Current progress in DARS model development for CFD Harry Lehtiniemi STAR Global Conference 2012 Netherlands 20 March 2012
Application areas Automotive DICI SI PPC Fuel industry Conventional fuels Natural gas Dual fuel Biofuels Synthetic fuels Environment Soot control NOx formation Unburned hydrocarbons Heavy, energy and chemical industries: Power generation Gas turbines Flames and burners Chemical vapor deposition 2
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 3
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 4
Aspects of chemistry choice Pressure Temperature Reactor Fuel 4 Mpa 700 K Const P N-heptane 5
Key Features of DARS-FUEL mechanisms Reduced stiffness Methodology to reduce the stiffness of a mechanism -> short computational time, even with detailed models Compact even in detailed format Only species which are important for the decomposition of the fuel are included Multiple formats The mechanisms are available in various formats Libraries for CFD software Pre-compiled libraries for direct use in simulations Compatibility with STAR-CD/CCM+ products Consistency Reaction Names Thermodynamic data Transport properties 6
Key Features of DARS-FUEL mechanisms Good documentation Rule based Semi Automatic Generation* Mechanism generation for larger alkanes based on rules Rules tested against several alkanes Extrapolation to fuels without experimental base Automatic graph theory based generation -> efficient and less error prone Soot and NOx Integrated modeling of soot precursors Integrated coupling to NOx formation Complete solution From mechanism development to table generation From detailed reaction schemes to highly reduced, special purpose mechanisms Constant development Constant improvement and development of new models *M. Hilbig, L. Seidel, X. Wang, F. Mauss, and T. Zeuch. Computer aided detailed mechanism generation for large hydrocarbons: n- decane. 23rd ICDERS, 2011. 7
Available DARS-FUEL mechanisms Group Chemistry Reference fuel for Oxygenated methanol, ethanol, propanol Gasoline, Biofuels Mono-aromatics toluene, m-xylene Gasoline, Diesel, Jet Large aromatics 1-methylnaphalene Diesel, Jet Linear alkanes n-heptane, n-decane Gasoline, Diesel, Jet Branched alkanes iso-butane, iso-butane, iso-pentane, iso-octane, iso-dodecane Gasoline, Diesel, Jet Methylester methyldecanoate Biodiesel (RME/SME) Dimethylether Other Emission DME methane, ethane, propane, butane, pentane, neo-pentane, ethylene, acetylene, propene, hydrogen and others NOx, soot, formaldehyde, unburnt HC and others Natural gas, Biomass to gas / liquid, turbines Combined for all fuels 8
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 9
On-line chemistry DARS-CFD & DARS-TIF Ease of setup One-panel setup for complex fuel chemistries GUI support for setup Easy to Use Fast Accurate Combustion and emissions Detailed chemistry mechanisms Tailored size reduced chemistry Methods based on real physics A complete range of models adding chemistry information to: IC Engines cylinder processes Gas emissions, Soot After-treatment Fuels Flames and burners Chemical vapor deposition Coverage Mechanisms optimized to affordable sizes 10
Modeling with online chemistry Main approaches for online chemistry in 3D CFD with STAR-CD/CCM+ and DARS models: DARS-TIF: diesel engine application DARS-CFD: generic species transport model Library based emission models can be coupled with both DARS-TIF, DARS-CFD or other combustion models 11
Online chemistry - enhancements DARS-CFD Load balancing Improved scalability Lower CPU cost Clustering Method allows for cluster selection based on any parameters Number of clusters determined on-line based on user set tolerances Gradient based Lower CPU cost DARS-TIF Solver enhancement Speed-up: 200-600 % PDF integration Speed-up: 30 % Improved scalar dissipation rate treatment Ignition and emission prediction improvement Improved PDF treatment Ignition and emission prediction improvement 12
Clustering method prel. test results Pressure trace Full chemistry vs. cell clustering CPU time scaling vs. # cells Number of active clusters versus number of particles User defined cluster parameters: Entropy, enthalpy, enthalpy of formation, mixture fraction, any species Evaluated for DARS SRM: SI engine : 3-4 times faster DI engine: 1.5 2 times faster 13
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 14
What tables and for which models Model PVM PVM/G/TIF Table Auto-ignition Auto-ignition, Laminar flame speed, Species at 95 % burned ECFM-3Z-TKI Auto-ignition ECFM Laminar flame speed Soot NOx Soot source term flamelet table NOx source term flamelet table 15
Chemistry requirements Auto-ignition The chemistry must capture low- and high temperature ignition chemistry for a wide range of conditions Laminar flame speed The chemistry must be validated for a wide range of operating conditions Emission modeling The chemistry must properly capture both minor and major species profiles for a wide range of operating conditions The chemistry must work properly with PAH formation sub-model (important for soot source term tables) The chemistry must work properly with NOx formation sub-model (important for NOx source term tables) 16
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 17
Homogeneous Progress Variable for AI Progress variable: h 298 (chemical enthalpy) Reflects the conversion from chemical energy to thermal energy The following is solved for the progress variable: ρ h 298 + ρvh t 298 = ρ D + D t h 298 + ρω chem + ρω spray D t calculated using turbulent viscosity and Schmidt number internally DARS solver solves ρψ = ρω t chem ω spray : Spray term is derived using chemical enthalpy, continuity and mixture fraction transport No active scalars Z, Zvar, Xi, YEGR, h298, soot + NOx (transport) Species of interest (internal) for output can be retrieved from the table 18
Library generation procedure Flamelet-like resolution in mixture fraction space (101 points) Loop over FFEGR, EGR, P, T ox, Z, Xi with constant fuel side temperature Steps of 5 % in EGR, EGR composition as stoichiometric combustion products, up to 50 % EGR (mole fraction) Logarithmic stepping in P Equidistant stepping in T ox, same for all EGR values Steps of 5 % in Xi (mass fraction since mass fraction Xi transported) Progress variable: h298 100 points selected adaptively The PV tables contain: h298 source term NASA polynomial coefficients and mean molecular weight Species (needed for TIF initialization) and species source terms Table for one fuel contains about 1 GB of data 19
Temperature [K] PRF 95 gasoline PV library Accuracy Progress variable model vs. detailed chemistry, 0D reactor P = 32 bar, T = 733 K, Z = 0.062, EGR = 0 %, Constant pressure Excellent accuracy dt = 1e-6 s (0.01 CA @ 1667 RPM) dt = 5e-5 s (0.50 CA @ 1667 RPM) 2500 PV-TABLE Online Chemistry 2500 PV-TABLE Online Chemistry 2000 2000 1500 1500 1000 1000 0 0.002 0.004 0.006 0.008 0.01 Time [s] 0 0.002 0.004 0.006 0.008 0.01 Time [s] 20
PRF 95 gasoline PV library HCCI SRM DARS SRM is used to evaluate libraries 21
STAR-CD / es-ice test case for PRF 95 CI Passenger car CI engine configuration Bore/stroke: 81/93 mm Compression ratio: 16 P / T @ CA -120: 1.5 bar / 500 K Single, m inj,tot 19.2 mg, block profile, SOI CA 22 BTDC, EOI CA 14 BTDC Fuel: Gasoline (PRF 95 iso-octane/n-heptane PV table) EGR: 0 % Numerics: Standard k/epsilon turbulence model PISO MARS for momentum, turbulence, enthalpy and scalars (only passive scalars defined) CD for density Homogeneous PV model used as combustion model 22
STAR-CD / es-ice test case for PRF 95 CI Pressure Temperature 23
STAR-CD / es-ice test case for PRF 95 CI Mixture fraction and temperature 24
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 25
ECFM-3Z-TKI tables New ECFM-3Z-TKI tables Progress variable defined using h 298 (chemical enthalpy) New progress variable sampling to better resolve ignition process We found it was not possible to only use only 7 points for c, 0.025-0.3 To have accuracy, 15 points between 0 and 1 are used. It is possible to enter the table with c=0 like in the PV model Analytical solver written for progress variable update step Knowing the shape of the progress variable source term function it is possible to directly update the progress variable given the CFD time step size 26
ECFM-3Z-TKI Progress variable solver Pressure: 40 bar, Tunb: 800 K, Z: 6.22e-2, PRF 95 27
ECFM-3Z-TKI Tabulation aspects Single fuel table (PRF 95) Temperature: 101 points between 600 K and 1600 K Pressure: 8 points between 10 bar and 80 bar Mixture fraction points: 6 points depending on stoichiometry EGR points: 5 values between 0 % and 90 % EGR Temperature segregation: 21 values between 0 and 1 Around 60 MB data Dual fuel table (ethanol-gasoline mixtures) Temperature: 41 points between 600 K and 1600 K Pressure: 11 points between 10 bar and 200 bar Mixture fraction points: 16 points depending on stoichiometry EGR points: 5 values between 0 % and 90 % EGR Temperature segregation: 21 values between 0 and 1 Fuel composition points: 11 values between 0 and 1 (fraction ethanol) Around 400 MB data 28
Outline Aspects of chemistry choice Models for on-line chemistry calculation DARS-CFD DARS-TIF New developments for tabulated chemistry in CFD Progress variable model (PVM) ECFM-3Z-TKI Flamelet source term library approach updates Concluding remarks 29
Flamelet source term library approach 30
Concluding remarks DARS-FUEL Consistency Accuracy DARS-CFD & DARS-TIF Continuous development to lower CPU cost Tabulated chemistry development PVM development & coupling with G/TIF Consistency between progress variables in PVM & ECFM New solver developments for tabulation methods assure correct prediction of autoignition timing New methodology for emission source term library production 31
DARS models for STAR-CD/CCM+ Method Description Application STAR-CD STAR- CCM+ DARS CFD Direct integration of chemistry Species transport General gas phase chemistry, combustion Surface - gas phase interaction x x DARS TIF Transient flamelet solver Diesel engine modeling x Flame speed library Calculated with DARS Basic using detailed chemistry Any combustion model requiring laminar flame speed G-equation ECFM PCFM Flamelet library Calculated with DARS Basic using detailed chemistry Any presumed PDF model PPDF Soot NOx PPDF PCFM NOx AI tables Calculated with DARS Basic using detailed cemistry PVM ECFM-3Z-TKI x x 32