Numerical Investigation of Negative Temperature Coefficient Effects on Sooting Characteristics in a Laminar Co-flow Diffusion Flame.

It is a common sense that diesel engines produce worse soot emission than gasoline engines, even though gasoline direct injection also brings about terrible sooting tendency. However, reports showed that diesel emits less soot than gasoline in laminar diffusion flames, which implies that soot emission is a combined effect of multiple factors, such as the combustion mode, physical properties of the fuel, and also fuel chemistry. This work, thus, conducted numerical calculations in laminar co-flow diffusion flames of fuels with different negative temperature coefficient (NTC) behaviors in an order of n-heptane > iso-octane > toluene to solely evaluate the chemical effect, especially the role of low-temperature combustion on soot formation. 2-Dimensional simulations were carried out to obtain the soot distributions, and 0-dimensional simulations were performed to analyze the chemical kinetics of polycyclic aromatic hydrocarbon (PAH) formation and low-temperature reaction sensitivities. The grids of the 2-D model converged at 80(r) × 196(z), and the boundary conditions of both models were set to eliminate the influence of physical factors as much as possible. The results showed that there were three main reactions associated to the formation of aromatic hydrocarbons A1 at the first-stage combustion in the n-heptane flame and the iso-octane flame, in which the reaction of C7H15 + O2 = C7H15O2 enhances the NTC behavior. The first two reaction pathways generated larger molecular hydrocarbons and were unfavorable by A1 formation and therefore inhabit the PAH formation, and 49.8% of C7H16 reacted through the large molecular pathways, while the percentage for C8H18, with weaker NTC behavior, was only 37%. Toluene with even weaker NTC behavior showed no low-temperature oxidation. Therefore, in a more general case, fuels with stronger NTC behavior smoke less, and this conclusion could be promising potential to reduce soot emission in future.

Introduction

Soot originates from the heterogeneous combustion of hydrocarbon fuels, which is extremely harmful to human health and the environment.[1−4] In the transportation sector, not only heavy-duty diesel engines suffer from their infamous soot emission reputation but also the recent emerging market GDI engines show significantly high soot emission tendency. Within this scenario, increasingly strict emission regulations were imposed all over the world and soot emission control has become a public priority.

Many researchers have conducted experimental observations on the sooting tendency among actual engines fueled with diesel, gasoline, and their blends.[5−7] It is agreed that the soot emissions from port fuel injection (PFI) gasoline engines is quite low[8] due to the comparatively high volatility that is suitable for the premix combustion mode. On the contrary, diesel engines incline to exhaust more soot emission since diesel with a lower auto-ignition temperature and low volatility is usually used in the compression–ignition (CI) engines and bears non-premix combustion.[9] Moreover, the optical results from Zheng et al.[10] proved that the addition of gasoline into diesel would reduce soot production as a result of an increased liftoff flame length and better air–fuel mixing. However, these observations do not necessarily imply that gasoline with a lighter average molecular weight emits less soot than diesel. In fact, some direct-injection gasoline engines emit just as many soot particles as unfiltered diesel cars did in the past.[11] In addition, in a fundamental flame research conducted by Liu et al.[12] that compared the sooting tendency of gasoline/diesel blend flames on a liquid burner with the diffuse 2D-LOSA technology, the 2-dimensional soot volume fraction value decreases as the proportion of diesel in the gasoline/diesel mixture increases, which indicates that diesel is less likely to produce soot than gasoline in a laminar diffusion flame. Therefore, it can be speculated that the sooting characteristics in an actual engine are influenced by the combined effects of multiple factors, such as the combustion mode, physical properties of the fuel, and also fuel chemistry.

As is known, the soot generation tendency of different fuels varies largely from fuel to fuel due to different molecular structures,[13−16] so the fuel chemistry plays a critical role in the soot formation. However, it is not easy to evaluate a single effect through an internal combustion engine or a constant volume chamber experiment since the combustion processes inside them are comprehensive physical–chemical combined processes. Thus, different kinds of fundamental flames have been wildly used to separate the chemical process out of other factors, and many academic fruits have been achieved on the soot formation.[17−19]

Polycyclic aromatic hydrocarbons (PAHs) are the most widely accepted precursors in the nucleation stage of soot particles.[20,21] There are many kinds of PAHs, including benzene (A1), naphthalene (A2), phenanthrene (A3), pyrene (A4), and other larger aromatic hydrocarbons. Different fuels own different molecular structures, which determine the PAH formation ability, and thus the respective sooting tendency. Oxygenated fuels, such as methanol, ethanol, butanol, and so forth, form less PAHs and soot due to the oxidation effect.[22−24] Aromatic hydrocarbon fuels, such as toluene, n-propyl benzene, and so forth, are easy to form PAHs and promote the soot nucleation.[25−28] Mechanisms for the formation of soot have also been extensively studied since the 1990s,[29−32] among which the hydrogen abstraction carbon addition (HACA) theory is widely accepted for the growth process from A1 to high-ring PAHs.[33,34] The mechanism argued that the first step in the soot formation process is the generation of the soot precursor, especially the generation of the first benzene ring (A1) trough C4H4 + C2H2 = C6H6 and 2C3H3 = C6H6,[35] and the speed of this step determines the subsequent soot generation. After precursor formation, the aromatic ring grows further following the steps precursor formation, particle inception, surface growth and agglomeration, and particle oxidation. High-ring PAHs condense to form soot. Zhang et al.[36] conducted simulation studies in the two-dimensional laminar CH4/air diffusion flame mixed with vaporized gasoline surrogates, and the addition of the PAH coagulation mechanism can make the results more accurate with the experimental data.

Liu et al.[15] in our research group investigated the sooting characteristics of gasoline/diesel blends on a laminar co-flow diffusion flame. They found that when the fuel with stronger low-temperature reaction is combusted, the cold flame zone increased and the amount of soot is reduced. The speculation is that the enhancement of cold flame combustion, due to its stronger negative temperature coefficient (NTC) behavior of the diesel fuel, is a cause of soot reduction. NTC behavior is a phenomenon that occurs during low-temperature to high-temperature reaction for certain fuels.[37] Its effects on auto-ignition have been extensively understood and researched.[38] However, its effects on soot emissions have rarely been researched.

By reason of the foregoing, this work focuses on revealing the chemical effects of the NTC behavior on the soot characteristics precisely. Since the diffusion flame test of the liquid fuel with a high-boiling point is highly sensitive to the supply fluctuation,[12] experimental studies are hard to carry out. This study therefore conducted numerical calculations of three different fuels with different NTC behaviors in the order of n-heptane > iso-octane > toluene. 2-D and a 0-D diffusion flame models were set up and the boundary conditions are set to eliminate the influence of physical factors as much as possible so that the study can focus on the influence of low-temperature chemical reaction and NTC behavior.

Numerical Methods

Chemical Mechanism

The chemical mechanism used in the simulations is a simplified toluene reference fuel (TRF, n-heptane/isooctane/toluene) mechanism developed by An et al.,[9] which includes 85 species and 232 reactions. This mechanism was validated against test results of ignition delays, premixed laminar flame speeds, profiles of key species, and PAH concentrations from the shock tube, adiabatic flat flame burner, and GDI engine data, respectively, and showed reliable and promising performances in the reproduction of auto-ignition and flame propagation characteristics and PAH predictions at both low and high pressures. Furthermore, its reduced size saves calculation time when coupled in 2-D and 3-D CFD simulations. In this work, the low-temperature region is defined as the flame temperature under 1000 K, and the high-temperature reaction starts from 1000 K. With this scenario, the mechanism contains low-temperature reactions as shown in Table and is able to simulate the NTC behavior accurately. If the initial boundary temperature is low, the fuel undergoes the low-temperature reactions before entering the high-temperature reaction stage.

Table 1

Low-Temperature Reactions in An’s Mechanism

R2	C7H15 + O2 = C7H15O2
R3	C7H15O2 = C7H14OOH
R4	C7H14OOH + O2 = O2C7H14OOH
R5	O2C7H14OOH ≥ C7KET + OH
R6	C7KET ≥ C5H11CO + CH2O + OH
R16	C8H17 + O2 = C8H17O2
R17	C8H17O2 = C8H16OOH
R18	C8H16OOH + O2 = O2C8H16OOH
R19	O2C8H16OOH ≥ C8KET + OH
R20	C8KET ≥ C6H13CO + CH2O + OH

2-D Simulation of a Laminar Diffusion Flame

The previous experiment work in our group indicated that the cool flame zone in the laminar co-flow diffusion flames varied as the fuel changed, and soot emission reduced as the cool flame zone enlarged.[39] In order to verify the relationship between cool flames, the low-temperature reactions, and the soot formation more intuitively with a theoretical basis, a 2-D Coflame numerical model was established to simulate laminar diffusion flames of n-heptane, iso-octane, and toluene. This model is capable of calculating the temperature field, the concentration field of each component, the volume fraction of soot, the nucleation rate, the number density of soot, and also the dynamic evolution of smoke particles. The simulation domain and boundary conditions were set according to Liu’s work.[39] As shown in Figure , the calculation domain represents a coaxial laminar diffusion flame. The inner diameter of the fuel outlet is 10.9 mm, the wall thickness of the pipe is 1 mm, and the air outlet is a concentric tube with an inner diameter of 58 mm. The flow rate of fuel vapor is 7 g/h and that of the carrier gas N2 is 0.3 L/min, so the fuel vapor takes up 11% in the fuel mixture. The oxidant air is composed of 79% nitrogen and 21% oxygen and flows through the outer tube with a rate of 200 L/min. The initial temperature is 523 K.

Figure 1

Boundary conditions and the grid in the 2-D simulation.

The diffusion and convection terms in governing equations are discretized using the center difference and upwind difference schemes, respectively. To speed up the convergence rate, the triangular matrix algorithm (TDMA) is adopted to solve the momentum, energy, and pressure correction equations, and the discrete equations of gas composition, soot mass fraction, and soot density are solved using the full coupling method. In addition, the parallel computation method is used to increase the calculation efficiency. The calculation domain is meshed to an 80(r) × 196(z) grid in the r and z directions, respectively, where z = 0 represents the outlet plane of the fuel and air and r = 0 represents the symmetrical plane of the diffusion flame. Non-uniform grids are adopted to lower the model error and reduce time consumption; the mesh inside the inner tube is 0.2725 mm in length, the interface between the fuel and air is meshed to 0.1 mm in length, and the outer domain filled with air is meshed to 0.2525 mm × 1.04, where n is the nth grid.

The free-slip boundary condition and zero gradient boundary condition are set at the axial and vertical directions. After a sensitivity analysis, the calculation domain is proved to be large enough and the boundary position does not affect the simulation results. To verify the convergence of the grid, more sets of denser grids up to 120(r) × 320(z) are used in the test calculations, and their corresponding results showed little differences from that of the 80(r) × 196(z) grids. Therefore, the 80(r) × 196(z) grids are believed to be good enough for the evaluations of flame characteristics. The convergence precision is set as 10–5.

0-D Simulation to Research the Mechanism

Even though the 2-D simulation has merits in visualization, it is deficient in conducting the further chemical analysis since it is still a physical–chemical coupled process. Therefore, a 0-D simulation is carried out to analyze the chemical factors that impact the formation of PAHs and soot. The constant-pressure homogeneous reactor in CHEMKIN 17.0 software was used to simulate the cool flame and low-temperature reaction process of the individual n-heptane and iso-octane flames. Toluene hardly reacts at low temperatures.

When analyzing the relationship between NTC behavior and PAH formation, the n-heptane and iso-octane flames are stoichiometric flames under atmospheric pressure. The initial temperature changes from 500 to 1000 K. When studying soot forming characteristics for rich flames of n-heptane and iso-octane, the pressure is set to be 30/60 atm and the equivalence ratio is set as 1.0/3.0. The initial temperature changes from 700 to 1300 K.

Results and Discussion

Phenomenon of NTC Effects on Soot Formation

The simulated 2-D temperature distribution around the root of the n-heptane and iso-octane flames is obtained through the Coflame software, as shown in Figure a,b. It is seen that there is a low-temperature zone ranging from 600 to 1000 K near the burner outlet, colored from blue to yellow surrounding the red high-temperature zone. It is believed that when the region temperature is below 600 K, the ignition temperature is not reached, so there is no flame in this region. When the temperature is above 700 K, the first stage of ignition occurs and the “cold flame” begins to develop. When the temperature goes up to around 1000 K, the second ignition take place and the hot-temperature reactions become intensified and produce a rather “hot flame”. Therefore, in this work, the low-temperature region is defined as the flame temperature under 1000 K, and when the flame temperature is above 1000 K, the hot temperature reaction starts. According to the temperature distribution, the flames around the outlet are hollow-cone-shaped with a “cool flame” surrounding the “hot flame” structure.

Figure 2

Temperature distributions near the root of in the n-heptane and iso-octane flames and the experiment flame structures of TRF individual fuels.[15] (a) n-Heptane, (b) iso-octane, and (c) experiment flame structure.[15] Image (c) is reprinted with permission from [Liu, F.; Hua, Y.; Wu, H.; Lee, C. Effect of Toluene Addition on the PAH Formation in Laminar Coflow Diffusion Flames of n-Heptane and Iso-octane. Energy Fuels.2018, 32(6), 7142–7152]. Copyright [2018] [Energy and Fuels.]. Photograph courtesy of “Liu, Fushui and Hua, Yang”. Copyright 2018. Permission is not needed since the image is sourced from another ACS journal Energy and Fuels.

The simulation is consistent with the previous experiment in our group,[15,39] as shown in Figure c. The experiment results show that there are two obvious flame regions in the flames. The bright “yellow” flames at the upper side of the flame are reflections of soot incandescence, which indicate a strong soot accumulation. However, the weak “blue” flames at the bottom indicate that the luminosity is dominated by the chemiluminescence, which implies that these zones are “soot-free”. On the other hand, the two distinct regions can be roughly seen as an indication of an obvious two-stage ignition of fuels. The start of the “blue” flame represents the first stage of ignition, where low-temperature oxidation occurs and the “cool flame” takes place, while the start of the “yellow” flame represents the second stage of ignition, where high-temperature oxidation and the “hot flame” dominate. The transition process from low-temperature oxidation to hot-temperature oxidation, where the chemical reaction rate slows down to accumulate OH+ radicals and heat, forms the NTC phenomenon. Therefore, the length of the “blue” flame can be considered as an indicator of the intensity of the NTC tendency of a fuel. That is to say, the stronger the NTC tendency is, the longer gap the two-stage combustion has and the lengthier the “blue” flame becomes, which obtains a boarder “soot-free” zone and contributes to the imaginable soot reduction.

Key Intermediate Distributions

Keto-peroxides C7KET and C8KET are the key intermediates that are produced in the n-heptane and iso-octane flames under low-temperature reactions for keto-peroxides are generally temperature-sensitive intermediate products that only exist at the low-temperature region. Figure shows the two keto-hydroperoxide distributions in their corresponding flame. The C7KET radicals in the n-heptane flame are generated through reaction R5 according to low-temperature reaction shown in Table , and it is relatively stable at low temperatures but soon dissociates into C5H11CO, CH2O, and OH radicals through reaction R6 as the temperatures rise above 800 K. Therefore, C7KET can indirectly reflect low-temperature reaction intensity.

Figure 3

Keto-peroxide C7KET and C8KET distributions in the n-heptane and iso-octane flames, (a) n-heptane flame and (b) iso-octane flame.

At HAB (height above burner) = 0, the fuel vapor just emerges from the fuel outlet to meet the oxidant air from the outer tube. The temperature at this time is quite low, and the low-temperature reactions take place at once. The reactions begin to generate the most C7KET around the outlet interface and make the temperature rise quickly. As the flame height increases, the flame temperature increases too and develops into the high-temperature region rapidly, and the C7KET radicals vanish dramatically as the flame height increases. At the same time, the oxidant air from the outer tube propagates to the flame center through diffusion, and the low-temperature reactions occur accordingly at the regions far from the outlet interface. It is reasonable that as the flame height increases, the fuel and air inter-diffuse more, so the low-temperature reaction band becomes more spread. Therefore, the C7KET distribution presents a tick-mark-shaped band.

The distribution of the keto-peroxides C8KET in the iso-octane flame is similar to that in the n-heptane flame. Comparing the two distributions, C7KET radicals extend further due to the longer transitioning period from low-temperature reaction to high-temperature reaction, that is, the NTC effects. The higher and wider C7KET region also corresponds to the phenomenon that the “blue” flame region of n-heptane is larger than that of iso-octane. Since toluene has little low-temperature oxidation, the corresponding products in the toluene fuel barely exist. This explains why the “blue” flame in the toluene flame has the minimum size.

The amount of the keto-peroxides of the n-heptane and iso-octane flames on the central axis is provided in Figure as a more quantitative piece of evidence. At an HAB of 2 mm, the flame of n-heptane started to generate C7KET, which implies that the air supplied from the outer tube has arrived the central axis. The amount of C7KET radicals increases rapidly along the central axis from then on and peaks at an HAB of 5 mm. As the low-temperature reaction keeps proceeding, the temperature increases gradually and low-temperature reactions are suppressed after that. Therefore, the mole fraction of C7KET radicals finally decreases to 0 at an HAB of 7 mm. The keto-peroxide C8KET distribution shows a similar trend to that of the C7KET, which offsets a bit higher due to the longer first-stage ignition delay and the higher auto-ignition temperature. This corresponds to the higher liftoff height of the isooctane flame in the experiment shown in Figure . The mole fraction of C8KET peaks at an HAB of 6 mm with a lower quantity, indicating that the low-temperature reaction intensity of the iso-octane flame is weaker along the central axis due to the faster transition from low-temperature reaction to high-temperature reaction, which is a weaker NTC effect.

Figure 4

Distributions of keto-peroxides of the n-heptane and iso-octane flames on the central axis.

When the temperature increases beyond 800 K, the keto-hydroperoxides C7H14OOH produce aldehydes rather than C7KET. Aldehydes form immediately at the beginning of the first stage of ignition and disappear dramatically when the second stage of ignition starts. Formaldehyde, CH2O, is the most abundant of all aldehyde products, so it is often seen as an important index to reflect the evolution process from low-temperature reaction to high-temperature reaction. The electronically excited formaldehyde illuminates a pale “blue” chemiluminescence, which makes the CH2O radicals a more direct indicator of the “blue” flame and NTC behavior.

The 2-D distributions of CH2O radicals in the respective n-heptane, iso-octane, and toluene flames are shown in Figure . The distribution of CH2O radical in the n-heptane flame is wider than that in the iso-octane flame, whereas only a very slight CH2O radical distribution is seen in the toluene flame due to the barely happened low-temperature reactions. Both n-heptane and iso-octane flames generate a large amount of CH2O in the area around r = 6 mm and HAB = 0–2 mm, which correspond to the distribution of low-temperature products C7KET and C8KET. As the height increases, CH2O diffuses to the center of the flame, and the concentration gradually decreases. The range of the distribution area of CH2O is consistent with the range of the cool flame, n-heptane > iso-octane > toluene.

Figure 5

CH2O distributions of the n-heptane, iso-octane, and toluene flames, (a) n-heptane, (b) iso-octane, and (c) toluene.

There are two peaks of CH2O on the central axis of the n-heptane and iso-octane flames, as shown in Figure . Their lower peaks appear at the same height around an HAB of 8 mm, which are produced by low-temperature reactions. The n-heptane flame has more CH2O radicals, indicating a longer low-to-high temperature reaction in the n-heptane flame that causes the CH2O accumulation. It is seen that the CH2O radicals decrease dramatically corresponding to the start of second ignition. The second peaks appear at round HABs of 25 and 27 mm, respectively. The CH2O radicals in this zone are produced through small-molecule reaction CH3OCH2 = CH2O + CH3 and CH3O (+M) = CH2O + H (+M), which only happens under high-temperature conditions. The toluene flame has only one CH2O peak and a lower mole fraction due to the unobtrusive low-temperature reactions.

Figure 6

Distributions of CH2O on the central axis of the n-heptane, iso-octane, and toluene flames.

PAH Distributions

The two-dimensional distributions of soot volume fraction for three fuels are shown in Figure . The soot distribution area and the overall amount of soot in the toluene flame are the highest. A large amount of soot is generated on the wing of the flame from about an HAB of 8 mm and almost covers the whole flame area. This simulation, to a certain extent, underestimates the soot along the centerline, and the problem exists in other numerical simulations as well.[22,28,40] This deficiency in the soot formation prediction is quite likely caused by the oversimplified assumption that soot nucleation is initiated by the collision of two pyrene (A4) molecules, whereas the soot nucleation process involves multiple PAHs of different sizes in reality. The flame of n-heptane, which has the strongest NTC behavior, generates the least soot particles and the smallest soot distribution area. The n-heptane flame starts to generate soot particles at around an HAB of 25 mm, and the amount is 2 orders of magnitude less than that of toluene. The amount of soot particles and the distribution area of iso-octane are in between those of the n-heptane and toluene flames since the low-temperature reaction intensity is also in between the other two flames.

Figure 7

Soot distribution of the n-heptane, iso-octane, and toluene flames, (a) n-heptane, (b) iso-octane, and (c)toluene.

The distributions of four aromatic hydrocarbons A1–A4 of n-heptane, iso-octane, and toluene flames on the central axis are presented in Figure . A1 is considered as the initial precursor of soot formation, and the amount of A4 can indirectly characterize the soot amount. Both n-heptane and iso-octane belong to alkane, but the PAH distribution of the two flames is quite different. The quantity of A1–A4 in the iso-octane flame is generally higher than that in the n-heptane flame, which is consistent with the soot emission results. Comparing the height of the first peak of A1 and the peak of A2–A4, the higher the number of rings, the higher the position of PAHs, which suggest that A1 evolve to high-ring soot precursors. For the four PAH distributions of the n-heptane and iso-octane flames, there are several peaks in the A1 distribution, while A2–A4 are unimodal.

Figure 8

Distributions of A1–A4 on the central axis of the n-heptane, iso-octane, and toluene flames.

The PAH growth mechanism used in this work follows these rules. All the ring growth reactions are reversible reactions. The collision of two A4 radicals will generate soot particles, and this reaction is irreversible. According to these two rules, when the radical pool concentration of A4 increases to a certain extent, the probability of pairwise molecule collision increases, so more soot is produced. Then, more low-ring PAHs convert to high-ring PAHs to supplement the A4 radical pool. After such a steady stream of transportation, eventually, all converted into soot. However, at some certain stage, where the radical pools of A2 and A3 are abundant, the reactions may change the direction and break down to form A1 to maintain the equilibrium. Therefore, the upper peaks of A1 may be due to the reverse reaction of PAH growth.

The distribution of A1 in the iso-octane flame in the range of HABs of 5–20 mm is quite different from that of n-heptane. The amount of A1 in the n-heptane flame gradually increases with the increase of the height, reaching its peak at an HAB of 27 mm, and then decreases rapidly, whereas the iso-octane flame reaches its first peak at an HAB of 15 mm with a higher amount. As the flame height can be equivalent to the flame development time, the higher the flame height, the longer time the flame develops. By a comprehensive comparison among the flame height ranges of the keto-peroxide, CH2O, and A1, a conversion path from keto-peroxide → CH2O → A1 can be concluded as each of these radicals peaks at HABs of 6, 8, and 15 mm for the first time. In addition, the A1 formation is seen to be closely related to CH2O radicals since their distributions on the central axis are quite similar. Toluene is an aromatic hydrocarbon fuel, which hardly reacts under low-temperature conditions. However, it contains a ring structure, which makes it is easy to generate a much larger amount of PAHs, compared with the other alkanes as seen in the figure, under high-temperature conditions. To make it easier to see, the value of A1 in the toluene flame is divided by 50, and the values of A2 and A3 are divided by 20.

Since the mechanism adopted in this work is a reduced mechanism, reactions related to A1 formation shown in Table are the filtrated ones which are of significant importance by sensitivity analysis and rate of production (ROP) analysis. According to these reactions, the formation of A1 is closely related to small molecular hydrocarbons C2–C4, which are C3H3, C4H5, C2H2, and C3H4. Toluene has a ring structure and generates a large amount of monocyclic aromatic hydrocarbon A1 through the path C6H5CH3 (toluene) + H = A1 (benzene) + CH3. Therefore, the formation of A1 in the toluene flame is less affected by small molecular hydrocarbons above. The respective distributions of these small molecular hydrocarbons on the central axis in the three flames are shown in Figure . The distributions of C3H3 in the three flames are compared in Figure a. For the alkanes, the C3H3 radicals show a two-hump trend, which start from an HAB of 10 mm and rapidly increase around an HAB of 20. This infers that the C3H3 radicals correspond to the A1 accumulations in the alkane fuels at around an HAB of 25 mm through the reaction R198, C3H3 + C3H3 = A1 but have little relationship with the A1 accumulation of the iso-octane flame at an HAB of 15 mm. In comparison, the amount of C3H3 in the toluene flame is large, but it is mainly distributed in the zone of HABs of 20–40 mm, even higher than the aggregation position of A1–A4. Therefore, it is inferred that the main source of A1 in toluene flames is not C3H3 but directly generated from toluene. The large amount of C3H3 comes from the reverse reaction of R198.

Table 2

Reactions about the Generation of A1 in the Adoped Mechanism

R33	C6H5CH3 + H = A1 + CH3
R198	2C3H3 = A1
R199	C3H4 + C3H3 = A1 + H
R200	C4H5 + C2H2 = A1 + H

Figure 9

Distributions of C2–C4 radicals on the central axis of the n-heptane and iso-octane flames.

As is shown in Figure b, the mole fractions of C4H5 in the alkane flames occur mainly in the lower zone of flames with a pretty lower order of magnitude of 10–8 compared to the other three radicals, while that of the toluene flame is even less. Although C4H5 produces A1 through R200, C4H5 + C2H2 = A1 + H, it is not the most important radical in the formation of soot. Figure c shows the comparison of C2H2 distribution. C2H2 is not only the source of A1 formation but also an important reactant for the conversion of A1 to higher-ring soot precursors in HACA theory. Since the amount of C4H5 radicals is low, the portion of C2H2 involved in the A1 formation is pretty small according to reaction R200. Most of the C2H2 radicals take part in the conversion from A1 to higher-ring soot precursors. The C2H2 in the iso-octane flame is generally higher than that in the n-heptane flame, which corresponds to the higher soot content of the iso-octane flame. For the toluene flame, the C2H2 amount is comparatively low for it is harder to produce C2H2 dissociation reactions; however, the large amount of aromatic hydrocarbons still promotes the soot formation rate.

The C3H4 distributions are shown in Figure d. It is seen that C3H4 radicals are about the same magnitude with C3H3 but emerge earlier, so the C3H4 radicals are a major source of C3H3. Compared to the even-carbon small molecules, the odd-carbon small molecules dominate the production of PAHs. It explains the reason that distribution of the iso-octane flame has two peaks on the central axis, while n-heptane has only one peak, which is similar to the A1 distribution. Comparing the distribution of these small radicals, it can be found that in the alkane flame, the low-temperature reactions reduce the amount of C3H4 produced in the low-temperature region by generating a large amount of CH2O, which makes A1 to lack reactants in the early reaction, leading to a reduction in soot emissions. In the toluene flame, the small molecular hydrocarbons C2–C4 have little influence to the formation of A1. However, the lack of C2H2 limits the process of aromatic hydrocarbon growth.

Chemical Analysis of the Low-Temperature Effect

To verify the universality of the above conclusions and understand the phenomenon in the perspective of chemical kinetics, a constant-pressure homogeneous reactor in CHEMKIN software is used to analyze the above three fuel flames. The ignition delay analysis is conducted for the three fuels, and the results are shown in Figure . The ignition delay of toluene is much longer than that of n-heptane and iso-octane. In addition, both n-heptane and iso-octane have NTC effects and low-temperature reactions. In the n-heptane flame, there is a significant NTC trend between 1000/T = 1.28 and 1.56, while the NTC trend in the iso-octane flame is between 1000/T = 1.34 and 1.6. Since toluene hardly reacts with air oxidants under low-temperature conditions and little NTC behavior is found, due to its distinct physical and chemical properties and reaction path from those of alkanes, the following text will only focus on the low-temperature reaction in the n-heptane and iso-octane flames.

Figure 10

Ignition delays of n-heptane, iso-octane, and toluene flames.

In order to analyze the impact of the chemical mechanism under the low-temperature combustion stage on the soot formation, a group of simulations with the initial reaction temperature at 700 K are carried out. Figure shows the temperature changes of n-heptane and iso-octane during the reaction. It is seen that both fuels have a two-stage combustion phenomenon, where the first stage of combustion is caused by low-temperature reactions at the beginning. Then, the two fuels enter their second-stage combustion at 0.6 and 3.2 s, respectively.

Figure 11

Temperature evolution processes in the flames of n-heptane and iso-octane.

Figure shows the main reactions associated to the formation of aromatic hydrocarbons A1 at the first-stage combustion in the n-heptane flame and the iso-octane flame. For the n-heptane flame, C7H16 first undergoes dehydrogenation reaction to produce C7H15, and then, C7H15 mainly reacts further through three main reaction paths. These three reactions all react under 1000 K, and all contribute to the first-stage combustion. The first path is the oxygen addition reaction C7H15 + O2 = C7H15O2. This reaction accumulates heat and heats up the system, and the subsequent products are CH2O, CO, and small molecular hydrocarbons. The second reaction path is a chain branching reaction C7H15 + O2 = C7H14 + HO2. C7H14 reacts with O2 to generate CH2O, HCO, and small molecular hydrocarbons; this chain branching reaction slows down the reaction rate and elongates the second-stage ignition, and the NTC behavior appears. The third path is that C7H15 is directly decomposed into C3H6, C2H5, and C2H4, which implies that the system is under a relatively higher temperature than the previous two.

Figure 12

Main reaction pathways to aromatic hydrocarbons A1 of n-heptane and iso-octane flames in the first combustion stage. (a) Reaction pathways of n-heptane under low-temperature conditions. (b) Reaction pathways of n-heptane under low-temperature conditions.

Although the reaction path distribution of iso-octane is similar to that of n-heptane, the proportions of the three paths affect the subsequent products. According to the proportion of each reaction path, 26.9% of C7H16 transforms to A1 through the first reaction path and 22.9% of C7H16 transforms to A1 through the second path. In total, the percentage for C7H16 which transforms to A1 through the first two paths, which produce relatively large molecules, is 49.8%, while only 37% C8H18 transforms to A1 through the two large-molecule reaction pathways. It can be reasonably inferred that more fuel burned through large-molecule reaction pathways, whose conversion rate is lower, will definitely result in a reduced soot formation. When the fuel burns near the NTC temperature, the flame can be suppressed by NTC to transit to high-temperature flame combustion and stay in the low-temperature zone.

The ROP of the three reaction paths of C7H15 in the n-heptane flame and that of C8H17 in the iso-octane flame are compared in Figure to evaluate the intensities of the low-temperature reactions, which affect the ignition delay of both the first- and second-stage combustion. The intensity of the low-temperature reaction C7H15 + O2 = C7H15O2 in the n-heptane flame is higher than that of the low-temperature reaction C8H17 + O2 = C8H17O2 in the isooctane flame. These two oxygen addition reactions generate large molecular radicals and slow down the heat release speed, thus intensifying the NTC effect. Small molecular hydrocarbons such as C3H3 and C2H2 are also the main sources of aromatic hydrocarbons, while the distribution trend of C3H4 is the same as that in the two-dimensional simulation.

Figure 13

Production rates of key branch reactions of C7H15 and C8H17 during the first-stage combustion of n-heptane flame and iso-octane flames, (a) n-heptane and (b) iso-octane.

Figure shows the mole fractions of key small radicals and A1 during the first-stage combustion of n-heptane and iso-octane flames. It is seen that the n-heptane flame with stronger low-temperature reaction and weak decomposition reaction will not significantly increase the amount of small molecular hydrocarbons as shown in Figure a, compared to the iso-octane flame with weaker low-temperature reaction and strong decomposition reaction, which generates a large amount of C3H4 in the first-stage combustion, as shown in Figure b. The mole fraction of A1 in the iso-octane flame is greater than that in the n-heptane flame. Therefore, the reason that the low-temperature reaction suppresses soot formation is that it inhibits the direct decomposition of large alkyl groups to small molecular hydrocarbons, which promote the PAH formation and the subsequent soot generation process. Figure presents the reaction pathway.

Figure 14

Mole fractions of key small radicals and A1 during the first-stage combustion of n-heptane and iso-octane flames, (a) n-heptane and (b) iso-octane.

Figure 15

Main reaction pathways of A1–A4 in the alkanes.

Comparison of Factors Influencing the Soot Forming Characteristics

The above argument proves that low-temperature reaction can reduce soot emissions in the perspective of chemical kinetics, which also confirms that under the laboratory conditions, diesel with stronger NTC behavior releases less soot than gasoline. However, since traditional gasoline engines mostly work in the oil–gas premix combustion mode due to its high volatility, the combustion mixture is relatively uniform, which makes the traditional gasoline engine a cleaner engine than diesel in terms of soot emission. In comparison, the spray combustion mode forms a highly heterogeneous concentration field in the practical diesel engines. Besides, the soot formation in an actual engine is also affected by the cylinder pressure. Therefore, a numerical comparison is conducted in this part to shed light on the weight of low-temperature effects on the soot forming characteristics when compared with other factors.

Therefore, four sets of simulations with different working conditions are conducted. They are stoichiometric n-heptane flames with different initial pressures, C7H16–30 atm−Φ1.0 and C7H16–60 atm−Φ1.0, rich n-heptane flame C7H16–60 atm−Φ3.0, and stoichiometric iso-octane flame C8H18–30 atm−Φ1.0, where Φ represents the equivalence ratio. Their relationships of the mole fraction of A1 versus the initial temperature are shown in Figure . The mole fraction of A1 of C7H16–60 atm−Φ1.0 is slightly higher than that of the C7H16–30 atm−Φ1.0 case when the initial temperature is blow 1050 K, which is generally under the low-temperature conditions, while the mole fraction of A1 of C7H16–60 atm−Φ1.0 drops slightly down below that of C7H16–30 atm−Φ1.0 in the high-temperature region. Comparing the results between C7H16–30 atm−Φ1.0 and C8H18–30 atm−Φ1.0 with the same reaction pressure and equivalence ratio, n-heptane with intensified low-temperature reaction and stronger NTC behavior produces less amount of A1. At last, the results of the n-heptane flame under C7H16–60 atm−Φ3.0 and C7H16–60 atm−Φ1.0 working conditions with the same reaction pressure but different equivalence ratios show that the rich mixture C7H16–60 atm−Φ3.0 generates a dramatically large amount of A1 particles, which is an order of magnitude higher than that produced in the stoichiometric flame. The aforementioned comparisons indicate that the equivalence ratio among the three factors has the greatest influence on soot emissions, whereas the low-temperature effect is more important than pressure.

Figure 16

Mole fractions of A1 under different test conditions.

Furthermore, as shown in Figure , the ROP of C7H15 at stoichiometric and rich flames varies greatly. The low-temperature reaction C7H15 + O2 = C7H15O2 and C7H15 + O2 = C7H14 + HO2 at the stoichiometric flame takes up a much larger portion than that in the rich flame Φ3.0. The result implies that the suppress effects of low-temperature reaction and NTC on the soot formation are more prominent under lean and stoichiometric conditions. The reason is that due to the lack of oxygen, the combustion in the fuel-rich zone trends to shift toward a decomposition reaction, which requires less oxygen participation. The proportion of high-temperature reaction C7H15 ≥ C3H6 + C2H5 + C2H4 increases. The amount of small molecular hydrocarbons soars up and promotes soot generation. Therefore, even though diesel has a stronger low-temperature reaction that may improve the soot emission, this effect is not as prominent as that of equivalence ratio and this merit can be degraded largely under fuel-rich conditions.

Figure 17

Production rates of three key branch reactions of C7H15 in the n-heptane flames at equivalence ratio = 1.0 and 3.0, (a) Φ1.0 and (b) Φ3.0.

Conclusions

By numerically simulating the 2-D distributions of the temperature field, soot precursors A1–A4, and the low-temperature reaction in the n-heptane, iso-octane, and toluene flames, the effect of NTC on the formation of soot precursor PAHs, especially the monocyclic-ring benzene (A1), is investigated. Through the comparison with another to physical factors, pressure and equivalence ratio, the weight of NTC and the low-temperature effect are evaluated. The main conclusions are as follows:

The laminar diffusion flame of n-heptane, which has a stronger NTC behavior than the iso-octane flame and toluene flame, generates the least soot particles, and the soot distribution area is the smallest among the three.

The keto-peroxide and CH2O can be considered as the indicators of NTC behavior. The mole fractions of these radicals are less in the iso-octane flame in the “soot-free zone” and are pieces of evidence of weaker NTC effects of iso-octane.

The low-temperature reactions suppress the formation of PAHs by taking the oxygen addition reaction C7H15 + O2 = C7H15O2 and C7H15 + O2 = C7H14 + HO2 to produce large molecular hydrocarbons instead of the disassociation reactions that produce small molecular hydrocarbons, such as C2H2, C3H3, C3H4, and so forth.

In the first combustion stage, 49.8% of C7H16 reacts through the two large molecular reaction pathways, which suppresses the formation of A1. However, only 37% of C8H18 reacts through large molecular reaction pathways. The more the portion of the fuel burnt through the two large molecular reaction pathways, the less the soot formed in total.

Toluene generates a large amount of monocyclic aromatic hydrocarbon A1 through the path C6H5CH3 + H = A1 + CH3. Therefore, the formation of A1 in the toluene flame is less affected by small molecular hydrocarbons.

The influence of equivalence ratio on soot emissions is much higher than the pressure and NTC effects, and the suppress effects of low-temperature reaction and NTC on the soot formation are more prominent under lean and stoichiometric conditions.