Combustion, Performance and Emission Analysis of an Oxygen-Controlling Downsized SI Engine
Combustion, performance et analyse des émissions d’un moteur essence suralimenté avec contrôle de l’oxygène admis
1
College of Mechanical and Electrical Engineering, Wenzhou University, Wenzhou
325035 – PR China
2
Université d’Orléans, laboratoire PRISME, 8 Rue Léonard de Vinci, 45072
Orléans – France
e-mail: zhoujxi@wzu.edu.cn – bruno.moreau@univ-orleans.fr – christine.rousselle@univ-orleans.fr – fabrice.foucher@univ-orleans.fr
* Corresponding author
In the present study, experiments were carried out in a single-cylinder downsized SI engine with different rates of oxygen (15% to 27% by volume in the total mixture of intake gases except fuel) and equivalence ratios (from 0.45 to 1). Therefore, the oxygen volume fraction is due to oxygen enrichment or nitrogen dilution. The study of the impact of controlling the oxygen concentration on the combustion characteristics and emissions was performed at 1 400 rpm, at several loads (Indicated Mean Effective Pressure (IMEP) from 400 to 1 000 kPa). For each operation point, the spark advance and the intake pressure were adjusted simultaneously in order to maintain the load and obtain a minimum value of the indicated Specific Fuel Consumption (SFC). The effect of the oxygen concentration on the engine combustion characteristics was simulated by using the commercial software AMESim, with the combustion model developed by IFP Energies nouvelles, and an adapted algorithm was used to avoid residual gas calibration. By implementing a correlation for the laminar burning velocity, the in-cylinder pressures were perfectly predicted with a maximum pressure relative error of less than 2% for almost all the operating points. The classification of engine combustion according to the Peters-Borghi diagram, gives a deeper insight into the interaction between turbulence and the flame front.
Résumé
Dans cette étude, des expériences sur monocylindre essence suralimenté ont été réalisées avec comme variable le taux d’oxygène (15 % à 27 % en volume dans le mélange total de gaz d’admission à l’exception du carburant) admis et la richesse (0,45 à 1). La fraction volumique d’oxygène représente en fait soit un enrichissement en oxygène de l’admission moteur ou de l’azote de dilution. L’étude de l’impact du contrôle de la concentration d’oxygène sur les caractéristiques de combustion et les émissions a été effectuée à 1 400 tr/mn, pour plusieurs charges, de 400 à 1 000 kPa. Pour chaque point de fonctionnement, l’avance à l’allumage et la pression d’admission ont été ajustées simultanément afin d’obtenir la charge maximale et la valeur minimale de consommation spécifique indiquée de carburant. L’effet de la concentration en oxygène dans la combustion du moteur a été simulé à partir du logiciel AMESim, avec le modèle de combustion développé par l’IFP Energies nouvelles, et un algorithme adapté a été utilisé pour éviter l’étalonnage de gaz résiduels. À partir d’une corrélation pour la vitesse de combustion laminaire, les valeurs de pression cylindre ont été prédites avec une erreur relative inférieure à 2 %, pour tous les points de fonctionnement. Le classement de la combustion selon le diagramme de Peters-Borghi a permis d’avoir une idée plus précise de l’interaction entre la turbulence et le front de flamme dans ces conditions.
© J.X. Zhou et al., published by IFP Energies nouvelles, 2016
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
List of Symbols
CFD: Computational Fluid Dynamics
EGR: Exhaust Gas Recirculation
IMEP: Indicated Mean Effective Pressure
Qcomb: Heat release by combustion
rbg: Mean radius of burnt gases
SFC: Specific Fuel Consumption
u′: Instantaneous velocity fluctuation
Vcyl: Displacement volume of cylinder
Ξ: Flame front wrinkling factor
Introduction
The objective of this paper is to study the combustion of a downsized SI engine in the case of controlled oxygen concentration at the intake. On the one hand, oxygen-enriched combustion can improve engine power density with the same intake pressure level. Thus, oxygen-enriched combustion can be used either as a booster to increase engine output or as a combustion enhancer when the engine operates at low loads or in cold-start conditions. Several studies performed with oxygen enrichment in SI engines showed positive effects such as reduction of Unburned HydroCarbon (UHC) fuel and CO [1–4], optimization of the downsized operating mode due to the direct effect on the combustion process and overall engine thermodynamics [5], and a decrease in the combustion phase [6–8]. On the other hand, a low oxygen concentration in the air (or N2 dilution) can be considered as an alternative to Exhaust Gas Recirculation (EGR), which has been widely investigated in recent years [9, 10].
The oxygen concentration in the air can be controlled by using elective permeation through a non-porous polymeric membrane. This kind of polymeric membrane is now extensively studied and was developed recently to solve a part of the CO2 sequestration problem for fixed-site power generation, for example [6, 11, 12]. It would deplete or enrich oxygen in the intake air while being compact, lightweight, embeddable in the vehicle and inexpensive by overcoming the current problems of a loop EGR. Different from the current EGR dilution system, air is pre-selected to meet the N2 dilution or O2 enrichment level as desired for the oxygen controlled by the membrane. Then, the air is recomposed with oxygen enrichment or nitrogen dilution for engine applications [13].
The first purpose of this paper is to investigate experimentally the impact of controlling the oxygen concentration on engine performance and pollutant emissions, and to give a global overview of the optimized engine operating conditions with the lowest Specific Fuel Consumption (SFC) and exhaust gas emissions. In order to provide more details about the specificity of combustion processes with oxygen control, such as turbulent flame velocity evolution during the engine cycle or flame wrinkling, the AMESim commercial software with the combustion model developed by Richard et al. [14] was used. The AMESim software provides the following advantages: first, it enables the various physical domains involved in vehicle and engine system simulation to be addressed with a high level of detail [15]. Thanks to this advantage, an AMESim model can be set up with reference to the experimental configuration; for example, intake and exhaust pipes, engine valves, turbo-compressors, and so on. Secondly, the combustion model developed by Richard et al. [14] gives an accurate description of the physical processes. Although this combustion model is based on a weak assumption for the mean flame geometry, the combustion parameters (i.e. flame surface, turbulent intensity, laminar and turbulent burning velocities) provided by the model give a realistic physical meaning for the in-cylinder combustion process.
1 Experimental Setup
1.1 Engine Characteristics
Experiments were carried out in a single-cylinder SI engine (PSA EP6) characterized by a four-valve pent-roof chamber with a displacement volume (Vcyl) of 399.5 cm3 and a compression ratio of 10.5. The bore, stroke and connecting rod lengths were 77 mm, 85.8 mm and 138.5 mm, respectively. The engine was driven by an electric motor at a fixed engine speed, and equipped with an optical encoder mounted on the main shaft, giving a 0.1 Crank Angle Degree (CAD) as resolution. A conventional spark plug with an electrode space of 1 mm was used. A timer card ensured synchronization of the various trigger signals and data acquisition systems. The engine sucks in the air through a thermal mass flow-meter and all other flows were evaluated from this reference flow. Oxygen, nitrogen and air flows were controlled by using thermal mass flow meters with an accuracy of ±0.7% for the instantaneous flow. Before the intake pipe, all the gases passed through a plenum volume, to avoid pressure oscillations inside the intake port. The isooctane quantity was regulated by using a 0-8 kg/h Bronkhorst Coriolis mass flow controller with a maximum combined standard uncertainty of ±1% for the minimum flow rate considered in the present work. To provide a premixed air-fuel mixture inside the intake pipe, the fuel was vaporized and mixed inside the intake pipe just after the plenum to obtain a homogeneous air-fuel mixture. The air-fuel mixture temperature was kept constant at 70°C by an electric heater. A scheme of the air-fuel injection system in the engine environment is presented in Figure 1.
Figure 1 Scheme of air-fuel mixture system for the single-cylinder engine. |
Cylinder pressure was recorded with a water-cooled AVL quartz pressure transducer connected to a charge amplifier at 0.1 CAD resolutions. The linearity (<±0.6%) of the transducer was verified starting from a maximum pressure of 10 MPa (> maximum pressure in the engine). In this work, the absolute cylinder pressure was deduced by equalizing the in-cylinder pressure at 20 CAD after inlet valve opening timing to the intake mean absolute pressure. Thus, 100 consecutive cycles of cylinder pressure data were acquired by a PC equipped with a National Instruments acquisition board.
The exhaust emissions of gases (NOx, HC, CH4, CO, CO2 and O2) were measured by a classical emission analyzer (from Environment S.A.) with an accuracy <2% of the measured value and 1% of full scale.
1.2 Operating Conditions
Oxygen control was obtained by adding oxygen or nitrogen to ambient air. For simplification, we assume that the air consists of 21% O2 and 79% N2. To provide an O2 percentage higher than 21%, O2 was supplied by a compressed O2 high-pressure tank. For an O2 percentage less than 21%, N2 was produced by an N2 generator. Here, the percentage is defined as the volume fraction:
with:
with:
The volume fraction of air, O2 and N2 can then be determined for a given O2 percentage as shown in Table 1. The equivalence ratio φ was also predefined during experiments based on the quantity of oxygen and not on the quantity of air:
Determination of the volume fraction of air, O2 and N2 for different cases
The global reaction of the isooctane/N2/O2 mixture can be defined as follows:
where M is the quantity of excess air or unburned fuel for lean or rich mixtures, respectively. Thus, for a given equivalence ratio , the fuel mass flow rate is directly linked to the O2 mass flow rate.
In order to compare the O2 rate with the EGR percentage for a fixed equivalence ratio, the equivalence between them was calculated by replacing the inert gases CO2 and H2O with N2. For O2 < 20.9%, combustion takes place with N2 dilution, and for O2 > 20.9%, with O2 enrichment (or ‘negative’ dilution). An example of equivalence between the O2 and N2 dilution rates for a fixed equivalence ratio is presented in Table 2. Negative N2 dilution corresponds to O2 enrichment, which also means that the N2 is ‘removed’ from the air. As many studies have been carried out with EGR, represented by N2 dilution in this study, Table 2 facilitates comparison between our data and the data reported by other studies.
Correspondence between the O2 rate and EGR percentage at the stoichiometric equivalence ratio
For the test, the IMEP was adjusted to keep a quasi-constant value, and the spark ignition was set at the optimum spark ignition timing, to allow lower SFC. The intake pressure was also adjusted during the operation, so that for a given equivalence ratio and O2 percentage, the IMEP is adjusted by advancing the spark timing and intake pressure. The covariance value of the IMEP was used as a criterion in this study to determine the lean and N2 dilution operating limits: a maximum of 5% was considered to be acceptable.
1.3 Studied Cases
The different operating points studied experimentally and numerically are presented in Figure 2a with various conditions of the IMEP and oxygen percentage for an air-fuel mixture at the stoichiometric equivalence ratio. In Figure 2b, for a fixed condition of 8 bar IMEP, the effect of different equivalence ratios was improved.
Figure 2 All different operating conditions were studied at fixed 1 400 rpm. a) In fixed stoichiometric conditions, b) in fixed 8 bar IMEP conditions. |
2 Description of The Model
The combustion model used in this work is a physical 0D model, obtained by reduction of the 3D Coherent Flame Model (3D CFM) [14]. CFM formalism distinguishes two zones: fresh and burned gases, which are separated by a flame front propagating from the burned gases toward the fresh gas mixture. Chemical reactions of fuel oxidation occur in the flame front, a very thin layer compared with all scales of the turbulent flow, and post-flame chemistry takes place in the burned gases. The main assumptions made to derive the 0D model are the following [14, 16]:
-
in each zone, the mixture composition is assumed to be homogeneous and hence to have uniform properties;
-
the pressure is uniform throughout the cylinder in both zones;
-
the gaseous mixture consists of 15 species (Fuel, N2, O2, H2, H2O, CO, CO2, NO, NO2, HC, NH3, soot, O, H and OH);
-
the gases considered are ideal gases;
-
the turbulent kinetic energy field is assumed to be uniform in the cylinder.
2.1 Heat Release Calculation
The enthalpy balance equations were described in detail in the previous study of Richard et al. [14]. The heat release by combustion is expressed as:(1)
where hfi is the formation enthalpy of species i, and and are the mass variations of this species, respectively, in the flame front (ff) and in the burned gases due to post-flame (pf) chemistry reactions. can be expressed as a function of the fuel consumption rate:(2)
where υi is the stoichiometric coefficient of gas i in the reaction of fuel oxidation through the flame, and Wi denotes the molar weight of this gas. The fuel consumption rate is then given by a flame propagation model described in the next paragraph. Finally, is computed using a reduced chemistry approach [14] accounting for CO and NO formation in the burned gas zone.
2.2 Flame Propagation Model
The consumption rate of fuel throughout the flame front depends on the turbulent flame surface and the fresh gas properties according to:(3)
where ρfg is the fresh gas density, is the fuel mass fraction in the fresh gases and is the laminar burning velocity computed from a correlation, a function of the pressure and the fresh gas composition and temperature, described in the subsection below. The turbulent flame surface is written as the product of a mean surface and the flame front wrinkling factor, ). The mean surface is computed by considering a spherical expansion of the flame from the spark plug to the cylinder walls in a realistic geometry. For this purpose, is a priori tabulated as a function of the burned gas volume and the piston position as described [14, 17]. The flame wrinkling factor evolution is described by a physical equation describing the progressive transition from the laminar kernel to the fully turbulent flame. This equation was obtained by reduction of the 3D equation for the flame surface density [18, 19] and includes the unsteady effect of the turbulent flow on the flame stretch through the efficiency function Γ proposed in [20]:(4)
u′ is the instantaneous velocity fluctuation, lt is the integral length scale, δl is the laminar flame thickness estimated using Blint’s correlation [21], is the thermal expansion rate, rbg = (3Vbg/4π)1/3 is the current mean radius of burnt gases and Ξequ corresponds to the Ξ value when an equilibrium is reached between turbulence and flame wrinkling Ξ [18]. The main unknowns of the problem finally correspond to the turbulent flow field properties u′ and lt.
2.3 Turbulence Model
The integral length scale is assumed to be close to the piston-cylinder head distance D, as proposed by many authors [17, 22], while is obtained from a classical two- equation model describing the evolution of the mean and turbulent kinetic energy (resp. K and k) [14]. This model accounts for energy transfer from the mean flow into turbulence during the compression stroke and for dissipation of k on the Kolmogorov scale. Such a 0D model cannot be entirely predictive as the complex flow field evolution highly depends on the engine geometry and cannot directly be resolved as in 3D CFD calculations. Classically, the parameters of the turbulence model such as the integral length scale and the tumble number are then usually adjusted.
2.4 Adaptation of the AMESim Model for the Case of an ‘Oxygen-Controlled Engine’
Figure 3 presents the sketch of the AMESim model. On the left is the intake part, in the middle the main engine block, containing the model of the intake and exhaust valves, the model of the 12 gases, the engine model, and on the right the exhaust part.
Figure 3 Scheme of AMESim modeling for an oxygen ratio-controlled engine. |
In this AMESim model, four parts of the subroutine (marked as ① to ④) were modified to adapt the oxygen control calculation using an advanced AMESim tool, called ‘AMESet’:
-
① and ② are pipe models which transform the mixture of 3 gases (air, fuel and burned gases) into 12 species (fuel, N2, O2, H2, H2O, CO, CO2, NO, NO2, HC, NH3 and soot) when gases pass though the intake or exhaust pipes to the engine block. Inversely, from the engine block to the intake or exhaust side, 12 gases are transformed into 3 gases. In the pipe model, an integer parameter for the O2 mole percentage was defined; thus, by changing this parameter to above or less than 21%, the air can be considered as oxygen-enriched or nitrogen-diluted, respectively;
-
③ is the model of the 12 gas engine fluid data, where instead of a stoichiometric air-fuel ratio, a stoichiometric oxygen-fuel ratio is calculated to determine the equivalence ratio based on oxygen, as defined in the previous section;
-
④ the engine combustion model, was changed by using the integer parameter for the O2 mole percentage to control the oxygen percentage in the air and as a variable for laminar burning velocity correlation and auto-ignition delay time correlation.
2.5 Laminar Burning Velocity Correlation
The laminar burning velocity for different quantities of O2 was experimentally investigated in previous studies [23, 24]. In order to take into account the effect of oxygen control on the engine combustion simulation, a correlation based on an experimental estimation [24] was implemented in the AMESim model:
In the above relationship, is a function of the equivalence ratio:
and are the reference temperature and pressure, respectively and the laminar burning velocity of these conditions. In this correlation, pressure, temperature and laminar burning velocity are respectively expressed in bar, K and cm/s.
In the study of Galmiche et al. [24], an empirical term was added to take into account both the effect of N2 dilution and O2 enrichment:
where , with two constant values of and . is the percentage in the air defined by:
-
in the case of N2 dilution:
-
in the case of O2 enrichment:
is the reference percentage of O2 in the air, i.e. the percentage of O2 in the synthetic air.
The parameters used in the correlation are listed in Table 3.
Correlation parameters used to estimate laminar burning velocity
In order to apply this correlation to the engine combustion calculation, the effect of residual gas can be defined by:
where is the residual gas proportion inside the cylinder.
2.6 Calibration Process
The calibrations were accomplished by coupling AMESim with a Matlab program, to input experimental results in the requisite format.
Figure 4 illustrates the simulation process for all the engine operating points. First, experimental data (such as intake pressure, intake temperature, exhaust pressure, exhaust gas temperature, equivalence ratio, advanced ignition crank angle and in-cylinder pressure) are read by a Matlab program. In this program, Matlab-AMESim coupling was used to optimize the in-cylinder data.
Figure 4 Global scheme of different simulation and model coupling and looping. |
The second part of this algorithm focuses on two-zone AMESim model simulation with oxygen control. In this part, the AMESim development tool, AMESet, was applied to induce laminar burning velocity as a sub-routine inside the main combustion model. In addition, the main two-zone combustion model was modified to adapt it to the oxygen control.
Multi-cycle simulations were performed with the initial ‘guess value’ of the integral scale and tumble value, and then the simulation results of in-cylinder pressure of the last simulation cycle were compared with the experimental data for a range of crank angles from 250 to 450 CAD. The discrepancy between the two values can be calculated by:
The second term of the above expression gives a more precise calibration with regard to the importance of for heat release and exhaust gas emission. The parameter gives a flexible value for considering the greater or lesser effect of . Once the error between simulation and experimental in-cylinder pressure data is minimized, the value of lt and Ntumble can be obtained by loop calculation. Finally, the above lt and Ntumble maps were used to analyze the simulation results.
The present algorithm differs from the calibration process methodology employed by Richard et al. [14] using ‘IFP-Optilab’. In the present study, the initial conditions that concern the mass fraction of air and of isooctane fuel, intake pressure and temperature are taken directly from experimental data. Residual burned gases, which have a significant impact on the combustion process (heat release, knock, pollutants, etc.), were estimated by a multi-cycle simulation with five real engine cycles. Only the results of the last cycle were used to calculate the error between in-cylinder simulation and experimental data and to analyze the results.
3 Results and Discussion
3.1 Experimental Results
3.1.1 Fuel Consumption
The evolution of indicated SFC versus the equivalence ratio is presented in Figure 5 for different IMEP (400 kPa, 600 kPa, 800 kPa and 1 000 kPa). In these figures (Fig. 5a-d), the O2 percentage in the air varies from 15 to 27%. For each O2 percentage, the equivalence ratio is decreased to reach the lean operating limits. With the decrease in the equivalence ratio, SFC was decreased for a fixed O2 percentage, although an SFC increase is usually observed at the operating limits.
Figure 5 Evolution of indicated SFC versus the equivalence ratio for different IMEP; a) 400 kPa, b) 600 kPa, c) 800 kPa, d) 1 000 kPa. |
For IMEP 400 kPa and 600 kPa (Fig. 5a, b), by increasing the O2 percentage in the air and decreasing the equivalence ratios, a lower SFC can be obtained. Compared with the conventional operating point (with 21% O2, and a stoichiometric case), SFC can be reduced by 15%. The lowest SFC is found at an O2 percentage of around 25% to 27% with an equivalence ratio of 0.5 to 0.55, which means that oxygen-enriched combustion with a low equivalence ratio can be a potential method for improving fuel efficiency.
For IMEP 800 kPa (Fig. 5c), and an O2 percentage from 15% to 19%, a lower SFC can be obtained by increasing the O2 percentage while reducing the equivalence ratio, which is similar to the cases of IMEP 400 kPa and 600 kPa. However, for an O2 percentage from 21% to 27%, this tendency is reversed. The lowest SFC for IMEP 800 kPa is found at 19% O2 with an equivalence ratio of 0.7.
For IMEP 1 000 kPa (Fig. 5d), the lowest SFC is obtained at 15% O2 with an equivalence ratio of 0.9. In this case, the dilution of N2 will be a better way than oxygen enrichment to limit fuel consumption. Compared with the regular operating point (with 21% O2, and a stoichiometric equivalence ratio), N2 dilution has the potential to achieve about 10% fuel economy.
It can therefore be concluded that for low IMEP (400 kPa, 600 kPa), oxygen enrichment is a potential way to improve fuel efficiency, whereas for high IMEP (800 kPa, 1 000 kPa), N2 dilution should be used to have a better fuel economy.
3.1.2 HC Emissions
Figure 6 represents the evolution of indicated specific HC emissions versus the equivalence ratio for two different IMEP (400 kPa, 1 000 kPa). The effect of oxygen enrichment on HC emissions is obvious. For each fixed equivalence ratio, the HC emission decreases with the increase in the O2 percentage and the lowest values are always found at 27% O2. Oxygen enrichment can decrease the quenching distance of the mixture. In the study by Friedman and Johnston [25], the quenching distance was found to be a decreasing function of the flame temperature, and it is well known that the flame temperature increases in the case of oxygen enrichment [23]. Mazas et al. [26] also showed that quenching effects at the flame front are reduced due to the increase in the flame temperature when the oxygen-enrichment ratio is increased. Thus, oxygen enrichment allows the flame to propagate much closer to the cylinder walls than the N2 dilution mixture.
Figure 6 Evolution of indicated specific HC emissions versus the equivalence ratio for different IMEP; a) 400 kPa, b) 1 000 kPa. |
The specific HC emission increases when the equivalence ratio varies from 0.9 to 1; however, the HC emission decreases from the lean limits to an equivalence ratio of 0.9. The lowest value can be found around an equivalence ratio of 0.9.
3.1.3 NOx Emissions
The evolution of NOx emissions versus the equivalence ratio is presented in Figure 7. Contrary to HC emissions, at a fixed equivalence ratio value, NOx emissions increase when the oxygen percentage is increased, with the exception of a few points which may be due to measurement error. As the formation of NOx depends mainly on the burned gas temperature, due to the thermal formation process, the higher the burned gas temperature, the higher the rate of NOx formation. Thus, the higher emission of NOx for oxygen enrichment can be explained by the high temperature of the burned gases. The stoichiometric engine operating conditions show a slight decrease in NOx emissions for the highest O2 volume fraction, contrary to the literature. This is probably due to different reasons: as the later spark timing changes the thermodynamic conditions, it may have a significant effect on NOx, especially at 10 bar IMEP, for 27% O2, where the spark timing is retarded to prevent the knock. Moreover, although the O2 enrichment produces a higher burned gas temperature and then causes higher NOx emission, a lower concentration of N2 is not conducive to NOx formation. In the case of a highly lean mixture, the leaner the mixture is, the less NOx emission it gets, even though the SFC is greatly increased (the lean limit points in Fig. 5). In the case of a lean mixture, the leaner the mixture is, the less NOx emission it gets; the interval of the operation points for the equivalence ratio is normally set to 0.05, which may be too high, and induces the fluctuation of NOx emissions. As shown in the lower part of Figure 7a, the filled circles represent the NOx with high SFC (the lean limit points in Fig. 5a). The closer the lean limit is touched, the lower the NOx emission.
Figure 7 Evolution of indicated specific NO x emissions versus the equivalence ratio with different IMEP; a) 400 kPa, b) 1 000 kPa. |
3.1.4 CO Emissions
Figure 8 represents the evolution of indicated specific CO emissions versus the equivalence ratio for different IMEP (400 kPa, 1 000 kPa). CO emissions are mainly affected by the equivalence ratio ; especially for near unit value, they decrease in lean conditions due to the increase in the O2 concentration, which helps the oxidation of CO into CO2. Compared with the equivalence ratio, the O2 percentage in the air does not present a clear tendency. In Figure 8a, the local slope of the equivalence ratio between 0.95 and 1 for 17% O2 can reach 240 (the closer it is to the unit value, the greater it is), which means that with one percent variation in the equivalence ratio, the CO emission will have more than 2.4 g/kWh discrepancy. Conversely, the biggest variation from 15% to 27% O2 is about 7 g/kWh. Thus, to ensure precise CO measurement, the equivalence ratio should be controlled and the precision of the mass flow meters of fuel and gases should be checked.
Figure 8 Evolution of indicated specific CO emissions versus the equivalence ratio for different IMEP; a) 400 kPa, b) 1 000 kPa). |
3.2 Engine Model Results
3.2.1 Engine in-Cylinder Pressure
Experimental and simulated in-cylinder pressure curves for different engine loads at 1 400 rpm are presented in Figure 9. The in-cylinder pressure evolution is plotted for different cases of the oxygen percentage at a stoichiometric equivalence ratio, 15% O2, 21% O2 and 27% O2, respectively. The discrepancies can be found in Figure 9a, especially for the cases of IMEP 8 and 10 bar, which may be due to the assumption of spherical flame propagation inside the combustion volume. For a high nitrogen dilution mixture, the flame evolution in the cylinder becomes irregular, as previously observed by Mounaïm-Rousselle et al. [9]. However, good calibration results can be found in Figure 9b, c for 21% and 27% of O2, respectively. The simulated results show good agreement with our experimental data, which indicates that the evolution of the burning velocity and therefore the heat release rate is well described for all cases [16].
Figure 9 Comparison between experimental and simulated in-cylinder pressure evolution for different engine loads. φ = 1, a) 15% O2, b) 21% O2, c) 27% O2. |
In Figure 10, the IMEP absolute error and maximum pressure relative error versus the O2 percentage for different cases of IMEP are presented. For all cases, IMEP relative errors do not exceed 3.5% and maximum pressure relative errors are less than 2%. The largest values of these errors are found at 15% O2 and the smallest at an O2 percentage above 21%. The maximum pressure relative error roughly decreases with increasing O2 percentage.
Figure 10 a) Relative error of IMEP and b) maximum in-cylinder pressure. |
In Figure 11, the integral length and the tumble number versus the O2 percentage are presented for IMEP from 400 to 1 000 kPa. As these two parameters are used to adjust the simulation, their values evolve: with the increase in the O2 percentage, they increase, except in the case of 15% O2 for the tumble number.
Figure 11 a) Integral length scale and b) tumble values obtained to optimize the in-cylinder pressure. |
Although some operating points are beyond the general tendency, the evolution of the integral length scale and the tumble number is smooth when the IMEP, equivalence ratio and O2 percentage are varied. Thus, reasonable values of lt and Ntumble can be obtained by interpolation of the data in look-up tables, in order to calculate transient engine operations.
3.2.2 Combustion Characteristics
The evolution of the calculated in-cylinder pressure at the stoichiometric equivalence ratio and IMEP = 1 000 kPa for different cases of the O2 percentage is presented in Figure 12. The in-cylinder pressure is started at the spark ignition degree with the operation conditions listed in Table 4. The increase in the O2 percentage results in a decrease in intake pressure while maintaining the same IMEP. However, for 27% O2, in order to maintain IMEP at 10 bar, the intake pressure was kept at the same value as for 25% O2 while increasing the spark timing to prevent knock. For relatively high O2 percentages, engine operation was limited by knock phenomena, which can be seen in the in-cylinder pressure trace in Figure 12. To avoid engine knock, the higher spark timing was set for higher oxygen enrichment.
Figure 12 Evolution of the mean in-cylinder pressure at the stoichiometric equivalence ratio and IMEP = 1 000 kPa. |
Operating conditions for combustion analysis
Figure 13 presents the laminar burning velocity evolution versus CAD with different O2 percentages. The laminar burning velocity increases as a function of the O2 percentage. Compared with 15% O2, the laminar burning velocity of 27% O2 is more than 8 times higher. The reasonable results of laminar burning velocity under engine operation conditions were obtained thanks to the reliability of the experimental correlation proposed by Galmiche et al. [24].
Figure 13 Laminar burning velocity versus CAD with different O2 percentages at the stoichiometric equivalence ratio and IMEP = 1 000 kPa. |
The evolution of the flame surface and flame wrinkling versus CAD for different O2 percentages at the stoichiometric equivalence ratio and IMEP = 1 000 kPa are plotted in Figure 14. A previous experimental study (Mounaïm-Rousselle et al. [9]) showed that dilution enhances flame-turbulence interactions and that the corrugation generated through dilution occurs on smaller scales than the integral length scales. This phenomenon is well reproduced in the present simulation, as shown in Figure 14a, for 15 and 17% O2, which correspond to 26.87% and 17.65% of EGR, respectively, as presented in Table 2. The flame surface is much higher than in the case of lower dilution or oxygen enrichment. This tendency is mainly due to the high flame wrinkling in the case of high dilution, as presented in Figure 14b.
Figure 14 a) Flame surface and b) flame wrinkling versus the crank angle with different O2 percentages at the unit equivalence ratio and IMEP = 1 000 kPa. |
Finally, Figure 15 shows the evolution of the turbulent flame velocity versus CAD. The flame kernel growth is initially laminar-like as described in [27], and once the flame has grown to more than 5 mm, it can become turbulent [28].
Figure 15 Evolution of turbulent flame velocity with different O2 percentages at the stoichiometric equivalence ratio and IMEP = 1 000 kPa. |
Thus, based on the above findings, as the laminar burning velocity is much lower in the case of dilution, the laminar flame state lasts longer than with oxygen enrichment: about 7.5 CAD for 15% O2, 5 CAD for 21% O2 and 2.5 CAD for 27% O2. Although substantial differences are found for laminar burning velocities (8 times higher with 27% O2 compared with 15% O2), the turbulent burning velocities show comparatively little difference (less than 2 times higher with 27% O2 compared with 15% O2). This is mainly due to the reverse evolution of flame wrinkling: the lower the O2 percentage, the higher the flame wrinkling.
In order to characterize the combustion regime in the Peters-Borghi diagram [9], the evolution of the characteristic length and velocity scale ratio versus CAD for different O2 percentages is plotted in Figure 16. As previously described, after the spark, the flame is initially ‘laminar-like’, and the duration of this phase is from about 2.5 to 7.5 CAD. Here, the curve starts at 10 CAD after spark timing (360 + ST + 10 CAD) for all the cases. A higher the value of is found for a higher value of the O2 percentage; the reverse is true for . Two CAD cases of these ratios (10 CAD and 20 CAD after spark timing) are shown in Figure 17.
Figure 16 Ratio of the characteristic length a) and velocity scales b) versus CAD for different O2 percentages. |
Figure 17 Ratio of the characteristic length a) and velocity scales b) versus the O2 percentage for two CAD cases. |
Finally, in Figure 18, the combustion is classified according the Peters-Borghi diagram, which gives a deeper insight into the interaction between turbulence and the flame front. According to the study of Linse et al. [29], engine combustion is normally located in the thin reaction zone and the corrugated flamelet zone. Generally, the combustion of the earlier generation of SI engines takes place in the corrugated flamelet regime, while the combustion of the new generation of engines takes place in the thin reaction zone [30]. A higher turbulent level and charge motion are created when turbocharging and/or direct injection are used in new generation engines, which is the main reason for the different regime between new and earlier generation engines [29].
Figure 18 Comparison of the combustion traces in the Peters-Borghi diagram for different O2 percentages. |
In Figure 18, from 15% O2 to 27% O2, the combustion trace passes from the broken reaction zone to the thin reaction zone, and finally to the corrugated flamelet zone. For the cases of 17% to 23% O2, combustion takes place in the thin reaction zone, which may occur when the tested engine is supercharged and the high turbulence level is produced. For the cases of 25% and 27% O2, the combustion traces are mainly in the corrugated flamelet zone, and for the case of 15% O2, the combustion occurs in the broken reaction zone. Mainly, flamelet theory is used for engine combustion simulation, as validated by Poinsot et al. [31] based on direct simulation. Their results show that the resistance of the flame front to vortices is underestimated by classical diagrams which neglect viscous, transient and curvature effects. The limit line of the flamelet regime determined by Poinsot et al. [31] is drawn as the v-formed green line in Figure 18. Therefore, for the case of 15% O2, the spherical propagation of the flame might not be well adapted, which certainly contributes to the imperfect agreement between simulated and experimental in-cylinder pressure (Fig. 9a).
The effect of N2 dilution on the combustion trace corresponds well with the previous experimental study of Mounaïm-Rousselle et al. [9]. Dilution moves the combustion trace toward the upper left side in the Peters-Borghi diagram, and the effect of O2 enrichment shows a reverse tendency compared with N2 dilution. For all the operating cases, the combustion trace has a ‘turning point’. Compared with the simulation results of Linse et al. [29], the ‘turning angle’ is smaller, which is caused by the simplification of the definition in the AMESim model. is supposed to be a constant value and equal to its value at spark timing during the whole combustion stroke in the combustion model [14], while normally depends on the distance between the piston and cylinder head. Nevertheless, the general tendency is in good agreement with the previous studies of Linse et al. [29] and Mounaïm-Rousselle et al. [9].
Conclusion
In this study, the effect of controlling the oxygen concentration in the air on the performance and emissions in a downsized engine is presented and discussed, using both experimental and simulation methods.
The experimental results indicate that, by increasing the O2 percentage in the air and decreasing the equivalence ratios, lower fuel consumption can be achieved for IMEP of 400 and 600 kPa. However, for 800 and 1 000 kPa, the lowest fuel consumption is obtained for 19% O2 at 0.7 of the equivalence ratio and 15% O2 at 0.9, respectively. It can be concluded that for low loads, oxygen enrichment has a real potential to improve fuel efficiency, whereas for relatively high IMEP, N2 dilution still remains the best way. For all equivalence ratios, the HC emission decreases with the increase in the O2 percentage and the lowest value can always be found at 27% O2 but in this case, NOx emissions increase.
The numerical simulation was conducted with the AMESim software by taking into account the effect of O2 control. In-cylinder pressure was well reproduced, although discrepancies can be found for high levels of dilution. The increase in the O2 percentage results in a decrease in intake pressure while maintaining the same IMEP. The laminar burning velocity increases as a function of the O2 percentage. Compared with 15% O2, the laminar burning velocity of 27% O2 is more than 8 times higher. The flame surface of high dilution cases is much greater than that with lower dilution or oxygen-enriched cases. Therefore, turbulent flame velocities show few differences compared with laminar ones. The combustion trace passes from the broken reaction zone to the thin reaction zone, and finally to the corrugated flamelet zone by varying the O2 percentage from 15% to 27%. Only for the case of 15% O2 does the combustion take place in the broken reaction zone, in which the spherical propagation of the flame may not be well adapted.
Acknowledgments
The authors acknowledge Stephane Richard for discussions to improve the AMESim software and the new procedure for residual gases. This study was financially supported by PSA Peugeot Citroën and Region Centre. This work was also financed by the MACDOC project of the French National Research Agency, TDM program, ANR-12-VPTT-0008. The authors are also grateful for the support from the Foundation of Zhejiang Educational Committee (No. Y201430810) and Natural Science Foundation of Zhejiang Province (No. LQ15E060001), China.
References
- Poola R.B., Ng H.K., Sekar R.R., Baudino J.H., Colucci C.P. (1995) Utilizing Intake-Air Oxygen-Enrichment Technology to Reduce Cold-Phase Emissions, Vol. 104, New York, NY, Society of Automotive Engineers. 15. [Google Scholar]
- Kajitani S., Clasen E., Campbell S., Rhee K. (1993) Partial-load and Start-up Operations of Spark-ignition Engine with Oxygen Enriched Air, SAE Technical Paper 932802. [Google Scholar]
- Poola R.B., Sekar R., Ng H.K., Baudino J.H., Colucci C.P. (1996) The effects of oxygen-enriched intake air on FFV exhaust emissions using M85, Argonne National Lab., IL, United States. [Google Scholar]
- Ng H.K., Sekar R.R., Kraft S.W., Stamper K. (1993) The potential benefits of intake air oxygen enrichment in spark ignition engine powered vehicle, Society of Automotive Engineers. [Google Scholar]
- Caton J.A. (2005) Use of a cycle simulation incorporating the second law of thermodynamics: results for spark-ignition engines using oxygen enriched combustion air, SAE Technical Paper 2005-01-1130. [Google Scholar]
- Poola R.B., Stork K.C., Sekar R., Callaghan K., Nemser S. (1998) Variable Air Composition with Polymer Membrane–A New Low Emissions Tool, SAE Paper 980178. [Google Scholar]
- Kajitani S., Sava N., McComiskey T., Rhee K. (1992) A spark ignition engine operated by oxygen enriched air, Society of Automotive Engineers. [Google Scholar]
- Quader A. (1978) Exhaust Emissions and Performance of a Spark Ignition Engine Using Oxygen Enriched Intake Air, Combustion Science and Technology 19, 1–2, 81–86. [CrossRef] [Google Scholar]
- Mounaïm-Rousselle C., Landry L., Halter F., Foucher F. (2013) Experimental characteristics of turbulent premixed flame in a boosted Spark-Ignition engine, Proceedings of the Combustion Institute 34, 2, 2941–2949. [CrossRef] [Google Scholar]
- Van Blarigan A., Seiser R., Chen J., Cattolica R., Dibble R. (2012) Working fluid composition effects on methane oxycombustion in an SI-engine: EGR vs. CO2, Proceedings of the Combustion Institute 34, 2, 2951–2958. [CrossRef] [Google Scholar]
- Favre E., Bounaceur R., Roizard D. (2009) A hybrid process combining oxygen enriched air combustion and membrane separation for post-combustion carbon dioxide capture, Separation and Purification Technology 68, 1, 30–36. [CrossRef] [Google Scholar]
- Koros W.J., Chern R.T. (1987) Separation of gaseous mixtures using polymer membranes, Handbook of separation process technology, pp. 862–953. [Google Scholar]
- Coombe H.S., Nieh S. (2007) Polymer membrane air separation performance for portable oxygen enriched combustion applications, Energy Conversion and Management 48, 5, 1499–1505. [CrossRef] [Google Scholar]
- Richard S., Bougrine S., Font G., Lafossas F.-A., Le Berr F. (2009) On the reduction of 3D CFD combustion model to build a physical 0D model for simulating heat release, knock and pollutants in SI engines, Oil & Gas Science and Technology – Rev. IFP 64, 3, 223–242. [CrossRef] [EDP Sciences] [OGST] [MathSciNet] [Google Scholar]
- IFP Energies nouvelles (2010) AMESim IFP Engine library user manual, LMS IMAGINE S.A. [Google Scholar]
- Bougrine S., Richard S., Veynante D. (2011) On the combination of complex chemistry with a 0-D coherent flame model to account for the fuel properties in spark ignition engines simulations: Application to methane-air-diluents mixtures, Proceedings of the Combustion Institute 33, 2, 3123–3130. [CrossRef] [Google Scholar]
- Bozza F., Gimelli A. (2004) A Comprehensive 1D Model for the Simulation of a Small-Size Two-Stroke SI Engine, SAE Technical Paper 2004-01-0999. [Google Scholar]
- Richard S., Colin O., Vermorel O., Benkenida A., Angelberger C., Veynante D. (2007) Towards large eddy simulation of combustion in spark ignition engines, Proceedings of the Combustion Institute 31, 2, 3059–3066. [Google Scholar]
- Colin O., Benkenida A., Angelberger C. (2003) 3d Modeling of Mixing, Ignition and Combustion Phenomena in Highly Stratified Gasoline Engines, Oil & Gas Science and Technology – Rev. IFP 58, 1, 47–62. [CrossRef] [EDP Sciences] [Google Scholar]
- Charlette F., Meneveau C., Veynante D. (2002) A power-law flame wrinkling model for LES of premixed turbulent combustion Part I: non-dynamic formulation and initial tests, Combustion and Flame 131, 1–2, 159–180. [Google Scholar]
- Blint R.J. (1986) The Relationship of the Laminar Flame Width to Flame Speed, Combustion Science and Technology 49, 1–2, 79–92. [Google Scholar]
- Poulos S.G., Heywood J.B. (1983) The effect of chamber geometry on spark-ignition engine combustion, SAE Technical Paper 830334. [Google Scholar]
- Zhou J.X., Cordier M., Mounaïm-Rousselle C., Foucher F. (2011) Experimental estimate of the laminar burning velocity of iso-octane in oxygen-enriched and CO2-diluted air, Combustion and Flame 158, 12, 2375–2383. [CrossRef] [Google Scholar]
- Galmiche B., Halter F., Foucher F. (2012) Effects of high pressure, high temperature and dilution on laminar burning velocities and Markstein lengths of iso-octane/air mixtures, Combustion and Flame 159, 11, 3286–3299. [CrossRef] [Google Scholar]
- Friedman R., Johnston W.C. (1950) The Wall-Quenching of Laminar Propane Flames as a Function of Pressure, Temperature, and Air Fuel Ratio, Journal of Applied Physics 21, 8, 791–795. [CrossRef] [Google Scholar]
- Mazas A.N., Fiorina B., Lacoste D.A., Schuller T. (2011) Effects of water vapor addition on the laminar burning velocity of oxygen-enriched methane flames, Combustion and Flame 158, 12, 2428–2440. [CrossRef] [Google Scholar]
- Heywood J. (1994) Combustion and its modeling in spark-ignition engines, in International symposium COMODIA. [Google Scholar]
- Boulouchos K., Steiner T., Dimopoulos P. (1994) Investigation of flame speed models for the flame growth period during premixed engine combustion, Society of Automotive Engineers. [Google Scholar]
- Linse D., Hasse C., Durst B. (2009) An experimental and numerical investigation of turbulent flame propagation and flame structure in a turbo-charged direct injection gasoline engine, Combustion Theory and Modelling 13, 1, 167–188. [CrossRef] [Google Scholar]
- Wirth M. (1993) Die turbulente flammenausbreitung im ottomotor und ihre charakteristischen längenskalen, in RWTH Aachen. [Google Scholar]
- Poinsot T., Veynante D., Candel S. (1991) Quenching processes and premixed turbulent combustion diagrams, Journal of Fluid Mechanics 228, 561–606, 230. [Google Scholar]
Cite this article as: J.X. Zhou, B. Moreau, C. Mounaïm-Rousselle and F. Foucher (2016). Combustion, Performance and Emission Analysis of an Oxygen-Controlling Downsized SI Engine, Oil Gas Sci. Technol 71, 49.
All Tables
Correspondence between the O2 rate and EGR percentage at the stoichiometric equivalence ratio
All Figures
Figure 1 Scheme of air-fuel mixture system for the single-cylinder engine. |
|
In the text |
Figure 2 All different operating conditions were studied at fixed 1 400 rpm. a) In fixed stoichiometric conditions, b) in fixed 8 bar IMEP conditions. |
|
In the text |
Figure 3 Scheme of AMESim modeling for an oxygen ratio-controlled engine. |
|
In the text |
Figure 4 Global scheme of different simulation and model coupling and looping. |
|
In the text |
Figure 5 Evolution of indicated SFC versus the equivalence ratio for different IMEP; a) 400 kPa, b) 600 kPa, c) 800 kPa, d) 1 000 kPa. |
|
In the text |
Figure 6 Evolution of indicated specific HC emissions versus the equivalence ratio for different IMEP; a) 400 kPa, b) 1 000 kPa. |
|
In the text |
Figure 7 Evolution of indicated specific NO x emissions versus the equivalence ratio with different IMEP; a) 400 kPa, b) 1 000 kPa. |
|
In the text |
Figure 8 Evolution of indicated specific CO emissions versus the equivalence ratio for different IMEP; a) 400 kPa, b) 1 000 kPa). |
|
In the text |
Figure 9 Comparison between experimental and simulated in-cylinder pressure evolution for different engine loads. φ = 1, a) 15% O2, b) 21% O2, c) 27% O2. |
|
In the text |
Figure 10 a) Relative error of IMEP and b) maximum in-cylinder pressure. |
|
In the text |
Figure 11 a) Integral length scale and b) tumble values obtained to optimize the in-cylinder pressure. |
|
In the text |
Figure 12 Evolution of the mean in-cylinder pressure at the stoichiometric equivalence ratio and IMEP = 1 000 kPa. |
|
In the text |
Figure 13 Laminar burning velocity versus CAD with different O2 percentages at the stoichiometric equivalence ratio and IMEP = 1 000 kPa. |
|
In the text |
Figure 14 a) Flame surface and b) flame wrinkling versus the crank angle with different O2 percentages at the unit equivalence ratio and IMEP = 1 000 kPa. |
|
In the text |
Figure 15 Evolution of turbulent flame velocity with different O2 percentages at the stoichiometric equivalence ratio and IMEP = 1 000 kPa. |
|
In the text |
Figure 16 Ratio of the characteristic length a) and velocity scales b) versus CAD for different O2 percentages. |
|
In the text |
Figure 17 Ratio of the characteristic length a) and velocity scales b) versus the O2 percentage for two CAD cases. |
|
In the text |
Figure 18 Comparison of the combustion traces in the Peters-Borghi diagram for different O2 percentages. |
|
In the text |