High-Pressure Kinetic Mechanisms for Hydrogen and Hydrogen Syngas

Transcription

1 High-Pressure Kinetic Mechanisms for Hydrogen and Hydrogen Syngas 1 st International Workshop on Flame Chemistry Warsaw, Poland July 28, 212 Michael P. Burke Chemical Sciences and Engineering Division, Argonne National Laboratory Frederick L. Dryer Department of Mechanical and Aerospace Engineering, Princeton University Other collaborators: Yiguang Ju, Marcos Chaos, Jeffrey Santner, Francis M. Haas Stephen Klippenstein, Lawrence Harding

2 Motivation Growing interest in computational engine design/testing Fluid mechanics and kinetics sub-models H 2 and H 2 /CO Synthesis gas (H 2 /CO/H 2 O/CO 2 ) from coal/biomass gasification Core sub-model for all fuels Advanced engine technologies High P, low T f Modeling difficulties for flames Mass burning rate (g cm -2 s -1 ) C 2 H 2 /air, φ=.43 T f ~ 16K (Shi & Reitz 21) USC-MECH II USC-MECH II with Davis H2 Subset USC-MECH II with GRI 3. H2 Subset Pressure (atm) (Burke et al. 211) Y. Shi, R.D. Reitz, Fuel 89 (21) M.P. Burke, M. Chaos, F.L. Dryer, Y. Ju, Combustion and Flame 157 (21) M.P. Burke, F.L. Dryer, Y. Ju, Proceedings of the Combustion Institute 33 (211)

3 Difficulty in predicting high-pressure flames Mass burning rate (g cm -2 s -1 ) H 2 /O 2 /He, φ=.85 T f ~16K Present experiments Tse et al. (2) Konnov (28) Li et al. (27) Saxena & Williams (26) Sun et al. (27) O'Connaire et al. (24) Davis et al. (25) GRI-MECH 3. (1999) Pressure (atm) Mass burning rate (g cm -2 s -1 ) H 2 /O 2 /Ar, φ=2.5 T f ~16K Present experiments Li et al. (27) O'Connaire et al. (24) Saxena & Williams (26) Davis et al. (25) Konnov (28) Sun et al. (27) GRI-MECH 3. (1999) Pressure (atm) Large variations among models None of the models capture pressure dependence across all conditions M.P. Burke, M. Chaos, F.L. Dryer, Y. Ju, Combustion and Flame 157 (21) M.P. Burke, F.L. Dryer, Y. Ju, Proceedings of the Combustion Institute 33 (211)

4 H mole fraction What controls high-p/low-t f flames? 1 atm 1 atm 2 atm Temperature (K) HO 2+O=O 2+OH HO 2+HO 2=H 2O 2+O 2 O+H 2 =H+OH HO 2+H=OH+OH HO 2+OH=H 2O+O 2 H 2+OH=H 2O+H H+O 2 =O+OH H+O 2(+M)=HO 2(+M) Sensitivity Coefficient H 2 /O 2 /He, φ=.3 T f ~14K 1 atm 5 atm 1 atm Flux (mole/cm 3 /s) H+O 2 M.P. Burke, M. Chaos, F.L. Dryer, Y. Ju, Combustion and Flame 157 (21) M.P. Burke, F.L. Dryer, Y. Ju, Proceedings of the Combustion Institute 33 (211) OH+O (R1) +M HO 2 (R2) increasing P More HO 2 more HO 2 +radical flux Flame zone shifts peak sensitivity at higher T s collision efficiencies of products More R1/R2 competition amplified sensitivity (Situation similar for H 2 /CO) 4

5 Complexity of the modeling problem HO 2 +O=O 2 +OH HO 2 +HO 2 =H 2 O 2 +O 2 O+H 2 =H+OH HO 2 +H=OH+OH HO 2 +OH=H 2 O+O 2 H 2 +OH=H 2 O+H H+O 2 =O+OH H+O 2 (+M)=HO 2 (+M) Sensitivity Coefficient H 2 /O 2 /He, φ=.3 T f ~14K 1 atm 5 atm 1 atm Uncertainty in all reactions of 1% burning rate uncertainty of 3% Realistic accuracy improvements for elementary reactions will not yield typical expected accuracies for global behavior Optimization against global targets necessary Mass burning rate (g cm -2 s -1 ) H 2 /O 2 /He, φ=.5, T f ~ 14K Li et al. Li et al. w/ (R7) Fit 1 Li et al. w/ (R7) Fit 2 OH+HO 2 =H 2 O+O Pressure (atm) Functional temperature dependence of OH+HO 2 =H 2 O+O 2 highly disputed/ unknown Parameter optimization techniques don t work if the functional dependence is not known 1. M.P. Burke, F.L. Dryer, Y. Ju, Proceedings of the Combustion Institute 33 (211)

6 Complexity of the modeling problem HO 2 +O=O 2 +OH HO 2 +HO 2 =H 2 O 2 +O 2 O+H 2 =H+OH HO 2 +H=OH+OH HO 2 +OH=H 2 O+O 2 H 2 +OH=H 2 O+H H+O 2 =O+OH H+O 2 (+M)=HO 2 (+M) Sensitivity Coefficient H 2 /O 2 /He, φ=.3 T f ~14K 1 atm 5 atm 1 atm Mass burning rate (g cm -2 s -1 ) H 2 /O 2 /He, φ=.5, T f ~ 14K Li et al. Li et al. w/ (R7) Fit 1 Li et al. w/ (R7) Fit 2 OH+HO 2 =H 2 O+O Pressure (atm) A rigorous modeling solution will likely require both: Empirical adjustments to rate constants Improved fundamental understanding of select processes Neither alone appears sufficient to solve the problem. 1. M.P. Burke, F.L. Dryer, Y. Ju, Proceedings of the Combustion Institute 33 (211)

7 Updated kinetic-transport models H 2 : Hong et al. (211) and Burke et al. (212)* HO 2 formation/consumption H+O 2 (+M) = HO 2 (+M) HO 2 +radical reactions H 2 O 2 reactions among others CO: Haas et al. (212) CO + OH = CO 2 + H, CO + HO 2 = CO 2 + OH HCO chemistry *Uncertainties remained: adjustments of rate parameters to improve predictions Z. Hong, D.F. Davidson, R.K. Hanson, Combust. Flame 158 (211) M.P. Burke, M. Chaos, Y. Ju, F.L. Dryer, S.J. Klippenstein, Int. J. Chem. Kinet. 44 (212) F.M. Haas, S. Vranckx, M. Chaos, R.X. Fernandes, F.L. Dryer (212) in preparation. 7

8 Model performance [O 2 ] τ ig (mol liter -1 µs) OH mole fraction (ppm) Skinner and Ringrose (1965) - 5 atm Schott and Kinsey (1958) - 1 atm Petersen et al. (1995) - 33 atm Petersen et al. (1995) - 64 atm Petersen et al. (1995) - 57 & 87 atm Present model Hong et al. 1/T (K -1 ) Hong et al. (21) Present model -.7ppm H Present model -.7ppm H, +/- 23K Hong et al. -.7ppm H Mass burning rate (g cm -2 s -1 ).25 a) φ =.7 Burke et al. (21, 211) b) φ = 1. c) φ = 2.5 Present model Hong et al. Burke et al. (212) Haas et al. (212) Time (ms) Pressure (atm) (Santer et al. 212, 5E1 on Friday) Hong/Burke perform similarly well against most targets Largest differences in flames Burke et al. within 2%, Hong et al. within 4% Parameter adjustments not unique uncertainties remain! Z. Hong, D.F. Davidson, R.K. Hanson, Combust. Flame 158 (211) M.P. Burke, M. Chaos, Y. Ju, F.L. Dryer, S.J. Klippenstein, Int. J. Chem. Kinet. 44 (212) F.M. Haas, S. Vranckx, M. Chaos, R.X. Fernandes, F.L. Dryer (212) in preparation. J. Santner, F.L. Dryer, Y. Ju, Proc. Combust. Inst. (212) in press, oral presentation : 5E1 on Friday. 8

9 Uncertainties remaining in 212 (for flames) Parametric uncertainties HO 2 + X reactions HO 2 + H = OH + OH H2 + O 2 H 2 O + O HO 2 + OH = H 2 O + O 2 HO 2 + HO 2 = H 2 O 2 + O 2 H + O 2 (+M) = HO 2 (+M) Pressure dependence 3 rd body efficiencies for H 2 O and CO 2 CO + O + M = CO 2 + M Model assumptions Nonlinear mixture rules M.P. Burke, M. Chaos, Y. Ju, F.L. Dryer, S.J. Klippenstein, Int. J. Chem. Kinet. 44 (212) F.M. Haas, S. Vranckx, M. Chaos, R.X. Fernandes, F.L. Dryer (212) in preparation. P. Saxena, F.A. Williams, 7th US National Combustion Meeting, Atlanta, GA,

10 Recall the complexity of the modeling problem and uncertainties in OH+HO 2 = H 2 O+O 2 HO 2 +O=O 2 +OH HO 2 +HO 2 =H 2 O 2 +O 2 O+H 2 =H+OH HO 2 +H=OH+OH HO 2 +OH=H 2 O+O 2 H 2 +OH=H 2 O+H H+O 2 =O+OH H+O 2 (+M)=HO 2 (+M) Sensitivity Coefficient H 2 /O 2 /He, φ=.3 T f ~14K 1 atm 5 atm 1 atm Mass burning rate (g cm -2 s -1 ) H 2 /O 2 /He, φ=.5, T f ~ 14K Li et al. Li et al. w/ (R7) Fit 1 Li et al. w/ (R7) Fit 2 OH+HO 2 =H 2 O+O Pressure (atm) A rigorous modeling solution will likely require both: Empirical adjustments to rate constants Improved fundamental understanding of select processes Neither alone appears sufficient to solve the problem. 1. M.P. Burke, F.L. Dryer, Y. Ju, Proceedings of the Combustion Institute 33 (211)

11 Modeling strategies Current kinetic models: sets of rate parameters Hierarchical, comprehensive modeling Westbrook & Dryer (1984) Optimization and Uncertainty Quantification Frenklach (1984), Frenklach,Wang,Rabinowitz (1992):Solution-mapping + optimization of A-factors Frenklach et al. (24), Sheen & Wang (29): Uncertainty Quantification of A-factors Turányi et al. (212), Sheen et al. (212): Uncertainty quantification of A-n-E a Require massive amounts of data to constrain full T/P/M-dependence of all k s Extrapolation outside the dataset very challenging Direct incorporation of theory useful Replaces fitting formulas with physical theories Common for extrapolation of data for a single reaction Imposes constraints spanning all T/P/M Multi-scale models: sets of molecular parameters Optimal use of information from ab initio calculations, k measurements, combustion measurements Theory fills in the gaps across all T/P/M 1. M.P. Burke, S.J. Klippenstein, L.B. Harding, Proceedings of the Combustion Institute (212) in press. 11

12 OH + HO 2 = H 2 O + O k (cm 3 mol -1 s -1 ) 1 13 OH + HO 2 = H 2 O + O 2 (R1) H 2 O 2 (+M) = OH+OH(+M) (R2) H 2 O 2 +OH = HO 2 +H 2 O (R3) HO 2 +HO 2 = H 2 O 2 +O 2 (R4) HO 2 +OH = H 2 O+O 2 (R5) OH+OH = O+H 2 O Peeters and Mahnen (1973) DeMore (1979) Lii et al. (198) Cox et al. (1981) Kurylo et al. (1981) Braun et al. (1982) DeMore (1982) Goodings & Hayhurst (1988) Keyser (1988) Hippler & Troe (1992) 1/T (K -1 ) Hippler et al. (1995) Kappel et al. (22) Srinivasan et al. (26) Hong et al. (21) Keyser (1988) Sivaramakarishnan et al. (27) Chaos & Dryer (28) - Hippler Chaos & Dryer (28) - Kappel Rasmussen et al. (28) Mass burning rate (g cm -2 s -1 ) H 2 /O 2 /He, φ=.5, T f ~ 14K Li et al. Li et al. w/ (R7) Fit 1 Li et al. w/ (R7) Fit Pressure (atm) 12

13 Multi-scale informatics set of molecular parameters informed by data across all scales I. Molecular data E, v s, v imag, TST, RRKM-ME, II. Rate constant measurements k n (T,P,M) -D reactor, 1-D flame, III. Combustion measurements [OH] vs. t,s u, Mathematical implementation (I) + (IV) Local surrogate model Least-squares error minimization Iterated until converged Sij = δ ij X = Optimization parameters: Molecular parameters + experimental conditions E, v s, v imag, + T, P, [M i ], F ( X ) ± Z i j = Yt, i i (II) (III) S ij = n S ij lnk ln k p, n ( Ti, Pi, M i ) = X p, n Fi ( T, P, M ) i i i j lnk p, n ( T, P, M ) i X j i i ~ S = Y ± ij ( X j X j ) j i Z i X j,opt and C X 13

14 Implementation for H 2 O 2 system Optimization variables H 2 O 2 (+M) = OH+OH(+M) A' (1), n (1), E (1) H 2 O 2 +OH = HO 2 +H 2 O E (2), v' all(2), v' tr(2), v' ss(2), v' imag(2), E w(2), η' H2O2, η' TS(2) 1 14 OH + HO 2 = H 2 O + O 2 HO 2 +HO 2 = H 2 O 2 +O 2 HO 2 +OH = H 2 O+O 2 OH+OH = O+H 2 O Shock-heated H 2 O 2 /H 2 O/O 2 /Ar Shock-heated H 2 O/O 2 /Ar Shock-heated H 2 O 2 /Ar Optimization Targets E (3), v' all(3), v' tr(3), v' ss(3), v' imag(3), E w(3), η' TS(3) E (4g), v' all(4), v' tr(4g), v' ss(4g), v' imag(4g), E w(4g), η' TS(4g) E (4e), v' TS(4e), v' tr(4e), v' ss(4e), η' TS(4e), f ' VRCTST,c(4) E (5g), v' all(5), v' tr(5g), v' ss(5g), v' imag(5g), E w(5g) E (5e), v' TS(5e), v' tr(5e), v' ss(5e) T' i, P' i, M' H2O2,o,i, M' H2O,o,i, M' O2,o,i T' i, P' i, M' H2O,o,i, M' O2,o,i, M' H,o,i T' i, P' i, M' H2O2,o,i, σ' 1,H2O2, σ' 2,H2O2, σ' 1,HO2, σ' 2,HO2 I. Molecular data: ab initio calculations (Klippenstein/Harding) II. Rate constant measurements: see paper III. Combustion measurements: OH(t), H 2 O(t) Shock-heated H 2 O 2 /Ar (Hong et al. 29,21) OH(t) Shock-heated H 2 O/O 2 /Ar (Hong et al. 21) abs 215nm (t) Shock-heated H 2 O 2 /Ar (Kappel et al. 22) k (cm 3 mol -1 s -1 ) 1 13 DeMore (198) DeMore (1982) Keyser (1988) Lii et al. (198) Cox et al. (1981) Kurylo et al. (1981) Braun et al. (1982) 1. M.P. Burke, S.J. Klippenstein, L.B. Harding, Proceedings of the Combustion Institute (212) in press. Kappel et al. (22) Srinivasan et al. (26) Hong et al. (21) T (K),, 14

15 Different interpretations for OH+HO Absorbance (a) Kappel et al. experiments Kappel et al. predictions A priori model Optimized Constrained model (b) Kappel et al. (22) (c) Much weaker T-dependence (Secondary reactions) Time (ms) Mole fraction (ppm) Hong et al. experiments Hong et al. predictions A priori model Constrained model Hong et al. (21) Time (ms) Lower magnitude (Arbitrary H atom doping) 1. M.P. Burke, S.J. Klippenstein, L.B. Harding, Proceedings of the Combustion Institute (212) in press. 15

16 Absorbance Mole fraction (ppm) Consistent description of OH+HO 2 = H 2 O+O 2 Kappel et al. experiments Kappel et al. predictions A priori model Optimized Constrained model Time (ms) Hong et al. experiments Hong et al. predictions A priori model Constrained model (a) (b) Kappel et al. (22) (c) Hong et al. (21) Time (ms) k (cm 3 mol -1 s -1 ) DeMore (198) DeMore (1982) Keyser (1988) Lii et al. (198) Cox et al. (1981) Kurylo et al. (1981) Braun et al. (1982) Kappel et al. (22) Srinivasan et al. (26) Hong et al. (21) A priori model Constrained model T (K) Single description consistent with: 1. Ab initio calculations 2. Low-T k measurements 3. High-T raw global data Milder T-dependence Minimum near 12 K 1. M.P. Burke, S.J. Klippenstein, L.B. Harding, Proceedings of the Combustion Institute (212) in press. 16

17 Absorbance Mole fraction (ppm) Consistent description of OH+HO 2 = H 2 O+O 2 Kappel et al. experiments Kappel et al. predictions A priori model Optimized Constrained model Time (ms) Hong et al. experiments Hong et al. predictions A priori model Constrained model (a) (b) Kappel et al. (22) (c) Hong et al. (21) Time (ms) k (cm 3 mol -1 s -1 ) Simultaneous weighting of diverse data types DeMore (198) DeMore (1982) Keyser (1988) Lii et al. (198) Cox et al. (1981) Kurylo et al. (1981) Braun et al. (1982) Kappel et al. (22) Srinivasan et al. (26) Hong et al. (21) A priori model Constrained model T (K) Theory guides experimental interpretations Raw data and careful documentation extremely powerful 1. M.P. Burke, S.J. Klippenstein, L.B. Harding, Proceedings of the Combustion Institute (212) in press. 17

18 Absorbance Mole fraction (ppm) Consistent description of OH+HO 2 = H 2 O+O 2 Kappel et al. experiments Kappel et al. predictions A priori model Optimized Constrained model Time (ms) Hong et al. experiments Hong et al. predictions A priori model Constrained model Time (ms) (a) (b) Kappel et al. (22) (c) Hong et al. (21) k (cm 3 mol -1 s -1 ) DeMore (198) DeMore (1982) Keyser (1988) Lii et al. (198) Cox et al. (1981) Kurylo et al. (1981) Braun et al. (1982) Hong et al. (212) Kappel et al. (22) Srinivasan et al. (26) Hong et al. (21) A priori model Constrained model T (K) Z. Hong, K.-Y. Lam, R. Sur, S. Wang, D.F. Davidson, R.K. Hanson On the rate constants of OH + HO 2 and HO 2 + HO 2 : A comprehensive study of H 2 O 2 thermal decomposition using multi-species laser absorption. Combustion Symposium: 5D11 M.P. Burke, S.J. Klippenstein, L.B. Harding A quantitative explanation for the apparent anomalous temperature dependence of OH + HO 2 = H 2 O + O 2 through multi-scale modeling. Combustion Symposium: 4D9 M.P. Burke, S.J. Klippenstein, L.B. Harding, Proceedings of the Combustion Institute (212) in press. Z. Hong, K.-Y. Lam, R. Sur, S. Wang, D.F. Davidson, R.K. Hanson, Proc Combust Inst (212) in press.,, 18

19 Conclusions High-pressure syngas flames Emphasize HO 2 pathways + collision efficiencies of CO 2 /H 2 O Inherently difficult to model Rigorous modeling solutions Empirical adjustments based on global targets Improved fundamental characterization Uncertainties remain in both 1) model parameters and 2) model assumptions Moving forward Incorporation of theory to fill in the gaps Raw data and careful documentation Characterization of non-idealities/uncertainties in experiments and theory 19

20 Acknowledgements This work was supported by: Director s Postdoctoral Fellowship from Argonne National Laboratory (MPB) U. S. Department of Energy, Office of Basic Energy Sciences under Contract No. DE-AC2-6CH11357 (SJK, LBH) U. S. Department of Energy, University Turbine Systems Research Program under Contract No. DE-NT752 (FLD,YJ) U.S. Department of Energy, Office of Basic Energy Sciences, Energy Frontier Research Center under Contract No. DE- SC1198 (FLD,YJ) 2

21 Thank you. Questions? 21

22 End 22