Investigation of Biodiesel–Diesel Fuel Blends on Combustion

Nov 12, 2012 - Department of Mechanical, Materials and Manufacturing Engineering, The University of Nottingham, Malaysia Campus, Jalan Broga,...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/EF

Investigation of Biodiesel−Diesel Fuel Blends on Combustion Characteristics in a Light-Duty Diesel Engine Using OpenFOAM Harun Mohamed Ismail,† Hoon Kiat Ng,*,† Suyin Gan,‡ Xinwei Cheng,† and Tommaso Lucchini§ †

Downloaded via UNIV OF LOUISIANA AT LAFAYETTE on January 22, 2019 at 15:01:29 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Department of Mechanical, Materials and Manufacturing Engineering, The University of Nottingham, Malaysia Campus, Jalan Broga, 43500 Semenyih, Selangor Darul Ehsan, Malaysia ‡ Department of Chemical and Environmental Engineering, The University of Nottingham, Malaysia Campus, Jalan Broga, 43500 Semenyih, Selangor Darul Ehsan, Malaysia § Department of Energy, Politecnico Di Milano, Via Lambruschini 4, 20156 Milano, Italy ABSTRACT: This paper reports the differences in combustion characteristics of fossil diesel and the methyl esters of coconut (CME), palm (PME), and soy (SME) over a range of engine conditions. The studies are conducted at a constant engine speed of 2000 rpm, and at engine load values of 0.5 kW (low), 1.5 kW (medium), and 2.5 kW (high). The investigated fuels are CME, PME, and SME at 0% diesel mixture (B100), 50% diesel mixture (B50), and fossil diesel (B0). Here, the OpenFOAM opensource computational fluid dynamics code is utilized to simulate the in-cylinder events. An in-house model for thermophysical and transport properties is employed, along with a mechanism comprising 113 species and 399 reactions with integrated NOx kinetics. Good levels of accuracy are achieved in the prediction of the ignition delay (ID) period, peak pressures, pressure traces, and heat-release rate profiles for all the test fuels. Biodiesel fuels are found to produce larger fuel droplet sizes, longer spray penetration, and lower vaporization rates compared to those of fossil diesel. In terms of the combustion behavior, the ID period decreases as the engine load is increased for neat CME, PME, and SME. However, the ID period increases as the load is raised for the B50 mixtures of CME, PME, and SME. All neat biodiesels and their blends except neat SME produce shorter ID periods than fossil diesel. Changing the fuel type from diesel to biodiesel alter the physical and chemical delay and, hence, the overall ID period. As a result, variations in the combustion behavior between the fuels are recorded.

1. INTRODUCTION In recent years, biodiesel is increasingly used as a substitute to fossil diesel mainly because of depleting petroleum reserves and more stringent emission standards.1 In addition, biodiesel can be used in existing compression ignition (CI) or diesel engines with minimal or no modifications because its physicochemical characteristics are very similar to those of fossil diesel.2 Neat biodiesel and its blends are known to reduce tailpipe concentrations of particulate matter, hydrocarbons (HCs), and carbon monoxide (CO).3 Relative increase in nitrogen oxide (NOx) emissions compared to diesel fuel is observed in some studies when biodiesel fuels are tested in unmodified engines.3,4 Nevertheless, there are also studies that demonstrated that biodiesel fuels such as coconut methyl ester (CME) reduce the formation of NOx.5,6 In light of these inconclusive findings, more comprehensive studies are required to improve the fundamental understanding of biodiesel combustion and emission characteristics as well as to conclusively establish the benefits and drawbacks of biodiesel usage in diesel engines.7,8 In diesel engines, the type of fuel (diesel, biodiesel and/or blends) substantially affects both the in-cylinder physical and chemical processes because of variations in the physicochemical properties. These in-cylinder processes dictate the combustion characteristics, which, in turn, influence the engine-out pollutants. For instance, at low load conditions, the main part of the combustion is premixed. Here, substantial fuel’s chemical energy is released over a short period of time in the premixed combustion (PMC) phase near the top-dead center (TDC), which significantly increases the local and average in-cylinder © 2012 American Chemical Society

temperatures. As a result, the NOx formation rate is increased, leading to higher engine-out NOx.9 On the other hand, during the mixing-controlled combustion (MCC) phase, the soot formation rate is increased because of fuel-rich combustion in the core of the spray jet whereby the equivalence ratio is between 1.5 and 4 and the local in-cylinder temperature is between 1400 and 2000 K.9 Thus, it is necessary to fully understand the combustion characteristics of various biodiesel fuels in order to better comprehend and improve the emissions of these fuels. To date, the changes with the spray structure and the impact on combustion characteristics due to various biodiesel fuels have not been extensively investigated through combustion modeling. Early initiatives in this field were carried out by Yuan et al.,10,11 who presented methods for predicting the physical properties for biodiesel that can be used for combustion modeling. Chakravarthy et al.7 estimated the thermophysical and transport properties of soy biodiesel for use in computational fluid dynamics (CFD) modeling to evaluate the extent to which combustion is modified when biodiesel is substituted for diesel. The preliminary results obtained showed that biodiesel ignition is significantly delayed because of slower liquid evaporation, with the lower vapor pressure and higher liquid heat capacity of biodiesel being the two key contributors to this observation. Another CFD modeling effort by Ra et al.8 investigated the effects of the soy physical properties on the diesel engine Received: July 12, 2012 Revised: October 9, 2012 Published: November 12, 2012 208

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

combustion characteristics. At various engine operating conditions, the fuel physical properties had significant effects on the ignition delay (ID) and burning rates, with no single physical property observed to solely determine the differences in the ID between diesel and biodiesel fuels. Nonetheless, the simulation results were most sensitive to the liquid fuel density, vapor pressure, and surface tension through their effects on the mixture preparation processes. Recent works by Brakora et al.12 as well as Yang and Golovitchev13 concentrated on the development of a reduced combustion kinetics mechanism for biodiesel to study the combustion characteristics of soy methyl ester (SME) and rapeseed methyl ester (RME) fuels. As explained here, these fundamental combustion studies focus mainly on SME and RME fuels. From the current available literature detailing this, an investigation into the combustion characteristics of biodiesel fuels such as CME and palm methyl ester (PME) and the complex interactions with the fuel properties has not been extensively conducted to date. Therefore, the objective of this study is to examine the combustion characteristics of various biodiesel fuel types and blends as compared to diesel fuel. A CFD approach is used alongside an experimental engine test-bed study to accurately capture the physical and chemical events of various neat biodiesels and their blends in a light-duty diesel engine. Here, a set of numerical experiments are performed to elucidate the key effects of CME, PME, and SME at 0% (B100) diesel mixture, 50% diesel mixture (B50), and diesel (B0) on the combustion characteristics under different engine operating conditions. For this reason, the rates of heat release are evaluated under different operating conditions in order to analyze the combustion behaviors of the biodiesel fuels. The heat release process can largely reflect the combustion and emission characteristics of a CI engine when additional information is available such as the ID period and engine load.14 Here, changes in the spray structure, ID, peak PMC, and MCC period between the tested fuels are extensively appraised.

Figure 1. Definition of combustion phasing utilized in this study.

the high peak HRR value. The MCC phase is short, and for some cases, it is negligible. Fuels with a shorter ID period or high load conditions typically result in a longer MCC phase. Because of the shorter time for fuel and air to mix (a short ID period) or longer fuel injection duration to accommodate the higher fuel mass required, the PMC phase is negligible compared to the MCC phase, as can be seen in Figure 1. Finally, the third category of combustion falls in between these two extreme features, with balanced PMC and MCC phases. The heat release due to combustion is determined using the First Law of Thermodynamics. Using the change in the pressure and volume per unit of crank angle (CA) degree, the net rate of heat release is calculated using expression (1) dQ net dθ

=

dp γ dv 1 p + V γ − 1 dθ γ − 1 dθ

(1)

where γ is the adiabatic exponent with a value of 1.35 employed in this study,9 p is the in-cylinder pressure, and V is the volume as a function of the CA. For the HRR computations, the following assumptions are made: • The HRR analysis is limited to within the closed cycle of the engine cycle between the intake valve closing (IVC) and exhaust valve opening (EVO). • The effect of blow-by is negligible. • The cylinder wall temperature is constant within the analysis interval. • The in-cylinder pressure is uniform within the control volume.

2. HEAT-RELEASE RATE (HRR) ANALYSIS This section explains the method of determining the HRR profile and the assumptions made. The heat-release profile varies in terms of duration, magnitude, and shape/phasing when the engine operating conditions are changed. In this study, the engine speed is maintained at a constant value of 2000 rpm. The engine load is varied from 0.5 kW to 1.5 and 2.5 kW. In addition, a total of seven different fuel blends is investigated. As such, significant variations in the heat-release profile are expected across the different test cases. Variations in the HRR profiles are classified under three distinct categories, as shown in Figure 1. In each category, the combustion is divided into two main features: PMC and MCC. The PMC phase is where the combustion rate is limited by the chemical reaction rate.15 The MCC phase is where the combustion process is significantly influenced by the locally available oxygen concentration, where a diffusion flame is observed because of an extended fuel injection timing. Here, the burn rate is controlled by the mixing process, which is dictated by the turbulent intensity. Typically, fuels with a long ID period tend to have more fuel mass burnt in the PMC phase. Similarly, fuels tend to undergo longer PMC phase relative to the MCC phase at low load because of the shorter fuel injection duration and the later start of the combustion event, as shown in Figure 1. Here, a higher rate of energy release takes place during the PMC phase, as indicated by

3. ENGINE EXPERIMENTAL SETUP AND CFD MODEL FORMULATIONS 3.1. Light-Duty Diesel Engine Experimental Setup. Simulations are carried out for a quiescent, bowl-in-piston lightduty diesel engine, which has a compression ratio of 19.1:1 and a displacement volume of 347 cm3. Four equally spaced injector holes deliver the required fuel. The air-cooled engine operates at speeds between 1500 and 3500 rpm and loads between 0.5 and 2.5 kW. The engine bore and stroke dimensions are given as 80 and 69 mm, respectively. Table 1 summarizes the specifications of the research engine used in this study. The test engine is coupled to an asynchronous dynamometer mounted on an engine test stand. For determination of the pressure trace, a Kistler type 601A quartz pressure transducer is used. A K-type thermocouple with an upper measurement limit of 1000 °C and a maximum error of ±4 °C is used for acquisition of the exhaust gas temperature.5,6 Other engine data such as torque, specific fuel 209

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

Table 1. Engine Specifications and Test Conditions for the CFD Simulation Study Specifications of the Nottingham Test Engine engine type piston type cylinder head type displacement volume per cylinder compression ratio stroke bore connecting rod length IVC EVO engine speed Validation Test Range load (kW) 0.5 0.5 0.5 1.5 1.5 1.5 2.5 2.5 2.5

speed (rpm) 2000 2000 2000 2000 2000 2000 2000 2000 2000

light-duty diesel bowl-in-piston flat cylinder head 347 cm3 19.1:1 69 mm 80 mm 114.5 mm −140° ATDC 140° ATDC 1500−3500 rpm

fuel B100

fuel B50

CME PME SME CME PME SME CME PME SME

CME PME SME CME PME SME CME PME SME

Figure 2. 90° sector mesh of the Nottingham test engine.

from the in-cylinder gases to the chamber walls is modeled using the Han and Reitz correlation.19,21 The liquid spray structures are predicted using the Huh injector model to represent the injector and fuel droplets primary spray breakup.18,19 Fuel spray atomization is modeled using the modified Huh and Gosman model.18,19,22 Secondary breakup is computed using the Kelvin-Helmoltz and Rayleigh-Taylor model.19 Heat transfer between the discrete and gas phases is determined using the Ranz−Marshall correlation.16,17 The Bai and Gosman correlation is used to predict the dynamics characteristics of droplet wall interactions on the piston bowl wall surface.19 Further details about the spray model constants as well as model implementations and validations are reported elsewhere.16−19,23 The spray characteristics of diesel fuel are defined by the thermophysical and transport properties of a C14H30 surrogate. The properties of CME, PME, and SME developed by Ismail et al.19 are utilized for neat biodiesel fuels. For the B50 biodiesel−diesel blends, the properties of neat biodiesel are mixed with that of C14H30 using a multicomponent modeling approach at a 50% blend ratio. For the combustion process, each computational cell is assumed to be a homogeneous system and a reduced chemical mechanism is used to estimate the PMC and MCC phases. The homogeneous, perfectly stirred reactor (PSR) system is used instead of the non-homogeneous, partially stirred reactor (PaSR) system because PaSR requires considerably higher computing time and produces greater level of numerical instability. It is also found that there is no significant increase in the prediction accuracy when the PaSR approach is used. As such, the PSR approach is adopted here where the chemical reaction rates are calculated by an ordinary differential equation (ODE) stiff solver using an operator-splitting technique. For each time step Δt, the ODE solver maps the initial composition ψ0 = ψ(t0) to the reacted value ψ (t0 + Δt), which is a unique function of ψ0 called the reaction mapping, R(ψ0). The composition array is defined by ψ = {YI, ...,YNs, p, T}, where Yi is the species mass fraction, Ns the number of species, T the temperature, and p the pressure. However, direct integration of detailed chemistry is computationally demanding. For this reason, the proposed combustion model works in combination with the tabulation of dynamic adaptive chemistry algorithm.17,19,24 Such an approach combines mechanism reduction and tabulation techniques that are both performed on-the-fly in each cell during the simulation. Significant speed-up factors compared to direct integration are

consumption, equivalence ratio, and volumetric efficiency are collected from the respective sensors contained within the engine test stand. A Bosch BEA350 gas emissions analyzer is employed to measure CO, CO2, HC, and NO levels in the exhaust stream, whereas a Bosch RTM 430 smoke opacimeter is used to determine the smoke capacity level.5,6 For the emissions analyzer, accuracy levels of ±0.001 vol. %, ±0.001 vol. %, ±1 ppm vol. and ±1 ppm vol. are achieved for the measurements of CO, CO2, HC and NO, respectively.5,6 A photodiode receiver in the smoke opacimeter is able to determine the smoke opacity up to an upper limit of 100% with an accuracy of ±0.1%.5,6 All experimental data are recorded at fully warm, steady-state engine conditions. Further details on the experimental measurements and tests can be found in the work reported by Ng et al.5,6 3.2. Computational Mesh and Initial Conditions. A 90° sector mesh is generated to take advantage of the symmetry imposed by the four equally spaced injector nozzle holes installed centrally. The hexahedral mesh type is employed to construct the engine mesh, as shown in Figure 2. The initial conditions of the intake pressure and temperature at IVC are 1.23 bar and 300 K, respectively. Here, the dynamic mesh layering utility of the mesh topological change library (Lib-ICE)16−18 is utilized for modeling of the compression and expansion strokes. The mesh with a cell size of 2.0 (radial) × 2.5 (axial) × 2.0 (azimuthal) mm is used throughout this study because it is found to give similar accuracy but with shorter computing time compared to a finer resolution mesh.19 Further details on the mesh configurations and grid-independent study have been reported by Ismail et al.19 All simulations reported here are performed for the closed part of the engine cycle only. 3.3. Computational Model Formulations. A description of the computational model used, which was implemented in OpenFOAM open-source CFD code under the Lib-ICE library developed by Tommaso et al.,18 is provided in this section. Here, the in-cylinder flows and turbulence effects are calculated using the widely used RNG k−ε turbulence model.19,20 Heat transfer 210

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

Figure 3. Pressure traces and HRRs for low, medium, and high loads at 2000 rpm for (a) CME (B50), (b) PME (B50), (c) SME (B50), (d) CME (B100), (e) PME (B100), and (f) SME (B100).

to be observed in the test results. The engine speed is maintained at 2000 rpm, and the engine load is varied from low (0.5 kW) to medium (1.5 kW) and high (2.5 kW) loads, as given in Table 1. A medium engine speed is chosen because it lies within the typical operating range of on-road vehicles and also within the engine’s drivability limits. Meanwhile, the chosen engine loads are between the idling and maximum loads achievable for all of the test fuels.6 Parts a−f of Figure 3 show a comparison of the in-cylinder pressure trace and HRR profile between experimental and simulated results of six test fuels at three different engine loads. Generally, good levels of accuracy are achieved in the prediction of the ID period, peak pressures, pressure traces, and HRR profiles for all of the test fuels. The change in the combustion behavior from the PMC to MCC phase because of a prolonged fuel spray injection period is also clearly captured for all of the simulations. Additionally, the ID period and peak pressure are accurately predicted for all tested loads within the range of ±0.2 °CA. The maximum error between the experimental and computed peak pressure values across the test range is limited to below 4.5%. Furthermore, the timing of the peak pressure varies within ±1.5 °CA of the experimental data for all of the tested fuels and engine loads. Parts a−f of Figure 4 illustrate the calculated and measured NOx and soot emissions at EVO time. The majority of the test cases have a maximum percentage error between predicted and measured emissions data within an acceptable limit of 8.7% for NOx and 13.5% for soot. Both the NOx and soot trends for all of the tests are also well-predicted. As shown in Figure 4, it is found that the highest soot emission is given by CME, followed by PME and lastly SME for both the neat and B50 blends of the biodiesel fuels. This is mainly attributed to the unique combustion behavior displayed by each fuel, the local in-cylinder temperatures, and the concentrations of freely available O and OH radicals for soot oxidation. At low load, the influence of the combustion characteristics on soot emissions is more apparent. Therefore, fuels undergoing mainly PMC, such as neat SME,

ensured (>300), allowing a semidetailed mechanism to be used for simulations of practical devices such as CI engines. For the combustion chemistry, the 113 species and 399 reactions mechanism (BOS-V2) with integrated NOx reaction kinetics proposed by Ismail et al.25 is utilized to represent the CME, PME, and SME fuel chemistry. In all of the simulations, the different fuels are modeled by varying the mass composition ratio of saturated and unsaturated components with respect to the individual fuel type.19 When the biodiesel fuel changed from CME to PME and SME, the level of unsaturation for the fuel component is varied from 20% to 50% and 80%, respectively, to match the typical unsaturation levels present in the biodiesel fuel. A modified Chalmers n-heptane mechanism19,25,26 is utilized to represent the combustion of diesel fuel. Finally, a phenomenological model is used to predict soot emissions. The Hiroyasu approach is used to describe the formation of soot, while its oxidation is estimated by the Nagle-Strickland and Constable model. In order to obtain comparable soot data sets, both the experimental and simulated data sets are normalized using expression (2), where Zi is the input or output variable and (λ1, λ2) is the range of the function.27 The data are normalized between the range of 0 and 1 such that the experimental and computed data are in the same scale. ⎛ Z − Z min ⎞ i i x i = λ1 + (λ 2 − λ1)⎜ max min ⎟ ⎝ Zi − Zi ⎠

(2)

4. RESULTS AND DISCUSSION 4.1. Validation of Simulated Engine Data against the Experimental Results. A total of 18 engine test points is utilized to validate and investigate the combustion characteristics of a total of six biodiesel fuel types and blends in comparison with fossil diesel. The validated fuels are CME, PME, and SME at 0% (B100) and 50% (B50) diesel mixture. The selection of a B50 biodiesel−diesel mixture is to allow a feasible number of test points to be conducted while still allowing sufficient differences 211

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

Figure 4. Emission indexes of NOx and normalized soot for (a) B50 blend at low load (0.5 kW), (b) B50 blend at medium load (1.5 kW), (c) B50 blend at high load (2.5 kW), (d) B100 at low load (0.5 kW), (e) B100 at medium load (1.5 kW), and (f) B100 at high load (2.5 kW).

models and model constants are able to accurately predict the incylinder combustion and emission characteristics with acceptable error tolerance limits. 4.2. Effects of Biodiesel Fuel Types and Blends on the Mixture Preparation Process. One of the important factors that determine the mixture preparation process before the SOC is the start of injection (SOI) timing. In this study, the SOI timing is advanced by 0.2° and 1.2 °CA when the diesel fuel is substituted with the B50 and B100 blends, respectively, for CME, PME, and SME. In this engine setup, the fuel injection system is based on a rotary-type fuel injection system. Here, actuation of the fuel injector valve is controlled by the fuel-line pressure, which is governed by the fuel pump. Typically, the system is calibrated for diesel fuel. However, diesel fuel has lower bulk modulus, surface tension, and viscosity than biodiesel fuel. Therefore, the fuel-line pressure increases at a higher rate when biodiesel fuel is used instead. As a result, the fuel injector valve opens at an earlier CA with biodiesel fuel compared to when diesel is used. When the SOI is advanced, the fuels tend to undergo early SOC before the TDC position during the compression stroke. This leads to higher local and average in-

have less soot emission because the soot formation rate is low, as is typically observed in a premixed flame. On the other hand, neat and a B50 blend of CME consistently produce a reduction of the exhaust NOx level, as well as give the highest percentage of NOx reduction across all of the engine loads tested. Among the B50 blends of biodiesel fuels, the use of B50 SME results in the highest NOx emission, followed by B50 PME and finally B50 CME. This trend is also applicable with the various neat biodiesel fuels for different engine load operations. Figure 4 also shows that an increase in the NOx level at EVO is observed when the engine load is reduced, with highest NOx emission recorded at low engine load for different test fuels. At low load, the fuel injection period is short where fuel delivery is mostly completed before the start of combustion (SOC). Here, the combustion process is predominantly PMC, where the fuel burns more completely and results in an elevated local in-cylinder temperature. This leads to greater NOx formation for all of the test fuels at low load compared to medium and high load operations. Further details of the CFD engine validations against the experimental results are reported by Ismail et al.19 On the basis of the validation results, it can be concluded that the selected CFD 212

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

to the B50 blends at low and medium loads. At high load, however, all of the fuels have similar levels of vaporized fuel mass. The reason for this is that almost all of the fuel is injected before the SOC at low load. Hence, the vaporisation rate is highly dictated by the fuel mass flow rate, individual fuel properties, and saturated evaporation point. On the other hand, the fuel injection period is prolonged at high load and, therefore, the vaporization rate is greatly enhanced by the combustion process for all of the test fuels. The fuel vaporization rate is predominantly controlled by the thermophysical properties prior to the SOC event.19 However, the vaporization rate is mainly regulated by the rate of heat exchange between the fuel droplets and high-temperature gas mixtures after autoignition. 4.3. Effects of Biodiesel Fuel Types and Blends on Combustion Characteristics. 4.3.1. Effects on the ID. The type of fuel utilized in a CI engine significantly influences the ID. In this study, it is found that the ID period is shortened when diesel is substituted with neat CME and PME, while the ID period is increased when diesel is substituted with neat SME, as illustrated in Figure 6. This trend is observed for all three engine loads. In contrast, ID periods shorter than that of diesel are obtained for B50 blends of CME, PME, and SME. The observed trend can be attributed to two factors: the first is the fuel thermophysical properties, and the second is the fuel chemical composition. The thermophysical properties typically control the droplet breakup rate, vaporization rate, spray length, vapor length, and fuel/air mixing prior to the SOC. This is collectively known as the physical delay. Meanwhile, the chemical composition dictates the onset of energy release for the ignition process. It is typically measured using the cetane number, where a higher cetane value indicates a shorter ID period. The chemistry effect is commonly known as the chemical delay. Normally, the effects of the chemical delay are more prominent than those of the physical delay.9 The chemical delay is governed by the molecular composition of each fuel. Among the biodiesel fuels, those with lower unsaturation levels tend to have shorter IDs. For example, neat CME with a 20% average unsaturation level by mass has the shortest ID, while neat SME has the longest ID with an average of 80%.19 Fuels with higher levels of unsaturation have longer ID periods because the carbon double bond is stronger than the single bond. Hence, more energy is required to break this bond such that the oxidation process can be initiated. As a result, fuels with higher levels of unsaturation require a longer time to gain sufficient energy to break all of the double bonds compared to those with lower levels of unsaturation. Apart from the fuel type, the fuel blend level also affects the ID. Among the tested biodiesel fuels, the magnitude of reduction in the ID period is enhanced when B50 blends of CME, PME, and SME are used instead of their B100 counterparts, as shown in Figures 6 and 7. Here, the combined effects of the physical and chemical delays are responsible for the observed trends. First, the B50 blends have much finer fuel droplets compared to their B100 counterparts because of better overall atomization. Second, the 50% lighter diesel component in the B50 blends undergoes preferential evaporation. This leads to a much higher vaporization rate of B50 blends compared to B100 fuels. Consequently, the overall physical delay of B50 blends is reduced significantly compared to B100 fuels, resulting in a reduced overall ID period for all B50 blends. Furthermore, variations in the ID are observed when the engine load is changed. The greatest reduction in the ID period occurs at high load, while the lowest occurs at low load when diesel fuel is substituted with neat biodiesel fuel, as can be seen in

cylinder temperatures and pressures because of the dual effects from both the combustion process and the reduction in the control volume size during compression stroke. This condition is ideal to promote NOx formation. The fuel spray and vapor structure are also significantly affected when diesel fuel is replaced with B50 or neat CME, PME, and SME during the mixture preparation stage. At low load, almost all of the fuel mass is injected before the SOC. From the simulation data, neat biodiesel fuels generally have longer spray penetration lengths than diesel, which follow the trend shown in Table 2 and Figure 5a. Similar trends are also observed at Table 2. Trends of the Liquid Spray Penetration Length for Various Test Fuels load rank

0.5 kW

1.5 kW

2.5 kW

1 (highest) 2 3 4 5 6 7 (lowest)

PME (B100) SME (B100) CME (B100) SME (B50) diesel PME (B50) CME (B50)

PME (B100) SME (B100) CME (B100) diesel SME (B50) PME (B50) CME (B50)

PME (B100) SME (B100) CME (B100) SME (B50) PME (B50) CME (B50) diesel

medium and high engine loads, as depicted in Figure 5b,c. However, when CME, PME, and SME are blended with 50 vol % of diesel fuel, the general trend as discussed above is not reproduced. Here, B50 blends of PME and CME exhibit lower liquid penetration lengths compared to diesel at low engine load. At medium load operation, all the B50 biodiesel blends have lower liquid penetration lengths. Finally, all of the biodiesel fuels produce longer liquid penetration lengths compared to diesel at high load. The factors governing the observed trend are the ID, engine load, fuel thermophysical properties, and timing at which the liquid length reaches the stabilization point. After the liquid length stabilization point, the majority of liquid droplets are fully vaporized rapidly due to combustion. The fuel that reaches this stabilization point earlier tends to have a shorter liquid penetration length. The important thermophysical properties that significantly affect the spray penetration length are determined to be the surface tension, viscosity, liquid density, and vapor pressure.19 Here, it is found that the atomization quality of biodiesel fuel is poorer compared to diesel. Biodiesel fuel droplets are less affected by the in-cylinder aerodynamics forces because of the higher surface tension and viscosity, which lead to larger droplets being formed. In addition, the normalized vaporization rate of a single droplet of B100 and B50 biodiesels is much lower than that of diesel fuel as a result of lower vapor pressure,19 given that the fuel mass injected into the system is identical. Consequently, the biodiesel fuel droplet loses mass slower as opposed to its diesel counterpart. As a result of the combined effects of the larger droplet size and lower vaporization rate, biodiesel fuel droplets have greater momentum to traverse across the combustion chamber before the SOC. Once autoignition takes place, however, the spray penetration length is now governed by the enhanced vaporization process because of the higher charge temperature. Here, the influence of the ID and time at which the liquid length reaches the stabilization point are more important in determining the spray penetration length of the fuel. As for the vaporization trends illustrated in Figure 6, neat biodiesel fuels have lower levels of vaporized fuel mass compared 213

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

Figure 5. Simulated liquid spray penetration lengths for the different test fuels and blends, with an engine speed of 2000 rpm at (a) 0.5 kW, (b) 1.5 kW, and (c) 2.5 kW.

mainly due to the biodiesel fuel thermophysical properties and fuel injection pressure. As stated before, neat biodiesel fuel has higher viscosity and surface tension compared to diesel. The fuel spray penetration length, atomization, droplet breakup rate, and

Figure 7. This means that, at high load, the use of neat biodiesel enhances the magnitude of the reduction in the ID period, whereas at low load, the reduction in the ID is moderated. Both Figures 7 and 8 illustrate this point. These observations are 214

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

rate at high load. Furthermore, better air utilization of the surrounding air is achieved as a result of the longer spray penetration length. Consequently, the physical delay is reduced significantly at high load. The overall magnitude of the reduction in the ID period is thus enhanced at high load compared to low load when diesel is substituted with neat biodiesel fuel. Nevertheless, the above-mentioned trend is reversed when diesel is substituted with the B50 blends of CME, PME, and SME. In these cases, greatest magnitude of ID reduction is achieved at low load, while the lowest is achieved at high load, as shown in Figure 7. This indicates that the use of the B50 fuel blends enhances the reduction of ID period at low load, while the reduction in the ID period is moderated at high load, as seen in Figure 8. Here, the surface tension and viscosity of B50 blends are lower than those of neat biodiesel because of mixing with diesel. At any load, the B50 blends tend to have greater atomization rate, fuel droplet breakup rate, and vaporization rate than neat biodiesels, leading to a shorter spray penetration length. When the fuel is injected at high injection pressure at high load, the B50 fuel droplets undergo enhanced atomization and droplet breakup rate and thus lose the momentum to travel further into the combustion chamber. Poorer fuel/air mixing occurs as a result, which eventually increases the physical delay compared to that of neat biodiesel. Thus, a lower magnitude of the reduction in the ID is achieved at high load when the diesel is substituted with B50 biodiesel−diesel blends. In summary, the engine load has a significant influence on the level of changes to the ID period when the fuel blend level is varied, with opposing trends for B100 and B50 fuels. 4.3.2. Effects on the Combustion Behavior. HRR Profile. For all loads, the combustion feature changes when diesel is substituted with CME (B100) and PME (B100). Parts a−c of Figure 9 show that longer periods of the MCC phase are obtained with more fuel mass burned in the diffusion flame when diesel is substituted with CME (B100) and PME (B100). However, the combustion behavior is not changed significantly when diesel is substituted with neat SME. Here, most of the fuel mass is burned in the PMC phase similar to that of diesel fuel. Similar trends are also observed when diesel fuel is substituted with B50 blends, as illustrated in Figure 9a−c. The trends are mainly caused by the differences in the ID period of the biodiesel fuels. Because of higher unsaturation levels, neat SME and the B50 SME blend tend to have longer ID periods than neat CME and PME as well as their B50 blends. Neat SME also has a longer ID period than diesel, as shown in Figure 6. As a result, SME fuels tend to have longer period of fuel injection before the SOC, which leads to most of the fuel mass to be burned during the PMC phase. Thus, the PMC phase is generally longer. Conversely, neat CME and PME as well as their B50 blends have significantly shorter ID periods than that of diesel, as shown in Figure 6. This leads to earlier SOC, which results in substantial fuel mass burned during the MCC phase and, thus, longer MCC phases at all loads. In addition to the fuel type, the engine load also significantly influences the HRR profile. From Figure 9a, it is found that all of the fuels have significantly longer PMC phases at low load compared to medium and high loads. Negligible MCC periods are observed for neat SME and diesel fuel at low load. However, there are clear changes in the combustion profiles when the load is increased from low to high, whereby a significant portion of the fuel is now burned during the MCC phase for all of the fuels. The dominance of the MCC phase as the load is raised is illustrated in Figure 9b,c. The reason for these observations is that, at low load,

Figure 6. Computed ID period and vaporized fuel mass at the SOC for all of the test fuels at three different engine loads.

Figure 7. Variations in the computed ID period when the base fuel (diesel) is substituted by biodiesels at different engine loads.

Figure 8. Variations of the computed ID period with the engine load for all of the test fuels at 2000 rpm.

vaporization are highly dependent on these properties. At high load, all of the tested biodiesel fuels have longer liquid spray penetration lengths because of higher fuel injection pressure. The fuel droplets also tend to undergo a greater atomization rate when exiting the fuel injector because of the higher injection pressure. The combined effects of improved atomization and longer spray penetration length result in a higher vaporization 215

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

Figure 10. Computed peak PMC values for the test fuels at 0.5, 1.5, and 2.5 kW.

vaporized mass up to the point of the SOC compared to neat biodiesel fuels at low load, as depicted in Figure 6. As such, B50 blends typically have higher peak PMC compared to their B100 counterparts. At medium and high load conditions, the fuels that exhibit longer IDs have higher mass of fuel burned (MFB) values during the PMC phase. As the ID period is increased, more combustible mixture is available to be consumed over a short period of time, thus giving rise to the high peak PMCs observed in Figure 9b,c. Generally, there is no real correlation between the engine load and peak PMC values. There is also no clear correlation between the engine load and the ability of biodiesel fuel to affect the peak PMC. Nevertheless, there are clear trends when the fuel type and blend level are varied. Here, the highest reduction in the peak PMC value is achieved by neat CME, followed by neat PME and, finally, neat SME, as shown in Figure 11. Similar trends are also

Figure 9. HRR profiles for the test fuels at (a) 0.5 kW, (b) 1.5 kW, and (c) 2.5 kW.

the fuel injection period is shorter. Almost all of the fuel mass is fully injected before the SOC; hence, a significant portion of the fuel mass is burned during the PMC phase. On the other hand, the fuel injection period is much longer at high load to accommodate the greater fuel mass required. This causes the fuel to be burned in the diffusion flame of the MCC phase. Additionally, the change in the combustion phasing from PMC to MCC is enhanced at low load when diesel is replaced with neat CME and PME as well as their B50 blends. The change in the combustion behavior is not significant at higher load because all the fuels tend to undergo the MCC phase due to the longer fuel injection period. Peak PMC Value. The peak PMC is affected by the vaporization rate of the individual fuel and also by the ID. At low load, the PMC phase is mainly influenced by the quantity of the available combustible mixture because almost all of the fuel is injected before the SOC. Fuels with higher vaporization rate have higher peak PMC and vice versa, as shown in Figures 9a and 10. All B50 blends are also observed to have marginally higher

Figure 11. Variations in the computed peak PMC value when the base fuel (diesel) is substituted by biodiesel at different engine loads.

observed when diesel fuel is substituted with B50 blends of CME, PME, and SME. This can be attributed to the chemical composition of biodiesel fuel. Both neat and B50 blends of SME have the highest unsaturation level, which therefore produces a longer ID period. A longer ID period leads to a greater quantity of vaporized fuel mass available before the SOC. At autoignition point, the higher quantity of the combustible mixture is burned over a short period, giving rise to the high peak PMC values seen in Figures 9−11. In contrast, neat and B50 blends of CME tend to have shorter ID periods because of the 216

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

Figure 12. Variations in the combustion characteristics of the test fuels: timings of (a) the peak PMC, (b) EOPMB, (c) CA50, and (d) CA90.

low unsaturation levels. Consequently, a lower amount of vaporized fuel mass is available for combustion at the SOC, resulting in a lower peak PMC. Combustion Efficiency. In CI engines, the highest energy efficiency is achieved when most of the heat release from combustion occurs near the TDC.9 The combustion efficiency of different fuels at various engine operating conditions can be identified using data extracted from the HRR profile, namely, the combustion duration, end of combustion (EOC), CA of 50% MFB (CA50), and CA of 90% MFB (CA90). Fuels with short combustion duration, high MFB near the TDC, and early CA50 and CA90 are regarded as fuels with higher combustion efficiency. Prolonged heat release in the expansion stroke leads to more heat loss through heat transfer between the hot gases and combustion chamber walls. In the expansion stroke, the hot gas mixture is exposed to the expanding control volume. This results in higher contact surface between the hot gases and chamber walls. With a higher wall surface exposed to the hot gases, a higher rate of heat loss occurs. Therefore, the combustion process is regarded as less efficient if a significant amount of energy is released at a relatively later stage of the expansion stroke. In order to facilitate the subsequent discussion, the definitions and conventions used to analyze the combustion efficiency are detailed first. The combustion duration is defined as the CA period between the SOC and 90% MFB. The combustion duration is utilized to determine the combustion speed of the fuels. Meanwhile, the EOC is defined as the CA position with 90% MFB (abbreviated here as CA90). The timings for 50% MFB (CA50) and 90% MFB (CA90) are utilized to map the combustion efficiency of each test fuel at different engine loads. In this study, neat biodiesels and B50 biodiesel blends are

compared against the combustion efficiency of diesel fuel at different loads. Referring to Figure 12a−c, the timings of the peak PMC, end of premixed burn (EOPMB), and CA50 are advanced when diesel is substituted with neat CME, PME, and SME, as shown for all loads. Similar observations are attained when diesel is substituted with B50 blends of CME, PME, and SME. These trends are mainly due to the varying ID periods and SOIs of the fuels. Here, all of the biodiesel fuels (neat and B50) except neat SME have shorter ID periods than diesel. In addition, all of the biodiesel fuels (neat and B50) have advanced SOI timings compared to diesel. On the other hand, CA90 is a function of both the engine load and fuel type. Here, CA90 is advanced at medium load, but at high load, CA90 is delayed for in all cases when diesel is substituted with biodiesel fuel. However, CA90 is delayed at low load when diesel is substituted with biodiesel fuel except for neat SME, as illustrated in Figure 12d. This may be due to the dependency of the burn speed on the combustion phasing of each fuel. In terms of the overall combustion duration or speed, all neat biodiesels and their blends except neat SME exhibit a slower combustion speed compared to diesel when the combustion process is predominantly PMC. Furthermore, biodiesel fuel burns at a lower rate than diesel when the combustion process is predominantly MCC. However, when the combustion process is balanced between PMC and MCC (typically at medium load for this study), the biodiesel fuels tend to burn faster than diesel fuel. This is because each fuel has different combustion behavior at different loads due to its chemical composition and thermophysical properties. Among the biodiesel fuels, the timings of the peak PMC, EOPMB, and CA50 are advanced the most when diesel is substituted with CME (B100) followed by PME (B100) and 217

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels



finally SME (B100), as seen in Figure 12a−c. The trends are also similar for the B50 counterparts. Here, the ID period is responsible and consistent with the observation that SME produces the longest ID period followed by PME and finally CME. It can be deduced that the CME fuels have the highest ability to reduce the peak PMC values. This attribute of CME may be one of the contributing factors to its ability to reduce exhaust NOx.

Article

AUTHOR INFORMATION

Corresponding Author

*Tel.: +603 89248161. Fax: +603 89248017. E-mail: hoonkiat. [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the Ministry of Science, Technology and Innovation of Malaysia under Grant 03-02-12-SF0045. The Faculty of Engineering at The University of Nottingham, Malaysia Campus, is also acknowledged for its support toward this project.

5. CONCLUSIONS This paper reports the mixture preparation process and combustion characteristics of neat and B50 blends of CME, PME, and SME biodiesels compared to diesel fuel. The differences in the fuel properties have significant effects on the in-cylinder mixture preparation process, thus affecting combustion and emission characteristics. It is found that the SOI timing is advanced for biodiesel because of the higher bulk modulus, surface tension, and viscosity, which collectively increases the incylinder temperatures and promotes higher NOx. Biodiesel fuels are found to have larger fuel droplet sizes, longer spray penetration, and lower vaporisation rate compared to diesel. The liquid penetration length is observed to be highly influenced by the timing of the SOC event. The vaporization rates of CME, PME, and SME are observed to be enhanced when B50 blends of the respective fuels are used compared to the neat biodiesels. The B50 blends of CME, PME, and SME have higher vaporization rates because of the faster evaporation of the lighter 50% diesel component in these multicomponent fuel blends. The change in the fuel type from diesel to biodiesel alters both the physical and chemical delays and, hence, the overall ID period. As a result, variations in the combustion phasing between the fuels are seen. Generally, fuels with longer ID periods tend to have higher fuel mass burned in the PMC phase and fuels with shorter ID periods have greater fuel mass burned in the MCC phase at a specific engine load condition. With the exception of neat SME, the ID is reduced when diesel is substituted with neat biodiesels or biodiesel blends at all loads. The degree of reduction in the ID period is enhanced when B50 blends of CME, PME, and SME are used compared to their B100 counterparts. The use of neat biodiesel enhances the magnitude of the reduction in the ID at high load, while at low load, the magnitude of the reduction in the ID is moderated. This trend is reversed when diesel is substituted with B50 blends of CME, PME, and SME. The ID period is reduced when the load is increased for neat CME, PME, and SME. However, the ID increases as the load is increased for the B50 blends of CME, PME, and SME. Even though engine load alters the combustion behavior, the combustion process itself is still very much dependent on the fuel type at each load. Fuels with long ID periods (i.e., neat and B50 blend of SME, diesel) have higher MFB values during the PMC phase. Conversely, neat and B50 blends of both CME and PME have higher MFB values during the MCC phase. It is found that neat and B50 blend of CME have the highest magnitude reductions in the peak PMC compared to diesel, whereas neat and B50 blend of SME have the lowest magnitude reductions in the peak PMC. The fuel burn duration or speed is found to be highly influenced by the combustion behavior, which is dictated by the fuel type and engine load. Biodiesel fuels burn at a lower rate compared to diesel under the PMC or the MCC phase. However, when the combustion periods of the PMC and MCC phases are similar, the burn speed is greater for biodiesel fuel compared to diesel.



NOMENCLATURE ALMR adaptive local mesh refinement B0 100% diesel B50 50% biodiesel−diesel mixture B100 100% biodiesel CA crank angle CA50 crank angle with 50% mass fraction burned CA90 crank angle with 90% mass fraction burned CFD computational fluid dynamics CO carbon monoxide CO2 carbon dioxide CME coconut methyl ester CI compression ignition HC hydrocarbon EOC end of combustion EOPMB end of premixed burn EVO exhaust valve opening HRR heat-release rate ID ignition delay IVC intake valve closing MCC mixing-controlled combustion MFB mass fraction burned N2 nitrogen NO nitrogen monoxide NOx nitrogen oxide oxygen O2 PaSr partially stirred reactor PMC peak premixed combustion PME palm methyl ester PSR perfectly stirred reactor SME soy methyl ester SOC start of combustion TDC top-dead center Xi normalized data Zi maximum/minimum data λ1 , λ2 normalization range



REFERENCES

(1) Ng, J. H.; Ng, H. K.; Gan, S. Recent trends in policies, socioeconomy and future directions of the biodiesel industry. Clean Technol. Environ. Policy 2010, 12 (3), 213−238. (2) Ng, H. K.; Gan, S. Combustion performance and exhaust emissions from the non-pressurised combustion of palm oil biodiesel blends. Appl. Therm. Eng. 2010, 30 (16), 2476−2484. (3) Ayhan, D. Biodiesel production from vegetable oils via catalytic and non-catalytic supercritical methanol transesterification method. Prog. Energy Combust. Sci. 2005, 31, 466−487. (4) Mueller, C.; Boehman, A.; Martin, G. An Experimental investigation of the origin of increased NOx emissions when fueling a

218

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219

Energy & Fuels

Article

heavy-duty compression-ignition engine with soy biodiesel. SAE Tech. Pap. Ser. 2009, 2009-01-1792. (5) Ng, J. H.; Ng, H. K.; Gan, S. Engine-out characterisation using speed−load mapping and reduced test cycle for a light-duty diesel engine fuelled with biodiesel blends. Fuel 2011, 90, 2700−2709. (6) Ng, J.-H.; Ng, H. K.; Gan, S. Characterisation of engine-out responses from a light-duty diesel engine fuelled with palm methyl ester (PME). Appl. Energy 2012, 90 (1), 58−67. (7) Chakravarthy, K.; McFarlane, J.; Daw, S. C.; Ra, Y.; Reitz, R. D.; Griffin, J. Physical properties of soy biodiesel & implications for use of biodiesel in diesel engines. SAE Tech. Pap. Ser. 2007, 2007-01-4030. (8) Ra, Y.; Reitz, R. D.; McFarlane, J.; Daw, S. C. Effects of fuel physical properties on diesel engine combustion using diesel and biodiesel fuels. SAE Tech. Pap. Ser. 2008, 2008-01-1379. (9) Heywood, J. B. Internal Combustion Engine Fundamentals; McGraw-Hill International: New York, 1988. (10) Yuan, W.; Hansen, A. C.; Zhang, Q. Predicting the physical properties of biodiesel for combustion modelling. Trans. ASAE 2003, 6 (46), 1487−1493. (11) Yuan, W.; Hansen, A. C.; Zhang, Q. Vapor pressure and normal boiling point predictions for methyl esters and biodiesel fuels. Fuel 2005, 84, 943−950. (12) Brakora, J.; Ra, Y.; Reitz, R. D.; McFarlane, J.; Daw, C. S. Development and Validation of a Reduced Reaction Mechanism for Biodiesel-Fueled Engine Simulations. SAE Tech. Pap. Ser. 2008, 200801-1378. (13) Yang, J.; Golovitchev, V. I. Construction of combustion models for rapeseed methyl ester biodiesel fuel for internal combustion engine applications. Biotechnol. Adv. 2009, 27, 641−655. (14) Usman, A.; Zheng, M. Fast heat release characterization of diesel engine. Int. J. Therm. Sci. 2008, 47, 1688−1700. (15) Uwe, H.; Rofl, E.; Johansson, B.; Andersson, Ö . Detailed Heat Release Analyses with Regard to Combustion of RME and Oxygenated Fuels in an HSDI Diesel Engine. SAE Tech. Pap. Ser. 2007, 2007-010627. (16) D’Errico, G.; Ettorre, D.; Lucchini, T. Comparison of combustion and pollutant emission models for DI diesel engines. SAE Tech. Pap. Ser. 2008, 2007-24-0045. (17) D’Errico, G.; Lucchini, T.; Atzler, F.; Rotondi, R. Computational fluid dynamics simulation of diesel engines with sophisticated injection strategies for in-cylinder pollutant controls. Energy Fuels 2012, 26, 4212−4223. (18) Lucchini, T.; D’Errico, G.; Ettorre, D. Numerical investigation of the spray/mesh/turbulence interactions for high-pressure, evaporating sprays at engine conditions. Int. J. Heat Fluid Flows 2011, 32, 285−297. (19) Ismail, H.; Ng, H. K.; Cheng, X.; Gan, S.; Lucchini, T.; D’Errico, G. Development of thermophysical and transport properties for the CFD simulations of in-cylinder biodiesel spray combustion. Energy Fuels 2012, 26, 4857−4870. (20) Ismail, H.; Ng, H. K.; Gan, S. Evaluation of non-premixed combustion and fuel spray models for in-cylinder diesel engine simulation. Appl. Energy 2012, 90 (1), 271−279. (21) Han, Z.; Reitz, R. D. A temperature wall function formulation for variable-density turbulent flows with application to engine convective heat transfer modeling. Int. J. Heat Mass Transfer 1997, 40, 613−625. (22) Bianchi, G. M.; Pelloni, P. Modeling the diesel fuel spray breakup by using a hybrid model. SAE Tech. Pap. Ser. 1999, 1999-01-0226. (23) Karrholm, F. P.; Nordin, N. Numerical investigation of mesh/ turbulence spray interaction for diesel applications. SAE Tech. Pap. Ser. 2005, 2005-01-2115. (24) Contino, F.; Jeanmart, H.; Lucchini, T.; D’Errico, G. Coupling of in situ adaptive tabulation and dynamic adaptive chemistry: An effective method for solving combustion in engine simulations. Proc. Combust. Inst. 2011, 33, 3057−3064. (25) Ismail, H.; Ng, H. K.; Gan, S.; Lucchini, T.; Onorati, A. Development of a reduced biodiesel combustion kinetics mechanism for CFD modelling of a light-duty diesel engine. Fuel 2012, DOI: 10.1016/ j.fuel.2012.10.015.

(26) Golovitchev, V. I.; Atarashiya, K.; Tanala, K.; Yamada, S. Towards Universal EDC-Based Combustion Model For Compression Ignited Engine Simulation. SAE Tech. Pap. Ser. 2003, 2003-01-1849. (27) Ismail, H.; Ng, H. K.; Queck, C. W.; Gan, S. Artificial neural networks modelling of engine-out responses for a light-duty diesel engine fuelled with biodiesel blends. Appl. Energy 2012, 92, 769−777.

219

dx.doi.org/10.1021/ef301162d | Energy Fuels 2013, 27, 208−219