixture fraction requires inversion of an integral expression; the method therefore has a natural synergy with the Conditional Source-Term Estimation (CSE) closure for the chemical source term in turbulent reacting flows, which calculates conditional averages of progress variables by integral inversion. contact: [email protected] Large eddy simulation of turbulent combustion with a dynamic second-order moment closure model K. Luo, J. Yang, Y. Bai and J. Fan Zhejiang University, China. A new dynamic second-order moment closure (DSMC) model is developed for large eddy simulation of turbulent combustion. The averaged reaction rate is directly closed in the form of Arrhenius law. The thirdorder fluctuation correlations are neglected, and the second-order fluctuation correlations are closed using the algebraic form. All the coefficients in the model are evaluated dynamically. A-priori validation using a DNS database and a-posteriori validation by LES of Sandia Flame D are performed. The results show the capacity of the new model to capture different flame regions and the efficiency to predict turbulent combustion. contact: [email protected] A dynamically adaptive combustion framework for the general description of complex flame configurations H. Wu, Y. See, Q. Wang and M. Ihme Stanford University, USA. A new combustion modeling framework is developed to dynamically adapt the local fidelity of the combustion model for reacting flows by combining a hierarchy of combustion models with different fidelity. The adaptation is achieved by dynamically assigning a combustion model under consideration of their accuracy and computational cost. The usage of each model is confined to the trust region whose size is specified by the user. The choice of combustion model is determined adaptively such that lower dimension manifold is used when it is deemed sufficiently accurate. To achieve this, a manifold assessment method for chemistry is utilized to estimate the model prediction accuracy of a given quantity of interest. This fidelity-adaptive model is applied to a triple flame to demonstrate its capability and the model performance is assessed through direct comparisons against a detailed numerical simulation. contact: [email protected] CP5 (2/6) Tuesday April 21st (10h00 – 12h00) Spectral model of premixed turbulent flame S. Rashkovskiy Institute for Problems in Mechanics of the Russian Academy of Sciences, Russia. We consider the hierarchical model of premixed turbulent flame and derive an equation which describes the propagation of curved premixed flame front in turbulent flow. We show that in space of wavenumber of flame disturbances it is possible to formulate the Cauchy problem and calculate the structure of turbulent premixed flame and its propagation velocity. We discuss the numerical method of solution of obtained equation and compare the results of simulation with some analytical solutions of this equation. contact: [email protected] Analysis of approximate deconvolution strategies and explicit filtering for LES of turbulent flames P. Domingo and L. Vervisch CORIA, CNRS & INSA de Rouen, France. Large eddy simulation of turbulent combustion using approximate deconvolution and explicit flame filtering is further discussed. In this approach, the closure of the unresolved non-linear terms relies on the explicit filtering of their exact expressions, computed after an approximate deconvolution of the resolved scalars. Various strategies may be envisioned to perform this deconvolution, which provides and approximation of the unresolved subgrid scale fluctuations in LES. Using DNS of a jet flame imbedded in LES, approximate deconvolution based on a series of approaches, as inverse discrete filtering, analytical inverse filtering or tabulated inverse filtering with detailed chemistry, are examined. contact: [email protected] Chemical source term modeling in LES of reacting flows using deconvolution method Q. Wang and M. Ihme Stanford University, USA. In this study, an LES model for chemical source term using deconvolution techniques is proposed. The model recovers the sub-grid scale scalars that are discarded by the LES filter from the information retained in the large-scale structures using a Wiener filter. However, like other linear deconvolution operations being used in literature, the Wiener filter has a limitation in constraining the physical bound of variables. To counter this limitation, a constrained minimum mean-square 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP5: Turbulent combustion optimization problem is solved locally. No additional assumptions and relaxation terms are required in the current model. Both a priori and a posteriori studies will be conducted. 57 turbulence fluctuations in the flame front and changes the turbulent flame thickness. contact: [email protected] contact: [email protected] Radiative properties of molecular combustion products and of some hydrocarbons: state of the art The framework of multiple mapping conditioning, its applications and prospects A. Klimenko1, M. Cleary2, B. Sundaram1 and L. Dialameh1 1 2 The University of Queensland, Australia. The University of Sydney, Australia. The framework of Multiple Mapping Conditioning (MMC) has undergone a noticeable evolution since the original MMC was suggested by Klimenko and Pope in 2003. The MMC approach evolved from deterministic to stochastic and generalized formulations of the model. In this more general interpretation, MMC is a PDF approach that allows for enforcing a set desired properties on mixing operations by conditioning this operator on MMC reference variables. In this form MMC is not a specific model but a framework introducing a set of principles that allow us to improve modeling through enforcing desired conditional properties on mixing. This understanding of MMC has resulted in a number of specific models including sparse-Lagrangian FDF, Shadow Position Mixing Model and MMC Partially Stirred premixed reactor. In context of premixed flames, the MMC approach allows for modeling of both distributed reactions and flamelet regimes (as well as any regime in between these limiting cases). This presentation will discuss major features and applications of generalized MMC. contact: [email protected] Radiation effects on a turbulent premixed syngas flame S. Karimkashi1, M. Bolla1, H. Wang1, E.R. Hawkes1, H.G. Im2, P.G.. Arias2 and M. Talei3 1 School of Mechanical and Manufacturing Engineering, University of New South Wales, Australia. 2 Clean Combustion Research Center, King Abdullah University of Science and Technology, Saudi Arabia. 3 Department of Mechanical Engineering, University of Melbourne, Australia. In this study, the effect of radiation on a freely propagating turbulent premixed syngas flame is investigated using two-dimensional direct numerical simulation (DNS). Both adiabatic and non-adiabatic (with radiation) cases are examined and compared. The radiative transfer equation (RTE) has been solved using the discrete ordinate method (DOM) and the radiation properties of the medium are assumed to be grey. The DNS results demonstrate the role of radiation absorption affecting the temperature and species concentrations. Furthermore, radiation moderates P. Rivière and A. Soufiani Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. We present in this communication the state of the art concerning radiative properties of light molecular species that appear either as final products or as small hydrocarbons in combustion applications. Discussion will start with the available up to date spectroscopic databases and their limitations; then various approximate models will be introduced and their accuracy studied in the context of combustion. The last part of the presentation will address the overlapping between different species absorption bands in order to highlight for instance the possibility of preheating of fresh gases by radiation from combustion products. contact: [email protected] CP5 (3/6) Tuesday April 21st (13h30 – 15h30) An a priori DNS study of molecular mixing models in a planar stationary premixed flame X. Zhao1, H. Kolla2 and J. Chen2 1 2 University of Connecticut, USA. Sandia National Laboratories, USA. A statistically stationary planar turbulent premixed methane flame is simulated using direct numerical simulations (DNS). The turbulent burning velocity of the flame is calculated as a volume integral of the fuel consumption rate across the flame, and through a feedback loop, maintains the mean inflow velocity to enforce stationarity. The DNS study includes parametric variations in Damköhler number and Reynolds number. Lagrangian tracer particles are advected along with the flow field in order to aggregate Lagrangian statistics. The statistics are then used to perform an a priori assessment of molecular mixing models in the context of transported probability density function methods. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 58 CP5: Turbulent combustion 3D DNS of methane-air turbulent premixed planar flame in thin reaction zones B. Yenerdag, Y. Naka, M. Shimura and M. Tanahashi Direct numerical simulation of spontaneous flame propagation under reactivity controlled compression ignition conditions A. Bhagatwala1, R. Sankaran2, S. Kokjohn3 and J. Chen1 Tokyo Institute of Technology, Japan. Three-dimensional direct numerical simulation of methane-air turbulent premixed planar flame propagating in homogenous isotropic turbulence is conducted to investigate local flame structure in thin reaction zones. A detailed kinetic mechanism (GRIMech 3.0) which includes 53 reactive species and 325 elementary reactions is used to represent methane-air reactions. For a better understanding of the local flame structure in thin reaction zones, distributions of mass fractions of major species, heat release rate and temperature are investigated. To clarify effects of turbulence on the local flame structure, the statistical characteristics of flame elements are also revealed. contact: [email protected] A peta-scale direct numerical simulation study on pollutant formation in turbulent premixed flames P. Trisjono and H. Pitsch 1 2 3 Sandia National Laboratories, USA. ORNL, USA. University of Wisconsin, USA. Auto-ignition and spontaneous flame propagation in a primary reference fuel mixture consisting of n-heptane and iso-octane at reactivity controlled compression ignition (RCCI) conditions were studied through direct numerical simulation. The simulations employed a novel mass source/sink term to replicate the effects of compression heating. The effect of temperature and fuel concentration gradients on the combustion mode was investigated. The effect of varying bulk temperature and the shape of the stratification were also simulated. It was found that higher n-heptane concentration and higher level of thermal stratification resulted in a greater degree of flame propagation, whereas lower nheptane concentration and higher pressure resulted in more prevalent autoignition. contact: [email protected] Institut für Technische Verbrennung, RWTH Aachen University, Germany. The development of high-fidelity modeling strategies for pollutant formation in turbulent premixed flames is often severely impeded by the lack of reliable data. In this work, a new direct numerical simulation of a lean turbulent methane-air flame at a jet Reynolds number of Re=9000 including complex NOx and CO pollutant formation chemistry is presented. The simulation configuration features a temporally evolving premixed jet flame that is discretized on almost 3 billion grid cells and the chemistry is described by a detailed chemical mechanism consisting of 32 species. This makes this DNS novel with respect to the level of detail of the interaction of turbulence and pollutant formation, and a powerful database for the understanding and modeling of NOx and CO. There are three objectives for the present talk. First, the computational approach and the numerical setup are presented. In this study, the highorder finite difference inhouse code CIAO is utilized, which features excellent primary and secondary conservation and scales up to 100.000 cores. Furthermore, the quality of the DNS database is discussed, comparing characteristic quantities of the DNS database against canonical DNS quality criteria and demonstrating that the presented DNS meets these criteria. Finally, statistics of formation rates of NOx and CO are analyzed and assessed in the context of modeling strategies for large-eddy simulations. Direct numerical simulation of a turbulent lifted DME jet flame in a heated coflow at elevated pressure Y. Minamoto and J. Chen Combustion Research Facility, Sandia National Laboratories, USA. Direct numerical simulation (DNS) of a threedimensional turbulent lifted dimethyl ether (DME) jet flame in a heated coflow is performed at 5 atmospheres. The jet Reynolds number is 11,700 and and a reduced chemical mechanism that is comprised of 30 species is employed to include the effects of negative temperature coefficient and low-temperature heat release on the flame dynamics. The near field of the jet including the lifted flame stabilization is investigated based on flame structure, propagation, and chemical analysis. The interaction between premixed flame propagation into a partially oxidized mixture of low-temperature autoignition intermediate species and the underlying jet coherent structures is also discussed. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP5: Turbulent combustion 59 Modal analysis of reacting jet in cross flow using direct numerical simulations LES combustion modelling of the dynamics of turbulent pulsed premixed flames T. Sayadi1, S. Lyra2, C. Hamman3, H. Kolla2, P. Schmid4 and J. Chen2 A. Chatelier1, R. Mercier2, N. Bertier3 and B. Fiorina2 1 University of Illinois at Urbana-Champaign, USA. 2 Combustion Research Facility, Sandia National Laboratories, USA. 3 Center for Turbulence Research, Stanford University, USA. 4 Department of Mathematics, Imperial College London, UK. Three-dimensional direct numerical simulations of a turbulent transverse jet in cross flow (JICF) are performed in order to study the influence of heat release on the jet shear layer instability. The numerical simulations consist of a H2/He jet of Reynolds number 2420, in vitiated cross-flow of methane combustion products, at a channel Reynolds number of 9840. The dominant modes of the system are then extracted using Dynamic Mode Decomposition (DMD). The DMD algorithm employs a parallel QR-factorization, which is further used as a basis of the underlying parallel SVD leading to the modal decomposition. The resulting parallel algorithm scales well on machines with a large number of processors and, therefore, allows the decomposition of very large data-sets (4.7 billion grid points). 1 ONERA / EM2C, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. 3 ONERA, France. 2 This work presents numerical investigations of turbulent flame dynamics. The simulations use a recent formulation of the model F-TACLES (Filtered TAbulated Chemistry for LES) to address flame front resolution issues. For that purpose, two filter sizes associated to the flow and the flame front, respectively, are introduced in the filtered progress variable balance equation. This new formalism designed to improve resolved flame turbulence interactions, is tested on a pulsed swirled flame stabilized in a non-adiabatic confined combustor. Comparison between experimental measurement and numerical prediction of flame transfer functions is presented. contact: [email protected] DNS of high turbulence intensity premixed methane-air flames in a 3D inflow-outflow configuration G. Nivarti and S. Cant contact: [email protected] University of Cambridge, UK. CP5 (4/6) st Tuesday April 21 (16h00 – 18h00) Turbulent diffusion flame modelling in large eddy simulation (LES) F. Shum-Kivan, B. Cuenot and E. Riber CERFACS, France. Diffusion flames appear in many practical systems and are particularly present in spray combustion. In the context of LES, the Thickened Flame approach initially developed for premixed combustion is here adapted to diffusion flames. A thickening factor based on the scalar dissipation rate is introduced, and illustrated on 1D flame calculations, using either reduced or detailed chemistry. To retrieve the correct flame wrinkling, an efficiency function is derived from Direct Numerical Simulation of a 3D reacting mixing layer. Finally, a posteriori validation is presented with LES simulations of the reacting mixing layer. contact: [email protected] Turbulent flame speed is central to several theories of turbulent flame propagation, but its experimentally observed variation with turbulence intensity remains unexplained. Until recently, DNS of turbulent reactive flows has been restricted to moderate turbulence intensities due to the prohibitive computational expense involved at higher intensities. We have conducted a parametric DNS study of premixed methane-air flames interacting with intense turbulence by varying the inflow turbulence intensity in a 3D inflow-outflow configuration. The bending curve observed through our simulations has been compared with recent experimental results and the underlying mechanisms have been investigated. contact: [email protected] Numerical simulation of NOx formation in syngas combustion H. Watanabe1, T. Kawai2 and S. Ahn3 1 Kyushu University, Japan. The University of Tokyo, Japan. 3 CRIEPI, Japan. 2 Characteristics of NOx formation in syngas combustion under O2/CO2 combustion condition is numerically investigated by means of homogeneous reactor model of CHEMKIN, a direct numerical simulation (DNS) with Arrhenius formulation (ARF) and DNS with the flamelet/progress variable approach (FPV) in comparison to air combustion condition. Results show 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 60 CP5: Turbulent combustion that the major pathways for NO formation drastically change with O2/CO2 combustion condition in the zerodimensional computation. The characteristics shown in the zero-dimensional computation is also confirmed in ARF. It is also revealed that the general feature of syngas combustion and the characteristics of NO formation can be precisely captured by FPV. CP5 (5/6) Wednesday April 22nd (13h30 – 15h30) Numerical and experimental analysis of laminar and turbulent oxy-fuel jet flames using a direct comparison of the Rayleigh signal F. Hunger1, M. Zulkifli2, B. Williams2, F. Beyrau3 and C. Hasse1 contact: [email protected] DNS and experiments of H2/He jets in a vitiated turbulent crossflow S. Lyra1, B. Wilde2, H. Kolla3, T. Lieuwen2 and J. Chen3 1 Combustion Research Facility, Sandia National Laboratories, USA. School of Mechanical Engineering, Georgia Institute of Technology, USA. 3 Combustion Research Facility, Sandia National Laboratories, USA. 2 Results are presented from a joint experimental and numerical study of H2/He jet injected into a turbulent, vitiated crossflow of lean methane combustion products. High-speed stereoscopic PIV and OH PLIF measurements are obtained alongside 3D DNS of inert and reacting JICF with detailed H2/CO chemistry. Under the investigated conditions a burner-attached flame initiates uniformly around the nozzle. Significant asymmetry is observed between the reaction zones located on the windward and leeward sides, due to the substantially different scalar dissipation rates. Vorticity spectra extracted from the windward shear layer reveal that the reacting jet is globally unstable. 1 2 3 Laminar and turbulent oxy-fuel jet flames are investigated both experimentally and numerically with emphasis on the direct comparison of the Rayleigh signal. The Rayleigh signal was measured for both flame conditions, especially correcting for background light. Equivalently, the signal was processed numerically based on the numerical species data and the temperature. The laminar flame was used for validating the procedure of processing the Rayleigh signal by using the actual species and temperature data. In the turbulent simulation, the Rayleigh signal was incorporated in the flamelet/progress variable approach in order to obtain the Favre-filtered Rayleigh signal which is of non-linear nature. contact: [email protected] Fuel effects on bluff-body stabilized turbulent premixed flames contact: [email protected] V. Katta1 and W. Roquemore2 1 LES flamelet-progress variable modeling of a turbulent piloted dimethyl ether flame S. Popp1, F. Hunger1, S. Hartl1, D. Messig1, B. Coriton2, J. Frank2, F. Fuest3 and C. Hasse1 1 Chair of Numerical Thermo-Fluid Dynamics, ZIK Virtuhcon, Technische Universität Bergakademie Freiberg, Germany. 2 Combustion Research Facility, Sandia National Laboratories, USA. 3 Department of Mechanical and Aerospace Engineering, The Ohio State University, USA. Considering complex fuels, e.g. dimethyl-ether (DME), is a current topic within the field of turbulent combustion. Recent experimental investigations showed significant differences in the flame structure when using DME instead of methane. The scope of this work is the analysis of the flame structure of a piloted DME-flame, based on the Sydney/Sandia piloted methane flame series, using a flamelet-progress variable approach within an LES framework. Numerical results are compared to a comprehensive set of experimental data, including velocity, temperature and species measurements. Furthermore, comparison of computed and directly measured LIF and Rayleigh signals is proposed. TU Bergakademie Freiberg, Germany. Imperial College London, UK. Otto von Guericke Universität Magdeburg, Germany. 2 Innovative Scientific Solutions Inc, USA. Air Force Research Laboratory, USA. Previous experiments and simulations for a bluff-body stabilized, turbulent lean premixed CH4/air flames revealed that atom balances are not conserved across the flame brush going from reactants to products. It was found that the imbalance in C/H ratio results from preferential diffusion. It is important to know whether this imbalance is particular to methane fuel or also occurs for aviation fuels such as JP8. In the present paper, reacting flows for the bluff-body burner with various fuel mixtures are simulated using a detailed CFD code and investigated the fuel effects on C/Hatom-ratio distributions. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP5: Turbulent combustion Preferential diffusion effects in LES of mild combustion with Flamelet Generated Manifolds S. Abtahizadeh, R. Bastiaans, J. van Oijen and P. de Goey Eindhoven University of Technology, The Netherlands. Preferential diffusion effects are investigated in flame stabilization of turbulent lifted flames using LES with a FGM-PDF model. The experimental test case is the Delft JHC burner in which methane based fuel has been enriched with various amounts of H2. A novel numerical model is proposed based on the Flamelet Generated Manifolds (FGM) to account for preferential diffusion effects in autoignition. Afterwards, this model is implemented in LES of the H2 enriched turbulent lifted flames. Main features of these turbulent lifted flames such as the formation of ignition kernels and stabilization mechanisms are analysed and compared with the measurements. contact: [email protected] Resolved flame LES of the Cambridge stratified burner using PFGM F. Proch and A. Kempf University of Duisburg-Essen, Germany. Large eddy simulation results are presented for the Cambridge stratified burner. The grid resolution is 100 µm, which is sufficient to resolve the progress variable field for the applied premixed flamelet generated manifolds approach without any kind of thickening, filtering or tracking of the flame. To validate the simulation, mean and rms radial profiles are compared against the available measurement data. Subsequently, an a-priori analysis of the progress variable- and velocity fields is performed to extract the related subfilter contributions for different filter widths. In addition, JPDFs of equivalence ratio and progress variable from simulation and experiment are compared. contact: [email protected] Large-eddy simulation/probability density function modelling of Cambridge stratified flame H. Turkeri and M. Muradoglu Department of Mechanical Engineering, Koc University, Turkey. 61 time and space. For detailed chemistry representation of the flame, an augmented reduced mechanism (ARM2) for methane oxidation, which involves 19 species and 15 reactions (including NO chemistry) is incorporated into LES/PDF calculations using ISAT algorithm. The results are in good agreement with the experimental data. contact: [email protected] Challenging modeling strategies for LES of nonadiabatic turbulent stratified combustion B. Fiorina1, R. Mercier1, G. Kuenne2, A. Ketelheun2, A. Advic2, J. Janicka2, D. Geyer2, A. Dreizler2, E. Alenius3 C. Duwig3, P. Trisjono4, K. Kleinheinz4, H. Pitsch4, S. Kang5, F. Proch6, F. Cavallo-Marincola6, A. Kempf6 1 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. TU Darmstadt, Germany. 3 Lund University, Sweden. 4 Sogang University Korea. 5 Lund University, Sweden. 6 Universität Duisburg-Essen, Germany. 2 This work provides a state-of-the-art picture of a selection of computational approaches for LES of stratified flame. Five simulations, which differ by modeling approach, CFD code, combustion chemistry, numerical techniques, computational meshes and user, are presented. Each of these computational strategies is designed to capture the filtered turbulent flame propagation speed. In addition, as the models account for non-adiabatic effects on the combustion chemistry, quenching phenomena induced by heat losses are captured. Encouraging similarities between the computations and the experiments are observed. It creates some confidence that these modeling approaches are leading in the right direction. contact: [email protected] CP5 (6/6) Wednesday April 22nd (16h00 – 18h00) Effects of preheat temperatures and turbulence intensities on the disruption of the reaction zone in high Karlovitz premixed flames S. Lapointe and G. Blanquart California Institute of Technology, USA. In this study, large-eddy simulation (LES)/probability density function (PDF) method is applied to Cambridge stratified flame in order to ascertain its ability to calculate the details of the effects of the stratification. In LES, large-scale motions are explicitly computed, whereas the effects of subgrid scale (SGS) motions on the large-scales are modelled. The interactions between turbulence and chemistry are modelled by the PDF method. LES/PDF equations are solved by OpenFOAM/HPDF code which is second-order in both Direct numerical simulations of premixed n-heptane/air flames at Karlovitz numbers above 200 are performed using detailed chemistry. Different unburnt temperatures and turbulence intensities are used and their effects on the flame structure are investigated. As the unburnt gases are preheated, the viscosity ratio across the flame is reduced and the effective Karlovitz number at the reaction zone is increased. Both the increase in the unburnt temperature and turbulence 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 62 CP5: Turbulent combustion intensity are found to lead to increased fluctuations in the local fuel consumption rates. A new criterion for the transition from the thin to broken reaction zones is proposed. contact: [email protected] LES/FDF of a lean premixed methane counterflow flame M. Rieth1, J. Chen2, F. Proch1, P. Lindstedt3 and A. Kempf1 1 Universität Duisburg-Essen, Germany. University of California at Berkeley, USA. 3 Imperial College London, UK. 2 Analysis of natural gas fired furnaces with different oxygen enrichments R. Prieler1, M. Demuth2 and C. Hochenauer1 1 2 Graz University of Technology, Austria. Messer Austria GmbH, Austria. A CFD modelling that integrates a steady flamelet model with a detailed chemical mechanism is applied to a natural gas fired lab-scale furnace under different oxygen enrichments. Results are compared to measured temperatures and heat fluxes inside the furnace. A skeletal mechanism which considered 25 reversible reactions and 17 species showed very good agreement with the measurement for all oxygen concentrations in the oxidizer with marginal differences in flame length and maximum temperatures. These results are also compared with those obtained by means of eddy dissipation concept (EDC). Both approaches calculated similar results for temperature and species concentrations. contact: [email protected] Lean premixed methane turbulent opposed jet flames in fractal turbulence with different equivalence ratios, experimentally investigated at Imperial College London, have been simulated with a large eddy simulation and transported filtered density function (LES/FDF) method. The LES/FDF are performed massively parallel with around 10 million particles on a BlueGene machine to study lean premixed combustion approaching extinction. The (conditional) statistics of velocity and progress variable are compared to the experimental data allowing for the assessment of turbulence-chemistry interaction as well as the overall performance of the LES/FDF predicting a turbulent premixed flame. contact: [email protected] Sensitivity of internal flow dynamics and boundary conditions on a turbulent opposed jet flame configuration A. Ruiz, G. Lacaze, B. Coriton, J. Frank and J. Oefelein Implicit large-eddy simulation of a Bunsen-like burner with multi-scale turbulent forcing S. Zhao1, N. Lardjane2 and I. Fedioun1 1 2 ICARE-CNRS, France. CEA-DAM-DIF, France. A methane-air Bunsen-like burner with multi-scale turbulent forcing is simulated with a 5th order WENO scheme at high resolution. A similar experiment is conducted at ICARE, in which the forcing is achieved using a system of three differently perforated grids. In the simulation, the experimental 1D spectrum is taken as an input for the generation of the turbulent flow field to be injected at the inlet. An original characteristic boundary condition has been derived to apply this prescribed turbulent inlet. The non reacting simulation matches quite well the experimental data. Reacting cases are in progress. Sandia National Labs, USA. Effects of internal flow dynamics and boundary conditions on a turbulent opposed jet flame configuration are analyzed using the Large Eddy Simulation technique. The geometry corresponds to that used in experiments by Coriton and Frank for studies of turbulence-chemistry interactions in canonically generated flames. Specifying accurate boundary conditions for simulations is challenging without resolving the internal flow in the nozzles. Results presented will provide a systematic characterization of internal nozzle flow interactions with emphasis on providing optimal flow conditions, detailed turbulence characteristics at respective nozzle exits, and an improved understanding of the effects these phenomena have on various flame structures. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP6: Coal combustion 63 CP6: Coal combustion Monday April 20th (16h00 – 18h00) Adaptive kinetic model for coal devolatilization in oxy-coal combustion conditions S. Iavarone1, C. Galletti2, F. Contino3, L. Tognotti2, A. Parente1 and P. Smith4 1 Aero-Thermo-Mechanical Department, Université Libre de Bruxelles, Belgium. 2 Department of Civil and Industrial Engineering, University of Pisa, Italy. 3 Department of Mechanical Engineering, Vrije Universiteit Brussel, Belgium. 4 Institute for Clean and Secure Energy, University of Utah, USA. The oxy-combustion is emerging as the most likely low-cost “clean coal” technology for both NOx and SOx emissions reduction and carbon capture. The use of CFD tools is crucial for cost-effective oxy-fuel technologies development and environmental concerns minimization. The coupling of detailed chemistry and CFD simulations is still prohibitive and the improvement of simple but reliable kinetic mechanisms is therefore necessary. This work presents a CFD study on coal pyrolysis, regarding implementation, validation and uncertainty quantification of a reduced adaptive kinetic model, able to predict the total volatile and tar yields, depending on heating rate, temperature, pressure, and coal type. contact: [email protected] A modified two-step model for devolatilization A. Richards and T. Fletcher Brigham Young University, USA. Although complex models of coal pyrolysis have been developed based on chemical structure, simple but accurate models are currently needed for incorporation into large boiler simulation models. Many investigators fit coefficients for a simple model to match predictions of the complex model over a narrow range of heating rate or temperature. In this paper, a modified two-step model for devolatilization is presented. It is used by combining the original two-step model with a corrective function developed as part of the Yamamoto one-step model. The original two-step model works well for predicting yield values of total volatiles and tar for one heating rate, but is not as accurate if used over a wider range of heating rates, meaning a new set of parameters must be generated for each new heating rate. Combining the form of the original two-step model with the corrective function from the Yamamoto model gives the modified two-step model the predictive capabilities of the original two-step model, but also the accuracies inherent in the Yamamoto model compared to the CPD model. The results for different types of coal are presented here. The modified two-step model is shown to be a very accurate model for calculating yields in the devolatilization process over a wide range of heating rates (5000 K/s to 1 million K/s). This gives the model flexibility to model yields at any heating rate inside that range. contact: [email protected] A consistent approach for coupling the devolatilization of coal particles with tabulated chemistry R. Knappstein1, A. Ketelheun1, G. Künne1, J. Köser2, A. Dreizler2, A. Sadiki1 and J. Janicka1 1 Institute of Energy and Power Plant Technology, TU Darmstadt, Germany. 2 Institute of Reactive Flows and Diagnostics, TU Darmstadt, Germany. Within this work the predictive capability of a FGM combustion modelling strategy coupled with coal devolatilization models is investigated. First, the coupling of the devolatilizing coal particle and the finite volume scheme is addressed. The verification assessing the consistency regarding mass, species and enthalpy conservation combined with the thermo-chemical states extracted from the three-dimensional chemistry table is outlined. Second, measurements will be used for validation of this approach in a quasi-one-dimensional configuration. At this, coal particles cross a laminar flat flame whose high temperature causes the volatile gases to leave the particle and react within their flammability limits. contact: [email protected] Macropore growth in char particles under different gasification conditions K. Wittig1, A. Richter1, M. Kestel1, P. Nikrityuk2 and B. Meyer1 1 2 TU Bergakademie Freiberg, Germany. University of Alberta, Canada. Macropore growth within a single char particle under different gasification conditions is studied numerically. Therefore, a numerical reconstruction method for a porous particle surface using an adaptive octree-based subdivision is applied. CFD calculations of single reacting particles were carried out in order to provide the temperature and species profiles across the particle. This approach allows for the pore growth study in several regimes. Developments of porosity and specific surface area over carbon consumption are compared for different points in an entrained-flow reactor. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 64 CP6: Coal combustion 1 2 How does turbulent fluid motion influence heterogeneous combustion? Application of a skeletal kinetic mechanism on LES of a pulverized coal jet flame J. Krüger1, N. Haugen2 and T. Løvås1 S. Ahn1, H. Watanabe2, K. Tanno1 and N. Hashimoto1 1 EPT, NTNU, Norway. Sintef Energy AS, Norway. 2 A reactive-particle conversion model, taking into account internal gradients of concentration, is integrated into the open-source DNS code called "The Pencil code". Char particle conversion at constant pressure in isotropic turbulence with zero mean flow is studied, using GRI 3.0 and a detailed 26-step mechanism to model the homogeneous and heterogeneous reactions, respectively. Mass, species and energy interaction between the solid and the fluid phases are accounted for. The focus of the study is on the effect of turbulent scale and intensity on the reaction rate of the char. This will be used to develop a turbulent combustion/ gasification model for heterogeneous reactions. CRIEPI, Japan. Kyushu university, Japan. A numerical calculation of a pulverized coal jet flame has been performed using large eddy simulation (LES) technique to analyze the reaction field in detail. A reduced skeletal mechanism generated from a detailed kinetic mechanism through a reasonable reduction process is applied in this calculation for more detailed analysis. The result is verified through a comparison with experimental results conducted in CRIEPI. Important characteristics in coal combustion such as distribution of temperature and gases, particle size and behavior, flow stream are analyzed and discussed in this work. contact: [email protected] contact: [email protected] CP7: Rocket engines, ramjets, scramjets Monday April 20th (16h00 – 18h00) Three-dimensional structure of hypergolic ignition process for hydrazine/nitrogen dioxide un-like doublet impinging gas jets Y. Daimon1, H. Tani1, H. Terashima2 and M. Koshi3 1 Japan Aerospace Exploration Agency, Japan. University of Tokyo, Japan. 3 Yokohama National University, Japan. 2 The understanding of hypergolic fuel ignition process is very important to develop the bipropellant liquid rocket engine. In this work, hydrazine (N2H4)/nitrogen dioxide (NO2) un-like doublet impinging gas jets is simulated to explore the hypergolic ignition processes in N2H4/N2O4 bipropellant liquid rocket engines. The Navier-Stokes equations with a detailed chemical kinetics mechanism are solved to reveal the influence of the chemical reaction, i.e. hydrogen abstraction by NO2. We discuss the differences of three-dimensional structure of hypergolic ignition process between 400 and 600 K of gas temperature. contact: [email protected] Computational modeling of boron particle fueled ducted rocket combustion chambers B. Kalpakli and E. Acar ROKETSAN Missile Ind., Turkey. A detailed computational model for boron particle combustion in a reactive and variable composition atmosphere is developed. The main application area of the model is Computational Fluid Dynamics (CFD) based simulations of solid propellant ramjet combustion chambers and it is aimed to construct a combustion model for boron containing solid propellants. This model includes most of the physical processes required to define a high fidelity particle combustion simulation. Combustion model consists of surface and gas phase reactions along with phase change processes including evaporation and boiling. The reaction rate modeling of some similar studies are also improved for ramjet combustion chambers. contact: [email protected] Numerical simulation of turbulent combustion between a hydrogen jet and a supersonic vitiated air crossflow A. Techer, G. Lehnasch and A. Mura Institut Pprime, UPR 3346 CNRS, ISAE-ENSMA, Poitiers, France. The future of high-speed transportation systems depends on the development of hypersonic air-breathing engines. In such conditions, the flow entering the combustor is maintained supersonic to reduce the excessive heating and dissociation of air. The combustor operation therefore concentrates important difficulties, which dictate the propulsive system development. We report LES of an under-expanded hydrogen jet released into a turbulent cross-flow of vitiated air. This configuration is representative of practical conditions, encountered for instance in the dual mode ramjet combustor of the ONERA-LAERTE 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP7: Rocket engines, ramjets, scramjets facility. We investigate the turbulent mixing and combustion processes downstream of the underexpanded jet for various operating conditions. contact: [email protected] Numerical investigation of transverse hydrogen jet into supersonic crossflow using detached-eddy simulation L. Zhaoyang, W. Zenguo, S. Mingbo and W. Hongbo National University of Defense Technology, China. Three-dimensional unsteady reacting flowfield that is generated by transverse hydrogen injection into a supersonic mainstream is numerically investigated using detached-eddy simulation and a assumed sub-grid PDF model. The inviscid flux of Naver-Stokes is discretized by ninth-order WENO scheme and the viscous flux is discretized by sixth-order center difference scheme. The temporal discretization is done by means of implicit LU-SGS method. Validation is performed for the jet penetration height, and the predicted result is in good agreement with experimental trends. The results indicate that jet vortical structures are generated as the interacting counter-rotating vortices become alternately detached in the upstream recirculation region. Although the numerical OH distribution reproduces the experimental OH planarlaser-induced fluorescence well, there are some disparities in the ignition delay times due to the restricted availability of experimental and numerical data. contact: [email protected] Numerical study of flame dynamic characteristics in a supersonic combustor with dual cavity Y. Yixin, W. Hoogbo, W. Zhenguo and S. Mingbo 65 process of flame stabilization in a supersonic combustor with tandem and parallel dual-cavity. The clipped Gaussian PDF of temperature and multivariate β-PDF of composition were implemented to calculate the subgrid chemical sources in the conservation equations. The sub-grid variances of temperature and composition were constructed based on the scale similarity approach. Simulation reproduced the significant movement and transformation of the flame structure during the stabilization process, which were observed experimentally. contact: [email protected] Numerical simulation on detonation initiation and propagation in supersonic combustible mixtures with nonuniform velocities and species X. Cai1, J. Liang1, Z. Lin1, R. Deiterding2 and Y. Che1 1 2 National University of Defense Technology, China. DLR, Germany. Detonation initiation and propagation using a hot jet in supersonic combustible mixtures with nonuniform velocities and species is investigated through highresolution simulations. For nonuniform velocities, when the velocity in the upper half channel V2=0.5Vcj, the Mach reflection on the upper wall results in the formation of a new Mach reflection near the interface, which is a locally normal detonation wave. The continuous collisions between the lower wall and the upper triple point play an important role in the eventual formation of a dynamic stable structures. When V2 is increased to 0.6Vcj, 0.7Vcj, 0.8Vcj and 0.9Vcj, the initiation process becomes more and more similar with that in uniform supersonic combustible mixtures. For nonuniform species, it makes a big difference when exchanging the locations of the species in the upper half channel and the lower half channel. contact: [email protected] National University of Defense Technology, China. A hybrid LES/assumed sub-grid PDF closure model was carried out to investigate the spatio-temporal 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 66 CP8: Ignition / quenching CP8: Ignition / quenching CP8 (1/2) The flame consumption speed – comparison of theory and numerical simulations Tuesday April 21st (10h00 – 12h00) Direct numerical simulations of kernel growth of turbulent premixed syngas/air flames 1 2 3 G. Giannakopoulos , C. Frouzakis , M. Matalon and A. Tomboulides1 1 2 3 University of Western Macedonia, Greece. Swiss Federal Institute of Technology, Switzerland. University of Illinois Urbana-Champaign, USA. The early kernel development of spherical expanding syngas/air flames is studied using parametric direct numerical simulations in homogeneous isotropic turbulent flow fields with detailed chemistry and transport. The effects of mixture composition (equivalence ratio and CO/H2 ratio) and turbulence parameters on the flame dynamics are investigated in circular (2D) and spherical (3D) domains. Emphasis is given on the local flame structure, the propagation characteristics and on the proper definition of flame speed that describes the growth rate of spherical turbulent flames. G. Giannakopoulos1, C. Frouzakis2, M. Matalon3 and A. Tomboulides1 1 Department of Mechanical Engineering, University of Western Macedonia, Greece. 2 Aerothermochemistry and Combustion Systems Laboratory, ETH Zurich, Switzerland. 3 Department of Mechanical Science and Engineering, University of Illinois, USA. The notion of “flame consumption speed", conveniently defined as the rate of fuel consumption in a premixed flame, is discussed and results from the asymptotic theory are compared to direct numerical simulations for planar and spherically expanding flames. Differences between "flame displacement speed" and "flame consumption speed" are identified and quantified, both theoretically and numerically. The discussion also includes the possible definition of a Markstein length based on the dependence of flame consumption speed on flame stretch rate, and its relation to the more common definition of Markstein length. contact: [email protected] contact: [email protected] A statistical model to predict ignition probability LES analysis of the influence of the turbulence intensity on the critical ignition energy of a lean methane/air flame S. Mouriaux1, O. Colin1, B. Renou2 and D. Veynante3 1 IFP Energies Nouvelles, France. CORIA, CNRS and INSA de Rouen, France. 3 EM2C, CNRS and Ecole Centrale Paris, France. 2 A key parameter to control the success of ignition in Spark Ignition (SI) engines is the Minimum Ignition Energy (MIE), minimum energy required to ensure the growth and the propagation of the flame kernel. Experimental studies by Cardin et al. highlighted two regimes for the MIE depending on the Karlovitz number. This work proposes quasi-DNS simulations of ignition in a Turbulent Homogeneous Isotropic (THI) velocity field with the same properties as in the experiment, showing the ability of the TF-LES model to reproduce the experimental results and explaining the dependency of the MIE to the turbulent intensity. contact: [email protected] L. Esclapez, E. Riber and B. Cuenot CERFACS, France. Based on a series of non-reacting LES solutions, a model is proposed to predict ignition probability taking into account the flame kernel motion. The probability of presence of flame elements at each location is evaluated and statistics of mixture and turbulence are used to evaluate the probability of ignition success. This reduced model then allows the use of Uncertainty Quantification (UQ), to evaluate the sensitivity of ignition probability to unknown parameters related to a spark discharge. contact: [email protected] A method for predicting sensitive rate coefficients with high accuracy tested for the H2/CO/O2-system T. Methling, M. Braun-Unkhoff and U. Riedel German Aerospace Center (DLR), Institute of Combustion Technology, Germany. An evolutionary strategy and a gradient based solver are presented, for the sophisticated optimisation of chemical kinetic reaction models. The novel approach is based on new variation methods of rate coefficients and new evaluation methods of the distance between target values (experimental) and simulation values. Both solvers are tested on OH*-chemiluminescence data of 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP8: Ignition / quenching 120 shock tube experiments. For validation, the target data is simulated and random errors are added. Both solvers fit the simulation and experimental data with high precision. Furthermore, the gradient based solver is capable of predicting sensitive rate coefficients with outstanding accuracy compared to present confidence intervals. contact: [email protected] CP8 (2/2) Wednesday April 22nd (10h00 – 12h00) Direct numerical simulation of an igniting, turbulent jet with global n-heptane chemistry at enginerelevant conditions A. Krisman and E. Hawkes The University of New South Wales, Australia. A direct numerical study of a three-dimensional, turbulent, temporally evolving slot-jet ignition was conducted. The fuel was n-heptane, represented by a four-step global scheme which was selected to reproduce the two-stages of autoignition of n-heptane while permitting a feasible computational cost. The results indicate a complex ignition process wherein distinct low and high temperature auto-ignition events lead to the formation of edge-flames. A combination of propagating edge-flames and auto-ignition produce a fully-burning nonpremixed flame. The edge-flames observed are of a hybrid premixed/autoignition structure previously observed in laminar flame studies at similar thermochemical conditions. contact: [email protected] Direct numerical simulation of investigation of autoignition with split fuel injection D. Shin and E. Richardson University of Southampton, UK. Direct Numerical Simulation is used to investigate the effects of split fuel injection on autoignition. Two pulses of gaseous n-heptane are sequentially injected. The fuel issues into hot quiescent air with a jet Reynolds number of 7200. The combustion is modelled with a global four-step scheme, and additional passive scalars are transported in order to characterise mixing between fuel injected at different times. By varying the pulse duration and pulse separation times, the entrainment of air; the mixing between the two pulses; and the effect on the ignition process are investigated. contact: [email protected] 67 Effects of hydrogen added on the ignition of isooctane in a diffusion layer Z. Li and Z. Chen SKLTCS, Department of Mechanics and Engineering Science, College of Engineering, Peking University, China. Effects of hydrogen addition on the ignition in a diffusion layer between cold iso-octane and hot air is examined using numerical simulation. Detailed chemistry and transport are considered. The focus is mainly on the influence of preferential mass diffusion of hydrogen over iso-octane. The ignition delay time is found to first decrease and then increase slightly as the hydrogen blending ratio increases. The ignition enhancement caused by hydrogen addition is shown to strongly depend on the mass diffusivity of each fuel component. Moreover, the movement of the ignition front is investigated. contact: [email protected] Predicting extinction in condensed phase combustion in a counterflow geometry S. Koundinyan, D. Stewart, M. Matalon and J. Bdzil University of Illinois - Urbana Champaign, USA. We present a two-dimensional combustion model of the Titanium-Boron (Ti-B) system, which is thermodynamically consistent. For diffusion, we have combined the traditional Fick’s law with MaxwellStefan diffusion coefficients for a three species model. We have analyzed the model in counter-flow diffusion geometry and predict the required diffusion coefficients that correspond to experimentally observed combustion temperatures. The numerical code is validated using asymptotic analysis in the limit of complete combustion. We predict the extinction strain rate for the Ti-B system and compare the results between constant and non-constant density approximations. The presented analysis can be applied to other condensed phase combustion systems. contact: [email protected] Large eddy simulation of extinction limits in twodimensional plane buoyant turbulent diffusion flames S. Vilfayeau, J. White and A. Trouvé Department of Fire Protection Engineering, University of Maryland, USA. The general objective of this project is to support the development and validation of large eddy simulation (LES) models used to simulate the response of fires to the activation of a suppression system (using gaseous agents or water sprays). The configuration in this joint experimental/computational study corresponds to a methane-fueled, two-dimensional, plane, buoyancy- 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 68 CP8: Ignition / quenching driven, turbulent diffusion flame driven to extinction when exposed to increasing levels of nitrogen dilution or water mist. The study is aimed at evaluating flame extinction/re-ignition models using FDS and FireFOAM, two advanced LES-based solvers developed by the National Institute of Standards and Technology and FM Global, respectively. contact: [email protected] CP9: Two-phase flows CP9 (1/2) Tuesday April 21st (10h00 – 12h00) Counter-flow spray pulsated flames: from experimental measurements to numerical simulation Numerical study of alkane-air spray combustion J. Brändle de Motta1, M. Boileau1, L. Zimmer1, B. Robbes1, F. Laurent1 and M. Massot2 C. Nicoli1, P. Haldenwang1 and B. Denet2 1 2 Laboratoire M2P2, CNRS and Aix-Marseille Université, France. IRPHE, CNRS and Aix-Marseille Université, France. 1 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France; CNRS, Fédération de Mathématiques de l'Ecole Centrale Paris, France. 2 This numerical work concerns spray-flame propagation. Spray is schematizes by alkane droplets located at the nodes of a face-centered 2D-lattice, surrounded by gaseous alkane-air mixture. The study focuses on sprayflame speed in rich spray: a simple reduced kinetics derived for high alkane air mixtures is used. The main parameters of the study are the lattice path, the liquid loading, and the surrounding gaseous equivalence ratio. Our results agree qualitatively with the experimental observations: spray flame can propagate faster than pure premixed flames with the same overall equivalence ratio, as soon as droplet radius is large enough. The scope of the present contribution is to set up and simulate a configuration of stationary and pulsated counterflow sprays flames, where the polydisperse injected sprays are described through both Lagrangian and Eulerian multi-fluid approach coupled to a gaseous flow field in the low Mach number limit. We present detailed comparisons with experimental measurements designed for this purpose and using dedicated diagnostics for quantitive comparisons. Such configurations are intended to become reference solutions in order to validate industrial code, as well as their models and numerical methods for spray and gas resolution contact: [email protected] contact: [email protected] Spray modeling using a method of moment approach with droplet size distribution reconstruction M. Pollack, S. Salenbauch and C. Hasse Large eddy simulations of polydisperse turbulent nheptane spray flames with a dynamic ATF procedure coupled with the FGM chemistry reduction method F. Sacomano Filho1, M. Chrigui2, A. Sadiki1 and J. Janicka1 TU Bergakademie Freiberg, Germany. In order to model sprays in an Eulerian framework, the Population Balance Equation (PBE) needs to be solved. An efficient approach is based on the Quadrature Based Methods of Moments (QMOM). However, evaporation and dissapearance of the smallest droplets cannot be accounted for accurately in QMOM. To overcome this drawback, the droplet distribution is reconstructed applying the recently developed 'Extended QMOM‘ [1] for sprays. This allows a very accurate closure of these fluxes. Finally, this approach is extended to a bivariate formulation in order to include the temperature as second internal coordinate. [1] C. Yuan, F. Laurent, and R. Fox: Journal of Aerosol Science 51, 2012, 1-23. contact: [email protected] 1 Institute of Energy and Power Plant Technology, Technische Universität Darmstadt, Germany. 2 Unité de Recherche Matériaux, Energie et Energies Renouvelables, Université de Gafsa, Tunisia. Investigations of the capability of a dynamic Artificially Thickened Flame (ATF) procedure to describe the behavior of turbulent partially prevaporized spray flames in the Euler-Lagrange context are performed. Focus is given on the interactions characteristics between both phases and on the effects of the flame front propagation within the multiphase flow. The chemistry is described by the Flamelet Generated Manifold (FGM) method using a detailed reaction mechanism for n-heptane including 88 species and 387 elementary reactions. The configuration analyzed consists of a spray jet flame stabilized by a pilot flame. 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP9: Two-‐phase flows 69 Good agreement with experimental data validates the proposed approach. contact: [email protected] Toward the fully realizable large eddy simulation of disperse phase flows: Eulerian kinetic modelling and associated numerical methods M. Sabat1, A. Vié1, A. Larat1 and M. Massot2 1 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France; CNRS, Fédération de Mathématiques de l'Ecole Centrale Paris, France. 2 The Large Eddy Simulation of disperse phase flows, like droplets in aeronautical combustors, still requires further research and developments to improve predictivity as well as efficiency. Eulerian moment methods are preferred because of their intrinsic HPC ability. Here we present and validate a complete LES strategy based on the use of an anisotropic Gaussian closure for the droplet velocity distribution, a filtering at the kinetic level, and high order realizable numerical strategies to preserve the mathematical properties of the resulting modelling approach with a sufficient precision. The model is evaluated on configurations of increasing complexity that exhibit its main features. contact: [email protected] Analysis of multi-phase material model for Al/AlOx droplet K. Lee and D. Stewart University of Illinois at Urbana-Champaign, USA. We present a multi-phase model of a spherical Aluminum (Al) droplet coated with Aluminum Oxide(AlOx). The effect of temperature variations on the droplet is analyzed up to 2300K. Solid species are considered as an elastic isotropic solid and liquid species are given by modified form of Birch/Murnaghan and Fried-Howard free energy EOS. We analyze the influence of phase change in Al on AlOx at the interface and compute localized radial and tangential stress variations. A constitutive mixture relation of solid Al and liquid AlOx is considered and compared with the dual-layered spherical model. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 70 CP9: Two-‐phase flows CP9 (2/2) st Tuesday April 21 (13h30 – 15h30) Development of a turbulent liquid flux model for Eulerian-Eulerian multiphase flow simulations S. Puggelli1, A. Andreini1, C. Bianchini1, B. Facchini1 and F. Demoulin2 1 Department of Industrial Engineering (DIEF), University of Florence, Italy. 2 CORIA (COmplexe de Recherche Interprofessionnel en Aérothermochimie), Rouen, France. Present work introduces an Eulerian-Eulerian solver for multiphase flows proposed to include slip velocity effects in the framework of quasi multiphase approach. Particular attention is hence devoted to the turbulent liquid flux modelling: an innovative second order closure is derived and implemented in the Eulerian solver derived from the Eulerian Lagrangian Spray Atomization (ELSA) model. A detailed analysis of the resulting solver capabilities is performed exploiting a jet in crossflow test case at two different density ratios. The comparison with the experimental results shows that the solver leads to physically consistent improvements in the description of liquid injection development. contact: [email protected] A monotonic mixing-describing composition-space variable for the flamelet formulation of spray flames B. Franzelli1, A. Vié2 and M. Ihme1 1 2 CTR - Stanford University, USA. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. The classical mixture-fraction formulation, commonly employed as describing variable for gaseous diffusion flames, cannot be used for spray flames due to its nonmonotonicity caused by the evaporation source term in the corresponding conservation equation. By addressing this issue, a new mixing-describing variable, called effective composition, is defined to enable the analysis of spray-flame structures in composition space. This effective composition variable is employed for the derivation of a spray-flamelet formulation. The structure of laminar one-dimensional counterflow spray flames is numerically solved in the new phase space and a comparison with the physical solution is performed to demonstrate model consistency. contact: [email protected] Numerical investigation of shear-driven primary breakup of liquid fuel flows developed and applied to primary breakup of liquid fuel flows. The Robust Conservative Level Set (RCLS) method uses high-order WENO schemes on mixedelement unstructured meshes to solve the transport equation for the level-set variable. The method is implemented within the framework of OpenFOAM, and is fully parallelised. The present study uses the new capability to understand the phenomenology observed during prefilming airblast atomisation of a planar liquid sheet, for commercial aircraft application, which is essential for effective control of fuel atomisation and cleaner, more efficient combustion processes. contact: [email protected] Numerical simulations of kerosene spray atomization with various hybrid breakup models T. Liu1, L. Hu2, Y. Wu3, D. Zhao3 and G. Wang4 1 School of Engineering Science, University of Science and Technology of China, China. 2 School of Mechanical Engineering, Shanghai Jiaotong University, China. 3 Hypersonic Aircraft Research Center, China Aerodynamics Research and Development Center, China. 4 School of Aeronautics and Astronautic, Zhejiang University, China. Numerical simulations of kerosene spray are performed to evaluate the droplet size and distribution during the spray process of a nozzle using OpenFoam. Six hybrid breakup models, i.e. the combinations of two primary breakup models (the hollow-cone injector model and the Blob Sheet Atomization model) and three secondary breakup models (the KHRT, TAB and ETAB models), and LISA model are utilized to describe the whole spray process and the accuracies of their predictions are evaluated by comparing with the experimental data. The collision and coalescence of droplets are considered using the trajectory model and the wall refection model. The Sauter mean diameter (SMD) of the droplets and its distributions are adopted to evaluate those models’ capabilities to predict the spray process of the nozzle. After comparing between experimental data and predicted results and analyzing the differences of predicted results from different breakup models, we can draw a conclusion that predicted results obtained using hollow-cone injector model and KHRT model are better than those obtained using other models for this nozzle. For the primary breakup models, the Blob Sheet Atomization model underestimated the breakup rate around the nozzle’s orifice. The predictions of KHRT model are the best among three secondary breakup models. In general, the predicted results show that the secondary breakup model plays a major role on the spatial distribution of droplet diameter. contact: [email protected] C. Bilger, T. Pringuey and R. Cant University of Cambridge, UK. A novel state-of-the-art numerical capability for efficient modelling of multiphase flows has been 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP9: Two-‐phase flows 1 2 71 Compressibility effects in the initial transient of high-pressure Diesel injection Modeling and simulation of a dense evaporating spray M. Arienti1 and M. Sussman2 F. Doisneau, M. Arienti and J. Oefelein Sandia National Laboratories, USA. Sandia National Labs, USA. Florida State University, USA. We discuss the turbulent dispersion of atomized spray in response to the timed opening of a high-pressure, sub-critical, non-cavitating Diesel injector. Of interest in this study is the capability of modeling compressible flow in both gas and liquid phases while maintaining a sharp interface between the two. The internal and external flows are seamlessly calculated across the injection orifice using the Moment of Fluid method, with an embedded boundary technique to represent the injector’s walls. A realistic equation of state, experimentally calibrated at high pressure for liquid ndodecane, is implemented to relate the transient spray characteristics to the orifice opening. We compute a dense spray as it is transported by and evaporates in a turbulent reacting flow. The configuration is representative of a sub-critical highpressure reacting Diesel jet. State-of-the-art models and methods for both gas and liquid phases are coupled in a way that optimizes the load balance and the accurate resolution of the couplings, leading to a novel architecture for spray combustion parallel codes. This work aims at assessing the capabilities of a massively parallel LES solver in performing industrial simulations of sub-critical injection and combustion, which is an important tool for the design of advanced engines. contact: [email protected] contact: [email protected] CP10: New technologies CP10 (1/2) Tuesday April 21st (10h00 – 12h00) much more regular. Our model describes the flame anchoring phenomena when the combustion wave is stabilized inside of the porous media for a range of inlet velocities. contact: [email protected] A discrete model of gas combustion in porous media F. Sirotkin1, R. Fursenko2 and S. Minaev1 1 Far Eastern Federal University, Russia. Khristianovich Institute of Theoretical and Applied Mechanics, Russia. 2 A discrete quasi three-dimensional model of gas combustion in porous media is proposed. The porous media is represented as the set of randomly placed solid grains which are connected trough the artificial solid plate to mimic thermal conductivity between grains. The propagation of the combustion wave is simulated in the frame of thermal-diffusion model with the prescribed flow field computed preliminary using Lagrangian particles method. Dependencies of combustion wave velocity on the inlet gas velocity for different equivalence ratios and porosities are obtained and compared with the continuous one-dimensional thermal-diffusion model. Our results imply that the flame propagation in the porous media can be considered as a collective process, when the actual combustion wave can be represented by a set of individual flame fronts propagating in the mutually connected micro-channels of the different diameter. The discrete structure of the combustion wave is most prominent when the inlet velocities are small and the flame propagation is accomplished with pulsations of flame front fragments. When the inlet velocities are high enough, the propagation of the combustion wave is The effect of the three-dimensionality of the flame front inside a radial-flow porous burner – a detailed DPLS numerical study I. Dinkov, P. Habisreuther and H. Bockhorn Karlsruhe Institute of Technology, Engler-Bunte-Institute, Division for Combustion Technology, Germany. In the present study combustion process inside a porous inert media (PIM) has been numerically investigated using a real porous inert structure obtained from computer tomography. The results showed that the flame structure is substantially affected by the topology of the PIM at the pore level. Therefore, significant spatial fluctuations of local properties such as temperature and velocity are presented along the space and made obvious the three-dimensionality of the flame front inside the PIM. A possible way to take these spatial fluctuations into account with an assumption analogous to the Boussinesq hypothesis is suggested. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 72 CP10: New technologies Lean hydrogen flames in a narrow adiabatic channel C. Jiménez, D. Fernández-Galisteo and V. Kurdyumov CIEMAT, Spain. We present results of direct simulations of lean hydrogen-air flames propagating in planar narrow adiabatic channels, using detailed chemistry. Our simulations show that, as already predicted by numerical analysis within a thermo-diffusive approximation and Arrhenius chemical kinetics, when Le<1 double solutions, symmetric and non-symmetric, can exist and coexist for the same set of parameters. Moreover, the propagation velocities of the symmetric and the non-symmetric flames are very different, and so is the flashback critical condition. This can have important safety implications in devices designed to burn hydrogen in a lean laminar premixed way, as, for example, in micro-combustion devices contact: [email protected] A numerical investigation of heterogeneous/homogeneous combustion of ndodecane over Pt for microcombustion E. Tolmachoff, A. Booth, I. Lee, W. Allmon and C. Waits ARL, USA. For the next generation of portable chemical-toelectrical power converters, high temperature operation with liquid fuels will be critical for high efficiency and energy density. This work investigates the combustion of n-dodecane in a porous platinum-coated foam reactor. A lumped parameter model describes mass transport between the gas and the catalyst surface. Homogeneous chemistry is described by a skeletal mechanism for low and high temperature n-dodecane combustion. Surface reactions of abundant homogeneous species at the Pt surface are treated as lumped and irreversible. The resulting combined mechanism satisfactorily predicts CO2 yield from experiments for moderate to high temperatures (> 650 C). contact: [email protected] Direct numerical simulation of MILD combustion U. Göktolga, J. van Oijen and P. de Goey Eindhoven University of Technology, the Netherlands. MILD combustion is a promising concept which combines dilution and preheating, and decreases harmful emissions like NOx. It has been applied to industrial furnaces for years, but the physical phenomena occurring are yet to be revealed. In this study, 2D and 3D DNS calculations of MILD combustion in the form of auto-igniting mixing layers were performed. Effects of different parameters like heat loss and preferential diffusion were investigated. We found that the heat loss effects are crucial in predicting ignition delay and flame structure. Furthermore, we observed that the preferential diffusion effects were present even in highly turbulent environment. contact: [email protected] Large-eddy simulation of a MILD combustion burner using conditional source-term estimation J. Labahn and C. Devaud University of Waterloo, Canada. Conditional Source-term Estimation (CSE) has been successfully applied to different turbulent combustion problems. In the current study, Large Eddy Simulations (LES) of the Delft-Jet-in-Hot-Coflow burner, with a two mixture fraction CSE formulation, are performed to predict temperature and velocity profiles, and lift-off heights for two different flames, DJHC-4100 and DJHC-8100. The mean temperatures and velocities are in good agreement with the experimental data, with a slight overprediction of the centerline temperature. The predicted lift-off height for the DHJC-4100 flame is found to be lower than the experimental value, suggesting that a third conditioning variable may be needed for improved predictions. contact: [email protected] CP10 (2/2) Tuesday April 21st (13h30 – 15h30) Numerical aspects of the shockless explosion combustion P. Berndt1, R. Klein1 and C. Paschereit2 1 2 Freie Universität Berlin, Germany. Technische Universität Berlin, Germany. The shockless explosion combustion is a novel approach to constant volume combustion in gas turbines. In contrast to pulsed detonation, it aims to avoid strong shock waves by utilizing homogeneous auto ignition, similar to the HCCI process in internal combustion engines. Recharging is handled analogous to pulsejets through the pressure waves in the combustion chamber. We introduce the process in detail and familiarize the audience with the numerical challenges associated with its simulation. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP10: New technologies 73 Steady laminar one-dimensional flame solvers for modelling ions in flames Simulation of turbulent premixed combustion with a self-consistent plasma model for initiation B. Lee, M. Cha, H. Im and S. Chung H. Sitaraman and R. Grout King Abdullah University of Science and Technology, Saudi Arabia. National Renewable Energy Laboratory, USA. For predicting the amount of ions and their distribution in flames with or without external electric fields, Cantera-based open-source solvers for steady laminar one-dimensional flames, PREMIX++ and OPPDIF++, are introduced. To model the transport of charged species in flames, existing C++ classes of Cantera for one-dimensional reacting flows are modified to incorporate drift flux terms and a Poisson's equation of electric potential. With an extended reaction mechanism for ion chemistry proposed to model both lean and rich premixed methane flames, validation against Goodings' experiments is presented. Prediction of current-voltage characteristics for an opposed-flow diffusion flame is demonstrated. Turbulent combustion simulations are typically initiated using a time-dependent heat source kernel (representing an electric spark) in the computational domain to activate chemical reactions. In reality, the spark not only increases temperature of the background gas via ohmic heating, but also creates reactive radical species via electron impact. A detailed multi-species plasma discharge model is coupled with turbulent combustion equations to self consistently capture ignition. A spectral deferred correction (SDC) based multi-rate integrator will be used in the combustion solver. The final work will compare the reaction pathways that result in ignition to the active pathways under conventional hot-spot ignition model. contact: [email protected] contact: [email protected] Application of a two-fluid plasma model to the simulation of premixed flames under electric fields Computational modeling of plasma assisted combustion in gas turbines for electric power generation T. Casey1, P. Arias2, J. Han2, M. Belhi2, F. Bisetti2, H. Im2 and J. Chen2 A. Ehn1, J. Zhu1, P. Petersson1, Z. Li1, M. Alden1, C. Fureby2, T. Hurtig2, N. Zettervall2, A. Larsson2 and J. Larfeldt3 1 University of California, Berkeley, USA. King Abdullah University of Science and Technology (KAUST), Saudi Arabia. 2 1 2 In the numerical studies, interested in combustion interaction with electric fields, the effects of nonthermal electrons are not considered. To account for those effects, a two-fluid plasma model is developed. The gas mixture and electron transport equations are derived separately from Boltzmann’s kinetic theory. Boundary conditions for the electric potential, charged species concentrations and electron energy and velocity are established. The model is implemented in a fully compressible DNS reacting flow code to calculate the distribution of charged species and electric field in a 1D-premixed flame. An analysis of the effect of electric field on the flame properties is proposed. contact: [email protected] 3 LTH, Sweden. FOI, Sweden. Siemens, Sweden. To reduce the risk of combustion instabilities, increase fuel flexibility and combustion efficiency, electrical energy can be deposited in the flame using e.g. microwave radiation. This will create a non-thermal plasma modifying the flame chemistry. Here, the development and validation of a finite rate chemistry Large Eddy Simulation (LES) model is described. The model is based on a skeletal methane-air mechanism, extended by O3-chemistry, singlet O2-chemistry, chemoionization, electron excitation, electron impact dissociation, ionization and attachment, and N2 reactions. The LES model is applied to a methane-air low-swirl stabilized flame. An applied electrical field increases the turbulent flame speed by 15%, making the premixed bowl-shaped flame move closer to the burner mouth. Good agreement between the experimental and computational results is observed both with and without microwave stimulation. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 74 CP11: Detonation CP11: Detonation CP11 (1/3) st Tuesday April 21 (13h30 – 15h30) Flame acceleration in channels with cold walls C. Dion1, B. Demirgok2, V. Akkerman2, D. Valiev3 and V. Bychkov1 FORCE method. The study is performed with the constant emission rate of igniter. The space-time evolution of flow variables for three different cases of propellant burning rate is examined. contact: [email protected] Symmetric Baer and Nunziato model 1 Dept. Physics, Umea University, Sweden. 2 Dept. Mechanical and aerospace Engineering, West Virginia University, USA. 3 Dept. Applied Physics, Umeå University, Sweden. The present work considers the problem of premixed flame front acceleration in micro-channels with smooth cold non-slip walls in the context of the deflagration-todetonation transition; the flame accelerates from the closed channel end to the open one. Recently, a number of theoretical and computational papers have demonstrated the possibility of powerful flame acceleration for micro-channels with adiabatic walls. In contrast to the previous studies, here we investigate the case of flame propagation in channels with isothermal cold walls. The problem is solved by using direct numerical simulations of the complete set of the NavierStokes combustion equations. We obtain flame extinction for narrow channels due to heat loss to the walls. However, for sufficiently wide channels, flame acceleration is found even for the conditions of cold walls in spite of the heat loss. Specifically, the flame accelerates in the linear regime in that case. While this acceleration regime is quite different from the exponential acceleration predicted theoretically and obtained computationally for the adiabatic channels, it is consistent with the previous experimental observations, which inevitably involve thermal losses to the walls. In this particular work, we focus on the effect of the Reynolds number of the flow on the manner of the flame acceleration. R. Saurel University Institute of France. The Baer and Nunziato (1986) is nowadays widely used in two-phase flow community and particularly in granular media deflagration and detonation modelling. However this model is not symmetric in the sense that a carrier phase is considered with a dispersed granular one. It implies that the interfacial pressure in the equations is approximated to the gas one while the interfacial velocity is assumed to be the solid one. In many situations this is confusing as both phases may have very high densities, such as detonation products and reacting condensed material for example. This is also limiting when deaingl with gas - granular materials interfaces. In a former contribution (Saurel et al., 2003) a symmetric variant of the BN model was derived for fluid mixtures; i.e. in the absence of intergranular effects. The addition of granular effects (contact pressure, intergranular stress) seemed impossible to insert. Thanks to a deeper analysis granular effects are inserted in a symmetric BN formulation, while preserving both hyperbolicity and agreement with the second law of thermodynamics. Computational results demonstrate capabilities of the method. contact: [email protected] 3D simulation of discrete combustion waves in mechanically activated powder mixtures contact: [email protected] Numerical simulation of shock wave dynamics in internal ballistics of gun N. Chirame, D. Pradhan and S. Naik Defence Institute of Advanced Technology, India. In this work, we have simulated a 1D two phase flow Baer-Nunziato model of internal ballistics to study the shock wave motion inside the gun chamber. The model consists of conservation equation of mass, momentum and energy for gas and solid phase with conservation equation for solid volume fraction. The constitutive equations are co-volume equation of state and mass transfer rate of solid propellant. First ORder CEntred method with operator splitting technique has been employed to solve the equations. The academic case is simulated to study the shock capturing nature of S. Rashkovskiy1 and A. Dolgoborodov2 1 Institute for Problems in Mechanics, Russian Academy of Sciences, Russia. 2 N. Semenov Institute of Chemical Physics, Russian Academy of Sciences, Russia. A method of numerical simulation of propagation of a chemical reaction in powder mixture is considered. The method involves two steps: a simulation of powder structure and simulation of propagation of reaction wave in this structure. We discuss and estimate the mechanisms of transferring of the thermal pulse between the particles. The parametric calculations and the animation of the results of calculations were made. It is shown that combustion of powder mixture proceeds non-uniformly in space and time. The results of simulations are compared with experimental data. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP11: Detonation Simulations of compressible gas-dust flows E. Oran and R. Houim University of Maryland, USA In dust flows of current interest, gas flow speeds can range from highly subsonic to supersonic, the grain packing can range from fully packed to completely dispersed in the gas, and both the gas and the dust can range from chemically inert to highly exothermic. Examples include coal dust in mines, regolith on the lunar surface, and desert or beach sand, corresponding to flows generated by explosions, asteroid impact, and soft aircraft landings. To cover the necessary parameter range, we solve coupled sets of Navier-Stokes equations describing the background gas and the dust. As an example, an reactive-dust explosion in air, one that results in a type of shock-flame complex, is described and discussed. contact: [email protected] Detonation shock propagation through nitromethane embedded metal foam B. Lieberthal and D. Stewart University of Illinois at Urbana-Champaign, USA. Explosive devices manufactured from nitromethane and metal foam have numerous practical military and industrial applications because the individual components are safe to transport and easy to assemble. We run hydrodynamic simulations of a shock wave propagating through metal foam using a reactive flow model in the ALE3D software package. We explore in depth the applied pressure and energy of the shock wave and its effects on the fluid and inert material interface. We sample foam models of varying pore size composed of low and high impedance metals and test their viability as a hybrid explosive material. contact: [email protected] CP11 (2/3) Wednesday April 22nd (13h30 – 15h30) Three-dimensional simulations of shock-induced combustion around a spherical projectile with various diameters Y. Sakuragi and A. Matsuo Keio University, Japan. We carried out three-dimensional simulations of ShockInduced Combustion around a spherical projectile flying at 1931 m/s into H2-Air mixture, with two-step reaction model. The unstable behavior in front of projectile is revealed by a series of simulations with various projectile diameters. The bigger diameter case (24 mm) shows the fully three-dimensional unsteady 75 behavior, but the axisymmetric oscillations with constant period are observed in the smaller case (15 mm). In the intermediates, the three-dimensional instabilities gradually develop in the flow field with projectile scale-up. Thus, the three-dimensional simulations succeed in obtaining unstable behavior observed in the previous experiments. contact: [email protected] Weakly nonlinear detonations: theory and numerics L. Faria1, A. Kasimo1 and R. Rosales2 1 2 KAUST, Saudi Arabia. MIT, USA. Starting from the compressible reactive Navier-Stokes equations, we derive an asymptotic theory of weakly nonlinear detonations with diffusive effects retained. Focusing on the inviscid limit, we show by linear stability analysis and direct numerical simulations that the asymptotic system captures several features of detonations, such as (1) existence and instabilities of ZND waves, (2) galloping and chaotic detonations in one dimension, and (3) cellular structures in two dimensions. Finally, we study the effect of dissipative processes on the structure and stability of the traveling waves, and compare the ideal predictions with their dissipative counterpart. contact: [email protected] A compressible LEM-LES strategy (CLEM-LES) for modeling turbulent detonation propagation B. Maxwell1, S. Falle2, G. Sharpe3 and M. Radulescu1 1 2 3 University of Ottawa, Canada. University of Leeds, UK. Blue Dog Scientific Ltd., UK. A turbulent combustion model based on the Linear Eddy Model for Large Eddy Simulation (LEM-LES) has been developed to study highly compressible and reactive flows involving very rapid transients in pressure and energy. In the current work, the model is used to simulate 2D unsteady and turbulent detonation propagation, in a narrow channel, which involves a premixed methane-oxygen mixture at low pressures. The results have shown that the simulated detonation propagates at an average velocity that corresponds to the theoretical Chapman-Jouguet (CJ) velocity. Furthermore, the model captures well the twodimensional cellular structure that is observed in corresponding shock tube experiments. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 76 CP11: Detonation Set-valued solutions for detonations with losses 1 1 1 R. Semenko , L. Faria , A. Kasimov and B. Ermolaev 1 2 2 KAUST, Saudi Arabia. N. N. Semenov Insitute of Chemical Physics, Russia. We consider the 1D model of gaseous detonation in the packed bed of inert solid particles. The model is given by the reactive Euler equations with losses of heat and momentum and the Arrhenius-type heat release. The goal is to see if this simplified model can accurately reproduce the well-known detonation velocity deficit due to losses. We study the dependence between detonation speed and the amount of losses for steadystate solutions. Particularly, it is shown that this dependence allows for multiple or even set-valued solutions. The latter means that there can be a continuous set of solutions for a given fixed set of problem parameters. contact: [email protected] Two-phase simulations of TNT/Aluminium afterburning during explosions E. Fedina and C. Fureby Swedish Defence Research Agency – FOI, Sweden. Present study investigates the afterburning from explosions of a TNT charge and a TNT charge containing aluminium particles (TNT/Al) at two heights above ground, to demonstrate that numerical simulations can facilitate evaluation of the performance of high-explosives. The simulations are conducted using two-phase finite rate chemistry Large Eddy Simulation model in Euler-Lagrange form, incorporating interaction between phases by a two-way coupling. The results indicate that pure TNT requires shock-mixing layer interaction to sustain afterburning, while TNT/Al generates mixing through particle-gas interactions; hence the afterburning mainly depends on the oxygen supply. The simulations results show good agreement with the experimental data. contact: [email protected] CP11 (3/3) Wednesday April 22nd (16h00 – 18h00) Is is possible to achieve DDT behind a continuously growing Mach stem? Y. Lv and M. Ihme Stanford University, USA. moving shock that is reflected at a wedge. This configuration allows to isolate all physical mechanisms that are of relevance in triggering detonation. Parametric studies will be performed to analyze effects of the wall and relevant gas properties. For the numerical approach, a high-order discontinuous Galerkin method is used, and its capability for addressing such challenging multi-physics flow problems will be demonstrated. contact: [email protected] Numerical investigation on cellular H2-O2-Ar detonation evolution in mixtures with different hydrogen concentrations Q. Xiao, W. Fan, H. Li, K. Wang and Q. He School of Power and Energy, Northwestern Polytechnical University, China. Two-dimensional numerical simulations of cellular detonation evolution in H2-O2-Ar mixtures at a low pressure of 6.67 kPa have been performed in a 20 mm by 200 mm tube with a detailed chemistry kinetic mechanism. The initial overdriven detonation has been directly onset by a circular pocket of high pressure and temperature. The present study mainly focuses on the whole process of how initial overdriven detonation evolves to the final stable cellular detonation. Also, the influence of varied spatial hydrogen and Ar concentration on cellular detonation propagation is also investigated. contact: [email protected] On the influence of instabilities on the direct blast initiation of two-dimensional detonations H. Ng1, C. Kiyand12, G. Morgan2 and N. Nikiforakis2 1 Concordia University, Montreal, Canada. Cavendish Laboratory, Department of Physics, University of Cambridge, Cambridge, UK. 2 Two-dimensional, high-resolution numerical simulations are carried out in this study to investigate the influence of instabilities on the direct initiation of weakly unstable detonations. The computations are performed using Euler’s equations with a one-step Arrhenius kinetic model by general-purpose computing on graphics processing units (GPGPU). The introduction of an initial sinusoidal perturbation on the surface of the high energy source region with different amplitudes and wavelengths to generate cellular instabilities is assessed for either promotion of direct detonation initiation or adverse effect. contact: [email protected] Combustion behind a Mach stem is a fundamental mechanism for front propagation of unstable detonation waves. In this study, the reflection of a Mach-shock in a premixed gas mixture will be considered to examine the detonation mechanism. The Mach-stem is created by a 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP11: Detonation 77 Detonation in a radial supersonic outflow with simplified and realistic chemistry Numerical investigation on detonation initiation and propagation in supersonic reactive flows S. Korneev and A. Kasimov J. Li, Q. Xiao and W. Fan KAUST, Saudi Arabia. We study cylindrical detonation in a radial supersonic expansion for various heat release models. At first, we considered simple Arrhenius kinetics. Two types of solutions were found which give different instability pattern: square-wave cylindrical detonation and cylindrical detonation with very thin induction zone. The square-wave solution, as it was seen numerically, tends to collapse to the source and the other type of steady-state solution tends to expand. It was found, that expanding solution can be stabilized by set of obstacles which surrounds the detonation. We also find that obstacles barrier placed into radial supersonic inert expansion can initiate the cylindrical detonation. We perform comprehensive parametric study of radius of the steady state solution on incoming flow parameters. Also we study cylindrical steady-sate solution for Arrhenius kinetics with local density dependence and realistic two stages kinetics for hydrogen-oxygen mixture. School of Power and Energy, Northwestern Polytechnical University, China. The initiation and propagation process of detonation waves in a supersonic combustible flow was numerically investigated. An auto-ignition detonation phenomenon was initiated by a wedged channel. The time-dependent two-dimensional Euler equation was used to describe the chemically reacting flow with detonation combustion. The influence of the incoming Mach number and configuration of detonation chamber on the detonation initiation and propagation process was systematically investigated. The existing form of detonation waves under different conditions, the interaction between complex waves and the transition condition between upstream propagating normal detonation and standing oblique detonation wave were also addressed. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 78 CP12: Industrial furnaces CP12: Industrial furnaces Tuesday April 21st (16h00 – 18h00) Effect of particle shrinkage on the biomass pyrolysis behavior in a high-temperature entrained-flow reactor X. Ku, T. Li and T. Løvås University of Science and Technology (NTNU), Norway. Effect of particle shrinkage on biomass pyrolysis behavior in a laboratory-scale entrained-flow reactor is numerically investigated by an Eulerian-Lagrangian CFD model proposed in our earlier paper. The operating temperature is high (1400 °C) and two shrinkage submodels are adopted. The two different sub-models consist of (1) constant particle volume, and (2) constant particle density. Simulation results are analyzed both qualitatively and quantitatively in terms of particle distribution, product gas composition, pyrolysis time and particle residence time. Moreover, the calculated results are compared with the experimental data available in the literature. contact: [email protected] Avoiding ring formation in cement kilns D. Lahaye TU Delft, The Netherlands. Avoiding the formation of rings in rotary kilns is an issue of primary concern to the cement production industry. We developed a numerical combustion model that revealed that in our case study rings are typically formed in zones of maximal radiative heat transfer. This local overheating causes an overproduction of the liquid phase of the granular material, which tends to stick to the oven's wall and to form rings. To counteract for this phenomenon, we propose to increase the amount of secondary air injected to reduce the peak temperature. Experimental validation at the plant has confirmed that our solution is indeed effective. For the first time in years, the kiln has been operating without unscheduled shut-downs, resulting in hugely important cost savings. contact: [email protected] Large eddy simulation of coal and biomass co-firing M. Rabacal1, M. Costa1 and A. Kempf2 1 2 Instituto Superior Tecnico, University of Lisbon, Portugal. University of Duisburg-Essen, Germany. A large eddy simulation of coal and biomass co-firing in a large-scale laboratory furnace is presented. The inhouse CFD solver PsiPhi, optimized for massive parallel execution, is used. Coal and biomass particles are tracked individually using a Lagrangian approach. The combustion steps of drying and devolatilization are modeled using a thin film approach and a single first order reaction model, respectively. Coal char combustion is modeled using the Smith model, whereas for biomass a diffusion-limited surface reaction rate model is used. Major gas species distribution, particle properties conditional to gaseous properties and particle combustion history are discussed. contact: [email protected] A laboratory-scale downsizing procedure for highly resolved LES of large-scale combustion system B. Farcy, D. Midou, L. Vervisch and P. Domingo CORIA, CNRS & INSA de Rouen, France. The study of both flame dynamics and combustion instabilities, as for instance in aeronautical combustion chambers, have benefited strongly from highly resolved LES. The lengths in these systems are of the order of half a meter. On the other hand, industrial combustion systems featuring length of the order of 10 meters cannot be addressed today with high resolution. A nonlinear downsizing procedure is discussed, which preserves characteristic turbulent flame numbers (Damkohler and Karlovitz), to simulate large furnaces with high resolution. The method is first validated using DNS and applied to a DeNOx SNCR system and a blast furnace coal injector. contact: [email protected] Development of an unstructured CFD solver for the modeling of industrial furnaces L. Romagnosi, J. Anker, K. Claramunt and C. Hirsch NUMECA International, Belgium. The modeling of furnaces can represent a challenging task since multiple physical processes interact and a realistic modeling of each of them is necessary to attain an overall reliable prediction of the heat transfer, combustion products and the pollutants. To model industrial furnaces it is necessary to reliably model the reactive flow as well as the conjugate and radiative heat transfer. The formation of soot in furnaces can be beneficial, since it can significantly enhance the radiative heat transfer. However, since soot it is a pollutant, attention must be paid that it is oxidized before leaving the furnace. For environmental reasons also the CO and the NOx emissions should be kept at a minimum without compromising the efficiency of the furnace. In this study an unstructured, multiphysics-multipurpose CFD solver is presented for the modeling of industrial furnaces. The numerical scheme as well as different models available in the solver are explained. 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP12: Industrial furnaces 79 Dedicated test cases for CHT in reactive flows, modeling of radiative heat transfer, modeling of combustion processes with large heat losses and pollutants modeling will be used to assess and discuss the suitability of the various submodels for the modeling of furnace combustion. The overall modeling framework is assessed on the glass melting furnace of IFRF, the BERL natural gas burner and an industrial furnace geometry comparing the computational results with available measurement data. contact: [email protected] CP13: Instabilities Wednesday April 22nd (10h00 – 12h00) Nonlinear stability study of a time-delayed thermoacoustic system Gas expansion and shear layer effects in turbulent premixed flames X. Yang1, A. Turan1 and S. Lei2 1 2 S. Schlimpert, A. Feldhusen, J. H. Grimmen, B. Roidl, M. Meinke and W. Schröder The University of Manchester, UK. Bell Labs, Alcatel-Lucent, Ireland. This work aims to investigate the various nonlinear behaviours inherent in a time-delayed thermoacoustic system, viz. the Rijke tube, using a continuation approach. Unlike the conventional approach by Galerkin method to investigate nonlinear behaviour, a dynamic system is developed by spatially discretizing the acoustic equations using standard finite difference schemes in a Method of Lines manner. Temporal evolution of pressure oscillations is studied to demonstrate the interaction of acoustics and heat release perturbations. Moreover, the subcritical Hopf and fold bifurcations are captured to delineate the linear, nonlinear stability boundary and limit cycles. Furthermore, comparison of numerical calculations with experimental data reveals that the current approach does yield very good predictions. Institute of Aerodynamics, RWTH Aachen University, Germany. Combustion instabilities can cause serious problems which limit the operating envelope of low-emission lean premixed combustion systems. Predicting the onset of combustion instability requires a description of the unsteady heat release driving the instability. This study focuses on the analysis of fully coupled two-way interactions between a turbulent flow field and a premixed flame by solving the conservation equations of a compressible fluid in a fully coupled massive parallel framework. Results of a turbulent jet flame will be presented at varying gas expansion ratio to analyze the impact of the hydrodynamic instability and shear layer effect. contact: [email protected] contact: [email protected] DNS investigation on thermoacoustic oscillation characteristics of turbulent swirling premixed flame in a cuboid combustor K. Aoki, M. Shimura, Y. Naka and M. Tanahashi Tokyo Institute of Technology, Japan. To investigate the relation between fluctuations of pressure, heat release rate and vortical motions, direct numerical simulations (DNS) of hydrogen-air turbulent swirling premixed flame are conducted and dynamic mode decomposition (DMD) is applied to the DNS results. DMD of pressure and heat release rate fields and spectral analyses of vortical motions reveal that not only the quarter wave mode of longitudinal acoustics but also transverse acoustic modes, which interact with periodic vortical motions, play important roles in thermoacoustic instabilities. Acoustic energy budgets are also investigated based on DMD results of the source term in the acoustic energy equation. Large eddy simulation of transverse modes in a swirled kerosene/air combustor A. Ghani1, T. Poinsot2 and L. Gicquel1 1 2 CERFACS, France. IMF Toulouse, France. Combustion instabilities due to longitudinal and azimuthal modes have been addressed extensively in the last years but less attention has focussed on transverse modes. These modes create strong high-frequency oscillations leading to high-pitched sound levels and fast destruction of the system. The prediction of transverse modes still remains a problem because no theory is available. LES and acoustic solvers are used in this work to study transverse modes in a swirled kerosene/air combustor and propose a modified flame transfer function framework adapted to these transverse modes. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 80 CP13: Instabilities Acoustically induced vortex core flashback in a staged swirl-stabilized combustor Dynamical modeling of combustion instabilities in liquid rocket engines C. Lapeyre1, M. Mazur2, P. Scouflaire2, F. Richecoeur2, S. Ducruix2 and T. Poinsot3 M. Gonzalez-Flesca, T. Schmitt, S. Ducruix and S. Candel 1 CERFACS, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. 3 IMFT, France. 2 This paper describes a joint experimental and numerical investigation of flashback mechanisms in a swirled turbulent burner. This fully premixed combustor terminated by a choked nozzle operating at 2.5 bars shows a flame stabilized either within the combustion chamber or flashbacked into the injection duct. Flashback can occur naturally, or be triggered either through acoustic forcing or self-excited thermoacoustic instability. In the LES, self-excited oscillations reach high levels rapidly, leading to flame flashback, as observed experimentally. These results also suggest a simple method to avoid flashback by using fuel staging, which is then tested successfully in both LES and experiments. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. In an effort to predict combustion instabilities in liquid rocket engines, a reduced order modeling is developed. It relies on the expansion of the pressure field on the eigenmodes of the system. This dynamical model provides a state space description and accounts for the main effects such as cavity coupling between dome and chamber, damping, response of injectors to pressure perturbations, combustion response to acoustic oscillations or acoustic forcing. The model is used to study a laboratory scale cold flow model as well as fullscale geometries. The ongoing work consists in the simulation of a liquid rocket thrust chamber featuring combustion instabilities under certain operating conditions, with comparisons with experiments. contact: [email protected] contact: [email protected] CP14: High performance computing / software engineering Wednesday April 22nd (10h00 – 12h00) Predictive fire simulations with the software ISIS Scalable parallel and computationally-conservative implementations of the conditional moment closure model in large eddy simulations C. Lapuerta, S. Suard, F. Babik and G. Boyer IRSN, France. The CFD software ISIS, developed at IRSN is based on a coherent set of models that can be used to simulate fires in mechanically ventilated compartments. This fire model enables to perform predictive fire simulations, either using accurate pyrolysis models either global approaches for complex applications. Specific boundary conditions are used to predict the aeraulic behavior of the compartment for which the ventilation network of the entire facility may be computed by the IRSN zone code SYLVIA. The latest developments concern two important topics in fire modelling: large eddy simulation for reactive flows and fire suppression by water mist. Documentations and the open-source code are available at: https://gforge.irsn.fr/gf/project/isis. contact: [email protected] H. Zhang1, A. Garmory2 and E. Mastorakos1 1 2 University of Cambridge, UK. Loughborough University, UK. The scalable parallel and computationally-conservative conditional moment closure (CMC) model in large eddy simulations (LES) is implemented toward accurately predicting the turbulence-chemistry interactions in practical burners. Three parallel strategies are developed to distribute the CMC workload: fully local CMC, uniform round-robin and uniform adaptive allocation based on the instantaneous chemistry computation overhead. Their parallel efficiencies and scalability are investigated for stiff/non-stiff chemical mechanisms and different flame dynamics, e.g. swirling non-premixed flames far from and close to blow-off. The effects of CMC cell reconstruction and numerical flux prediction in physical space are also studied, which demonstrate the pronounced influence on calculating critical combustion phenomena. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP14: High performance computing / software engineering A methodology for the integration of stiff chemical kinetics on GPUs and application to the LES-PDF simulation of a turbulent non-premixed flame F. Sewerin and S. Rigopoulos Imperial College London, UK. In numerical schemes for reacting flows, it is common to isolate the chemical kinetics model by applying the method of fractional steps. Here, we explore the possible use of Graphics Processing Units (GPUs) in order to accelerate the solution of the reaction step for an implicit integration algorithm of order 5 (Radau5). The performance of the novel GPU implementation is evaluated for different reaction mechanisms, numbers of grid points, time steps and convergence tolerances. Moreover, strategies for parallelizing the reaction step across a CPU-GPU pair are investigated and the implementations are compared for an LES-PDF simulation of the Sandia D flame. contact: [email protected] Load balancing, dynamic repartitioning, and data migration in turbulent reactors' simulation P. Pisciuneri, E. Meneses, A. Zheng, A. Labrinidis, P. Chrysanthis and P. Givi University of Pittsburgh, USA. Different approaches for maximizing the scalability of parallel Lagrangian Monte Carlo solvers for turbulent combustion are assessed. The issues of load balancing, (re)partitioning, and data migration are considered via 81 three approaches: Zoltan (Sandia National Laboratories), Charm++ (University of Illinois at Urbana-Champaign), and (P)ARAGON (University of Pittsburgh). These approaches are compared in terms of weak and strong scaling in simulation of turbulent reactor via reduced order, finite-rate kinetics. contact: [email protected] PoKiTT: an efficient, platform agnostic package for thermodynamics, kinetics, and transport properties in reactive flow simulations N. Yonkee and J. Sutherland Department of Chemical Engineering, The University of Utah, USA. We introduce PoKiTT, a portable, lightweight library for performing data parallel calculations of thermochemical quantities commonly encountered in turbulent reactive flow simulations. The capabilities of PoKiTT include ideal gas property estimations, thermodynamic polynomial evaluations (NASA and Shomate), mixture average transport property calculations (diffusion coefficients, viscosity, and thermal conductivity), and detailed chemical kinetics computations. PoKiTT uses Nebo, a domain specific language, to provide a platform agnostic implementation which will be ready for future exascale architectures. We demonstrate the performance benefits of using PoKiTT over Cantera in the context of a PDE solver for both CPU and GPU executions. contact: [email protected] CP15: Kinetics Wednesday April 22nd (10h00 – 12h00) A modeling study of the low-temperature oxidation of n-hexane, 2-methyl-pentane and 3-methyl-pentane Z. Serinyel1, O. Herbinet2, R. Fournet2, V. Warth2 and F. Battin-Leclerc2 1 2 contact: [email protected] Kinetic modeling of the combustion of 2-methyl-2-butene CNRS-ICARE, France. CNRS-LRGP, France. Detailed chemical kinetic mechanisms were generated using software EXGAS for the oxidation of three isomers of hexane: n-hexane, 2-methyl-pentane and 3methyl-pentane. These mechanisms considered a new set of kinetic parameters for low temperature reactions obtained using quantum calculations and were used to simulate jet-stirred reactor experimental results. Simulations showed satisfactory agreement for fuel conversion, as well as the formation of many products, including the isomers of cyclic ethers (with rings including 3, 4, 5 and 6 atoms) which were experimentally quantified. Incorporation of the new set of rate constants into EXGAS improved model predictions to an important extent. C. Westbrook1, P.A. Glaude2, F. Battin-Leclerc2, O. Herbinet2, O. Mathieu3 and E. Petersen3 1 LLNL, USA. CNRS-LRGP, France. 3 Texas A&M University, USA. 2 2-Methyl-2-butene is an intermediate chemical species produced during the combustion of branched alkanes and a model molecule of the light alkenes chemistry. A detailed kinetic mechanism has been developed and validated against low and high temperature data obtained in a jet stirred reactor and in a shock tube. 2methyl-2-butene produces mostly resonance stabilized 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 82 CP15: Kinetics radicals, which can slowly decompose into more unsaturated products such as isoprene, 1,2-butadiene and 2-butyne, or rather react by combinations with HO2 radical. The prompt decomposition of the hydroperoxydes produced by these latter reactions enhance the reactivity and can be explained the formation of the main primary oxygenated products. combustion model for light aromatics, have been developed for each compound. Models predict well the conversion of reactants and the formation of the main products, such as aromatic tertiary tars, which are soot precursors. contact: [email protected] contact: [email protected] Comparison of the performance of several recent hydrogen and syngas combustion mechanisms C. Olm1, I. Zsély1, R. Pálvölgyi1, T. Varga1, T. Nagy2, H. Curran3 and T. Turányi1 1 Institute of Chemistry, Eötvös University (ELTE), Hungary. MTA Research Centre for Natural Sciences, Hungary. 3 Combustion Chemistry Centre, NUI Galway, Ireland. 2 A large set of experimental data was accumulated for hydrogen and syngas combustion, including ignition measurements in shock tubes and rapid compression machines, concentration–time profiles in flow reactors, outlet concentrations in jet-stirred reactors, and flame velocity measurements. The 7000 datapoints cover wide ranges of temperature, pressure and equivalence ratio. The performance of 19 hydrogen and 16 syngas combustion mechanisms was tested against these experimental data, and the dependence of accuracy on the types of experiment and the experimental conditions was investigated. For both the hydrogen and syngas combustion, these recently published mechanisms could be sorted to the categories of good, average and poor performers, but the performance of a mechanism depended very much on the conditions where it was applied. The influence of poorly reproduced experiments on the overall performance was also investigated. contact: [email protected] Kinetic study of the combustion of phenolic tars from biomass M. Nowakowska, O. Herbinet, A. Dufour and P.A. Glaude CNRS-LRGP, France. Tars are by-products of wood combustion and gasification, which cause fouling, pollutant formation and catalyst deactivation in processes. A significant fraction of aromatic tars comes from lignin pyrolysis mainly composed of guaiacol-type and syringol-type units. The reactions of methoxybenzene (anisole) and methoxyphenol (gaiacol) were studied as surrogates of these phenolic tars from lignin. Detailed kinetic mechanisms for the pyrolysis and oxidation, based on a Transition state theory based semi-automatic generation of surface reaction mechanisms P. Kraus and P. Lindstedt Imperial College London, UK. The present work focuses on developing a theoretical framework of estimating Arrhenius parameters in heterogeneous reaction systems, leveraging the UBIQEP method for estimating barrier heights, and a combined collision/transition-state theory based method for estimation of pre-exponential factors. The method is validated for the case of partial ethane oxidation over platinum, assessing the impact of key parameters (oxygen/carbon ratio, inlet velocity, site density, heat losses). The developed method reproduces experimental data with an accuracy comparable to the collision theory approach without the reliance on conjectural experimental parameters, facilitating an efficient generation of novel heterogeneous reaction mechanisms. contact: [email protected] Tabulation strategies of partially premixed dimethyl-ether flames S. Hartl, A. Zschutschke, D. Messig and C. Hasse TU Bergakademie Freiberg, Germany. Dimethyl-Ether (DME) is a promising alternative diesel-fuel. In contrast to methane, significant amounts of intermediate hydrocarbons are found, both close to the reaction zone and in the fuel-rich parts, where DME-pyrolysis takes place. These differences are important for combustion models, which are based on flamelet-like approaches. These flamelet results are usually tabulated for later retrieval in CFD-calculations. In this work strategies are proposed for numerical CFDsimulations with tabulated chemistry. Here, we look at the flamelet-progress-variable approach for DME flames. Direct-chemistry CFD results of a partiallypremixed and a non-premixed DME-flame are validated against experiments and further compared to different tabulation approaches. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP16: Laminar flames 83 CP16: Laminar Flames Wednesday April 22nd (13h30 – 15h30) Transient response of a laminar premixed flame to a radially diverging/converging flow Diffusive-thermal instabilities in premixed flames: stepwise ignition-temperature kinetics M. Sahafzadeh1, L. Kostiuk2 and S. Dworkin1 L. Kagan1, I. Brailovsky1, P. Gordon2 and G. Sivashinsky1 1 2 Ryerson University, Canada. University of Alberta, Canada. Laminar flamelets are often used to model turbulent flames. The libraries used for flamelets are typically based on counter-flow, strained and curved laminar flames under steady conditions. This research seeks to further understanding of the current techniques in this area by proposing a model for the less-considered laminar flame dynamics of a cylindrically-symmetric radially advected premixed flow. This model, which is developed for a transient premixed flame, studies the flame response to a diverging/converging flow, as the flame grows or shrinks without being aerodynamically strained in order to isolate these affects. contact: [email protected] The analysis diffusional-thermal stability of rich hydrogen-air combustion waves in model with detailed kinetic mechanisms 1 2 The study is motivated by the observation that in combustion of hydrogen-oxygen/air and ethyleneoxygen mixtures the global activation energy appears to be high at low enough temperatures and low at high enough temperatures, reflecting the complex nature of the underlying chemistry. Numerical simulations of a planar premixed flame propagation controlled by a stepwise ignition-temperature kinetics (representing the activation energy temperature-dependence) is carried out. It is shown that, for all its schematic nature, the diffusive-thermal model based on the ignitiontemperature kinetics reproduces successfully the basic features of both cellular and pulsating instabilities typical of low and high Lewis number mixtures. contact: [email protected] Direct numerical simulation of circular expanding premixed flames in a lean quiescent hydrogen-air mixture: phenomenology and detailed flame front analysis V. Gubernov1, A. Korsakova1, A. Kolobov1, V. Bykov2 and U. Maas2 1 P.N. Lebedev Physical Institute of the Russian Academy of Sciences, Russia. 2 Institute of Technical Thermodynamics, Karlsruhe Institute of Technology, Germany. The diffusional-thermal stability of rich hydrogen-air combustion waves under ambient and elevated pressures is numerically investigated. The model includes the detailed transport and detailed kinetic mechanism of hydrogen oxidation. Three different kinetic mechanisms are employed: GRI3.0, Warnatz, San-Diego. The critical conditions for the onset of pulsating instabilities are found in the equivalence ratio vs pressure parameter plane in each case. It is demonstrated that the boundaries of stability differ significantly depending on the specific kinetic mechanism. This suggests that finding the critical parameters for the onset diffusional-thermal instabilities experimentally offers a new way for the verification of kinetic mechanisms. contact: [email protected] Tel Aviv University, Israel. The University of Akron, USA. C. Altantzis1, C. Frouzakis2, A. Tomboulides3 and K. Boulouchos2 1 2 3 Massachusetts Institute of Technology, USA. ETH Zurich, Switzerland. University of Western Macedonia, Greece. The combined effect of the hydrodynamic and the thermodiffusive instability mechanisms on flames propagating in gravity-free, quiescent mixtures leads to the distortion of the smooth front by the appearance of small-scale cellular structures, which in turn influence the propagation dynamics and induce self-acceleration. The evolution of a 3D spherical lean H2/air flame, subject to the above-mentioned mechanisms, expanding in a laminar environment is studied employing direct numerical simulation with detailed chemistry and transport. Detailed local analysis of geometric and propagation characteristics of the flame together with statistical analysis is used to elucidate the phenomenology of cellularity and flame dynamics. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 84 CP16: Laminar flames Propagation of premixed flames in long narrow channels from a closed end: transition from constant speed to rapid acceleration V. Kurdyumov1 and M. Matalon2 1 Department of Energy, CIEMAT, Spain. Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, USA. 2 Propagation of a premixed flame in narrow channels from the closed end is considered analytically and numerically. An asymptotic approach shows that in sufficiently narrow channels there are two distinct solutions with a constant flame velocity corresponding to slow and fast flames. The solution with the highest velocity is unstable and therefore cannot be observed in experiments. The two solutions merge when the channel width is increased to a critical value, suggesting that the slow propagation with a constant velocity changes to a rapid exponentially-like acceleration. This has been corroborated by direct numerical simulations. The results thus provide a clear criterion for the conditions and onset of exponentially-like acceleration in long channels. Influence of heat losses and differential diffusion (Lewis number) are also reported. contact: [email protected] CP17: Real gas effects Wednesday April 22nd (13h30 – 15h30) Diffuse interface methods for diffusion flame from subcritical to supercritical pressure contact: [email protected] CP17: LES of transcritical LOx/GH2 injection in a rocket engine P. Gaillard, V. Giovangigli and L. Matuszewski ONERA, France. L. Matuszewski, P. Gaillard and V. Giovangigli This study is focused on the subcritical to supercritical regime transition encountered during the ignition of a liquid-propellant cryogenic rocket engine. In this work, both subcritical diphasic and supercritical monophasic regimes are addressed with a diffuse interface method allowing the investigation of the structure of a laminar strained LOx/GH2 diffusion flame from subcritical to supercritical pressure. Non-ideal thermodynamics and multi-component transport may lead at low temperature to instabilities and sharp interfaces separating immiscible thermodynamical states even at pressure higher than the critical pressure of the two propellant. ONERA, France. contact: [email protected] contact: [email protected] The absence of a dense potential core in transcritical injection Large eddy simulations of a naturally unstable transcritical coaxial injector D. Banuti and K. Hannemann A. Coussement1, T. Schmitt1, S. Ducruix1, S. Candel1 and Y. Nunome2 German Aerospace Center (DLR), Germany. Injection of cryogenic nitrogen into a chamber at supercritical conditions has become a popular test case for real gas CFD validation. However, there is a systematic discrepancy: a measured immediate drop of density upon entering the chamber could not be found with theory or CFD. It is shown here that heat transfer to the cryogenic injectant does take place and may explain the exhibited behavior. In a numerical study, different break-up modes could be reproduced by varying heat transfer inside the injector. This thermal break-up is essentially a new break-up mechanism, strongly associated with pseudoboiling, a supercritical state transition from liquidlike to gaslike. Turbulent dispersion and mixing in the wake of the injectors influence greatly the shape of the macroscopic flame in liquid propellant rocket engines. This study is focused on the effects on these phenomena of the stiff density gradients arising from the transcritical LOx injection. This work is based on a simplified multifluid method which relaxes the mesh size constraints required to describe the dense to dilute transition. This approach is used to perform a Large Eddy Simulation of a typical LOx/GH2 rocket engine injector. 1 2 Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Japan Aerospace Exploration Agency, Japan. In the context of liquid rocket combustion instabilities, a naturally unstable test-bench from JAXA operating in transcritical conditions is studied numerically, using the AVBP-RG solver jointly developed by CERFACS and EM2C. Two cases are considered: the first one corresponds to experimentally stable injection conditions, while the other one feature large pressure oscillations in the combustion chamber. First numerical results indicate a more dynamic behaviour of the unstable case, in good qualitative agreement with experimental data, compared with the stable case. The current state of our simulations, a possible scenario 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP17: Real gas effects 85 explaining the presence of the instability and ongoing analysis are presented. contact: [email protected] contact: [email protected] Large eddy simulation of transverse acoustic activity in a 42-injector rocket engine A. Urbano1, L. Selle1, G. Staffelbach2, B. Cuenot2, T. Schmitt3, S. Ducruix3 and S. Candel3 Oscillatory combustion in a sub-scale liquid rocket chamber T. Shimizu1, Y. Morii1, Y. Mizobuchi1, Y. Nunome1, T. Tomita1, H. Kawashima1 and M. Hishida2 1 2 Japan Aerospace Exploration Agency, Japan. Ryoyu systems Corp., Japan. Oscillatory combustion in a sub-scale liquid rocket chamber is studied by a time-dependent numerical analysis based on Large Eddy Simulation approach. The dense fluid properties for H2/O2 combustion at supercritical pressure and the laminar flamelet model for combustion, are implemented in an unstructured flow solver developed at Japan Aerospace Exploration Agency (JAXA). The obtained pressure oscillation agrees well with the experiment. The amplitude of pressure oscillation in the first longitudinal mode increased when the inlet hydrogen temperature becomes low. It is found that the flame length is the key to decide whether an oscillatory combustion in the short sub-scale chamber occurs or not. 1 2 3 IMFT, France. CERFACS, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Large Eddy Simulations of a liquid-propellant rocket engine are carried out. The configuration is an experimental combustion chamber, called BKD and operated by DLR, which comprises 42 coaxial injectors. This engine exhibits high-frequency instabilities under transcritical injection conditions. The aim of the study is to gain understanding in the physical mechanisms that trigger this transverse combustion instability, which is also a high-performance computing challenge given the number of injectors. The influence of operating conditions is investigated on coarse (50 million elements) and fine (500 million elements) meshes, both under forced and self-excited conditions, for the first transverse acoustic mode. contact: [email protected] CP18: Gas turbine combustion Wednesday April 22nd (16h00 – 18h00) Numerical computation methods of hot gases radiation in a turbofan combustion chamber: performance comparison for a design process Detailed analysis of light-round in an annular multiple-injector combustor R. Daviot and F. Loureiro M. Philip, R. Vicquelin, M. Boileau, T. Schmitt and S. Candel SAFRAN - SNECMA, France. Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. In turbofan combustion chambers, hot gases radiation plays a major role on the temperature reached by the wall, and therefore on its life limit. During chamber design, determination of this radiation by a numerical computation must combine accuracy of solution and speed of computation. This paper introduces a comparison of the results achieved through a MonteCarlo method and a Discrete Ordinates one. A comparative study has been made on an industrial configuration concerning the influence of gases radiation models such as 26 bands High Pressure Box Model and Weighted Sum of Grey Gases on the solution at different engine operating conditions. contact: [email protected] Recent large eddy simulations of the ignition transient in an annular combustor equipped with sixteen swirling injectors have shown remarkable agreement with experimental data in terms of flame topology and full combustor ignition time. In the present study, a detailed analysis of the mechanisms that control the light-round sequence is carried out for two sets of operating conditions. The ignition dynamics is characterized in terms of flame tip trajectory, flame surface at resolved and sub-grid scale levels, absolute and relative flame displacement velocity. This is then used to analyze the rate of displacement of the mean flame brush that determines the overall ignition time. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 86 CP18: Gas turbine combustion Modelling chemistry in coupled combustor and turbine CFD simulations Direct numerical simulation of premixed flames flashback in turbulent channel flows with fuel stratification R. Eggels A. Gruber1, E.S. Richardson2 and J. Chen3 Rolls-Royce Deutschland, Germany. 1 The interaction between engine components, such as combustor and turbine is getting more important. To be able to model the combustor and turbine nozzle guide vanes simultatenously a compressible reacting flow capabiltiy is required. A method has been developed to model the combustion process using the Flamelet Generated Manifold approach, which has been extended to model compressible flows where both the temperature and pressure are varying. The changing pressure and enthalpy and captured with an addional controlling variable, assuming an isotropic expansion. Furthermore, additional transport equations for the enthalpy, NOx and CO are taking into account. Additinal transport equations for NOx and CO taking into account, while the NOx and CO chemistry is slow compared to the change of pressure and temperature within the NGV. The method has subsequently been applied to an industrial gas turbine. contact: [email protected] Combustor effusion cooling design parameters analysis and modeling N. Savary, R. Hervé and G. Cottin Turbomeca, France. Meeting the challenge of designing gas turbine more money and fuel efficient requires among many other aspects to design cheap combustor able to handle increasing pressures and temperatures. Effusion cooling is one of the best value for money combustor cooling system as well as one of the most complex to model, due to the high range of scales of the phenomena involved. This numerical study explores the effect of the typical design parameters on effusion cooling efficiency. In particular, it proposes laws describing the cooling performances as a function of hole diameter, inclination and arrangement. contact: [email protected] SINTEF Energy Research, France. University of Southampton, UK. 3 Sandia National Laboratories, USA. 2 Among the challenges in enabling lean pre-mixed combustion technology for gas turbine applications, accurate prediction of flame flashback in mixing ducts stands out as a key requirement, in particular for highly reactive fuels containing hydrogen. Complete understanding of the complex physical processes that control flashback is of fundamental importance in gas turbine burner design and optimization. The present study, based on computationally demanding and highly resolved petascale direct numerical simulation (DNS), focuses on providing unprecedented insight into the turbulence-chemistry interaction that characterizes flashback of preheated hydrogen-air flames inside of mixing ducts. Bringing further previous experimental and modelling work, the present DNS introduces fuel stratification in the reactants' flow within the turbulent channel configuration and has shown that the turbulent flame shape, its propagation mechanism and its combustion regime are all considerably affected by fuel stratification. contact: [email protected] Large eddy simulation of a two-stage liquid fueled swirl burner: influence of injection strategies B. Cheneau, A. Vié and S. Ducruix Laboratoire EM2C, CNRS and Ecole Centrale Paris, France. Large Eddy Simulations of a two-stage swirling burner fueled with liquid dodecane studied experimentally at EM2C laboratory are performed. For a chosen lean operating point at constant power, fuel proportion is varied between the two injection devices, respectively central hollow cone and peripheral multipoint injections. Results highlight a dramatic influence on the flame stabilisation processes: by varying the staging factor, the flame takes various shapes associated with different combustion regimes and dynamics, as well as exhibiting hysteresis phenomena. The numerical simulations provide additional information that help in understanding the parameters and physics that trigger bifurcation and cause the hysteresis loop. contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. CP19: Fires 87 CP19: Fires Wednesday April 22nd (16h00 – 18h00) CFD simulations of fire-induced doorway flows in a small scale enclosure S. Suard, A. Koched, H. Pretrel and L. Audouin IRSN, France. CFD simulations of a fire in a small scale enclosure with an open doorway, are presented. Three heat release rates of 10.6, 15.5 and 21.7 kW, provided with a gas burner, are numerically studied and compared to temperature and velocity measurements at the doorway and temperature measurements inside the enclosure. Velocity measurements are performed using a SPIV technique allowing a full comparison with CFD results. For the three heat release rates, the simulation results agree well with experimental measurements. General flow patterns, provided by simulations, are reported and supply useful information for understanding enclosure fires. contact: [email protected] Mathematical modelling of the interaction between crosswind and aviation-fuel fire engulfing a full scale aircraft G. Wang and H.Y. Wang Institut P’, Fluides-Thermique-Combustion, CNRS, ENSMA, Université de Poitiers, France. Large fully turbulent fires, which result as a consequence of an aircraft accident, pose a severe hazard to the occupants and cargo. This numerical study focuses on the fire phenomenology associated with the presence of a full scale aircraft immersed within a large aviation-fuel fire in a moving fluid medium. Large Eddy Simulation for the fluid dynamic equations of three-dimensional elliptic flow is coupled with an Eddy Dissipation Concept, allowing to the simulation of fire growth and smoke (CO and soot) stratification in progressively vitiated environment. The prediction indicates that interaction between the large object and fire environment combined with the influence of wind conditions affects dramatically location of the continuous flame zone and heat flux distribution. formed by the vegetation and the surrounding atmosphere, in assimilating the vegetation as a sparse porous media and performing a homogenization-like procedure. After a detailed presentation of the physical and mathematical model, some numerical simulations are presented, illustrating various configurations of propagation of fire (surface fire, backfire, counter fire …), in various ecosystems (grass, shrubs), and under various external conditions (slope, wind). The problem of fire safety engineering in the wildland urban interface (WUI) is also treated. contact: [email protected] LES modelling of burning wildland fuels M. El Houssami1, A. Lamorlette2, D. Morvan2 and A. Simeoni1 1 2 University of Edinburgh, UK. Laboratoire M2P2, CNRS and Aix-Marseille Université, France. Thermal degradation and ignition of wildland fuels is simulated using a multiphase formulation. The physical model is implemented in FireFOAM, a Large Eddy Simulation (LES) solver for fire application, derived from the C++ object-oriented toolbox OpenFOAM. Fuel degradation is represented in three different steps: water evaporation, gas pyrolysis and char combustion. Several strengths and weaknesses of the current submodels (combustion, radiation, convection and turbulence) are identified by comparing simulations to experimental results, such as mass loss rates and heat release rates of burning samples of dry pine needles exposed to a radiative heat flux. contact: [email protected] contact: [email protected] How simulating the behaviour of wildland fires? D. Morvan Laboratoire M2P2, CNRS and Aix-Marseille Université, France. This communication demonstrates in what manner the behaviour of wildland fires can be simulated from a multiphase formulation. This approach consists in calculating the time evolution of the coupled system 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015. 88 CP19: Fires Evaluation of a sensor-driven wildlandfire spread modeling strategy using the FireFlux experiment 1 1 1 1 C. Zhang , M. Durand , W. Tang , M. Gollner , A. Trouvé1, M. Rochoux2, S. Ricci3, B. Cuenot3, J. Filippi4 and C. Clements5 Two and three dimensional numerical simulations of wildland fires A. Lamorlette and D. Morvan Laboratoire M2P2, CNRS and Aix-Marseille Université, France. 1 Department of Fire Protection Engineering, University of Maryland, USA. 2 Meteorological Research Division, Environment Canada, Canada. 3 CERFACS, SUC-URA1875, CNRS, France 4 Laboratoire Sciences Pour l’Environnement, CNRS and Université de Corse, France. 5 Department of Meteorology and Climate Science, San Jose State University, USA. The objective of this project is to develop a data-driven simulator capable of forecasting wildfire spread dynamics. The prototype simulator features the following components: a fire propagation solver that adopts a regional scale viewpoint, treats wildfires as propagating fronts, and uses a description of the local rate of spread of the fire as a function of vegetation, topographical and meteorological properties; a series of observations of the fire front position; and a data assimilation algorithm based on an ensemble Kalman filter. The prototype simulator is evaluated in a validation study corresponding to a controlled grassland-fire field experiment called FireFlux. Wildland fire propagation are simulated using a multiphase formulation. The physical model is implemented in FireFOAM, a Large Eddy Simulation (LES) solver for fire application, derived from the C++ object-oriented toolbox OpenFOAM. Two dimensional and three dimensional simulations are carried out in typical grassfire conditions, for different wind intensity, in order to enlighten three dimensional effects on both fire rate of spread and fire residence time. The role of radiation and convection ahead of the fire front is also evaluated, allowing to discuss the assumptions usually admitted for large scale fire modelling. contact: [email protected] contact: [email protected] 15th International Conference on Numerical Combustion, Avignon, France, April 19th – 22nd, 2015.