Premixed Burn Fraction: Its Relation to the Variation in NOx Emissions

Jun 13, 2013 - This involved sweeping the injection timing of an instrumented 2 L diesel engine from 14 BTDC (before top-dead-center) to 3 ATDC (after...
0 downloads 7 Views 1MB Size
Article pubs.acs.org/EF

Premixed Burn Fraction: Its Relation to the Variation in NOx Emissions between Petro- and Biodiesel D. M. Peirce,*,† N. S. I. Alozie,† D. W. Hatherill,‡ and L. C. Ganippa† †

Centre for Advanced Powertrain and Fuels Research, School of Engineering and Design, Brunel University, Uxbridge, UB8 3PH, United Kingdom ‡ Finning Power Systems, 688-689 Stirling Road, Slough, SL1 4ST, United Kingdom ABSTRACT: It is commonly reported in the literature that NOx emissions from a diesel engine increase when fuelling with biodiesel. However, some studies report varying or opposite results. This work scrutinized operating conditions known to yield both increases and decreases in NOx emissions when running on biodiesel. This involved sweeping the injection timing of an instrumented 2 L diesel engine from 14 BTDC (before top-dead-center) to 3 ATDC (after top-dead-center), under loads of 40 Nm and 80 Nm (equating to BMEP (brake mean effective pressure) of 2.5 bar and 5 bar, respectively), using ultralow sulfur diesel (ULSD) and rapeseed methyl ester (RME). Under a 40 Nm load, RME consistently generated lower NOx emissions than ULSD, whereas, under an 80 Nm load, RME generated higher NOx emissions at all but the most advanced/retarded injection timings. This behavior was linked to differences in combustion duration, ignition delay (ID), and the relative size of the premixed burn fraction (PMBF). Combustion tended to progress more quickly overall for the fuel that generated highest NOx emissions at most operating conditions. ID was always reduced when fuelling with RME, and hence PMBF was also reduced. Thus, reduced ID exerted conflicting influences over relative RME NOx emissions; a tendency to increase NOx, due to advanced start of combustion (SOC), and a tendency to decrease NOx, due to reduced PMBF. Additionally, calculations indicated that for the same SOC and PMBF RME would normally be expected to generate higher NOx emissions than ULSD. However, as the level of premixing increased, the magnitude of the ceteris paribus RME NOx increase appeared to decline. That is, as PMBF increases, the impact of the inherent factors beyond advanced SOCthat lead to higher NOx emissions when fuelling with biodiesel appear to be reduced. This may be related to variations in soot radiative heat losses. Changes in operating PMBF may therefore explain some of the variety that exists in the literature relating to the effects of biodiesel fuelling on NOx emissions.

1. INTRODUCTION 1.1. Biodiesel. Biodiesel is an alternative fuel composed of fatty acid alcohol esters and produced by the transesterification of organic oils, which can be derived from any of a wide range of renewable feedstocks.1 In their natural form, the major components of plant and animal oils are triglycerides (i.e., three fatty acids attached to a glycerine backbone). Although straight vegetable oils (SVOs) can be used as a fuel source,2 they have a number of drawbacks that make them ill-suited for use with existing diesel engine technology,3 primarily a viscosity many times greater than that of the traditional petrodiesel on which the majority of compression ignition engines are designed to run. The removal of the glycerine component and subsequent division of the oil molecules into more mobile single chain fatty acid esters by transesterification4 provides a fuel with improved flow properties, better suited to use in commercial diesel engines. There are, however, still many notable differences between the physicochemical properties of biodiesel and those of petrodiesel. The density, bulk modulus, viscosity, surface tension, cloud point, flash point, boiling point, and specific heat capacity of biodiesel tend to be higher, as well as the average hydrocarbon chain length, oxygen content, and cetane number; conversely, the degree of saturation, aromatic content, sulfur content, and lower heating value tend to be lower than those of petrodiesel.5−7 That said, the strong influence of biodiesel feedstock over the properties of the resulting fuel makes generalization tricky. In the majority of cases the common © 2013 American Chemical Society

changes listed above hold true, but for certain feedstocks, coconut or babassu oils, for example, which are very highly saturated, the magnitude of any particular change may be reduced or even the direction reversed.8 Among the many political, economical, technical and environmental advantages of biodiesel, improved exhaust emissions are very widely reported, with reductions in net carbon dioxide, carbon monoxide, unburned hydrocarbon and particulate emissions being anticipated in most instances.1,9−12 Emissions of nitrogen oxides, on the other hand, are commonly reported as being increased.13−15 1.2. NOx Emissions. Emissions of nitrogen oxides, or NOx, have become an increasingly critical issue in recent years due to the implementation of ever more stringent legislation that curtails the allowable limits to a fraction of their former levels. The concern surrounding NOx emissions in particular comes as a result of the role that they play in the formation of acid rain, ground-level ozone (a component of photochemical smog) and other air toxicity concerns.16,17 The term ‘NOx’ encompasses all oxides of nitrogen, but engine-out NOx emissions are comprised in the largest part of NO, and the remainder is almost all NO2, with only trace quantities of other molecules such as N2O being emitted.18 The formation of NOx occurs through several main Received: April 15, 2013 Revised: June 12, 2013 Published: June 13, 2013 3838

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

(particularly aromatics) having higher adiabatic flame temperatures than the majority of biodiesel components.36 Studies that have calculated the adiabatic flame temperature for representative fuel surrogates seem to suggest that the stoichiometric adiabatic flame temperature for biodiesel may be slightly lower36,37 or approximately equal to38,39 that of petrodiesel. Despite this, there may still be significant differences in adiabatic flame temperatures between petro- and biodiesel, but possibly being brought about by changes in local oxygen equivalence ratio.38 In fuel-rich regions the oxygen content of biodiesel would lead to an equivalence ratio relatively closer to unity and, thus, a higher adiabatic flame temperature than petrodiesel under the same fuel-air mixing conditions.37,38,40 Fuel-rich regions of the flame are not ordinarily expected to be where the bulk of NOx is generated,26,38 due to the relatively low temperatures at which rich flames burn and so it would seem that if these localized increases in flame temperature are important then it is because of some knock-on effect that they have (e.g., increased early reaction rates and higher downstream temperatures37) rather than their slightly increased temperature alone. It has been reported that there is a direct correlation between decreasing oxygen equivalence ratio in the autoignition zone and increasing NOx emissions.41 Local reduction in oxygen equivalence ratio, beyond its effects on adiabatic flame temperature and direct additional provision of oxygen for the NOx formation process itself, also serves to reduce in-cylinder soot concentrations. A reduction in soot concentrations has been observed in biodiesel flames,42,43 and in engines operating on both biodiesel44 and other oxygenated fuels.45 This reduction in soot leads to a reduction in soot radiative heat losses; soot radiative heat losses have been identified as a significant cause of NOx reduction when operating on traditional petrodiesel, due to the associated temperature reduction.46 Since in-cylinder soot quantities are lower when fuelling with biodiesel, the magnitude of this temperature reduction may decline, leading to increased temperatures and higher NOx emissions.38 In addition to increased thermal NOx due to a reduction in radiative heat losses, it has also been hypothesized that a reduction in soot formation may be chemically associated with increased NOx formation via the prompt NO mechanism.47,48 That is, because soot formation consumes C2H2 and hence reduces the availability of the hydrocarbon radical CH, a reduction in soot formation may increase CH availability and therefore increase the reaction rate that limits the primary prompt NO formation route. Sooting is reduced when fuelling on biodiesel because of the oxygen content of the fuel and the chemical structure of the hydrocarbons of which it is composed. Fuel-bound oxygen reduces sooting by producing CO and CO2 directly and hence diminishing the quantities of fuel-carbon available for the formation of soot precursors like C2H2, C3H3, and benzene,49 from which polycyclic aromatic hydrocarbons (PAHs) form and grow, ultimately becoming large enough to develop into soot particle nuclei.50 It is also thought that fuel-bound oxygen increases the concentration of oxygen radicals (O, OH, HCO, etc.) in fuel-rich regions of the flame.49 The effect of these radicals is 3-fold: first, they encourage more complete combustion, oxidizing available fuel-carbon to CO and CO2; second, they oxidize simple aromatics, curtailing the development of PAHs; and third, in the leaner periphery of the jet, they precipitate the oxidation of any soot that has formed.51 It has been seen in experimental52 and theoretical studies53 that fuelbound oxygen reduces soot emissions more efficiently than

mechanisms. In a diesel engine, the extended Zeldovich or thermal mechanism is generally considered to be dominant.19 The thermal mechanism is a temperature dependent process; formation rates become increasingly significant past approximately 1800 K, but beneath that, they are relatively low.20,21 The rate-limiting first equation of the mechanism, in which molecular nitrogen reacts with atomic oxygen to yield NO and a nitrogen free radical, has a high activation-energy and is essentially inactive at low temperatures, due to the high stability of the N2 triple-bond. In addition to this, the equilibrium O radical concentration, another important factor connected to the rate of NO formation, is also positively correlated with temperature.21−24 As well as being influenced by temperature changes, thermal NOx is dependent upon mixture stoichiometry and high-temperature residence times. Mixture stoichiometry affects NOx through its relation to adiabatic flame temperature but is also important because the rate at which NOx can form is reliant upon the availability of oxygen and nitrogen.25 As a consequence of this, it is in the fuel-lean environment surrounding the nearstoichiometric diffusion flame that NOx formation rates are normally expected to be highest.26,27 However, the thermal mechanism is known to proceed slowly relative to the time scales involved in the combustion process, and thus, much of the thermal NOx formation takes place in the hot postflame gases.20,25 Although thermal NOx formation is slow, formation rates can be increased by superequilibrium concentrations of the requisite radicals (as can be found in flame-fronts) and other faster acting NOx formation mechanisms. Among these, the prompt NO route is most important, whereby NOx is produced via hydrocarbon free radicals, primarily CH.28−31 This route is particularly pertinent to a discussion of biodiesel emissions, because it has been speculated that hydrocarbon free radical chemistry32 and prompt NO33,34 may bear responsibility for some portion of the apparent NOx increase. Prompt NO is an important consideration in premixed fuel-rich hydrocarbon combustion and in close proximity to the diffusion flamefront,18 because this is where concentrations of the CH radical are highest.23 In the majority of reported cases, it has been observed that biodiesel generates increased NOx emissions compared to petrodiesel, and a variety of explanations have been posited for this ‘biodiesel NOx increase’.10−15 Fundamentally, the main causal factors at the root of these explanations can be considered to center around the following properties: compared to petrodiesel, biodiesel has a higher oxygen content (around 11% for biodiesel, depending upon feedstock,6 but practically nil for petrodiesel), an altered hydrocarbon compositionnotably, a reduced proportion of low ignition quality aromatic hydrocarbons (practically nil in biodiesel, historically 20−40%35 in petrodiesel, although an upper limit of 11% is now specified by EN5905)and physicochemical properties that are generally less conducive to mixing. Most of the explanations put forward in the following introductory sections are drawn directly from the literature, others are based upon interpretation, and a minority of propositions are informed by data that will be presented and discussed in the body of this paper. 1.3. NOx Emissions from Biodiesel. It has been suggested that, since unsaturated hydrocarbons have higher adiabatic flame temperatures than saturated species and biodiesel tends to contain a greater proportion of unsaturated elements than petrodiesel, a portion of the biodiesel NOx increase may be the consequence of a de facto higher adiabatic flame temperature for biodiesel.19 However, although petrodiesel consists primarily of saturated hydrocarbons, it also contains constituents 3839

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

at 15 °C, 833/879 kg·m3; heat capacity, 1750/2000 J·kg−1·K−1; 10−90% boiling range, 220−300/320−340 °C.5,6,71,72) In relation to PMBF, it is also important to consider the oxygen content of biodiesel, which reduces the quantity of air that the fuel-jet is required to entrain before a combustible mixture is generated,37 and the reduced energy density of biodiesel (indicative values: ULSD, 43 MJ·kg−1; RME,6 37.6 MJ·kg−1), which necessitates that a greater quantity of fuel be premixed in order to form a PMBF containing an equal proportion of the total energy. Additionally, reduced energy density increases the necessary injection duration, reducing the amount of energy available in-cylinder at any moment between the SOI and end of injection (EOI) and limiting the potential for premixing. In relation to energy density, it should be noted that although biodiesel has a reduced heating value compared to petrodiesel, it is also more dense; considered volumetrically, the density increase will mitigate the effects of heating value reduction, but the volumetric energy density of biodiesel remains lower than that of petrodiesel. A larger PMBF is generally understood to correlate with higher NOx emissions (all other things being equal).73,74 Explanations proposed for the PMBF/NOx correlation include the following: • For a given SOC, the greater the PMBF the more heat will be released earlier in the cycle. Although the premixed flame temperature may be lower than that of the diffusion flame, the fact that temperatures and pressures rise earlier means that more compressionheating (and/or less expansion-cooling) is able to occur before the point of maximum pressure, leading to higher peak burnt-gas temperatures. Theoretical peak burnt-gas temperatures have been found to correlate well with PMBF and measured NOx emissions under some conditions.74,75 • The high temperature products of the premixed flame may have less immediate access to the cooler regions of in-cylinder gas than the products of the diffusion flame. Therefore, they may remain at high temperatures for longer as a result of being less quickly diluted into the cool surrounding gas.74 • A larger PMBF is expected to generate less in-cylinder soot, and therefore radiative heat losses would be reduced, leading to higher actual temperatures.74 • Increased NO formed via the prompt mechanism may also contribute.74,76,77 In addition to hypotheses relating changes in prompt NO formation to sooting differences between petro- and biodiesel (as mentioned earlier), there is some suggestion in the literature that biodiesel may generate higher prompt NO due to higher concentrations of hydrocarbon radicals that exist in the flames as a consequence of the increased quantity of unsaturated molecules present in the fuel.78,79 1.4. Objectives. Past and ongoing work undertaken at the Brunel University Centre for Advanced Powertrain and Fuels Research (CAPF) has found both slight increases and decreases in NOx emissions, dependent upon injection timing and load, when fuelling with biodiesel.80 Although the literature does account for this kind of variety,11−13 it is still an inconsistency that requires thorough explanation. The engine on which tests have been run is typically operated at relatively low loads, and connections have previously been made in the literature between load and NOx emissions from biodiesel fuelled

oxygen enrichment of the intake air, which is consistent with the notion that it is the rate of formation and growth of the initial rings that determines overall sooting propensity54,55 and suggests that this effect of oxygen provision during the early stages of formation is a critical factor in biodiesel soot reduction. Studies have reported that the sooting tendency of hydrocarbon species runs in the following descending order: aromatics > alkynes > alkenes > alkanes, under diffusion burning conditions and under premixed conditions, aromatics > alkanes > alkenes > alkynes.54 That is, aromaticity in particular tends to increase the sooting tendency of a fuel, as does increasing length of the hydrocarbon chain.56,57 The absence of aromatics in biodiesel, then, can be seen to be a critical point. Aromatics also have low cetane numbers,58 contributing the typically longer ignition delay (ID) of petrodiesel when compared to biodiesel. However, the ID of biodiesel varies widely depending on feedstock;1,59,60 cetane number increases with fatty acid chain length and reduced degree of unsaturation.19,61 Despite the often large proportions of polyunsaturated molecules in biodiesel, which may themselves have longer ignition delays than petrodiesel,58,62 the longer and more highly saturated chains present in biodiesel generally compensate for this deficit. (For indicative purposes, cetane numbers estimated from the literature for the fuels employed in the experiment documented in this paper are ULSD, 51, and RME, 54.5,6) The shorter ID of biodiesel advances the start of combustion (SOC) relative to petrodiesel under the same injection conditions, causing earlier increases in in-cylinder temperatures, longer high-temperature residence times, and hence promoting higher NOx emissions. Advanced SOC also advances the angle at which peak in-cylinder pressure occurs, tending to move it into closer proximity to TDC where, because the temporary volume is smaller, pressures and temperatures become higher. This effect is compounded when engines using a mechanical pump-line-nozzle (PLN) injection system are fuelled using biodiesel, because the increased bulk modulus of the fuel causes an advance in start of injection (SOI) timing, further advancing SOC.7,63 This effect is largely nullified in modern engines utilizing common-rail injection systems64 (as in the case of the study documented in this paper). Although the shorter ID of biodiesel leads to an advanced SOC, tending to increase NOx, it also reduces the size of the premixed burn fraction (PMBF) tending to reduce NOx. This is related to the time available for fuel-air mixing to occur. In addition to reduced time for fuel-air mixing, the physicochemical properties of biodiesel may also detract from the efficacy of mixture preparation. It is reported that, at the point of ignition, the jet length of biodiesel is approximately similar to that of petrodiesel,37 despite the shorter ID; this can be attributed to the fact that biodiesel spray breakup occurs later and, postbreakup, the biodiesel jet penetrates at a greater rate than petrodiesel due to greater droplet diameters65 (although the magnitude of this difference is likely to be reduced at higher incylinder temperatures). However, the jet is also narrower and therefore entrains less air. These factors (i.e., delayed breakup, larger droplet diameters (poorer atomization), and shallower spray plume angle) can be attributed, at least in part, to the higher viscosity, surface tension, density, heat capacity, and boiling point (slower evaporation) of biodiesel.65−70 (For indicative purposes, the following are properties estimated from the literature for ULSD/RME: viscosity at 40 °C, 3.3/4.5 mm2·s−1; surface tension at 20 °C, 25.8/30.6 mN·m−1; density 3840

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

engines.12,13,81−87 The objective of this study was to monitor operational conditions that were expected to yield both positive and negative changes in NOx emissions when fuelling with biodiesel and to identify possible combustion parameters contributing to the variations in relative NOx emissions between the two fuels.

apparent steady state; this typically took around 90 min under an 80 Nm load. At each operational condition data collection was postponed until all emissions had reached apparent consistency when viewed over a 180 s duration. In-cylinder pressure data were collected over 100 engine cycles per measurement, and the measurement was repeated 5 times for each point in the experimental matrix. The experimental matrix was completed twice in full, to confirm repeatability. Emissions data were recorded over 60 s intervals, twice for each point in the experimental matrix. Again, this process was repeated for confirmatory purposes. The emissions were measured using a Horiba MEXA 7170 DEGR emissions analyzer, which was recalibrated before work commenced on all occasions. 2.2. Data Analysis. Raw data collected using LabView were loaded into MATLAB and batch processed to retrieve the pertinent information. In-cylinder pressure was processed to extract relevant datapeak pressure, angle of peak pressure, angle between start of combustion (SOC) and peak pressureand to calculate apparent heat release rate (AHRR) data using the traditional first law heat release model,25 without any modeling of heat transfer or crevice effects, and using an assumed constant specific heat ratio of 1.35. This is a very basic approach, but it is held to be sufficient for the purposes of comparison. Approximations of cp/cv were made from the logarithm of pressure versus logarithm of volume charts in order to ensure that a constant value would represent both fuels equally well. The calculated ratios of specific heats were found to be essentially consistent between ULSD and RME under varying conditions, and so an assumed value of 1.35 was taken to be adequate. The definitions on which calculations of AHRR related parameters were based are illustrated in Figure 1 and explained in the accompanying text. Each 100 cycle pressure data set was used to generate a single average pressure trace, in order to reduce noise while maintaining the essential characteristics of combustion. It should also be noted that although pressure data was only logged by the data acquisition system once per crank angle degree, all values were interpolated to one decimal place by the MATLAB code. All heat release parameters were calculated from the AHRR curve without filtering or averaging, except for the end of combustion, which was defined on the basis of the moving average of AHRR in order to improve consistency, and the end of premixed burn, which was calculated from the second derivative of AHRR.

2. METHODOLOGY 2.1. Experimental Setup. Details of the experimental setup are given in Table 1. Concisely, the experiment that was performed

Table 1. Details of Engine and Experimental Setup engine type no. cylinders swept volume compression ratio injection system injection pressure engine speed fuels loads injection timings dynamometer emissions analyzer pressure transducer DAQ card shaft encoder

Ford Duratorq (PUMA) high speed direct injection (HSDI) 4 1.998 L 18.2:1 common-rail 800 bar 1800 rpm ultralow sulfur diesel (ULSD)/rapeseed methyl ester (RME) 40 N·m (2.5 bar BMEP)/80 N·m (5 bar BMEP) −14 ATDC to 3 ATDC (at 1 degree increments) Schenck W130 eddy-current dynamometer Horiba MEXA 7170 DEGR Kistler Piezostar 6125A (single cylinder) National Instruments PCI-6070E Lenord+Bauer GEL244 pickup with 360 toothed encoder wheel

involved the collection of in-cylinder pressure and emissions data under 40 Nm and 80 Nm loads (BMEPs of 2.5 bar and 5 bar, respectively), running at a constant speed of 1800 rpm, fuelling on ultralow sulfur diesel (ULSD) and rapeseed methyl ester (RME), sweeping start of injection (SOI) timing from −14ATDC to 3ATDC in one degree increments. The load was changed by an alteration in fuelling achieved by an elongation of the single injection event. The engine is also equipped with a turbocharger, an intercooler, and an exhaust gas recirculation (EGR) system, but none of these were employed for the purposes of this study. An in depth specification of the fuels used was not available (some indicative values have been provided in the introductory sections). After start-up, the engine was allowed to warm up until hydrocarbon emissions (the slowest pollutant to stabilize) had settled to an

1. Start of injection (SOI) was defined from the commanded SOI set within the engine management software. Any potential difference between commanded and actual SOI, due to solenoid delay for instance, should be consistent between measurements, since engine speed was held constant. 2. Ignition delay (ID) was defined as the difference between commanded SOI and calculated SOC. 3. Start of combustion (SOC) was defined as the point at which the AHRR curve crossed the x axis; that is, the heat release rate became positive.

Figure 1. Labeled plot of heat release and the derivatives used to calculate combustion criteria. 3841

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 2. NOx emissions as a function of start of injection timing for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads.

Figure 3. In-cylinder pressure (A and B) and apparent heat release rate (C and D) charts for ULSD and RME under 40 Nm and 80 Nm loads at −8 ATDC injection timing. 4. Premixed burn fraction (PMBF) was defined as the integral of the AHRR curve between SOC and EOPMB, divided by the integral of the AHRR curve between SOC and EOC. 5. End of premixed burn (EOPMB) was defined as the first point at which the second differential of heat release rate reached a local maximum following the global minimum. Under most conditions, this approximately corresponds to the position at which the AHRR reaches a first local minimum following the global maximum, but the second differential was used instead because under low loads there is not always a clear local minimum in the AHRR curve, as can be seen in Figure 1. 6. End of combustion (EOC) was defined as the first point at which the moving average of heat release rate dropped below zero. A moving average was used to minimize problems due to noise, while still being representative of the general tendencies of the data. Other values calculated from the in-cylinder pressure data included total apparent heat release, peak AHRR, angle of peak AHRR, and angle between SOC and peak AHRR, PMBF, 10−90% burn fraction intervals, duration of partial burn fraction intervals, and average burn rates through partial intervals. Emissions data were averaged over the 60 s durations recorded.

3. RESULTS AND DISCUSSION 3.1. NOx Emissions. Figure 2 shows the relationship between NOx emissions and SOI for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads. As can be seen in Figure 2A, NOx emissions were consistently lower for RME than for ULSD under a 40 Nm load, but under an 80 Nm load (Figure 2B), the relationship was less consistent. With SOI 14 degrees before TDC, NOx emissions for both fuels were approximately equal under an 80 N m load; as injection timing was retarded the RME NOx emissions became higher than those from ULSD, with the peak difference being recorded at −8ATDC; from there, the difference declined until both fuels recorded almost equal emissions again at −1ATDC; and past this point, further retardation led to a relative reduction in RME NOx emissions, to levels below those of ULSD. (Note that although the differences between the two fuels appear small under some conditions, the trends observed were observed consistently; the average standard deviation of repeated measurements was less than 25 ppm.) 3842

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 4. Ignition delay as a function of start of injection timing for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads.

Figure 5. NOx emissions as a function of start of combustion timing for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads.

3.2. Pressure and Heat Release. Taking the −8ATDC SOI position as an initial point of focus (because it was the position at which NOx emissions from RME were highest relative to ULSD under an 80 Nm load), plots of pressure and AHRR for both fuels at both loads can be seen in Figure 3. The pressure traces (Figures 3A and B) do not display any vast differences, but there are important points to be drawn from them. The first feature to note is that the RME curves rise earlier in both cases, but they are advanced to a slightly greater degree under the 80 Nm load. This assertion is supported by the heat release plots (Figures 3C and D), which show the same trend, and by the ID data presented in Figure 4, which indicate that the ID of RME was always shorter than that of ULSD and was reduced to a greater extent by the increase in load. (Note that the average standard deviations of ID values were less than 0.1 CA under the 40 Nm load and less than 0.2 CA under the 80 Nm load.) When operating temperatures are higheras they are at higher load (Figure 4B)vaporization of biodiesel will occur more readily than at lower temperatures, reducing the physical component of ID time.25,88,89 Therefore, it is probable that the larger observed reduction in ID at higher load is related to the impact of temperature change upon the physicochemical properties of biodiesel (its higher viscosity, density, heat capacity, and surface tension, reduced vapor pressure, etc., which make the fuel generally more resistant to vaporization). Although the change in temperature would accelerate the vaporization of petrodiesel, too, it may close the physical delay gap between the two fuels, making the chemical aspect of ID more clearly prominent. Further to the discussion of ID, Figure 5 shows NOx emissions plotted against SOC (as opposed to SOI in Figure 2), which serves to demonstrate the immediate impact of

RME’s shorter ID. Comparing the 80 Nm plots (Figures 2B and 5B), it can be seen that much of the difference between the two fuels can be directly attributed to the advance in SOC. As a function of SOC, the RME NOx emissions curve is effectively shifted left, moving it into closer concord with the ULSD curve under the increased load. RME NOx emissions remain above the ULSD trend for SOI conditions between −10 ATDC and −5 ATDC in Figure 5B (as a function of SOC and equating to SOC timings between approximately −8 ATDC and −2 ATDC), where in Figure 2B (as a function of SOI) they had exceeded the ULSD trend from −13 ATDC to −2 ATDC. Although it seems that advanced SOC plays a critical, primary role in the increase in NOx emissions from RME under an 80 Nm load, other secondary differences remain in the absence of this effect. The second feature to note from the pressure traces (Figures 3A and B) is that the peak in-cylinder pressure reached is higher, albeit modestly, for RME in both instances. This holds true for most injection timings under both load conditions. However, although under an 80 Nm load the difference between peak pressure for ULSD and RME is quite well correlated with the difference in NOx emissions (with relatively higher peak pressures occurring alongside relatively higher NOx), at the lower load this is not the case. This can be seen by comparing Figures 6A and 2A with Figures 6B and 2B. This may be because at lower load high temperatures are less widely spread, and so a global pressure value provides a less valuable temperature indicator in terms of the peak temperatures responsible for NOx formation; for instance, small localities of very high temperature can generate higher NOx emissions than the same amount of heat-energy evenly distributed, although the two scenarios could yield identical pressure data. It is 3843

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 6. Maximum in-cylinder pressure as a function of start of injection timing for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads.

Figure 7. Cumulative burned fraction (A) as a function of partial combustion duration (between SOC and CA90) and (B) as a function of crank angle position (between SOC and CA50) for ULSD and RME under 40 Nm and 80 Nm loads at −8 ATDC injection timing.

fuels, but for ULSD, they become steeper and remain steeper for longer than is the case for RME (indicative of a larger premixed burn fraction). The important difference to note between the 40 Nm and 80 Nm curves is that, following the initial sharper ascent of ULSD, the RME curve remains behind the ULSD curve at 40 Nm but ‘overtakes’ the ULSD curve at 80 Nm. Figure 7B shows absolute combustion intervals through the first half of combustion in terms of crank angle, rather than relative to SOC, and thus includes the direct influence of ID. This affords a clearer examination of differences in the actual combustion progress of the fuels, with RME beginning ahead in both cases and falling behind through the premixed burn phase. (A similar trend, biodiesel combustion being ‘overtaken’ by petrodiesel through the early burn phase, only to later catch up, has been previously reported by Bittle, Knight, and Jacobs.64,81) Here, the brevity of the period for which RME falls absolutely behind ULSD under an 80 Nm load at this injection timing can be seen clearly (with RME trailing only at the 40% combustion interval). This short period of recession seems to be generally characteristic of the SOI timing positions that produced relatively higher NOx emissions for RME under an 80 Nm load. How combustion progress varies across the full SOI range is qualitatively illustrated in Figure 8, which consists of binary plots identifying the fuel that reached each combustion interval earliest under given operational conditions. Figure 8, examined from the left to right, serves to clarify the way in which RME starts out ahead, due to its shorter ID, with SOC and 10−20% burn fraction intervals being reached earlier in all cases. Under the 40 Nm load (Figure 8A), ULSD catches up and overtakes RME; under the 80 Nm load (Figure 8B),

noteworthy that peak pressure was normally higher when fuelling with RME, even under conditions where NO x emissions were lower. This is likely to be related to RME’s earlier SOC. The third feature to note is that, on the 40 Nm pressure trace in Figure 3A, the RME curve remains noticeably higher than that of ULSD throughout the decline, indicating that RME combustion continued for longer, lagging behind that of ULSD. This can also be seen in the heat release plot (Figure 3C), which shows that ULSD initially released heat more intensely, reaching a higher peak AHRR. On the 80 Nm pressure trace (Figure 3B), the RME curve does not lag behind the ULSD curve in the same way. Although, just as at 40 Nm the peak heat release rate is higher for ULSD, the difference is that at 80 Nm a more substantial diffusion burn phase occurs after the premixed burn is completed, as can be seen from the AHRR plot in Figure 3D. This means that, despite the fact that heat is not released as quickly in the premixed phase, under increased load an increased rate of post-premixed heat release essentially compensates for the reduction in premixed AHRR. This can be seen more clearly in Figures 7, 8, and 9, which show partial combustion durations, relative combustion progress, and average partial heat release rates, respectively. Figure 7A confirms that at the −8 ATDC injection timing combustion proceeded more slowly and took longer to complete (completion here being represented by the 90% burn fraction interval) for RME under the 40 Nm load and for ULSD under the 80 Nm load. That is to say, generally, combustion was more prolonged at each load for the fuel that generated lowest NOx emissions. It can be seen, under both loads, that the burn fraction curves begin similarly for the two 3844

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 8. Relative combustion progress of ULSD and RME over partial combustion intervals under (A) 40 Nm and (B) 80 Nm loads across injection timings from −14 to 3 ATDC.

Figure 9. Average percentage of total fraction burned per degree over partial combustion intervals for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads at −8 ATDC injection timing.

plotted for the SOI timing of −8 ATDC in Figure 9. Since these plots are essentially derived from the AHRR data presented in Figures 3C and D, there is a clear similarity in form. As seen in Figure 9B, under the 80 Nm load, ULSD appears to be faster through the 10−40% burn intervals and slower through 40−70%, which, given the extended duration of the later burn fractions, gives RME adequate time to compensate for its slower early burn rate. From Figure 9A, the ULSD burn rate under the 40 Nm load can be seen to be significantly higher between the 10% to 70% burn intervals, without the RME having any period of similar advantage during which to ‘catch up.’ Largely, it appears that the difference in heat release rates and combustion durations may be correlated with the relative size of the PMBF that each fuel forms under any particular operational condition. At low load, where no fast mixingcontrolled burn phase occurs (referring to a period subsequent to the culmination of the premixed burn wherein an obvious increase in AHRR occurs), the fuel with the largest PMBF (i.e., ULSD) invariably combusts more quickly. At higher load, where a more significant diffusion burn phase presents itself, the secondary increase in heat release rate through this diffusion phase tends to cause a relative reduction in the total

ULSD catches up, overtakes, but then slows down and falls behind again. While at the injection timings between −8 ATDC and −4 ATDC it can be seen that ULSD combustion catches up with RME combustion only briefly under the 80 Nm load, at highly advanced and highly retarded timings the period for which ULSD remains ahead seems to be maximized, to some extent correlating with the reduced relative NOx emissions from RME that were recorded at these positions, as can be seen in Figures 2B and 5B. At injection conditions from 11 to 12 BTDC at 40 Nm, and 9−11 BTDC and 14 BTDC at 80 Nm, it appears that behavior contrary to the wider trends that have been described occurs during the final (80−90%) interval. This is unexpected, but without any good reason to regard the data as erroneous, it can only be suggested that this inconsistency found at some advanced injection timings may come about as a result of lateburn oscillatory fluctuations in heat release data, such as those visible in Figure 3C and D, which can induce variation into the calculation of later burn intervals. An important difference between the combustion behavior of the two fuels seems to be related to the gradient profile of the curves in Figure 7, which can be thought of as the average fractional burn (or fractional heat release) rate; this data is 3845

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 10. Difference in NOx emissions between ULSD and RME as a function of injection timing under an 80 Nm load with representative average fractional burn rate diagrams.

RME under low temperature conditions, detracting from the equivalence ratio reducing effects of the fuel-bound oxygen content and playing a part in reduced RME heat release rates under the 40 Nm load. Additionally, there may be important differences in the nature and effects of spray-wall impingement and liquid wall film formation. Figure 10 shows how the appearance of the fractional burn rate plot varies with injection timing, allowing for consideration of how this variation may be related to the difference in NOx emissions from the two fuels under an 80 Nm load. Examining burn rate Figures 10A (SOI at −14 ATDC) and 10B (−1 ATDC), allows for consideration of two conditions under which the differences in NOx emissions (as a function of SOI) between ULSD and RME were similar, that is, close to zero. In Figure 10A, the RME curve follows a shape that closely resembles that of the ULSD curve, only lower, implying that if the two curves were brought into line then RME would be expected to have relatively higher NOx emissions, all other things remaining the same. In part, this is because combustion begins sooner and therefore much of the heat being released is released slightly earlier. In Figure 10B, it can be seen that the peak burn rates for both fuels are considerably lower than in Figure 10A (on account of reduced PMBF) and that the RME peak is also smaller relative to the ULSD peak. A late increase in burn rate (beyond the 40% interval, as in Figure 9B) is visible on the RME plot, and since the relative NOx emissions are very similar (between the points illustrated in Figures 10A and B), it appears that this late increase might compensate for some of the reduction in peak burn rate. However, the ID difference between the fuels in Figure 10B is considerably larger than in Figure 10A, meaning that the effect of advanced SOC on NOx emissions will be more significant at B than at A. That is to say, based on burn rate profile alone, relative RME NOx emissions would be expected to be lower at B than at A, but RME’s more greatly advanced SOC at this point makes up for the difference.

combustion duration (SOC to CA90) and partial combustion duration (diffusion phase) of the fuel with a smaller PMBF (i.e., RME). Elsewhere,81 it has been suggested that the diffusion burn rate of biodiesel might be higher than that of petrodiesel, which is a tempting and plausible hypothesis. Under the conditions studied here, where the duration of the diffusion burn phase was always limited, it seemed that both fuels followed broadly similar fractional burn rate patterns when PMBF was approximately the same. That is, the RME diffusion burn rate did not necessarily appear to be any higher, per se, but because RME always had a smaller PMBF (due largely to its shorter ID) and underwent a relatively pronounced diffusion burn phase, while ULSD transitioned more directly between premixed and slow ‘late combustion’ phases,25,90 post-premixed combustion of RME did occur more quickly. However, the correct controls were not in place to allow a direct comparison between the diffusion burning rates of the two fuels, absent of differences in PMBF and combustion phasing, and hence, it is not possible to draw any definitive conclusions about burn rate one way or the other. As has previously been touched upon with regard to ID, the difference between the efficacy with which the two fuels vaporise and entrain air is likely to be larger at low load, due to the cooler in-cylinder conditions and their effect upon the properties of the fuels. It is therefore possible that differences in fuel atomization, vaporization, and the stoichiometry of the resulting mixture also contribute to the variations observed in heat release rates. Less complete evaporation and poorer distribution of fuel prior to ignition could lead to more locally rich regions throughout the premixed charge, in which total oxidation is not possible. This would be likely to delay heat release and reduce heat release rates through the premixed phase (as well as reducing flame temperatures, generating soot precursors, and otherwise affecting exhaust emissions). Since RME evaporates less readily than ULSD, these locally rich regions may be relatively more prevalent when fuelling with 3846

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 11. (A) Premixed burn fraction as a function of ignition delay and (B) NOx emissions as a function of premixed burn fraction for ULSD and RME under 40 Nm and 80 Nm loads.

via injection sweeps has the advantage that it can be used to induce changes in combustion behavior without making any modification to the properties of the fuels themselves, but it is an approach inevitably encumbered by the inclusion of effects related to changing SOC; predominantly, increased peak incylinder pressures and prolonged high-temperature residence times as injection timing is advanced. If, for the sake of simplicity, it is taken that NOx emissions vary with injection timing due to only three main factors - peak pressure, residence times, and combustion profile - then in order to observe the effects of any element alone, it is necessary to take steps to exclude the influence of the others. This has been achieved here retroactively, by the following process: 1. Two injection timing conditions were identified at which the combustion profiles were similar for each fuel under each load (the same two conditions, −7 ATDC and −2 ATDC, were used for both fuels and under both loads. Quantification of the similarity: ID, maximum variation of ID between −7ATDC and −2ATDC of 0.1 CA [where the values were 2.5 CA and 2.6 CA], with a mode difference of 0 CA; PMBF, maximum condition-tocondition variation of 0.02 [between values of 0.35 and 0.37], with a mean difference of 0.015; peak AHRR: maximum condition-to-condition variation of 1.3 J/deg [45 J/deg and 46.3 J/deg], with a mean difference of 0.65 J/deg). 2. The simplistic assumption was then made that, given their similar combustion profiles, all differences in NOx emissions between these two points were due to an approximately linear variation in peak pressure and residence time caused by the change in SOC. On this crude basis, a general SOC related NOx gradient was calculated at each load by dividing the change in NOx emissions between the two similar points by the change in SOC. 3. A compensation value was defined using this calculated gradient for every operational condition, as a function of SOC, and this value was subtracted from the measured NOx emissions to calculate ‘SOC compensated relative NOx emissions.’ SOC compensated relative NOx emissions derived by this means can be seen in Figure 12. There are regions in which these compensated emission values drop below zero, which is obviously unrealistic in an absolute sense, but these values are only intended to possess relative validity. The utility of this

Reflecting back on Figure 9B (−8 ATDC), it can be seen that the RME peak burn rate is not relatively lower by such a great extent as in Figure 10B, and the late burn rate is more markedly increased. Consider also that the ID difference at −8 ATDC is smaller than that at point B (see Figure 4B), indicating that the relative increase in NOx emissions between −8 ATDC (Figure 9B) and −1 ATDC (Figure 10B) is dependent upon additional factors associated with combustion behavior beyond advanced SOC alone. In Figure 10C (3 ATDC), the peak magnitudes for both fuels are higher than in Figure 10B but lower than in Figure 10A. Here, it is apparent that the RME peak is relatively very much lower than that of ULSD (indicating a large difference in PMBF between the two fuels), without any late pickup in burn rate, which is a profile consistent with the relatively reduced NOx emissions from RME at this point. 3.3. Premixed Burn Fraction. Figure 11A shows PMBF as a function of ID, and Figure 11B shows NOx emissions as a function of PMBF. From Figure 11A, it can be seen that ID and PMBF correlate positively with a fair degree of linearity (R2 values: ULSD (40 Nm), 0.88; RME (40 Nm), 0.93; ULSD (80 Nm), 0.92; and RME (80 Nm), 0.96) for each fuel but that RME tends to generate a smaller PMBF for the same ID. It is anticipated that this is the result of the reduced energy density, poorer mixing properties, and slower evaporation of biodiesel. (Note also that the −14 ATDC data point under a 40 Nm load, which had an apparent PMBF of 0.95 and can be seen in Figure 11B, has been omitted from Figure 11A, and from later PMBF plots; at this point, combustion duration for ULSD was consistently too short to allow accurate calculation of PMBF.) In Figure 11B, it can be seen that at all tested conditions PMBF was lower for RME than for ULSD. It is also clear that PMBF values are largest toward the highly advanced and highly retarded injection timing positions and lowest toward the center of the timing sweep; these trends are related to ID. Additionally, although at low load the trends for both fuels are generally consistent throughout, under the 80 Nm load there is a marked difference in the PMBF trend toward the most retarded injection timings. At these points, PMBF increases more dramatically for ULSD, leading to a large difference in combustion profiles between the fuels (as seen in Figure 10C)a difference large enough to lead to a significant relative reduction in RME NOx emissions. The nature of the relationship between PMBF and NOx emissions is not made clear by Figure 11B; the curves take on ‘C’ shapes due to the influence of SOC timing. Collecting data 3847

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 12. Start of combustion compensated relative NOx emissions as a function of start of combustion timing for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads.

Figure 13. Start of combustion compensated relative NOx emissions as a function of premixed burn fraction for ULSD and RME under (A) 40 Nm and (B) 80 Nm loads.

hypothetical NOx/PMBF relationships for the two fuels extrapolated on the basis of the trends observed in Figure 13. The reason for the correlation gap observed here, and the way that it reduces with increasing PMBF, may be related in part to soot radiative heat losses. It has been reported that as the relative proportion of diffusion burning increases soot emissions also increase.57,90,91 This being understood, and assuming at least some degree of correlation between soot emissions and in-cylinder soot, it is possible to conceive of the following hypothesis: • For a large PMBF, in-cylinder soot quantities might be low when fuelling with both ULSD and RME, meaning comparable radiative heat losses, and the tendency to generate lesser differences in NOx emissions for similar combustion behavior. • For a reduced PMBF, in-cylinder soot quantities would be expected to be higher for ULSD than RME, leading to increased heat losses and lower NOx emissions when fuelling with ULSD. Although the data collected in this study does not bolster the hypothesis that the diffusion burn rate itself is higher when fuelling with biodiesel, if such were the case, as has been suggested,64,81 then that would provide an alternative (or complementary) explanation for the correlation gap that has been described. That is, as the PMBF becomes smaller the diffusion burn phase becomes larger; the biodiesel combustion duration would be further reduced relative to that of petrodiesel and, therefore, heat released earlier due to the increasing prominence of this accelerated diffusion phase; hence, relative biodiesel NOx emissions would be expected to increase with declining PMBF. Additional factors bearing possible responsibility for the correlation gap may include changes in

approach can be seen by comparing Figure 11B with Figure 13, which shows compensated NOx emissions as a function of PMBF. In Figure 13, a fair degree of correlation is apparent between PMBF and the compensated NOx emissions values, particularly at the higher load (R2 values: ULSD (40 Nm), 0.97; RME (40 Nm), 0.94; ULSD (80 Nm), 0.98; and RME (80 Nm), 0.97). From these approximately linear relationships, it can be seen that SOC compensated relative NOx emissions increase with increasing PMBF. Under an 80 Nm load (Figure 13B) RME clearly generates higher NOx emissions for the same PMBF, while under a 40 Nm load (Figure 13A) the tendency appears to be similar but is less clear-cut. Interestingly, it appears that, as PMBF increases, the gap between NOx emissions from the two fuels narrows. Toward PMBF values of around 0.8, the NOx gap is nullified completely (based upon extensions of the calculated linear fits). Bearing in mind that RME always had a reduced PMBF, it can be proposed that, beyond the primary effect of advanced SOC when fuelling with biodiesel, the difference in NOx emissions between the two fuels will be influenced by X. The difference between the PMBF of the two fuels. Y. The size of the NOx/PMBF correlation gap between the two fuels in the operational PMBF range. Under all conditions except for those yielding extremely high PMBFs, X and Y will tend to offset each other; a large X and/or a small Y would reduce relative RME NOx emissions, a small X and/or a large Y would increase relative RME NOx emissions. X is determined directly from the PMBF of the two fuels. Y is conceptualized as a function of PMBF, being large when PMBF is low and becoming smaller as PMBF increases. Figure 14 clarifies this conceptual explanation graphically, with regard to 3848

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

Figure 14. Conceptual illustration of the relationship between the premixed burn fraction of the two fuels and the relative NOx emissions anticipated.

operational temperatures that alter rates of fuel evaporation and local mixture stoichiometry and changes in combustion chemistry, radical concentrations, flame temperatures, and heat transfer. PMBF appears to exert influence over, or be correlated with, the extent to which other inherent differences between petro- and biodiesel are able to affect relative NOx emissions levels, although the nature of any correlation lines that might be calculated will certainly vary from engine to engine and depend upon the specific fuels that are utilized.

4. CONCLUSIONS 1. The NOx emissions from RME were lower than those from ULSD under a 40 Nm load, but at an increased load of 80 Nm they were higher at the majority of injection timing conditions. 2. Under all measured conditions, ULSD had a longer ID and a larger PMBF than RME. Additionally, for the same ID, ULSD generated a larger PMBF than RME, indicative of the reduced energy density, inferior mixing behavior, and slower evaporation associated with RME. 3. The immediate effect of reduced ID is that SOC is advanced; when NOx emissions were considered as a function of SOC timing rather than SOI timing, the disparity between NOx emissions tendencies for the two fuels under an 80 Nm load was considerably reduced, suggesting that advanced SOC was responsible for a significant proportion of the increase in NOx emissions when fuelling with RME. 4. Under most operational conditions, the fuel that combusted most quickly, releasing heat earliest, was the fuel that generated the highest NOx emissions. a. The rate at which combustion progressed appeared to be related to the size of the PMBF. b. At low load, the PMBF for both fuels was so large that no fast mixing-controlled diffusion burn phase occurred. Under these conditions, ULSD combusted most quickly overall because it had a larger PMBF. c. At increased load, the PMBF for RME was often low enough that a significant fast diffusion burn phase did occur, compensating for the larger, faster PMBF of ULSD. Under these conditions, RME combusted most quickly overall. 5. When the effects of changes in SOC were compensated for, NOx emissions appeared to correlate fairly linearly (positively) with PMBF at each load.



a. Correlations were different for each fuel: RME tended to generate higher NOx emissions than ULSD for the same PMBF. b. Correlations were different at each load: the magnitude of the difference between RME and ULSD NOx emissions for the same PMBF was larger under an 80 Nm load than under a 40 Nm load. c. As PMBF increased, the calculated difference between RME and ULSD NOx emissions for a given PMBF became smaller. d. Some of the differences in correlation may plausibly be related to soot radiative heat transfer, local mixture stoichiometry, and any extant differences in diffusion burn rate. 6. In-cylinder pressure and heat release data provide useful practical feedback on engine behavior, and a valuable means for inferring information about in-cylinder phenomena. However, a complete understanding of the petrodiesel/biodiesel NOx difference will ultimately require more fundamental knowledge of the physical and chemical processes that govern the manner in which these fuels vaporize and combust.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Gratitude is due to Finning (U.K.) Ltd. for providing the funding and direction for this research. Thanks are also extended to technicians Andrew Selway, Kenneth Antiss, and Clive Barrett, and to colleagues Fanos Christodoulou, Gholamreza Abbaszadeh Mosaiiebi, Mohammed Abahussain, Prasad Boggavarapu, Dr. Stephen Hemmings, Dr. Yan Zhang, and Dr. Radu Beleca for their assistance.



REFERENCES

(1) Knothe, G.; Van Gerpen, J.; Krahl, J. The Biodiesel Handbook; AOCS Press: Champaign, IL, 2005. (2) Sidibe, S. S.; Blin, J.; Vaitilingom, G.; Azoumah, Y. Use of crude filtered vegetable oil as a fuel in diesel engines state of the art: Literature review. Renewable Sustainable Energy Rev. 2010, 14, 2748− 2759.

3849

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

(3) Tyson, K. S.; McCormick, R. L. Biodiesel Handling and Use Guidelines; Report No. NREL/TP-540-43672; National Renewable Energy Laboratory: Golden, CO, 2006. (4) Van Gerpen, J.; Shanks, B.; Pruszko, R.; Clements, D.; Knothe, G. Biodiesel Production Technology; Report No. NREL/SR-510-36244; National Renewable Energy Laboratory: Golden, CO, 2004. (5) Automotive Fuels, Diesel, Requirements and Test Methods; BS EN 590:2009+A1:2010; British Standards Institute: London, U.K., 2010. (6) Hoekman, S. K.; Broch, A.; Robbins, C.; Ceniceros, E.; Natarajan, M. Review of biodiesel composition, properties, and specifications. Renewable Sustainable Energy Rev. 2012, 16, 143−169. (7) Tat, M. E. Investigation of oxides of nitrogen emissions from biodiesel-fueled engines. Ph.D. Dissertation, Iowa State University, Ames, IA, 2003. (8) Sanford, S. D.; White, J. M.; Shah, P. S.; Wee, C.; Valverde, M. A.; Meier, G. R. Feedstock and Biodiesel Characteristics Report; Renewable Energy Group: Ames, IA, 2009. (9) Atabani, A. E.; Silitonga, A. S.; Badruddin, I. A.; Mahlia, T. M. I.; Masjuki, H. H.; Mekhilef, S. A comprehensive review on biodiesel as an alternative energy resource and its characteristics. Renewable Sustainable Energy Rev. 2012, 16, 2070−2093. (10) Szybist, J. P.; Song, J.; Alam, M.; Boehman, A. L. Biodiesel combustion, emissions, and emission control. Fuel Process. Technol. 2007, 88, 679−691. (11) Lapuerta, M.; Armas, O.; Rodriguez-Fernandez, J. Effect of biodiesel fuels on diesel engine emissions. Prog. Energy Combust. Sci. 2008, 34, 198−223. (12) Xue, J.; Grift, T. E.; Hansen, A. C. Effect of biodiesel on engine performances and emissions. Renewable Sustainable Energy Rev. 2011, 15, 1098−1116. (13) Sun, J.; Caton, J. A.; Jacobs, T. J. Oxides of nitrogen emissions from biodiesel-fuelled diesel engines. Prog. Energy Combust. Sci. 2010, 36, 677−695. (14) Varatharajan, K.; Cheralathan, M. Influence of fuel properties and composition on NOx emissions from biodiesel powered diesel engines: A review. Renewable Sustainable Energy Rev. 2012, 16, 3702− 3710. (15) Hoekman, S. K.; Robbins, C. Review of the effects of biodiesel on NOx emissions. Fuel Process. Technol. 2012, 96, 237−249. (16) NOx: How Nitrogen Oxides Affect the Way We Live and Breathe; Report No. EPA-456/F-98-005; United States Environmental Protection Agency, Office of Air Quality Planning and Standards: Research Triangle Park, NC, 1998. (17) Nitrogen Oxides (NOx), Why and How They Are Controlled; Report No. EPA-456/F-99-006R, United States Environmental Protection Agency, Office of Air Quality Planning and Standards: Research Triangle Park, NC, 1999. (18) Glassman, I.; Yetter, R. A. Combustion, 4th ed.; Elsevier Academic Press: Burlington, MA, 2008. (19) Ban-Weiss, G. A.; Chen, J. Y.; Buchholz, B. A.; Dibble, R. W. A numerical investigation into the anomalous slight NOx increase when burning biodiesel; A new (old) theory. Fuel Process. Technol. 2007, 88, 659−667. (20) Turns, S. R. An Introduction to Combustion: Concepts and Applications, 2nd ed.; Tata McGraw-Hill: New Delhi, India, 2012. (21) Modeling Pollutant Formation. In ANSYS Fluent, Fluent 6.3 Documentation; Fluent Inc.: Lebanon, NH, 2006; Chapter 20. (22) Sawyer, R. F. The formation and destruction of pollutants in combustion processes: clearing the air on the role of combustion research. Proc. Combust. Inst. 1981, 18, 1−22. (23) Dean, A. M.; Bozzelli, J. W. Combustion chemistry of nitrogen. In Gas-Phase Combustion Chemistry; Gardiner, W. C., Ed.; SpringerVerlag: New York, 2000; pp 125−343. (24) Hill, S.; Douglas Smoot, L. Modeling of nitrogen oxides formation and destruction in combustion systems. Prog. Energy Combust. Sci. 2000, 26, 417−458. (25) Heywood, J. B. Internal Combustion Engine Fundamentals; McGraw-Hill: New York, 1988.

(26) Dec, J. E. A conceptual model of DI diesel combustion based on laser-sheet imaging. SAE Technical Paper 970873, 1997, 10.4271/ 970873. (27) Flynn, P. F.; Durrett, R. P.; Hunter, G. L.; zur Loye, A. O.; Akinyemi, O. C.; Dec, J. E.; Westbrook, C. K. Diesel combustion: An integrated view combining laser diagnostics, chemical kinetics, and empirical validation. SAE Technical Paper 1999-01-0509, 1999, 10.4271/1999-01-0509. (28) Fenimore, C. P. Formation of nitric oxide in premixed hydrocarbon flames. Proc. Combust. Inst. 1971, 13, 373−380. (29) Miller, J. A.; Bowman, C. T. Mechanism and modeling of nitrogen chemistry in combustion. Prog. Energy Combust. Sci. 1989, 15, 287−338. (30) Bowman, C. T. Kinetics of pollutant formation and destruction in combustion. Prog. Energy Combust. Sci. 1975, 1, 33−45. (31) Konnov, A. A. Implementation of the NCN pathway of promptNO formation in the detailed reaction mechanism. Combust. Flame 2009, 156, 2093−2105. (32) McCormick, R. L.; Alvarez, J. R.; Graboski, M. S. NOx Solutions for Biodiesel; Report No. NREL/SR-510-31465; National Renewable Energy Laboratory: Golden, CO, 2003. (33) McCormick, R. L.; Williams, A.; Ireland, J.; Brimhall, M.; Hayes, R. R. Effects of Biodiesel Blends on Vehicle Emissions; Report No. NREL/ MP-540-40554; National Renewable Energy Laboratory: Golden, CO, 2006. (34) Hess, M. A.; Haas, M. J.; Foglia, T. A.; Marmer, W. N. Effect of antioxidant addition on NOx emissions from biodiesel. Energy Fuels 2005, 19, 1749−1754. (35) Graboski, M. S.; McCormick, R. L. Combustion of fat and vegetable oil derived fuels in diesel engines. Prog. Energy Combust. Sci. 1998, 24, 125−164. (36) Glaude, P. A.; Fournet, R.; Bounaceur, R.; Moliere, M. Adiabatic flame temperature from biofuels and fossil fuels and derived effect on NOx emissions. Fuel Process. Technol. 2010, 91, 229−235. (37) Mueller, C. J.; Boehman, A. L.; Martin, G. C. An experimental investigation of the origin of increased NOx emissions when fueling a heavy-duty compression-ignition engine with soy biodiesel; SAE Technical Paper 2009-01-1792, 2009, 10.4271/2009-01-1792. (38) Cheng, A. S.; Upatnieks, A.; Mueller, C. J. Investigation of the impact of biodiesel fuelling on NOx emissions using an optical direct injection diesel engine. Int. J. Engine Res. 2006, 7, 297−318. (39) Hashimoto, N.; Ozawa, Y.; Mori, N.; Yuri, I.; Hisamatsu, T. Fundamental combustion characteristics of palm methyl ester (PME) as alternative fuel for gas turbines. Fuel 2008, 87, 3373−3378. (40) Adi, G.; Hall, C.; Snyder, D.; Bunce, M.; Satkoski, C.; Kumar, S.; Garimella, P.; Stanton, D.; Shaver, G. Soy-biodiesel impact on NOx emissions and fuel economy for diffusion-dominated combustion in a turbo-diesel engine incorporating exhaust gas recirculation and common rail fuel injection. Energy Fuels 2009, 23, 5821−5829. (41) Ye, P.; Boehman, A. L. Investigation of the impact of engine injection strategy on the biodiesel NOx effect with a common-rail turbocharged direct injection diesel engine. Energy Fuels 2010, 24, 4215−4225. (42) Tran, M. K.; Dunn-Rankin, D.; Pham, T. K. Characterizing sooting propensity in biofuel-diesel flames. Combust. Flame 2012, 159, 2181−2191. (43) Bejaoui, S.; Lemaire, R.; Therssen, E.; Desgroux, P. Study of the effect of rapeseed methyl ester addition in diesel fuel on soot formation. In Proceedings of the European Combustion Meeting: Cardiff, U.K., 2011. (44) Klein-Douwel, R. J. H.; Donkerbroek, A. J.; Van Vliet, A. P.; Boot, M. D.; Somers, L. M. T.; Baert, R. S. G.; Dam, N. J.; ter Meulen, J. J. Soot and chemiluminescence in diesel combustion of bio-derived, oxygenated, and reference fuels. Proc. Combust. Inst. 2009, 32, 2817− 2825. (45) Mueller, C. J.; Pickett, L. M.; Siebers, D. L.; Pitz, W. J.; Westbrook, C. K.; Martin, G. C. Effects of oxygenates on soot processes in DI diesel engines: Experiments and numerical 3850

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

simulations. SAE Technical Paper 2003-01-1791, 2003, 10.4271/200301-1791. (46) Musculus, M. P. B. Measurements of the influence of soot radiation on in-cylinder temperatures and exhaust NOx in a heavy-duty DI diesel engine. SAE Technical Paper 2005-01-0925, 2005, 10.4271/ 2005-01-0925. (47) Guo, H.; Smallwood, G. J. The interaction between soot and NO formation in a laminar axisymmetric coflow ethylene/air diffusion flame. Combust. Flame 2007, 149, 225−233. (48) Ren, Y.; Li, X. Numerical simulation of the soot and NOx formations in a biodiesel-fuelled engine. SAE Technical Paper 201101-1385, 2011, 10.4271/2011-01-1385. (49) Cheng, A. S.; Dibble, R. W.; Buchholz, B. A. The effect of oxygenates on diesel engine particulate matter. SAE Technical Paper 2002-01-1705, 2002, 10.4271/2002-01-1705. (50) Tree, D. R.; Svensson, K. I. Soot processes in compression ignition engines. Prog. Energy Combust. Sci. 2007, 33, 272−309. (51) Kosaka, H.; Aizawa, T.; Kamimoto, T. Two-dimensional imaging of ignition and soot formation processes in a diesel flame. Int. J. Engine Res. 2005, 6, 21−42. (52) Schmidt, K.; Van Gerpen, J. The effect of biodiesel fuel composition on diesel combustion and emissions. SAE Technical Paper 961086, 1996, 10.4271/961086. (53) Zannis, T. C.; Pariotis, E. G.; Hountalas, D. T.; Rakopoulos, D. C.; Levendis, Y. A. Theoretical study of DI diesel engine performance and pollutant emissions using comparable air-side and fuel-side oxygen addition. Energy Convers. Manage. 2007, 48, 2962−2970. (54) Glassman, I. Soot formation in combustion processes. Proc. Combust. Inst. 1988, 22, 295−311. (55) Sidebotham, G. W.; Glassman, I. Flame temperature, fuel structure, and fuel concentration effects on soot formation in inverse diffusion flames. Combust. Flame 1992, 90, 269−283. (56) Ladommatos, N.; Rubenstein, P.; Bennett, P. Some effects of molecular structure of single hydrocarbons on sooting tendency. Fuel 1996, 75, 114−124. (57) Eastwood, P. Particulate Emissions from Vehicles; John Wiley and Sons Ltd.: Chichester, U.K., 2008. (58) Murphy, M. J.; Taylor, J. D.; McCormick, R. L. Compendium of Experimental Cetane Number Data; Report No. NREL/SR-540-36805; National Renewable Energy Laboratory: Golden, CO, 2004. (59) Bamgboye, A. I.; Hansen, A. C. Prediction of cetane number of biodiesel fuel from the fatty acid methyl ester (FAME) composition. Int. Agrophys. 2008, 22, 21−29. (60) Van Gerpen, J. Cetane number testing of biodiesel. In Proceedings of the Third Liquid Fuel Conference, Nashville, TN, 1996. (61) Hellier, P.; Ladommatos, N.; Allan, R.; Payne, M.; Rogerson, J. The impact of saturated and unsaturated fuel molecules on diesel combustion and exhaust emissions. SAE Technical Paper 2011-011922, 2011, 10.4271/2001-01-1922. (62) Schonborn, A.; Ladommatos, N.; Williams, J.; Allan, R.; Rogerson, J. The influence of molecular structure of fatty acid monoalkyl esters on diesel combustion. Combust. Flame 2009, 156, 1396−1412. (63) Szybist, J. P.; Boehman, A. L. Behavior of a diesel injection system with biodiesel fuel. SAE Technical Paper 2003-01-1039, 2003, 10.4271/2003-01-1039. (64) Bittle, J. A.; Knight, B. M.; Jacobs, T. J. The impact of biodiesel on injection timing and pulsewidth in a common-rail medium-duty diesel engine. SAE Technical Paper 2009-01-2782, 2009, 10.4271/ 2009-01-2782. (65) Boggavarapu, P.; Ravikrishna, R. V. A comparison of jatropha methyl ester and diesel non-evaporating sprays. In 12th International Conference on Liquid Atomization and Spray Systems, Heidlberg, Germany, 2012. (66) Ra, Y.; Reitz, R. D.; McFarlane, J.; Daw, C. S. Effects of Fuel physical properties on diesel engine combustion using diesel and biodiesel fuels. SAE Technical Paper 2008-01-1379, 2008, 10.4271/200801-1379.

(67) Higgins, B. S.; Mueller, C. J.; Siebers, D. L. Measurements of fuel effects on liquid-phase penetration in DI sprays. SAE Technical Paper 1999-01-0519, 1999, 10.4271/1999-01-0519. (68) Kuti, O. A.; Zhu, J.; Nishida, K.; Wang, X.; Huang, Z. Characterization of spray and combustion processes of biodiesel fuel injected by diesel engine common rail system. Fuel 2013, 104, 838− 846. (69) Lee, C. S.; Park, S. W.; Kwon, S. I. An experimental study on the atomization and combustion characteristics of biodiesel-blended fuels. Energy Fuels 2005, 19, 2201−2208. (70) Som, S.; Longman, D. E.; Ramirez, A. I.; Aggarwal, S. K. A comparison of injector flow and spray characteristics of biodiesel with petrodiesel. Fuel 2010, 89, 4014−4024. (71) Chhetri, A. B.; Watts, K. C. Surface tensions of petro-diesel, canola, jatropha and soapnut biodiesel fuels at elevated temperatures and pressures. Fuel 2013, 104, 704−710. (72) Rakopoulos, C.; Antonopoulos, K.; Rakopoulos, D. Multi-zone modeling of diesel engine fuel spray development with vegetable oil, bio-diesel, or diesel fuels. Energy Convers. Manage. 2006, 47, 1550− 1573. (73) Lee, R.; Pedley, J.; Hobbs, C. Fuel quality impact on heavy duty diesel emissions: A literature review. SAE Technical Paper 982649, 1998, 10.4271/982649. (74) Musculus, M. P. B. On the correlation between NOx emissions and the diesel premixed burn. SAE Technical Paper 2004-01-1401, 2004, 104271/2004-01-1401. (75) Ladommatos, N.; Parsi, M.; Knowles, A. The effect of fuel cetane improver on diesel pollutant emissions. Fuel 1996, 75, 8−14. (76) Love, N.; Parthasarathy, R. N.; Gollahalli, S. R. Concentration measurements of CH and OH radicals in laminar biofuel flames. Int. J. Green Energy 2011, 8, 113−120. (77) Hayhurst, A. N.; Vince, I. M. Nitric oxide formation from N2 in flames: The importance of prompt NO. Prog. Energy Combust. Sci. 1980, 6, 35−51. (78) Garner, S.; Brezinsky, K. Biologically derived diesel fuel and NO formation: An experimental and chemical kinetic study, Part 1. Combust. Flame 2011, 158, 2289−2301. (79) Som, S.; Longman, D. E. Numerical study comparing the combustion and emission characteristics of biodiesel to petrodiesel. Energy Fuels 2011, 25, 1373−1386. (80) Labecki, L. Combustion and emission characteristics of biofuels in diesel engines. Ph.D. Dissertation, Brunel University, Uxbridge, U.K., 2010. (81) Bittle, J. A.; Knight, B. M.; Jacobs, T. J. Interesting behavior of biodiesel ignition delay and combustion duration. Energy Fuels 2010, 24, 4166−4177. (82) Patterson, J.; Hassan, M. G.; Clarke, A.; Shama, G.; Hellgardt, K.; Chen, R. Experimental study of DI diesel engine performance using three different biodiesel fuels. SAE Technical Paper 2006-01-0234, 2006, 10.4271/2006-01-0234. (83) McCormick, R. L. Effects of Biodiesel on NOx Emissions; NREL/ PR-540-38296. California Air Resources Board Biodiesel Workgroup: Sacramento, CA, 2005. (84) Hoekman, S. K.; Broch, A.; Robbins, C.; Ceniceros, E. Investigation of Biodiesel Chemistry, Carbon Footprint, and Regional Fuel Quality; CRC Report No. AVFL-17a; Coordinating Research Council: Alpharetta, GA, 2011. (85) Giakoumis, E. G.; Rakopoulos, C. D.; Dimaratos, A. M.; Rakopoulos, D. C. Exhaust emissions of diesel engines operating under transient conditions with biodiesel fuel blends. Prog. Energy Combust. Sci. 2012, 38, 691−715. (86) Dorado, M. P.; Ballesteros, E.; Arnal, J. M.; Gomez, J.; Lopez, F. J. Exhaust emissions from a diesel engine fueled with transesterified waste olive oil. Fuel 2003, 82, 1311−1315. (87) Armas, O.; Gomez, A.; Cardenas, M. D. Biodiesel emissions from a baseline engine operated with different injection systems and exhaust gas recirculation (EGR) strategies during transient sequences. Energy Fuels 2009, 23, 6168−6180. 3851

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852

Energy & Fuels

Article

(88) Vaughn, T.; Hammill, M.; Harris, M.; Marchese, A. J. Ignition delay of bio-ester fuel droplets. SAE Technical Paper 2006-01-3302, 2001, 10.4271/2006-01-3302. (89) Obert, E. F. Internal Combustion Engines and Air Pollution; Harper and Row Publishers: New York, 1973. (90) Stone, R. Introduction to Internal Combustion Engines, 3rd ed.; Macmillan Press Ltd.: London, U.K., 1999. (91) Han, S. B.; Hinze, P. C.; Kwon, Y. J.; Mun, S. S. Experimental investigation of smoke emission dependent upon engine operating conditions. SAE Technical Paper 971658, 1997, 10.4271/971658.

3852

dx.doi.org/10.1021/ef4006719 | Energy Fuels 2013, 27, 3838−3852