Development of an LNT reaction mechanism for the BMW 120i LNT catalyst, Phase 1. NOx Adsorption and Oxygen Storage

Transcription

1 Development of an LNT reaction mechanism for the BMW 10i LNT catalyst, Phase 1 NOx Adsorption and Oxygen Storage

2 Acknowledgements This work was a joint collaboration among Authors : Dominik Artukovic (GTI Stuttgart Gmbh) Dick Blint (N Kinetics Research, LLC) Stuart Daw (ORNL) Ryan Dudgeon (GTI) Enrico Pautasso (GTI Torino) Josh Pihl (ORNL) Syed Wahiduzzaman (GTI) 5/9/013

3 Background ORNL purchased a MY008 BMW 1-series 10i lean gasoline engine vehicle with a N43B0.0-liter, 4-cylinder engine with direct injection The vehicle was characterized on a chassis dynamometer Engine measurement results are available on the CLEERS database A core from the LNT was made available for ORNL laboratory testing Lean Gasoline Engine Reductant Chemistry During Lean NOx Trap Regeneration James E. Parks, Vitaly Prikhodko, William Partridge, Jae-Soon Choi, Kevin Norman, Shean Huff and Paul Chambon Oak Ridge National Laboratory SAE, /9/013 3

4 Includes water and CO so WGS is active CLEERS LNT Protocol Oxygen Storage Nominal fixed conditions: 30,000 1/hr SV. Objective: Establish inherent oxygen storage (as reflected in reductant demand) exclusive of NOx. Frequency: Run at the beginning of protocol testing for each material. Run No.* Temp (deg C) + Gas Mix ++ Lean period (s) Injected Reductant Regen peak (ppm) Regen period (s) C1 T max M1 0 H 10, C T max M 60 CO 10, C3.75(T max -T min )+T min M 60 CO 10, C4.5(T max -T min )+T min M 60 CO 10, C5.5(T max -T min )+T min M3 60 CO 10, C6.5(T max -T min )+T min M4 60 CO 10, C7.5(T max -T min )+T min M 60 CO 10, C8 T min M 60 CO 10, No. of cycles Footnotes: * Calibration data consists of measured post-reactor values for CO and UEGO output for runs C-C8 as the inlet gas mixture is cycled between rich and lean conditions. + T max = 500 C and T min = 150C. ++ Gas inlet mix M1= 5% HO, 5% CO, balance N Gas inlet mix M= 5% HO, 5% CO, 10% O, balance N Gas inlet mix M3= 5% HO, 5% CO, 5% O, balance N Gas inlet mix M4= 5% HO, 5% CO, 1% O, balance N /Lean- NOx/files/ LNTmap_7_18 _05.pdf Long cycle Cycles Run Temp (deg C) + Gas No. Mix ++ SV (1/hr) Lean period (s) Reductant* Max 10 minute lean cycle for protocol Regen peak (ppm)* * Regen period (s)# 1 T max 1 30,000 0 H 1, T max /3 30, CO/H 1, T max /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min /3 30, CO/H 1, (T max -T min )+T min /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min /3 30, CO/H 1, (T max -T min )+T min /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min /3 30, CO/H 1, (T max -T min )+T min /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min A/3 30, CO/H 1, (T max -T min )+T min A/4 30, None T max 1 30,000 0 H 1, T min /3 30, CO/H 1, T min /4 30, None T max 1 30,000 0 H 1, T min A/3 30, CO/H 1, T min A/4 30, None T max 1 30,000 0 H 1, T max /3 30, L 1, T max /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min /3 30, L 1, (T max -T min )+T min /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min /3 30, L 1, (T max -T min )+T min /4 30, None T max 1 30,000 0 H 1, (T max -T min )+T min /3 30, L 1, (T max -T min )+T min /4 30, None T max 1 30,000 0 H 1, T min /3 30, L 1, T min /4 30, None /9/013 4 No. of cycles

5 Protocol Issues Original protocol evaluation, Epling, Currier and Yezerets Temperatures ranging from C equally spaced in 1/T [K] to facilitate better reaction modeling 150 C, 09 C, 86 C, 393 C, 550 C For this application 600 seconds lean was not sufficient to get complete NOx storage at all temperatures Josh reran the long POx protocol and adjusted storage times to reach full NOx storage 5/9/013 5

6 New Generation Lean GDI LNT (benchmarking against CLEERS reference catalyst) CLEERS reference Lean GDI, 004, provided by Umicore Mg & Al-rich Cordierite Cordierite New LNT Lean GDI, 009, from BMW 10i vehicle Ce & Zr-rich Ba-rich Mg & Al-rich Al-rich Void: monolith channel Ce & Zr-rich New generation is more highly dispersed for most components

7 CLEERS LNT versus BMW LNT FULL SIZE BRICKS Element CLEERS ref BMW Units Prop erty cell density CLEERS ref BMW Units cells/in cells/cm mass g area cm length cm volume cm L density g/l Ceria slightly less PGM similar Ba g/l Ce g/l Zr g/l La g/l Pt g/l Pd g/l Rh g/l PGM g/l K g/l Sr g/l Common NOx storage components Dispersion and cell density most significant differences Na g/l Ca N/A 0.09 g/l 5/9/013 7

8 ORNL Laboratory Reactor Architecture Flow simulations were used to calculate measurement delays Measurement delays were calculated to vary from 0.1 second to 0.7 seconds For efficiency, the full flow architecture was not used for mechanism development 5/9/013 8

9 ORNL Laboratory Setup JOSH

10 Modeling Approach GT-SUITE software 1D, quasi-steady flow solution GTI s Advanced Adaptive chemistry solver Ref: SAE /9/013 10

11 GT-SUITE Advanced Adaptive Chemistry Solver Chemical kinetics creates stiff systems. To solve these systems robustly and accurately, the solver must be highly The solver uses adaptive : Time steps Axial steps Non-linear iterations ADAPTIVE Highest stability standards are applied: Fully coupled, no lagging BDF (Backward Differentiation Formula) integration methods Numerically efficient: Especially helpful for transients times faster than real time depending on kinetics and inlet boundary conditions 5/9/013 11

12 Water-Gas-Shift Reaction Calibration WGS calibrated at 86 C and 393 C to match steady-state CO/H ratio Inactive at 150 C and 09 C; at chemical equilibrium at 550 C Ahrrenius plot of ln(rate) vs. 1/T gave kinetic rate constant fit y 8090x 4.18

13 Oxygen Storage Protocol Measurements 60 second 10% O storage, 30 second reduction by 1% CO Balance of OSC reaction and WGS shows a temperature-dependent oxygen storage capacity Water-gas-shift reaction calibrated to match steady-state CO/H ratio Thick lines: measured Thin lines: model prediction

14 Oxygen Storage and Reduction Modeling I am not sure that Oxygen storage capacity increases with increasing temperature due to there is any balance with WGS reactions substantiation in the literation for this Maximum storage estimated from extrapolation 1.5 mol/m 3 Temperature-dependent equilibrium coverage controls storage capacity 339 Ce ( ) 1 O T 4 max( T,339) y 70.1x 1.5 y 339.1x 1

15 NOx Storage Experimental Data Measurements available for 7 different cases Not Inlet: 10% O equilibrated 5% CO at 5% 10 H O minutes N excess 300 ppm NO 10 minutes not sufficient to stabilize 300 ppm NO NO concentration 5/9/013 15

16 NO Oxidation from NO Inlet Experiments Protocol does not directly provide NO oxidation information NO /NOx reaches near steady-state in all cases NO oxidation kinetics were calibrated using NO /NOx ratio

17 NO Oxidation from NO Inlet Experiments SV = 30,000 1/hr 5/9/013 17

18 NOx Storage Modeling Equilibrium coverage method for NOx storage modeling has two major flaws NO O NO BaO Ba(NO 3 ) R kc NOx C O 0 C O sign When, R = 0, which is not supported by the data eq ( T) NOx clearly releases from the surface when oxygen is absent eq ( T) NOx When, R < 0, net rate is NOx release from surface NOx C NOx The net rate is still dependent on, which is not physical, eq ( T) NOx 5/9/013 18

19 NOx Storage Modeling NO inlet experiments calibrated first Where does the model get NO at time-0?? Reverse NO oxidation does not occur at 150 C; negligible at 09 C One-step disproportion reaction accounts for outlet NO at 09 C; 150 C approaches theoretical value at steady-state BaO 3NO Ba( NO3 ) NO 5/9/013 19

20 3 site? NOx Storage Modeling Three-step disproportion needed to match peak NO outlet concentration at 150 C BaO NO BaO NO BaO 3NO Ba( NO3 ) NO BaO NO BaO NO BaO NO Ba( NO3 ) Two site storage needed for delayed breakthrough Quite possibly the low temperature adsorption is on ceria, see Chuan et. al, Applied Catalysis B: Environmental (01) rate-limited at low temperature 5/9/013 Thick lines: measured Thin lines: model prediction 0

21 NOx Storage Modeling NO Inlet Experiments Current mechanism Two storage sites Storage + release of both NO and NO Retains satisfactory results with NO experiments Additional work needed Release at 150 C Breakthroughs at 86 C and 393 C Measured Model prediction 5/9/013 1

22 Equilibrium NOx Storage Capacity Current model predicts storage capacity both quantitatively and qualitatively Experiment Model

23 Current Mechanism R1: NO 0.5O R : BaO NO R3: BaO NO R4 : BaO R5 : BaO NO 1.5O R8: BaO 3NO R9 : BaO NO 1.5O R1 : H R13: Ce R14 : Ce R15: Ce O CO O O O NO R6 : Ba( NO ) R10 : Ba( NO ) R11: Ba( NO ) R7 : Ba( NO ) O CO Ce H NO BaO NO BaO BaO NO 1.5O BaO NO BaO NO 1.5O BaO NO H Ce Ce CO O O NO 3 3 O Ba( NO ) Ba( NO ) CO H Ba( NO ) 0.5O NO O 3 3 Ba( NO ) 0.5O Active Site Densities (moles/m 3 total catalyst volume) PGM.0 Ba 64 Ba 11 cerium 84 5/9/013 3

24 Current Mechanism All rates are turnover number based All concentrations in mol/m 3 Reaction A T a (K) R1 6.56E7 51 R R R4 1.1E R R R7. 50 R R R R R E R See Rate Form R See Rate Form R See Rate Form R1 Ae R j Ae Ta T R1 Ae R13 AC R14 AC R15 AC Ta T O CO H Ta T C C max 0,1 e e T NO H Ce T C C i 0.5 O O All other rate forms: C C K CO / G max( T,344.5) i NO eq CH C K CeO4 O4 eq 1.64 CO CeO 4 CeO 4 / G (order 1 concentrations and coverages) Ce O 4 5/9/013 4

25 Summary Experimental results at 150 o C provide significant additional information about the low temperature storage process The oxygen storage protocol is quite sufficient for developing kinetics; however, a two site storage model might be needed especially when the short cycle effects are modeled A three site model for NOx storage is needed perhaps due to NOx storage on ceria For this catalyst lean cycle times greater than 600 seconds are needed to reach steady state. In some cases times in excess of 700 seconds seem to be needed At low temperature NO and NO adsorption do not seem to be strongly connected through the NO oxidation step. NO and NO have vastly different storage pathways Prediction of the temperature dependent NO and NO storage appears to be good High temperature NOx adsorption appears to be a one site process This mechanism appears to describe both the NO and the NO adsorption at low temperature; however, experiments for NO storage in the absence of O would be very informative 5/9/013 5

26 Future Work TPR experiment for NO oxidation for better transient modeling Refine the NOx storage kinetic parameters with additional modeling Calibrate NOx reduction by CO, H, and C 3 H 6 Model the short cycle lab experiments for transient kinetics development Simulation of the engine results may be considered 5/9/013 6

27 Current Mechanism G T (1 K K K K K y K 65.5exp(961/ T ) 080 exp(361/ T ) 3.98exp(11611/ T ) exp( 3733/ T ) y ) (1 K 1 CO C3H6 3 CO C3H6 y y )(1 K 4 y NO ) y j : mole fraction 5/9/013 7