Effects of Hydrogen Addition on NOx Emissions in Hydrogen-Assisted Diesel Combustion

29 International Multidimensional Engine Modeling Users Group Meeting Detroit, MI, 19 April 29 Abstract Effects of Hydrogen Addition on NOx Emissions in Hydrogen-Assisted Diesel Combustion H. Zhang, G.K. Lilik, A.L. Boehman and D.C. Haworth The Pennsylvania State University, University Park, PA 1682 Computational fluid dynamics (CFD) has been used to explore the changes in NOx emissions with hydrogen substitution that have been observed experimentally in hydrogen-enriched diesel combustion over a range of operating conditions. In the experiments, it is found that engine-out NO tends to decrease while NO 2 increases with increasing levels of H 2 substitution. In spite of the significant simplifications and approximations that are made in the CFD model, the model is able to reproduce the experimentally observed trends for some operating conditions. A model that explicitly accounts for turbulence-chemistry interactions using a transported probability density function (PDF) method does somewhat better than a model that ignores the influence of turbulent fluctuations on mean chemical production rates, although the importance of the fluctuations is not as strong as has been reported in some other recent modeling studies. The CFD results confirm that temperature changes alone are not sufficient to explain the observed reduction in NO and increase in NO 2 with increasing H 2. The CFD results are consistent with the hypothesis that in-cylinder HO 2 levels increase with increasing H 2, and that the increase in HO 2 enhances the conversion of NO to NO 2. 1. Introduction NOx emissions from diesel engines remain a significant environmental concern. In a recent experimental study of hydrogen-assisted diesel combustion [1], a modest increase of NOx was reported with increasing levels of hydrogen addition. A particularly interesting finding was that NO 2 emissions tended to increase with increasing hydrogen enrichment, while NO emissions tended to decrease. The engine-out NO/NO 2 ratio is important because NO and NO 2 interact differently with exhaust aftertreatment systems. It was hypothesized that the in-cylinder HO 2 level increased with increasing hydrogen enrichment, and that this facilitated conversion of NO to NO 2 via the reaction NO + HO2 NO2 + OH. Three-dimensional time-dependent computational fluid dynamics (CFD) can complement experimental engine measurements. In contrast to engine experiments, detailed spatially and temporally resolved information on multiple physical quantities is readily extracted from a CFD simulation. On the other hand, significant simplifications and approximations are inherent in a CFD model. These include simplifications in the geometric configuration, physical models that must be introduced for phenomena including liquid fuel sprays, hydrodynamic turbulence, and combustion, and inaccuracies that are associated with the numerical algorithms and discretization. A judicious blend of experiment and CFD modeling can yield deeper insight than either tool used in isolation. The purposes of this modeling exercise are two-fold. First, the ability of a CFD-based model to capture the experimentally observed NOx emissions trends is assessed. Second, the CFD model is used to test a specific hypothesis that has been proposed concerning the mechanism for the observed trend of reduced NO/NO 2 ratio with increasing hydrogen enrichment. 1

2. Physical Models and Numerical methods 2.1. Liquid fuel sprays A stochastic Lagrangian formulation is used for liquid fuel sprays [2, 3]. The principal models that are available include deformation, breakup, drop drag, turbulent dispersion, collision and coalescence, vaporization and spray-wall impingement. Here the deformation, drop drag, turbulent dispersion, vaporization and spray-wall impingement models were enabled while the breakup, collision and coalescence models were disabled. In this study only TAB deformation was considered. The drag model of [4] was adopted. The spray model parameters and fuel-injector characterization were taken from Kung [5, 6]; these are representative of a modern light-duty direct-injection diesel engine. 2.2. Chemical mechanism A 71-species n-heptane/nox mechanism has been used [5, 6, 7] which is based on a 4-species skeletal n-heptane mechanism from Chalmers [8], together with NOx chemistry from Glarborg et al. [9]. The mechanism includes multiple NOx formation pathways: the thermal NO mechanism, the N 2 O intermediate mechanism and the prompt (Fenimore) NO mechanism. 2.3. Turbulence-chemistry interactions Results from two sets of simulations will be reported. In the first set, the effects of turbulent fluctuations in composition about their local mean values are ignored. In the second set, a transported probability density function (PDF) method has been used [1], with standard models for turbulent transport (gradient diffusion) and mixing (pair-exchange models). Earlier modeling studies of HCCI engines [11] and direct-injection diesel engines [5] have shown that the influence of turbulence-chemistry interactions can be significant, particularly for emissions. 2.4. Numerical algorithms The CFD code uses an unstructured, deforming mesh and a finite-volume discretization [12, 13, 14]. The discretization is implicit and first-order in time and up to second-order in space (central differencing). An iteratively implicit, pressure-based, sequential (segregated) solution procedure is used to solve the coupled system of governing equations; the pressure algorithm is patterned after SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) [15] and PISO (Pressure-Implicit Split Operator) [16, 17]. Additional Lagrangian particle algorithms and particle/mesh coupling strategies are used to solve the modeled composition PDF transport equation, in cases where the transported PDF method has been used [1, 11, 18, 19]. 2.5. Geometric configuration and operating conditions Detailed dimensioned drawings and/or CAD data for the actual engine geometry were not available for mesh generation. A fuel-injector characterization also was not available. Therefore, a highly idealized geometric representation was used for the CFD study, together with a fuel-injector specification that was adapted from a recent modeling study of a different modern light-duty direct-injection diesel engine [5, 6]. For these reasons, quantitatively accurate agreement between the CFD model and the experimental measurements is not expected. Rather, the focus will be on the ability of the model to capture the trends in NOx emissions with different levels of hydrogen enrichment that were observed experimentally. A 45-degree sector mesh of 54 cells was used to represent a generic bowl-in-piston diesel that is 2

representative of a modern small-bore automotive diesel engine. A piston top-ring-land crevice has been included, as this has been shown to be important for emissions [11, 2, 21]. The simulations begin post-intake-valve closure (138 crankangle degrees before top-dead-center - TDC) and are carried through bottom-dead-center (BDC) of expansion. The initial in-cylinder pressure and temperature are obtained from the experiments, and the initial composition is specified to correspond to the desired amount of premixed hydrogen, air, and EGR. Here EGR is simulated using as CO 2 and H 2 O. All wall temperatures are set at 45 K. The fuel-injection profile in the simulations is specified such that the total amount of liquid fuel injected is the same as in the experiment for each operating condition, and such that the instantaneous fuel-injection rate is proportional to the experimentally measured injector needle-lift profile. In all cases, there is a small pilot injection followed by the main injection. The CFD model was calibrated to match the measured pressure trace for the baseline operating condition without hydrogen by adjusting the initial conditions (pressure and temperature) and clearance height (or compression ratio). The agreement between model and measurement remains imperfect because of the several simplifications that have been made, but is expected to be satisfactory for the purposes of this study. 3. Results In this section, measured engine-out NOx emissions are compared with computed in-cylinder values at BDC. This implicitly neglects the influence of reactions that occur post-exhaust-valve opening. 3.1. Computed and measured NOx without hydrogen enrichment Computed results are from two sets of simulations in (Fig. 1): one where the influence of turbulent fluctuations on mean chemical rates has been ignored ( FV ), and one where the influence of turbulent fluctuations has been included ( PDF ). The computed NO values are lower than the measured NO values for all six modes; the quantitative agreement is best for the conventional diesel (CD) light-load cases (Mode 1 and Mode 3), and the differences between FV and PDF models are relatively small (2-3%). Computed NO 2 values are higher than the measured NO 2 values for all cases except Mode 3 (CD/36 rpm/25% max load) and Mode 6 (HECC/18 rpm/25% max load); NO 2 values from the PDF model are consistently lower than those from the FV model. NO dominates for the conventional diesel cases, but the NO/NO 2 ratio is smaller for the low temperature combustion (LTC) and high efficiency clean combustion (HECC) combustion modes. The quantitative agreement between model and measurement is far from perfect, and this is to be expected, given the significant simplifications in the geometric configuration, injector characterization, and chemistry models. Overall, the best agreement is for the light-load conventional diesel cases (Modes 1 and 3) with the PDF-based model. 3

12 12 1 NO_fv NO_pdf 1 NO2_fv NO2_pdf 8 8 NO [ppm] 6 NO2 [ppm] 6 4 4 2 2 mode1 mode2 mode3 mode4 mode5 mode6 mode1 mode2 mode3 mode4 mode5 mode6 Figure 1. Computed and measured NO and NO 2 for % H 2 for six modes. 3.2. FV model with hydrogen enrichment We next explore the ability of the model to capture the experimentally observed trends in NO and NO 2 with up to 15% hydrogen enrichment. Figure 2 show the computed and measured percentage changes (relative to % H 2 ) in NO and NO 2 with H 2 substitution levels of 2.5%, 5%, 7.5%, 1% and 15%. The model follows the experimental trends fairly well for the two conventional diesel, light-load cases (Modes 1 and 3). Results for the higher load and for the unconventional combustion modes are not as good. In particular, the model fails to capture the significant increases in NO 2 with hydrogen enrichment that are observed experimentally. Sensitivity of computed results to variations in clearance height, initial temperature, and initial pressure was explored. There was a modest improvement in the computed NO 2 versus H 2 trend with a.5 mm increase in clearance height (not shown). 2 2 2 15 1 5 15 1 5 15 1 5-5 -5 5 1 15-5 5 1 15 5 1 15 (a) Mode1: CD/18 rpm/ 25% max load (b) Mode 2: CD/18 rpm/75% max load (c) Mode3: CD/36 rpm/25% max load 5 2 2 4 3 2 1 15 1 5 15 1 5 5 1 15-5 5 1 15-5 5 1 15 (d) Mode4: CD/36 rpm/75% max load (e) Mode 5: LTC/18 rpm/25% max load (f) Mode6: HECC/18 rpm/25% max load Figure 2. Computed (FV model) and measured % changes (wrt/% H 2 ) in NO and NO 2 w/h 2 addition for six modes. 4

3.3. PDF model with hydrogen enrichment Some improvement with respect to the FV model was found for Mode 2 (CD/18 rpm/75% max load), and perhaps smaller improvements in NO for the two advanced combustion modes (Modes 5 and 6), with consideration of turbulent fluctuations using a transported PDF method (not shown). 3.4. Discussion Hypotheses that were raised in [1] with respect to the decreasing NO/NO 2 ratio with increasing H 2 enrichment can now be assessed using the CFD model results. Because the greatest consistency between model behavior and experimental results was found for the light-load conventional diesel cases, we focus our attention on the CD/18 rpm/25% max load case (Mode 1). The computed global in-cylinder HO 2 levels were found to increase with increasing H 2 (not shown). This lends credence to the hypothesis that the conversion of NO to NO 2 is enhanced with increasing hydrogen enrichment by the path HO 2+NO NO 2+OH. The bulk mean temperature, maximum temperature, and amount of mixture having a temperature greater than 17 K (an approximate threshold temperature above which thermal NO formation becomes important) vary little with hydrogen enrichment. Together with the HO 2 results, these suggest that the observed changes in NOx with H 2 addition are not the result of changes in thermal NO, but rather are the result of more subtle kinetics effects. 4. Conclusion CFD has been used to explore the changes in NOx emissions with hydrogen substitution that have been observed experimentally in hydrogen-enriched diesel combustion over a range of engine operating conditions. Significant simplifications were invoked in the CFD model compared to the real engine. These include an idealized geometric configuration, assumed fuel-injector characteristics, and simplified chemical kinetics and other physical models. In spite of these approximations, the model is able to reproduce the experimentally observed trend of decreasing NO and increasing NO 2 with increasing H 2 levels for some operating conditions. A model that explicitly accounts for turbulence-chemistry interactions using a transported PDF method does somewhat better than a model that ignores the influence of turbulent fluctuations on mean chemical production rates, although the importance of fluctuations is not as strong as has been reported in some other recent modeling studies [7, 11]. The CFD results confirm that temperature variations alone are not sufficient to explain the observed reductions in NO and increases in NO 2 with increasing H 2. The CFD results are consistent with the hypothesis [1] that in-cylinder HO 2 levels increase with increasing H 2, and that the increase in HO 2 enhances the conversion of NO to NO 2. Acknowledgement This research was supported by the U.S. Department of Energy (Instrument No. DE-FC25-O4NT42233). Reference [1] Gregory K. Lilik. Hydrogen assisted diesel combustion. MS Thesis, The Pennsylvania State University, University Park, PA 28. [2] A.A. Amsden, P.J. O'Rourke, and T.D. Butler. KIVA-II, a computer program for chemically reactive flows with sprays. Los Alamos National Laboratory Technical Report No. LA-1156-MS, 1989. [3] A.M. Lippert, S.M. Chang, S. Are, and D.P. Schmidt. Mesh independence and adaptive mesh refinement for advanced engine spray simulations. SAE Paper No. 25-1-27, 25. 5

[4] Alex B. Liu, Daniel Mather, Rolf D. Reitz. Modeling the effects of drop drag and breakup on fuel sprays, SAE Paper No. 9372 1993. [5] E.H. Kung and D.C. Haworth. Transported probability density function (tpdf) modeling for direct-jnjection internal combustion engines. SAE Paper No. 28-1-969 (28). [6] E.H. Kung. PDF-based modeling of autoignition and emissions for advanced direct-injection engines. Ph.D. Thesis, The Pennsylvania State University, University Park, PA 28. [7] E.H. Kung, S. Priyadarshi, B.C. Nese, and D.C. Haworth. A CFD investigation of emissions formation in HCCI engines, including detailed NOx chemistry. 26 Multidimensional Engine Modeling Users' Group Meeting, Detroit, MI, 26. [8] V.I. Golovitchev. http://www.tfd.chalmers.se/~valeri/mech.html. Chalmers University of Technology, Gothenburg, Sweden, 2. [9] P. Glarborg, M.U. Alzueta, K. Dam-Johansen, and J.A. Miller. Kinetic modeling of hydrocarbon/ nitric oxide interactions in a flow reactor. Combust. Flame, 115:1-27, 1998. [1] D. C. Haworth. Progress in probability density function methods for turbulent reacting flows. Prog. Energy Combustion Science. 29. Submitted for publication. [11] Y.Z. Zhang, E.H. Kung, and D.C. Haworth. A PDF method for multidimensional modeling of HCCI engine combustion: Effects of turbulence/chemistry interactions on ignition timing and emissions. Proc. Combust. Inst., 3:2763-2771, 25. [12] Haworth, D.C., EI Tahry, S. H. and Huebler, M. S. A global approach to error estimation and physical diagnostics in multidimensional computational fluid dynamics. International J. for Numer. Meth. In Fluids, 17:75-97, 1993. [13] Haworth, D.C., EI Tahry, S. H., Huebler, M. S. and Chang, S. Multidimensional port-and-cylinder flow calculations for two- and four-valve-per-cylinder engines: Influence of intake configurations on flow structure. SAE Paper No. 9257, 199. [14] Khalighi, B., EI Tahry, S. H. Haworth, D. C. and Huebler, M. S. Computation and measurements of flow and combustion in a four-valve engine with intake variations. SAE Paper No 95287, 1995. [15] Patankar, S. V. Numerical Heat Transfer and Fluid Flow. McGraw-Hill, New York, 198. [16] Issa, R. I. Solution of the implicitly discretised fluid flow equations by operator splitting. J. Comput. Phys., 62:4-65, 1986. [17] Issa, R. I., Gosman, A. D. and Watkins, A. P. The computation of compressible and incompressible recirculating flows by a non-iterative implicit scheme. J. Comput. Phys., 62:66-82, 1986. [18] Y.Z. Zhang and D.C. Haworth. A general mass consistency algorithm for hybrid particle/finite-volume PDF methods. J. Computational Physics, 194: 156-193, 24. [19] S. Subramaniam and D.C. Haworth. A probability density function method for turbulent mixing and combustion on three-dimensional unstructured deforming meshes. Internl. J. Engine Res, 1:171-19, 2. [2] M. Christensen, A. Hultqvist, and B. Johansson. The effect of piston topland geometry on emissions of unburned hydrocarbons from a homogeneous charge compression ignition (HCCI) engine. SAE Paper No. 21-1-1893, 21. [21] S.M. Aceves, D.L. Flowers, F. Espinosa-Loza, J. Martinez-Frias, R.W. Dibble, M. Christensen, B. Johansson, and R.P. Hessel. Piston-liner crevice geometry effect on HCCI combustion by multi-zone analysis. SAE Paper No. 22-1-2869, 22. 6