ECN 7 DIESEL COMBUSTION IGNITION CHEMISTRY Nick Killingsworth

ECN 7 DIESEL COMBUSTION IGNITION CHEMISTRY Nick Killingsworth, LLNL CONTRIBUTORS § § § § 1 Evatt Hawkes, UNSW Goutham Kukkadapu, LLNL Leonardo Pachano, CMT Qiyan Zhou, Poli. Mi Emma Zhao, ANL Fabien Tagliante, SNL Wai Tong Chung, Stanford Hesheng Bao, TUe LLNL-PRES-812028

Spray D Spray A SPRAY IGNITION DEPENDS ON NOZZLE DIAMETER 2 HT ignition close to spray tip HT ignition upstream of spray tip

CONTENTS § Motivation and objectives § Conclusions from previous ECN meetings § Discussion of n-dodecane chemical mechanisms § Comparison of ID from spray experiments and simulations § Conclusions and recommendations

CONCLUSIONS ABOUT IGNITION CHEMISTRY FROM ECN 5 § Large variation in NTC region § Newer mechanisms tend to have shorter ignition delays over NTC region compared to shock tube experiments § Use of Yao mechanism found to result in advanced timing at low temperatures § Cai mechanism found to result in late combustion at low-temperatures. § Polimi is good but found to be too stiff for several codes § Need for high quality fundamental data to validate mechanisms § Need more participation from chemical kinetic experts § Need to create a roadmap to aid development of future reduced mechanisms for other fuels 4 φ=1 21% O 2

MANY POSSIBLE SOURCES FOR DIFFERENCES BETWEEN 0 -D IGNITION DELAY CALCULATIONS AND SHOCK TUBE EXPERIMENTS § Experiments less sensitive to carbon chain length than kinetic models which is unexpected considering that the cetane number typically increases with carbon chain length indicating increased reactivity [1] § There is some uncertainty in the pressure history of shock tube experiments which is not modeled in 0 -D simulations [2] [3] § The equation of state can have considerable influence on ignition delay times with increasing pressure, particularly in the negative temperature coefficient region [3] 5 [1] Kukkadapu et al. , "An updated comprehensive chemical kinetic model of C 8 -C 20 n-alkanes, " in 10 th U. S. National Combustion Meeting, 2017. [2] Zhukov et al. , Autoignition of n-decane at high pressure, Combustion and Flame, Volume, 2008. [3] Kogekar et al. "Impact of non-ideal behavior on ignition delay and chemical kinetics in high-pressure shock tube reactors. " Combustion and Flame 189 (2018): 1 -11.

UPDATED DETAILED LLNL ALKANE MECHANISM IS MORE REACTIVE AT LOW TEMPERATURES § 2121 species § 8268 reactions [1] [2] § Incorporates the recent changes to the detailed n-heptane mechanism § Utilizes new experimental data and quantum chemical calculations to generate a new set of consistent reaction rate rules for large alkanes § Species above C 12 were removed in addition to reactions involving these species § Manuscript describing mechanism is in preparation 6 [1] Sarathy et al. "Comprehensive chemical kinetic modeling of the oxidation of 2 -methylalkanes from C 7 to C 20. " Combustion and flame, 2011 [2] Kukkadapu et al. , "An updated comprehensive chemical kinetic model of C 8 -C 20 n-alkanes, " in 10 th U. S. National Combustion Meeting, 2017.

TWO HYBRID MECHANISMS HAVE MANY SIMILARITIES Core mechanism (C 0 -C 4) Yao (54 species/269 reactions) 32 species/ 191 reactions (Vie et al. C&F 2015) Hybrid mechanism LLNL-Hybrid (65 species/352 reactions) 62 species/ 342 From Detailed Aramco 2. 0 mechanism Added C 4 H 7 and p. C 4 H 9 and associated reactions to accommodate the C 5 -C 12 sub -mech. Low Semi-global: 4 species/ 18 Semi-global: 3 species / 8 temperature reactions sub-mech Adapted from n-decane mech. Based on simplifying low by Bikas and Peters C&F temperature oxidation of 2001 alkanes from Battin-Leclerc Prog. Energy Comb. Sci. 2008 High 18 species/ 60 reactions temperature sub-mech Reduced from 60 species/ 522 reaction mech by You et al. Proc. Comb. Inst 2009 7 3 species / 8 reactions Shares all 3 species and 6 reactions with low temperature sub mech

TWO HYBRID MECHANISMS UTILIZED DIFFERENT TUNING PROCEDURES TARGETING DIFFERENT DETAILED MECHANISMS § Yao mechanism tuned 2 times using sensitivity analysis at the following conditions, 1. p = [20, 50, 80] bar, T = 800 K, f = [0. 5, 1. 0, 2. 0] 2. p = 20 bar, T = [800, 900] K, f = [0. 5, 1. 0, 2. 0] § Yao mechanism validated against spray A simulations with WM combustion model LLNL-Hybrid mechanism § LLNL-Hybrid fuel specific reaction parameters tuned to minimize error in 0 -D ignition delay between reduction and detailed mechanism for 1 st and 2 nd stage ignition – Tuning: 10 reaction rates and 7 stoichiometric coefficients 8 [1] Yao et al. , A compact skeletal mechanism for n-dodecane with optimized semi-global low-temperature chemistry for diesel engine simulations, Fuel, 2017 [2] Lapointe et al. 2019 https: //doi. org/10. 1016/j. proci. 2018. 06. 139 Mechanism available at https: //combustion. llnl. gov/mechanisms/alkanes/n-dodecane

NTC BEHAVIOR IS DIMINISHED AS EQUIVALENCE RATIO INCREASES, LESS SO FOR REDUCED MECHANISMS 900 K - Detailed Mechanism tign: DT=400 K tign: DT=50 K 900 K - Hybrid Mechanism 900 K - Yao Mechanism

HEAT RELEASE IN RICH MIXTURES LEADING TO HIGHER TEMPERATURES AROUND PERIPHERY AND HEAD OF SPRAY Detailed - Temperature [K] Hybrid - Temperature [K] § Converge 2. 3 RANS CFD simulations — Combustion Model • Zero-RK - WMR – 2 D Multi-Zone • Temperature = 5 K • Progress equivalence ratio = 0. 05 — Spray model • KH-RT break up • Frossling Evaporation model — Mesh details • • • 10 2 mm base cell size Minimum cell size 0. 25 mm Nozzle embedding of 0. 25 mm Velocity and Temperature AMR Peak cell count: ~240, 000

MORE TEMPERATURE RISE AT RICHER MIXTURES WITH DETAILED MECHANISM 900 K - Hybrid Mechanism 900 K - Detailed Mechanism 11 Progress Equivalence Ratio (-)

LIST OF CONTRIBUTORS Numerical Institution Spray Chem. Experimental Comb. Turb. Institution A ANL C A Yao WM RANS SNL C D A CMT WM / UFPV RANS LLNL-hybrid WM LES Yao / Frassoldati TWM / TFPV RANS A Yao FGM LES A Yao/ LLNL-hybrid/ LLNL-detailed WM RANS A D SNL D Yao D A POLIMI C D TUe LLNL 12 D Spray Different markers Technique High-speed schlieren and PLIF WM - solid markers TCI – hollow markers Institute-Mech. -Comb. model- Turb. model Yao LLNL Detailed Frassoldati RANS LES: SNL&TUe

FAIRLY GOOD AGREEMENT OF IGNITION DELAY AT SPRAY A CONDITIONS Spray A 13

MORE SPREAD IN IGNITION DELAY AT SPRAY C CONDITIONS Spray C 14

SIMULATIONS MORE SENSITIVE TO COMBUSTION MODEL AT 800 K AND 1100 K Spray D 15

SIMULATED IGNITION DELAY AVERAGED FOR EACH NOZZLE AND COMBUSTION MODEL Comparison of Spray A, C, & D 16

CONCLUSIONS § Ignition delay well predicted at Spray A conditions, by most groups – More spread for models that incorporate TCI § Ignition delay not captured as well at 800 K and 1100 K and with larger nozzles § Ignition delay earlier for the smaller nozzles, which is captured by all models § Combustion models with TCI tend to predict earlier ignition at 800 K and later at 1100 K, for all nozzles versus WM combustion models § Yao and LLNL hybrid mechanisms have similar behavior, are too reactive at 800 K, both not reactive enough at 750 K 17

RECOMMENDATIONS § Need to better understand discrepancies between shock tube data and simulations of n-dodecane over NTC region § Should try targeting intermediate species especially soot precursors. Need to determine what features are most critical. § The method used to create the LLNL-hybrid mechanism could be adapted to other fuels 18

THANK YOU FOR YOUR ATTENTION! § Please answer the questions on the Beekast survey. LLNL-PRES-812028 19 This work was partially performed under the auspices of the U. S. Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC 52 -07 NA 27344. Lawrence Livermore National Security, LLC