Soot formation in laminar counterflow flames

Abstract

Many practical soot-emitting combustion systems such as diesel and jet engines rely on diffusion flames for efficient and reliable operation. Efforts to mitigate soot emissions from these systems are dependent on fundamental understanding of the physicochemical pathways leading from fuel to soot in laminar diffusion flames. Existing diffusion flame−based soot studies focused primarily on over-ventilated coflow flame where the fuel gas (or vapor) issues from a cylindrical tube into a co-flowing oxidizer, and counterflow flame, where a reacting zone is established between two opposing streams of fuel and oxidizer. As a canonical diffusion flame configuration, laminar counterflow diffusion flames have been widely used as a highly controllable environment for soot research, enabling significant progress in the understanding of soot formation for several decades. In view of the possibility of fuel/oxidizer premixing in practical systems, counterflow partially premixed flames have also been studied. In the present work we intend to provide a comprehensive review of the researches on various aspects of soot formation utilizing counterflow flames. Major processes of soot formation (formation of gas phase soot precursors, soot inception and surface reactions, as well as particle-particle interactions) are examined first, with focus on the most recent (post-2010) research progress. Experimental techniques and associated challenges for the measurement of soot-related properties (some knowledge of which is helpful for full appreciation of the experimental data to be reviewed) are then introduced. This is followed by a detailed description of soot evolution in counterflow flames, which is complemented by a discussion on the similarity and differences of the sooting structures between counterflow and coflow diffusion flames. Parametric studies of the effects of fuel molecular structure, fuel additive, partial-premixing, pressure, temperature, stoichiometric mixture fraction, and residence time on soot formation in counterflow flames will then be addressed in detail. This review concludes with a summary of the knowledge and challenges gathered and demonstrated through decades of research, and an outlook on opportunities for future counterflow flame−based soot research towards a more complete understanding of soot formation and the development of novel techniques for soot mitigation in practical combustion devices.

Introduction

Rapid population growth and progressive industrialization of developing countries assure rising worldwide energy demands in the coming decades: A 30% increase in energy consumption is projected from 2020 to 2040, even though continuous improvement in efficiency of energy use is anticipated [1]. Although alternative sources such as wind, solar, geothermal, and nuclear energies are increasingly important in the world energy mix, conventional petroleum-based fuels are still expected to play a dominant role in the foreseeable future. This is especially true in the transportation sector, where the society relies heavily on internal combustion (IC) engines fueled by natural gas, gasoline, kerosene, and diesel. Note that IC engines use 70% of the 86 million barrels of crude oil the world consumes every day [2]. Energy from hydrocarbons comes with a price. Apart from contributing to the greenhouse effects and thus to climate change, the use of hydrocarbon fuel also leads to the formation of various hazardous pollutants, including−but not limited to−nitric oxide (NOx), carbon monoxide (CO), unburned hydrocarbons (UHCs) and soot. Here soot is defined as carbonaceous particles resulting from pyrolysis or incomplete combustion of hydrocarbon fuels. In terms of elemental composition, soot is largely carbon and may also contain small amounts of hydrogen and oxygen. Being very fine particles with diameters typically less than a micrometer, soot is known to be an important contributor to PM_2.5 (particulate matter with diameters smaller than 2.5 µm).

As far as reciprocating IC engines are concerned, NOx, CO, and UHC can be effectively removed using a three-way catalyst in spark-ignition (SI) engines operating with stoichiometric fuel-air mixtures. Soot emission is generally not an issue in these conventional port-injection SI engines. But with the advent of gasoline direct injection (GDI) technology (intended mainly to improve fuel efficiency), soot particles can be produced at locally rich regions as a result of charge inhomogeneity. Although the total soot mass emitted from GDI engines may not be significant, its number density can still be tremendous, since GDI engines are likely to produce very fine soot particles [3], [4]. Compression ignition (CI) diesel engines are conventional producers of NOx and soot, and worse, a trade-off relationship exists between NOx and soot formation in the diesel combustion process. Many measures designed to reduce NOx emission result in inhibiting the oxidative removal of soot, signifying the importance of soot suppression during its formation stage. As environmental regulations on soot emission become increasingly stringent, the incentive for combustion scientists and engineers to develop innovative and effective methods for soot mitigation grows with it.

Concerns about soot emission are many and justified. Being the major condensed-phase by-product, soot originates from the incomplete combustion of hydrocarbon fuels. In most cases, the presence of soot indicates poor combustion conditions, where not all the fuel molecules can be fully oxidized to form CO₂ and H₂O for maximum thermal energy recovery. Soot particles contain chemical energy in the form of C-H and C-C bonds, which could be released as desirable thermal energy if soot formation could be avoided altogether, or if all the initially formed soot particles were fully oxidized in a later stage. Therefore, the emission of soot from practical combustion devices can be translated directly to combustion inefficiency.

More importantly, soot emission into the atmosphere has dire consequences to human health and the environment. After being inhaled, ultrafine soot particles with diameters less than 100 nm (typical for flame generated soot) can travel deeply into the lungs, deposit in the alveoli, and cross the cells to enter the circulatory system [5]. These ultrafine particles can then be translocated to other organs like the liver, heart, and brain [6], causing potential health issues such as cardiovascular diseases. Polycyclic aromatic hydrocarbons (PAHs), the main organic component of soot, have also been associated with carcinogenic and mutagenic effects [7]. Schwartz and coworkers [8], [9] demonstrated positive correlations between PM concentrations and mortality rates that reflect the serious health issues caused by soot emission.

Regarding environmental implications, soot is a major contributor to global warming, second only to CO₂ [10]. Such efficacy is believed to be caused by its influence on snow albedo as well as its atmospheric stability at high altitudes [11]. Similarly, soot deposited on ice glaciers can promote melting, raising the sea level with all the accompanying problems. Soot particles, with graphitic carbon as the primary constituent, are the biggest contributor to light absorption by atmospheric aerosols [12]. The reduction of visibility in densely populated urban areas due to the presence of soot could cause widespread societal concerns. Recent severe haze events in China, affecting nearly 800 million people, have been closely linked to fossil fuel use. It has been suggested that both primary particulate and secondary aerosol precursors from fossil fuel and biomass combustion can be controlled to minimize the negative impacts [13]. Because of its contribution as cloud condensation nuclei, soot in the upper atmosphere is also known to affect weather in the form of clouds and precipitation [14], [15], [16].

In most cases, soot is the unwanted by-product of hydrocarbon combustion and its formation needs to be suppressed as much as possible. However, there are situations in which soot production is intentional. In utility and industrial boilers, for example, released combustion heat needs to be effectively transferred to the feedwater. One important heat transfer mode in such large-scale systems is radiation, which can be significantly enhanced by the presence of soot particles [17]. Of course, these particles need to be eliminated before the flue gases are emitted to the environment. In addition, carbon black is an important raw material for the reinforcing filler in tires and the industry strives for maximum soot yield for a given quantity of hydrocarbon fuel.

The need to suppress soot emission, or to produce soot with special properties, necessitates active control of the soot formation processes. This in turn requires a fundamental understanding of the physicochemical pathways from fuel to soot. Soot formation is one of the most complex phenomena in combustion, involving complicated interactions between combustion chemistry, fluid mechanics, mass/heat transport, and particle dynamics, so that despite decades of active research, many gaps still remain in our understanding of soot [18]. In the attempt to bridge these gaps, the underlying mechanisms of soot formation are the objectives of combustion scientists worldwide, evident from the large body of literature about soot published in recent years. For instance, soot, as a keyword, is present in the title or abstract of 45 articles published in Combustion and Flame in the year of 2015, accounting for more than ten percent of the 383 total articles. This number is even higher for 2016, with 51 out of the 374 articles investigating soot formation.

A survey of the literature reveals that a significant portion of soot studies conducted to date are based on laboratory-scale laminar flames at atmospheric, or even lower pressures. This may seem inappropriate, at first glance, as combustion modes in most soot-emitting practical combustion devices are turbulent, and under high pressure. Such a mismatch is certainly not because soot formation in high-pressure turbulent flames can be well represented by those in laminar flames. One only needs to consider turbulence-chemistry interactions and the pressure-dependence of reaction rates to realize the drastic differences. However, the following facts limit our ability to investigate in detail the soot characteristics of high-pressure turbulent flames. First, the highly transient and inhomogeneous nature of turbulent flames presents a challenge for many soot-related experimental techniques in terms of providing sufficient temporal and spatial resolutions. Second, a high peak value and wide dynamic range of soot loadings prevent many high-pressure turbulent flames from being reliably diagnosed. Related issues include, for example, probe clogging for intrusive methods and signal attenuation/beam steering for optical methods [19]. Third, it is somewhat difficult to establish well-controlled high-pressure flames that are suitable for fundamental studies of soot formation and oxidation processes; several researchers have observed flame stability issues at high pressures [20], [21]. With an increase in pressure, the gas-solid conjugate heat transfer between flame gas and burner nozzle becomes increasingly important and complicates the specification of flow boundary conditions [22].

In addition to the above technical constraints, the fact that physicochemical formation pathways of soot are still not fully understood−even in zero-dimensional systems−may provide enough rationale for the wide interests in soot studies with simple flow configurations. In this regard, various laboratory-scale setups have been employed, including constant volume combustion chambers [23], [24], [25], [26], [27], [28], [29], [30], shock tubes [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], well-stirred reactors [41], [42], [43], [44], [45], burner-stabilized flat premixed flames [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], coflow diffusion flames [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], and counterflow diffusion flames (CDF) [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88], [89], [90], [91], [92]. The conditions for soot formation can be very different among these experimental configurations and conclusions regarding the sooting processes obtained from one configuration may not be directly applicable to others. For example, the trend that soot loadings increase with the increase in flame temperature [93], [94], [95], [96], as observed in diffusion flames, can be reversed in some premixed flames [97], [98], [99]. This can be explained (at least partly) by noting that unlike in a diffusion flame, a very rich fuel pyrolysis region, where soot precursors are formed without oxidation, usually does not exist in a premixed flame. In addition, certain fuel additives may play different roles, sometimes with opposite effects on soot formation in different reacting systems. The doping of small amounts of ethanol, or dimethyl ether, was seen to boost soot volume fractions in ethylene diffusion flames [100], [101]; but it inhibited soot formation in ethylene/air premixed flames [102], [103].

Constant volume/pressure combustion chambers are typically used to study liquid fuel spray combustion [23], [24], [25], [26], [27], [28], [29], [30], a close representation of the combustion process in CI engines. Unfortunately, soot formation during the burning of fuel sprays depends not only on soot chemistry but on many other physical processes such as spray penetration, droplet size distribution, and velocity field of the entrained air. It is challenging to isolate these factors to obtain quantitative information on soot behaviors. Note that a shock tube can provide homogeneous, constant pressure/temperature conditions, ideal for studying the chemistry of fuel pyrolysis and subsequent soot formation. However, the time scale of the sooting processes in shock tube experiments is typically 1∼2 orders of magnitude shorter than that in flames [31]. For this reason, shock tubes are primarily used for investigation of early stage soot formation; while slower processes such as particle coagulation and aggregation cannot be effectively examined [31]. A jet-stirred reactor (JSR) is an alternative way to reach constant pressure and temperature conditions, and the residence time can be varied by changing the reactants’ flow rates. However, external heat sources are needed and typical temperature achievable in JSR experiments is only at the lower end of flame environments. Moreover, the reactants in shock tubes or JSR are highly diluted by inert gas (e.g., argon) in most cases, differing noticeably with practical flame conditions. Note that the effect of reactant dilution on soot formation can be significant [95], [104].

The details of soot formation in turbulent diffusion flames entail complex study, and most combustion systems rely on turbulent diffusion flames for operation. Fortunately, the concept of laminar flamelets could provide a link between turbulent and laminar diffusion flames [105], [106]. Fundamental investigations of the pathways from fuel molecules to soot particles in laminar diffusion flames can provide essential information toward clarifying the sooting process in practical flames.

Soot formation characteristics in laminar diffusion flames have been extensively studied in over-ventilated coflow diffusion flames, where the fuel gas (or vapor) issues from a cylindrical tube into co-flowing oxidizer [107], and counterflow flames, where a reaction zone is established between two opposing streams of fuel and oxidizer [108], [109]. The general features of the flow configurations differ between these two types of flames. A coflow flame has a two-dimensional flow field so that the species concentrations, temperatures, and flow velocities vary in both axial and radial directions. A counterflow flame, on the other hand, is quasi-one-dimensional along the normal direction of the flame, a feature that significantly facilitates the analysis of flame/sooting structures. This becomes particularly relevant if models with detailed chemical kinetics and particle dynamics are to be employed for the simulation of these flames, as the decreased dimensionality of counterflow flames significantly reduces the computational cost. To give an example, a recent computational work [110] on an ethylene/air coflow flame using a semi-detailed chemical reaction mechanism [111], [112] and a sectional soot model was reported to take a wall time of more than 150 hours on a 400 CPU Linux cluster (i.e., 60,000 CPU hours) for convergence; similar work for one-dimensional counterflow flames typically takes no more than several hours on a PC. Also, it is generally believed that a reacting zone of premixed nature is responsible for the stabilization and attachment of the flame to the nozzle in a coflow diffusion flame. This premixed zone exists due to the inter-diffusion of fuel and oxidizer in the vicinity of the burner nozzle, where chemical reactions are inhibited by heat and radical losses to the wall [109], [113]. Therefore, coflow diffusion flames are also affected by the characteristics of this stabilizing premixed edge flame. For this reason, some fundamental aspects of diffusion combustion are thought not to be well represented by coflow diffusion flames, while counterflow diffusion flames are recognized as pure diffusion flames [109]. Finally, the residence time in coflow flames is relatively long and sometimes difficult to parameterize. In counterflow flames, residence time and stretch rate can be adjusted by varying fuel and/or oxidizer flow velocities, providing a unique way to study soot chemistry with variable residence time. Also, its much shorter residence time is more representative of the turbulent processes in practical flames [114]. Despite the above merits, it is important to point out that counterflow flames also have disadvantages. For instance, a counterflow flame is usually rather confined in space, necessitating high spatial resolution for its diagnostic. In contrast, coflow flames are more spread out spatially which makes probe sampling of the gas phase or particulates much easier.

Major differences exist in the sooting characteristics between coflow and counterflow diffusion flames. Soot formed in the fuel-rich region of a normal coflow flame is always convected downstream, towards the high temperature flame front where fuel and oxidizer are mixed stoichiometrically. As a result, oxidation of soot by oxygen and hydroxyl radical is inevitable. In fact, the widely referred smoke point condition is a critical condition where soot formation is balanced just by its oxidation [104], [115], [116], [117]. But for counterflow diffusion flames, the absence or presence of soot oxidation depends on the relative position between the stagnation plane and the flame front. By adjusting the dilution ratio of the fuel and oxidizer streams (and thus the stoichiometric mixture fraction), it is possible to establish a soot formation (SF) flame where soot particles, once formed, are convected away from the flame without further oxidation [73]. In this way, the soot formation process, as well as the physicochemical features of the flame-generated soot, can be investigated with little interference from soot oxidation. Detailed descriptions of the sooting processes in diffusion flames are covered in a later section, but it can be seen from the above discussion that counterflow flames provide a well-defined, canonical configuration to complement coflow diffusion flames for soot research on diffusion flames.

The literature on soot formation is extensive and excellent reviews [18], [31], [34], [99], [107], [118], [119], [120], [121], [122], [123], [124], [125], [126], [127], [128] were periodically published to summarize research progress. Many of these reviews had a specific focus. For example, in addition to providing an excellent theoretical discussion on the general kinetic pathways of soot formation, experimental data reviewed by Wang [18] were largely based on studies of burner-stabilized premixed flames. The work of Eremin [31] focused on the formation of soot particles in shock tubes while Tree and Svesson [124] analyzed soot processes in compression ignition engines. D'Anna [125] insightfully examined both numerical and experimental aspects of nanoparticle formation in laminar flames, focusing on premixed and coflow diffusion flames, while with limited coverage on counterflow diffusion flames. Karataş and Gülder [107] offered a dedicated review on high-pressure coflow diffusion flames. Despite the importance of soot studies in laminar counterflow flames, and the many important conclusions drawn from them, there has been no dedicated review in the open literature summarizing soot research in these flames, which dates back to 1980s and continues to contribute to our expanding knowledge on soot. The special features of soot processes in counterflow flames certainly deserves such efforts. Motivated by this, we intend to provide a comprehensive survey of research outcomes resulting from investigations of soot formation in counterflow flames, complementing existing reviews on soot research in many other flame/reactor configurations.

Following is the structure of this survey of soot research in laminar counterflow flames. First, an overview of the soot formation pathways is revisited, with special attention paid to the most recent progress. The discussion follows the generally acknowledged major processes of soot formation [122]: (1) gas-phase precursor formation; (2) soot nucleation; (3) particle surface growth, particle coalescence and agglomeration; and (4) soot oxidation and fragmentation. Experimental methods for soot diagnostics are then briefly discussed to provide necessary background for following discussions on the general sooting structures of diffusion and partially-premixed flames. Similarities and differences in the soot evolution processes between coflow and counterflow flames are also highlighted. Experimental and numerical counterflow flame-based soot studies are then extensively reviewed and the materials are organized based on the various factors affecting soot formation, including fuel molecular structure, fuel additives, pressure, temperature, residence time, external fields, and flame unsteadiness. The discussion on fuel molecular structure also summarizes various quantitative indices developed to represent sooting tendencies of different fuels, which may find application in assessing and formulating surrogate fuels for sooting tendencies. These indices include the conventional smoke point, the coflow flame-based yield soot index and the newly proposed, CDF-based sooting temperature index and sooting sensitivity index. This review concludes with a brief summary and an outlook on research needs, with respect to both experimental and numerical aspects, for promoting fundamental understanding of soot formation in counterflow flames.