Revisit of Diesel Reference Fuel (n-heptane) Mechanism Applied to Multidimensional Diesel Ignition and Combustion Simulations

Transcription

1 Seventeenth International Multidimensional Engine Modeling User's Group Meeting at the SAE Congress, April,, Detroit, Michigan Revisit of Diesel Reference Fuel (n-heptane) Mechanism Applied to Multidimensional Diesel Ignition and Combustion Simulations Feng Tao * Department of Applied Mechanics, Chalmers University of Technology, Göteborg, SE- 9, SWEDEN Rolf D. Reitz and David E. Foster Engine Research Center, University of Wisconsin Madison, Madison, WI, U.S.A. ABSTRACT - Two reduced n-heptane mechanisms, previously developed in 998 ( species and reactions) and in ( species and reactions) respectively, were revisited and the mechanism improvement was pursued. In the newly revised version, the mechanism consists of 8 species and reactions. The mechanism was validated against the auto-ignition delay times of n-heptane/air mixture measured in shock-tubes (equivalence ratio ranging from. to. at bar, from. to. at. bar and. bar), the free propagating speeds of laminar premixed flames ( atm and atm), and the chemical structure of a burner-stabilized laminar premixed flame ( atm). The mechanism was applied to the three-dimensional simulations of Cummins heavy-duty diesel engine, Sandia HSDI light-duty diesel engine, and Sandia high-pressure constant-volume chamber. All the simulations were performed without any modification of the mechanism rate constants. The results demonstrate that the mechanism is able to reproduce the cool flame phenomena under the influences of temperature and EGR. The revised mechanism could provide a building block for the future improvement/development of realistic fuel mechanisms and engine soot modeling. INTRODUCTION The Chalmers University of Technology has been working on the development of reduced n-heptane mechanisms for over years. The first work dates back to the early 99s [], but the first attempt to incorporate an n-heptane mechanism directly into the KIVA-V calculation was in 99 []. Nevertheless, the sizes of the mechanisms used in both studies are small, consisting of only species and 9 reactions [] or species and reactions [], respectively. In the autumn of 99, I (Feng Tao) joined Prof. Jerzy Chomiak s team as a PhD student and soon worked together with Assoc. Prof. Valeri I Golovitchev to re-develop an n-heptane mechanism for spray ignition, combustion and emissions formation studies. The goal of the work was to develop a computationally-affordable, kinetically-sound chemical mechanism. Partly, this work was inspired by a very comprehensive n-heptane mechanism of Curran et al. [] that was published in 998. A very early version of our mechanism appeared in late 998. This mechanism, consisting of species and reactions, is known today as the Chalmers mechanism at the Engine Research Center (ERC), University of Wisconsin - Madison. n-heptane was chosen as the fuel of the work because its cetane number (CN~) is rather close to that of typical diesel fuels (CN~). Meanwhile, a relatively wide variety of experiments were performed and data published in literature, providing a good support for the mechanism evaluation and the spray combustion modeling validation. Since ignition is the first important event of diesel reaction (ignition and combustion) processes, and since the auto-ignition of n- heptane/air mixture demonstrates a unique negative temperature coefficient (NTC) behavior, we focused our attention on the predictability of auto-ignition of n-heptane/air mixtures during the n-heptane mechanism development. The measured auto-ignition delay times of n-heptane/air mixture in shock-tubes [] were used for the purpose. This -species mechanism has since become a protocol for the larger-size mechanisms involving poly-aromatic hydrocarbons (PAHs) formation (e.g., species and reactions []) that were used in a series of studies on n-heptane spray ignition [], flame liftoff and combustion zone structure [] and diesel soot formation model development [8]. The simulations of spray ignition and combustion were performed and validated heavily relying on the IDEAL high-pressure constantvolume rig [9]. The -species mechanism was also an important element that was used to construct a primary reference fuel (PRF) mechanism [], a diesel surrogate mechanism [-], or a smaller-size mechanism [] aimed at HCCI study. During the period between -, when I worked as a postdoc researcher at ERC, I experienced a certain success when I applied the mechanism to engine simulations but, meanwhile, observed some peculiar phenomena when the * Corresponding author, ftao@chalmers.se

2 mechanism was used to study the low temperature combustion or HCCI-like combustion where advanced injection strategies or heavy EGR supplies were used. To cope with these issues, I had to content myself with very uncomfortable practice by modifying the rate constants of the n-heptane mechanism. Such practice, which broke the harmony and balance of the chemical mechanism, was essentially inconsistent to the philosophy of the mechanism development and, eventually, led me to revisit of the mechanism. Figure compared the auto-ignition delay times of n-heptane/air mixture in shock-tubes and the free propagating speeds in premixed flame environments predicted using the -species mechanism to measurements. The mechanism performance in the shock-tube auto-ignition prediction is basically acceptable (Figure a), but the predicted flame speeds are far less than the measured ones (Figure b). This indicates that further mechanism improvement is in need. A preliminary attempt to improve the mechanism was reported []. T/K 8 Ignition Delay Time, τ ign /ms - Experimental Data (Aachen) T~- [K] P =. [bar] φ =.~. Calc. n-heptane/air φ =. n-heptane/air φ =. n-heptane/air φ =. - Laminar Burning Velocity [cm/s] 9 8 (Exxon Mobil) Calc. ( species, reactions) K/T (a) Equivalence Ratio (b) Figure : Comparison of (a) the predicted (lines) auto-ignition delay times of n-heptane/air mixture in shock-tubes at bar and (b) the predicted (line) free propagating speeds of premixed laminar flames at atm and K with measurements (symbols). The mechanism consists of species and reactions. This paper summarizes the continued effort on the improvement/development of diesel reference fuel (n-heptane) mechanism. The revision was motivated not only by needs in multi-dimensional engine modeling but also by recent progress in chemical kinetics. The strategy adopted in this study was that all the multidimensional engine simulations for conventional and non-conventional operations should be performed using the same mechanism without any modification of the rate constants. The results and discussion will be briefly given in the following sections. MECHANISM DEVELOPMENT The SENKIN and PREMIX codes of the CHEMKIN package were used for the mechanism development. The species thermodynamic data and transport properties were compiled from a variety of literature sources and some were updated recently. In the currently revised mechanism, 8 species and reactions were involved. The validation of the mechanism were carried out through comparison between the predictions and the measured auto-ignition delay times of n- heptane/air mixture in shock-tubes (equivalence ratio ranging from. to. at bar, from. to. at. bar and. bar), the free propagating speeds of laminar premixed flames ( atm and 98 K; atm and K), and the chemical structure of a burner-stabilized, laminar premixed flame ( atm and 98 K). The rate constants of some reactions were adjusted to ensure good agreement between measurements and predictions. Details of the chemical mechanism will be documented elsewhere []. Figure (a-c) shows that the NTC behavior of auto-ignition of n-heptane/air mixture in shock-tubes are well predicted and they are in better agreement with measurements than the predictions performed using the -species mechanism. For the lean (φ=.) mixture, this new mechanism demonstrates even better performance than the LLNL mechanism [] except in the low-temperature (< 8 K) regime. The predicted propagating speeds of premixed laminar flames at atm and K (Figure d) and the chemical structure of a burner-stabilized, rich laminar premixed flame (Figure ) are also very satisfactory in comparison to the measured data.

3 T/K 8 T/K 8 Ignition Delay Time, τ ign /ms - Experimental Data (Aachen) T ~ - [K] P =. [bar] φ =.~. Calc. n-heptane/air φ =. n-heptane/air φ =. n-heptane/air φ =. n-heptane/air φ =. - Ignition Delay Time, τ ign /ms - Experimental Data (Aachen) T~- [K] P =. [bar] φ =.~. Calc. n-heptane/air φ =. n-heptane/air φ =. n-heptane/air φ = K/T (a) K/T (b) Ignition Delay Time, τ ign /ms T/K 8 - Experimental Data (Duisberg-Essen) T ~ - [K] P ~ - [bar] φ =.~. Calc K/T (c) n-heptane/air φ=. n-heptane/air φ=. n-heptane/air φ=. n-heptane/air φ=. - Laminar Burning Velocity [cm/s] 9 8 (Exxon Mobil) Calc. (8 species, reactions) Equivalence Ratio (d) Figure : Comparison of the predicted (lines) auto-ignition delay times of n-heptane/air mixture in shock-tubes (a) at. bar, (b) at. bar, (c) at bar, and (d) the predicted (line) free propagating speeds of premixed laminar flames at atm and K with measurements (symbols). The mechanism consists of 8 species and reactions. SANDIA HIGH-PRESSURE, CONSTANT-VOLUME VESSEL The experimental conditions of Pickett and Idicheria [] were used for the study. The experiments were performed in an optically-accessible, high-pressure constant-volume combustion vessel at Sandia Lab. The vessel has a cubical chamber, 8 mm on one side, and it was designed to simulated quiescent diesel engine conditions. In their study [], ambient gas compositions, density and temperature were controlled to simulate the effects of EGR, ambient pressure and temperature on spray ignition, flame liftoff and soot formation. The detailed descriptions of the experiments can be found in Ref. [] or relevant publications. The 8-species mechanism was applied to the three-dimensional simulations, which were carried out using the KIVA-V code of ERC version. In order to evaluate the mechanism over a wide range of conditions, effects of ambient oxygen concentrations, temperature, density and injection pressure on the n-heptane spray ignition and combustion were the focus of the present study. As an example, Figure illustrates the comparison of the predicted spray penetration, formaldehyde (CH O) iso-surface at the concentration of g/m and temperature distribution for different ambient oxygen concentrations at the instant. ms. The effects of different oxygen concentrations on the flame temperature, flame liftoff, and formaldehyde formation are noticeable, and comparison indicates that all the simulations are in good agreement with experimental observations. More detailed description on the simulated results will be provided elsewhere []. Nevertheless, the animated results will be illustrated at the conference site.

4 CO.. O. CO. n-c H x -.x - H O.x - -C H.x - -C H..8 H.x -.x -.x -.x x -...x - -C H...x -. -C H 8. C H.x CH C H. C H Figure : Comparison of predicted (lines) and measured (symbols) mole fractions profiles in the rich n-heptane/o/n flame. The equivalence ratio is.9, the ambient pressure atm and temperature 98 K. The mechanism consists of 8 species and reactions.

5 CUMMINS HEAVY-DUTY DIESEL ENGINE The engine installed at the ERC labs is a single-cylinder, direct-injection (DI), -stroke, Cummins N-series production engine, which is representative of modern heavy-duty size-class diesel engines. It features low-swirl, simulated turbocharging, valves, and a centrally located direct-injection combustion system. The engine has a bore of mm and a stroke of mm, with a compression ratio of.:. It has a piston with a scalloped design and a cylinder head with a flat combustion face (non-scalloped head). A summary of the engine specifications was listed in Table of Ref. [8]. Extensive discussions and motivations for studies on this engine can also be found in [8]. A o -sector mesh (Figure in [8]) was used for the simulation of the present study. The benchmark experiment (Table in [8]) is representative of conventional diesel operation and the modeling has already been performed using the standard ERC KIVA-V code, in which the Shell ignition model and the characteristictime-combustion (CTC) model were used. In the current study, the n-heptane mechanism coupled with diesel fuel properties was incorporated into the KIVA-V simulation. Although such a coupling is still an open question, the performance of the mechanism in the simulation is very satisfactory as illustrated in Figure, in which the predicted and measured pressure and heat release rate curves are compared. Possibly, the prediction of ignition timing and combustion phasing might be further improved if a real diesel mechanism would have been used in the simulation. SANDIA LIGHT-DUTY DIESEL ENGINE This Sandia light-duty diesel engine with optical access had been a subject of low-temperature combustion study in a previous work [9] where the simulations were carried out using the ERC standard KIVA-V code. The engine has a bore of 9. mm and a stroke of 8 mm, with a compression ratio at 8.. The engine specifications and engine operations were listed in Table - of Ref. [9]. A o -sector mesh (Figure in [9]) was used for the simulations. The effect of EGR was one of the major focuses in the previous study. It was observed that, when the EGR level is low, the predictions were in good agreement with measurements. Nevertheless, when the EGR level becomes higher above %, the discrepancies between the predicted and measured pressures around ignition time becomes substantial, and at very high EGR (i.e., % and 8%), almost unacceptable. The cause of such a discrepancy was mainly due to the Shell ignition model, which is incapable of reproducing the cool-flame ignition reactions. In the present study, modeling of cool-flame ignition behavior under the influence of various EGR levels was the major focus. Figure shows comparison between the predicted and measured pressure and heat release rate traces. The red curves are the predicted results [9] obtained previously using the standard ERC KIVA-V code, and the blue curves corresponds to those obtained using the newly revised 8-species and -reaction mechanism. Among all the eight cases, the first event of diesel reactions, i.e., the normal ignition at low/medium EGR and/or the cool-flame ignition at high EGR, was well reproduced by the simulations, though further improvements on the proper prediction of the main combustion processes are still needed. CONCLUSIONS The previously developed n-heptane mechanism was revisited and its improvement was pursued. The newly revised mechanism consists of 8 species and reactions. The revision of the mechanism was based on comparison against the auto-ignition delays of n-heptane/air mixture in shock-tubes, the free propagating speeds of laminar premixed flames, and the chemical structure of a burner-stabilized, rich laminar premixed flame. The evaluation shows that the mechanism is affordable for multidimensional diesel engine simulations. It has been successfully applied to simulate the n-heptane spray ignition, flame-liftoff and combustion in Sandia high-pressure constant-volume vessel under a variety of EGR, ambient temperature and density conditions, the conventional diesel ignition and combustion processes in Cummins heavy-duty diesel engine, and the cool-flame ignition and lowtemperature combustion in Sandia HSDI diesel engine. The revised mechanism has demonstrated good potential of providing a building block for the future improvement and/or development of realistic fuel mechanisms and engine soot modeling.

6 (a) (b) (c) (d) (e) Figure : The simulated temperature distributions showing the lifted flames of a single spray in a high-pressure constantvolume vessel for the five different ambient oxygen concentrations: (a) 8% (b) % (c) % (d) % (d) %. The simulations were performed using a chemical mechanism of 8 species and reactions. The fuel is n-heptane, the orifice diameter μm, and the injection pressure difference bar. The other ambient conditions are: temperature K, and density.8 kg/m. The iso-surface (colored by temperature) downstream of the spray tip corresponds to formaldehyde (CH O) at a concentration of g/m. The time is. ms. Cylinder pressure [MPa] Firing pressure curve Motoring pressure curve Heat release rate Apparant Heat Release [J/deg] Temperature OH - CH O (a) (b) Figure : (a) Comparison of the predicted (red line) and measured (black line) cylinder pressures and heat release rates, (b) the side-view snapshots of temperature, OH and CH O distributions at -. [CA ATDC]. The engine specifications and the benchmark experimental conditions are described in Tables - in [8]. The TDC temperature and pressure are 99 K and ~ MPa, respectively. The engine speed is rev/min. The mechanism used in the simulation consists of 8 species and reactions.

7 w/o EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-. o CA ATDC Rs=. rpm % EGR SOI=-9. o CA ATDC Rs=. rpm Figure : Comparison of computed and measured in-cylinder pressures and the heat release rates for different EGR levels: (A) %, (A) %, (A) %, (A) %, (A) %, (A) %, (A) %, (A8) 8% (see Table in Ref. [9]). The mechanism consisting of 8 species and reactions was used in the simulations (indicated by blue curves). The red curves were the results simulated using the standard ERC KIVA-V code in which the Shell ignition model and the characteristic-time-combustion (CTC) model were used.

8 Acknowledgements: Half of the work was performed at University of Wisconsin - Madison and the rest continued at Chalmers University of Technology. Comments from Profs. Ingemar Denbratt, Jerzy Chomiak and Valeri I. Golovitchev are gratefully appreciated. References: [] Karlsson, J.A.J., Modeling Auto-Ignition, Flame Propagation and Combustion in Non-Stationary Turbulent Sprays, Chalmers University of Technology, PhD dissertation (99). [] Nordin, P.A.N., Golovitchev V.I., Numerical Evaluation of n-heptane Spray Combustion at Diesel-Like Conditions, The seventh International KIVA Users Meetings at the SAE Congress, Detroit, Michgan (99). [] Curran, H.J., Gaffuri, P., Pitz, W.J., Westbrook, C.K., A Comprehensive Modeling Study of n-heptane Oxidation, Combust. Flame, Vol., pp.9- (998). [] Ciezki, H.K., Adomeit, G., Shock-Tube Investigation of Self-Ignition of n-heptane/air Mixtures under Engine Relevant Conditions, Combust. Flame, Vol.9, pp.- (99). [] Feng Tao, Numerical Modeling of Soot and NOx Formation in Non-Stationary Diesel Flames with Complex Chemistry, Chalmers University of Technology, PhD dissertation (). [] Feng Tao, Valeri I. Golovitichev, Jerzy Chomiak, Self-Ignition and Early Combustion Process of n-heptane Sprays under Diluted Air Conditions: Numerical Studies Based on Detailed Chemistry, SAE Paper --9 (). [] Feng Tao, Jerzy Chomiak, Numerical Investigation of Reaction Zone Structure and Flame Liftoff of DI Diesel Sprays with Complex Chemistry, SAE Paper -- (). [8] Feng Tao, Valeri I. Golovitichev, Jerzy Chomiak, A Phenomenological Soot Model for the Prediction of Soot Formation in Diesel Spray Combustion, Combust. Flame, Vol., pp. -8 (). [9] Koss, H.J., Brüggemann, D., Wiartalla, A., Bäcker, H., Breuer, A., Investigations of the Influence of Turbulence and Type of Fuel on the Evaporation and Mixture Formation in Fuel Sprays, IDEA Subprogramme A Report (99). [] Ogink, R., Computer Modeling of HCCI Combustion, Chalmers University of Technology, PhD dissertation (). [] Rente, T., Injection Strategies for Heavy Duty DI Diesel Engines, Chalmers University of Technology, PhD dissertation (). [] Edman, J., D Modeling of Conventional and HCCI Diesel Combustion, Chalmers University of Technology, PhD dissertation (). [] Amar Patel, Song-Charng Kong, Rolf D. Reitz, Development and validation of a reduced reaction mechanism for HCCI engine simulations, SAE Paper --8 (). [] Feng Tao, Rolf D. Reitz, Werner Willems, Numerical Characterization of CO and HC Formation Structure in an HSDI Diesel Engine Combustion under Part-Load, Split-Injection Conditions, THIESEL Conference on Thermoand Fluid Dynamic Processes in Diesel Engines, September -,, Valenci, Spain. [] Feng Tao, Comprehensive Modeling of n-heptane Auto-Ignition and Flames in Idealized Environments, Manuscript in Preparation (). [] Pickett L.M., Idicheria, C.A., Effects of Ambient Temperature and Density on Soot Formation under High-EGR Conditions, THIESEL Conference on Thermo- and Fluid Dynamic Processes in Diesel Engines, September -,, Valenci, Spain. [] Feng Tao, Rolf D. Reitz, Dave E. Foster, Lyle Pickett, Manuscript in Preparation (). [8] Feng Tao, Dave E. Foster, Rolf D. Reitz, Soot Structure in a Conventional Non-Premixed Diesel Flame, SAE Paper --9 (). [9] Feng Tao, Yi Liu, Bret A. RempelEwert, David E. Foster, Rolf D. Reitz, Dae Choi, Paul C. Miles, Modeling the Effects of EGR and Injection Pressure on Soot Formation in a High-Speed Direct-Injection (HSDI) Diesel Engine Using a Multi-Step Phenomenological Soot Model, SAE Paper -- (). 8