Kinetic Modeling Study of the Ignition Process of Homogeneous

Feb 11, 2013 - The surrogate fuel is modeled as a blend of n-heptane, toluene, and cyclohexane. Detailed mechanisms consisting of 1140 species and 459...
0 downloads 4 Views 2MB Size
Article pubs.acs.org/IECR

Kinetic Modeling Study of the Ignition Process of Homogeneous Charge Compression Ignition Engine Fueled with Three-Component Diesel Surrogate Gan Xiao, Yusheng Zhang, and Jing Lang* School of Energy & Power Engineering, Huazhong University of Science and Technology, Wuhan 430074, China S Supporting Information *

ABSTRACT: An improved surrogate diesel fuel composition has been proposed to simulate the autoignition time of diesel fuel under homogeneous charge compression ignition (HCCI) engine conditions. The surrogate fuel is modeled as a blend of nheptane, toluene, and cyclohexane. Detailed mechanisms consisting of 1140 species and 4590 reactions were constructed by merging well-developed available chemical kinetics substructures for each chemical species. The optimal ratio of the selected diesel surrogate fuel components, n-heptane/toluene/cyclohexane = 8:1:1, was determined using trial-and-error blend methods. Numerically, the modeled heat-release rate obtained from a zero-dimensional single-zone code for the proposed new model was intensively validated against detailed single- and two-component kinetic models together with the referenced experimental engine data. The obtained results show that the new model provides a remarkable agreement with the obtained experimental data and can capture the autoignition angle and the whole combustion process effectively. Sensitivity analysis and flux analysis were further conducted to understand the roles of the different hydrocarbon classes in diesel fuels, and the key elementary reactions involved in ignition inhibition upon the addition of cyclohexane were identified. It is demonstrated that cyclohexane is a highly reactive species and can be used as a “tuning knob” to control the delay between the two stages of combustion.

1. INTRODUCTION Even with the expected development of new and cleaner sources of energy, it is still believed that the combustion of liquid fuels will remain the main source of energy for transportation for the coming decades.1 It is therefore highly important to limit the environmental impact and reduce the polluting emissions of these fuels. During the past few decades, great efforts have been made to develop cleaner and more efficient engines.2 In particular, more attention has been paid to low-temperature combustion (LTC) regimes including premixed charge,3 reactivity controlled,4 and homogeneous charge compression ignition (HCCI) engines.5 These combustion modes are due to highly premixed combustion and are mainly controlled by fuel kinetics. Lower combustion temperatures result in NOx reduction because of the high activation energy of NO formation reactions.6 In addition, use of long ignition delay times allows adequate time for mixing prior to the start of combustion, so that rich regions are reduced and soot formation is inhibited while good thermal efficiency is maintained because of higher specific heat ratio and lower wall heat-transfer losses.7 LTC is largely controlled by the details of the fuel ignition chemistry, and the role of turbulent mixing is of secondary importance compared to that of chemical kinetics. Therefore, it is essential to have a thorough knowledge of the effects of fuel composition, physical properties, and fuel chemistry on the combustion process to improve engine performance and design fuel characteristics for advanced-concept combustion engines. Chemical kinetics is widely used to study autoignition, the subsequent heat-release process, and emission characteristics.8 In the past few decades, increasingly sophisticated and detailed chemical kinetic models that combine potentially hundreds or © 2013 American Chemical Society

even thousands of elementary reaction steps have been developed that describe the pyrolysis and oxidation of a large number of fuels with an emphasis on the quantitative prediction of combustion phenomena.9 Because practical engine fuels consist of hundreds of individual chemical components, it is challenging to measure the exact composition of each fuel batch that comes from a refinery and infeasible to consider the oxidation chemistry of all compositions for modeling targets. As a consequence, the scientific world has adopted the approach of surrogate fuels.10,11 A surrogate fuel is defined as a fuel composed of a smaller number of known concentrations of selected pure compounds whose behavior matches certain characteristics of the target fuel. The problem of finding the properties of real fuel and its chemical reaction mechanisms becomes intractable because fuel consists of a complex mixture of hundreds of medium- and high-molecular-weight hydrocarbons. The composition of the surrogate fuel is chosen such that certain important targets are close to those of real fuel. However, certain guidelines are available for the choice of surrogates based on their intended applications. These guidelines include feasibility (known chemical kinetic mechanism), simplicity (computational capability), similarity (real fluid physical and chemical properties, density, heating value, evaporation characteristics, chemical composition, C/H ratio, and ignition delay), cost, and availability.12−14 For HCCI purposes, the surrogate mixture should simulate, as accurately as possible, the ignition characteristics of the fuel (fuel reactivity Received: Revised: Accepted: Published: 3732

December 12, 2012 February 3, 2013 February 11, 2013 February 11, 2013 dx.doi.org/10.1021/ie303406k | Ind. Eng. Chem. Res. 2013, 52, 3732−3741

Industrial & Engineering Chemistry Research

Article

nisms, as well as engine experimental data. In addition, kinetic analysis was used to understand the characteristics of the surrogate constituents in controlling key processes leading to autoignition and to investigate the roles of the different surrogate components and their impact on the ignition propensity of the mixture. Such a computational study is also expected to provide insight into the fundamental understanding of the choice of neat constituents for the surrogate mixture and their relative proportions.

and ignition time), and it is also desirable that all of the chemical classes of hydrocarbons in the original fuel are represented. n-Heptane is often used as a single-component surrogate diesel fuel because it has a cetane number (CN) comparable to that of European diesel.15,16 However, CN alone does not seem to be representative enough for the ignition characteristics of diesel fuel. This is probably because the experimental determination of cetane number does not cover a real and wide range of engine conditions and because the physical delay time can also affect ignition timing.17 Hence, previous studies have suggested the consideration of a substitute that includes the linear paraffin content and aromatic compound mixture representing diesel fuel. Hernandez et al.18 selected a surrogate fuel consisting of a mixture of 50% n-heptane and 50% toluene (by mass), which was proposed to simulate the autoignition time of diesel fuel in HCCI combustion. Machrafi et al.19 investigated the autoignition of PRF (primary reference fuel) and TRF (toluene reference fuel) at inlet temperatures from 25 to 120 °C, equivalence ratios from 0.18 to 0.53, and compression ratios from 6 to 13.5. It appears that both isooctane and toluene delayed the ignition with respect to nheptane, with toluene having the strongest effect. Recently, Luo et al.20 studied n-heptane/toluene/1-hexene under low-temperature combustion conditions for a wide range of intake oxygen concentrations ranging from 21% to approximately 10%. The experimental results indicated that toluene had larger effects on ignition delay at lower intake oxygen concentrations, and the TRF20/1-hexene mixture had a better match in autoignition delay and emissions in both conventional combustion and lowtemperature combustion processes. However, until now, little research has been carried out as to whether other important fuel components, especially cycloalkanes, affect diesel HCCI autoignition and combustion processes. As a practical issue, cycloalkanes are a significant component of conventional diesel fuel (up to approximately 35%), jet fuels (up to 20%), and gasoline (up to 10%).21 Because of the increasing use of oil sands and shale, which have higher cycloalkane contents than conventional crude oil sources, the role of cycloalkane should be more important for transportation fuel in the near future. Cycloalkanes can also raise soot emission levels because they are known to dehydrogenate and produce aromatics, which can lead to the production of polycyclic aromatics that are thought to be inception sites for soot growth.22 On the other hand, cyclohexane is one of the simplest cycloalkanes, and recently, Vranckx et al.23 validated the different cyclohexane mechanisms under elevated pressures in the low-temperature regime pertinent to advanced compression combustion engine technologies such as HCCI. In that regard, cyclohexane should be used as a surrogate for this class of hydrocarbons to study its performance under diesel HCCI conditions. In the present study, n-heptane, toluene, and cyclohexane were chosen as representatives of alkanes, aromatics, and cycloalkanes, respectively. By merging well-developed detailed mechanisms, we develop a three-component kinetic model for diesel HCCI combustion. The proposed mechanism was implemented in CHEMKIN software,24 and the mixing ratio was obtained by means of a trial-and-error approach. Numerically, the results obtained from a zero-dimensional kinetic code for the proposed three-component model were intensively validated against single-component (n-heptane) and two-component (n-heptane/toluene) detailed kinetic mecha-

2. DETAILED CHEMICAL KINETIC MODELING DEVELOPMENT The main target of this development work was to predict the exact ignition timing of different diesel surrogates with detailed mechanisms under diesel HCCI conditions and analyze the chemical behavior of different components. Hence, we designed six types of surrogate fuels, including single-component, twocomponent, and three-component surrogates with different ratios. Table 1 lists the fuel properties of diesel fuel and its Table 1. Composition and Properties of the Diesel Surrogates Investigated in This Study

fuel n-heptane n-heptane + toluene surrogate A surrogate B surrogate C surrogate D diesel

n-heptane/ toluene/ cyclohexane (mass ratio, %)

cetane number

C/H mass ratio

lower heating value (MJ/kg)

density at 20 °C (g/mL)

100 50:50

56 32.5

5.25 7.1

44.3 42.32

0.679 0.773

80:10:10 70:20:10 70:10:20 60:30:10 commercial diesel fuel

47 42.3 42.7 37.6 52.1

5.6 5.9 5.6 6.3 6.8

43.82 43.45 43.73 43.03 42.65

0.709 0.727 0.718 0.745 0.838

surrogates, including cetane number, C/H ratio, lower heating value, and density. Each surrogate fuel has its own function, and the optimum ratio of two-component surrogate fuel was determined as 50:50 percent by mass, based on the work of Hernandez et al.18 According to the method developed by Cancino et al.,25−27 the “adaptation process” of the kinetic model was used to construct the diesel surrogate fuel model. It is based on merging available chemical kinetics substructures for each chemical species used in the surrogate mixture to generate a detailed kinetics model. The blending procedure is timeconsuming. When two chemical species have the same chemical formula, it is necessary to check thermodynamic databases to compare their enthalpies h, heat capacities cp, and entropies s at several temperatures. If their thermodynamic properties have the same (or similar) values, then the two chemical species are identical despite the different names or abbreviations used in the original kinetic models. Then, one of the duplicate species is removed, and the occurrences of that species is replaced by the equivalent one. Automatic mechanism checking is used to simplify the process. In this work, the initial kinetics model tailored for the studied surrogates was the n-heptane model of Curran et al.28 Then, the reactions needed to describe the oxidation of toluene proposed by Pitz et al.29 were used and introduced into the n-heptane model of Curran et al. In addition, cross reactions based on the work of Andrae et al.30 between the oxidation of alkane and 3733

dx.doi.org/10.1021/ie303406k | Ind. Eng. Chem. Res. 2013, 52, 3732−3741

Industrial & Engineering Chemistry Research

Article

3.2. Determining the Best Mixing Ratio for a New Surrogate. One of the aims of the proposed work was to determine the optimum ratio of the three single-component surrogate fuels. The current study employed a trial-and-error blend method for the formulation of surrogate mixtures based on different component autoignition propensities, chemical kinetic modeling calculations, and HCCI experimental data. The entire procedure consisted of two parts: the qualitative comparison of certain components in autoignition processes and the quantitative description the minimum value of the statistical mean square error (MSE) of the start of combustion (SOC) for all diesel surrogate models. Because alkanes have a short ignition delay compared with the other fuel components and because aromatics have been used for a long time as octane improvers and have been shown to suppress the low-temperature heat release typical of alkanes,35 the ignition characteristics of cycloalkanes are also between those of alkanes and aromatics.14 Therefore, surrogate blending with a higher n-heptane concentration was simulated first, and the n-heptane concentration was then reduced and the toluene concentration increased gradually. As can be seen in Table 1, the n-heptane mass fraction of surrogate A was 80%, that of surrogates B and C was 70%, and that of surrogate D was 60%. The compositions of many mixed fuels of various toluene and cyclohexane combinations were calculated and compared, but the results of the representative four cases including the experimental data for the normalized heat-release rates are shown in Figure 1 for certain engine operating

toluene are also of importance and were considered in the detailed model. This two-component model, consisting of 644 species and 2796 reactions, is called the “n-heptane/toluene” model in this article. The three-component mechanism was constructed by combining the two-component n-heptane/ toluene mechanism into a cyclohexane mechanism containing 1081 species and 4269 reactions reported by Silke et al.22 Finally, the proposed detailed kinetics model of threecomponent diesel surrogate fuel model was composed of 1140 chemical species including 4590 elementary reactions.

3. DIESEL HCCI EXPERIMENT AND SUGGESTIONS FOR NEW SURROGATES 3.1. Diesel HCCI Experiment and Simulation. Experiments were conducted in a four-stroke direct-injection singlecylinder diesel engine, with four valves and low compression ratio. The details of the experimental setup and the characteristics of the HCCI engine and error assessment have been discussed elsewhere.18,31,32 Some essential configurations will be reviewed in this section. Diesel HCCI conditions were achieved with very early fuel injection, an ultramultihole injector, and medium or heavy exhaust gas recirculation (EGR), which allow the fuel to mix completely with air before ignition and reduce wall wetting and the onset of knocking as much as possible. Consequently, the diesel fuel/air mixture was assumed to be homogeneous from the end of the injection process, because the physical ignition delay can be considered negligible compared to the chemical ignition delay. The engine conditions were selected to study the most effective parameters affecting the oxidation kinetics of any fuel, such as the engine load [quantified by the indicated mean effective pressure (IMEP)], the engine speed, and the charge composition (quantified by the EGR percentage). All of these parameter combinations can be used to present a variety of levels of the in-cylinder pressure and temperature, resulting in different autoignition behaviors. Further details of the engine specifications and the operating conditions can be found in the Supporting Information. Because of the large size of the detailed mechanisms and because the main objective of this work was to study the chemistry pertaining to the phasing timing, the engine system was modeled by means of simplified physical representation with detailed chemical kinetics. The simplifications have already been successfully applied to the study of HCCI combustion because the high level of dilution and the efficient mixing of the air/fuel mixture achieved in this type of engine guarantees that the autoignition process of the system is mainly driven by the kinetics.33,34 The engine was simulated by the single-zone model of the CHEMKIN 4.1 program package,24 which assumes that the temperature, pressure, and species concentrations are uniform in the combustion chamber and ignores heat transfer, turbulence effects, and charge inhomogeneities. The in-cylinder instantaneous pressure signal was used as input data for a thermodynamic diagnostic model to obtain the experimental heat-release rate (HRR), and then the same engine conditions were simulated to obtain the modeled HRR. The usefulness of the single-zone assumption is a simple tool for studying some fundamental aspects related to HCCI combustion, such as the prediction of the fuel autoignition time. The in-cylinder composition at the beginning of the cycle was computed for each set of operating conditions based on the engine equivalence ratio and the EGR rates. The equivalence ratio was estimated from the exhaust gas composition.

Figure 1. Experimental and modeled HRRs for different threecomponent diesel surrogate fuels at n = 1500 rpm, EGR = 40%, and IMEP = 4.1 bar.

conditions as an example. It can be clearly seen that the different tested surrogate fuels displayed two-stage heat release or a two-stage combustion process involving a low-temperature oxidation (LTO) stage followed by a high-temperature oxidation (HTO) stage separated by a time delay between them attributed to negative temperature coefficient (NTC). The modeled HRR evolution shows very sharp and narrow peaks as a consequence of the single-zone assumption of the CHEMKIN subroutine. By reducing the n-heptane mass ratio and increasing both the toluene and cyclohexane mass ratios, the LTO and HTO peaks can be reduced and retarded, respectively. Therefore, this phenomenon highlights the fact that the qualitative composition of the fuel can be determined 3734

dx.doi.org/10.1021/ie303406k | Ind. Eng. Chem. Res. 2013, 52, 3732−3741

Industrial & Engineering Chemistry Research

Article

evolutions for four selected engine conditions and three surrogate compositions. Both the experimental and modeled HRRs are characterized by two peaks, corresponding to the onset of cool flame (low-temperature chemistry) and complete combustion (high-temperature oxidation). Because of the single-zone character of the CHEMKIN subroutine, the model HRR evolution shows a very sharp shape (as also observed in the cumulative HRR curve) and narrow peaks. It can be seen that the three-component model was most effective at capturing the location of the peak heat release for the LTO and HTO events for the optimal surrogate blend with changing engine operating conditions; however, the calculated value for the single-component model was always earlier than the experimental data, and two-component model had the opposite effect. On the other hand, according to the cumulative HRR process, it can be observed that our suggested model captured autoignition timing and the whole combustion process effectively. In conclusion, the surrogate A model could be used as a new surrogate in modeling the n-heptane/toluene/cyclohexane fuel mixture for HCCI combustion chemistry with good accuracy.

by a simple analytical analysis providing an estimate of the alkane, cycloalkanes, and aromatic contents. In the second part, the optimum ratio of the three singlecomponent surrogate fuels was determined as that minimizing the statistical mean square error (MSE), defined as N

MSE =

∑i = 1 (SOCi(exp) − SOCi(mod))2 N

where N is the number of engine simulation runs and SOC(exp) and SOC(mod) are the experimental and modeled autoignition angles, respectively. The value of SOC is negative if ignition occurs before top dead center (TDC). In this article, as in other works,36 the start of combustion (autoignition angle) is defined as the crank angle corresponding to 10% of the cumulative heat released during the stage of the high-temperature combustion process. The optimal composition ratio of the suggested components of diesel surrogate fuel was determined by a quantitative methodology involving MSE. The MSE values were calculated for each of the six diesel surrogate fuels represented in Table 1 under 16 different engine conditions (Supporting Information). The selected criterion of the optimal n-heptane/toluene/ cyclohexane composition was calculated by comparing the modeled ignition delay angles with experimental values obtained from single-cylinder engine tests. The obtained results are presented in Figure 2, which provide the lowest MSE value

4. ANALYSIS OF KINETIC MECHANISMS To understand the kinetics involved in the ignition inhibition of the three single-component mechanisms, we divided the overall analysis into three parts: (1) sensitivity analysis to determine the key elementary reactions controlling the chemistry leading to autoignition; (2) OH flux analysis to explain the advantages of the three-component surrogate model compared to the twocomponent model, especially focused on the addition of cyclohexane; and (3) further numerical analysis to investigate the effects of compositional changes in the surrogate mixture. 4.1. Sensitivity Analysis. To further examine the fuel chemistry in the low-temperature region prior to autoignition, a sensitivity analysis was carried out under HCCI conditions. The influence of a model parameter on the selected output was evaluated through the sensitivity coefficient, defined as S̃ =

∂ϕ ∂α

where φ and α represent the output and the parameter investigated, respectively. In the present study, the sensitivity coefficients of the pre-exponential factor k of each reaction rate constant on the species concentration c was calculated in its normalized form to allow for comparison between values. The normalized matrix of the sensitivity coefficients is defined as

Figure 2. MSE values for different diesel surrogate fuel compositions.

for the suggested ratio for surrogate A, which consisted of 80% n-heptane, 10% toluene, and 10% cyclohexane by mass. Also, the MSEs of the single- and two-component surrogate fuel models are compared with those of the three single-component models, illustrating that the prediction accuracy of autoignition timing can be improved by nearly 17% relative to that of the optimum n-heptane/toluene two-component surrogate model. 3.3. Further Validation Based on Single- and TwoComponent Surrogate Fuels and Experimental Data. To provide more insight into the certainty analysis of our suggested mechanisms, the obtained results from a zero-dimensional kinetic model for our proposed three-component model were intensively validated against single- and two-component detailed kinetic models in addition to different engine experimental data. Figures 3 and 4 display, as examples, the experimental and the modeled normalized HRRs and the cumulative HRR

S̃ =

kj ∂ci ∂ ln ci = ci ∂kj ∂ ln kj

where kj is the jth reaction rate constant, ci is the ith species concentration, and ∂ci/∂kj is the sensitivity coefficient of the ith species concentration to the jth reaction rate constant. Because OH radical is a good marker for the onset of the oxidation process, the normalized sensitivity coefficient of OH radical with respect to the pre-exponential factor for surrogate A was carried out at the very early stage for the low combustion temperature. As shown in Figure 5, the most relevant reactions in this case are related to low-temperature mechanisms of nheptane and cyclohexane in the surrogate fuel mixture. At the onset of heat release, OH radicals are consumed mainly by cyclohexane abstraction reactions and n-heptane NTC-region reactions, whereas their production is primarily driven by the 3735

dx.doi.org/10.1021/ie303406k | Ind. Eng. Chem. Res. 2013, 52, 3732−3741

Industrial & Engineering Chemistry Research

Article

Figure 3. Experimental and modeled HRRs for the evolution of different diesel surrogate fuel chemical kinetic mechanisms at different engine conditions: (a) n = 1500 rpm, EGR = 40%, IMEP = 4.1 bar; (b) n = 1500 rpm, EGR = 40%, IMEP = 2.8 bar; (c) n = 2000 rpm, EGR = 60%, IMEP = 3.2 bar; and (d) n = 2000 rpm, EGR = 60%, IMEP = 5.6 bar.

Figure 4. Experimental and modeled cumulative HRRs for the evolution of different diesel surrogate fuel chemical kinetic mechanisms at different engine conditions: (a) n = 1500 rpm, EGR = 40%, IMEP = 4.1 bar; (b) n = 1500 rpm, EGR = 40%, IMEP = 2.8 bar; (c) n = 2000 rpm, EGR = 60%, IMEP = 3.2 bar; and (d) n = 2000 rpm, EGR = 60%, IMEP = 5.6 bar.

low-temperature branching path (mainly by n-heptane). Toluene, on the contrary, is a single-stage fuel with very limited reactivity at low temperatures. The most favored abstraction reactions involve the formation of the benzyl radical, which is thermally stable because of electron delocalization, and its reaction with oxygen at autoignition temperatures is thermodynamically less favored. Interestingly, it

is noted that cross reactions between n-heptane and toluene have positive sensitivity coefficients, which can increase the overall reaction rate for building up the radical pool as they transform benzyl radicals into the more reactive heptyl radicals. Taken together, the sensitivity bars of the reactions involved in the formation and decomposition of the fuel ketohydroperoxides, key low-temperature branching agents, reveal how the 3736

dx.doi.org/10.1021/ie303406k | Ind. Eng. Chem. Res. 2013, 52, 3732−3741

Industrial & Engineering Chemistry Research

Article

Figure 5. Sensitivity coefficients of the OH radical with respect to pre-exponentials at the onset of the first-stage pressure rise (in cylinder, T = 822 K, P = 41 bar, CAD = 18 BTDC).

Figure 6. Calculated HO2 and OH molar fractions during the ignition of two-component and three-component surrogates with initial conditions (in cylinder, T = 822 K, P = 41 bar, CAD = 18 BTDC).

both n-heptane/toluene two-component fuel and surrogate A fuel. It is seen that, during the early part of the first-stage activity, at