Identification of Main Reaction Path of Soot Formation from Primary Pyrolysis Products in Coal Gasification

Abstract

In coal gasification, soot forms from volatile matter (VM) and it influences the gasification efficiency. Recently, biomass or plastic wastes with higher VM than coal are used for gasification processes and the importance of soot is increasing. For predicting soot formation, elementary step models are often used. However, defining main reaction paths is difficult because the models include huge species and reactions. In this study, the polyaromatic hydrocarbon formation path was analyzed via an elementary step-like model with 257 species and 1107 reactions to understand soot formation. The main reaction paths were extracted. Initially, 257 species were classified into non-aromatic species, one-ring, two-ring, and three- or more-ring aromatics. Ten species in each group with high maximum concentrations were considered as the major species. The reactions were extracted as follows: extracting reactions according to their contribution to the major species mass balance; adding fast reactions; and identifying the main reaction path. Consequently, the main reaction path extracted (MRE) model for the combustor condition was reduced to under 100 reactions for two kinds of conditions. Simulation of coal gasification reaction with the MRE model successfully described soot formation behavior in experiments with a drop tube furnace.

Keywords:

• Coal gasification
• Soot formation
• Skeletal model
• Polyaromatic hydrocarbon
• Drop tube furnace

1. Introduction

The integrated coal gasification combined cycle (IGCC) is an advanced power generation system that efficiently utilizes coal. In Japan, an air-blown entrained-flow-type gasifier was developed for IGCC (Asano 2012). An O₂/CO₂/H₂O-blown gasifier was also developed for an oxyfuel-type IGCC (Oki et al. 2016). Furthermore, the gasification technology is currently extended to a poly-generation system that co-produces power and chemical products with CO₂ capture (Umemoto et al. 2021). These gasification technologies are useful for utilizing coal and other solid fuels, such as biomass and plastic wastes in the current situation of high demand for CO₂ emission reduction. For the development of gasifiers, investigation of chemical reaction mechanism is important.

Coal particles, which are blown into these gasifiers, are decomposed into char and volatile matter (VM) through primary pyrolysis. Char gasification is the rate-determining step, which has been extensively investigated (Miura et al. 1989; Roberts and Harris 2007; Takarada et al. 1985; Umemoto et al. 2013). The VM also influences the coal gasification performance (Hosokai et al. 2022). For example, Li et al. (2023) measured the VM components in primary pyrolysis and found that they were C1 to C16 species. They easily volatilize and react with other oxidizing agents, such as O₂, O, OH, etc., otherwise, they convert to polyaromatic hydrocarbon (PAH) or soot. Furthermore, other feedstocks like biomass and plastic wastes which attract attention for decreasing fossil fuel utilization have a higher content of VM than coal (Madanikashani et al. 2022; Okumura 2021). In the case of using these feedstocks with coal, reactions of the VM become more important. The gasification reactivity of soot is lower than that of char, and soot formation reduces the carbon conversion efficiency (Umemoto et al. 2016).

The authors recently developed a primary pyrolysis model by extending the chemical percolation devolatilization (CPD) model (Umemoto et al. 2017), and modified it for application to various coal types (Umemoto et al. 2023). The original CPD model describes the coal structure using abstract components, such as aromatic ring clusters, links, and peripheral groups (Fletcher et al. 1992; Grant et al. 1989). The extended CPD (Ex-CPD) model assigns a specific chemical composition to the coal structures to describe specific chemical species of the primary pyrolysis products in gas and tar. Then, the gas and tar reactions in the gas phase can be calculated using elementary step-like reaction models which have been still refined for accurate prediction of soot formation reactions (Jin et al. 2023; Martin et al. 2022). In the previous literatures, the authors used an elementary step-like reaction model developed by Richter and Howard (2002) to describe the formation of PAHs that develop into soot in coal gasification. The calculated results of the PAH formation behavior agreed with that of soot under several experimental gasification conditions. Tar and acetylene consumption by oxidizing agents decreased the soot yield rather than gasifying the soot. However, clearly defining the main reaction path is challenging, because the elementary step-like reaction model includes numerous reactions. Therefore, extracting critical reactions is necessary to define the main reaction paths.

Studies on PAH formation or VM reactions are fewer than those on char gasification and coal primary pyrolysis. However, methods of numerical analysis have been developed, and this type of gas-phase reaction has received increasing attention. For example, Sakurai et al. (2013) experimentally identified coal VM components to propose a method for analyzing the reaction behavior by adding several reactions to an elementary step-like reaction model. Soot formation is a type of chemical vapor deposition (CVD) reaction. The reactions of hydrocarbons or liquidized fuels that resemble PAH have been analyzed using elementary step-like reaction models (Kousoku et al. 2014). Another model regards chemical species as virtual substances, similar to primary pyrolysis models that consider coal structure by aromatic ring clusters, links, and peripheral groups. Niksa (2017) applied the concept of the primary pyrolysis model, specifically, a FLASH-CHAIN model, to the gas-phase reaction; the reaction model size was considerably smaller than that of general elementary step-like reaction models. These methods are essential to analyze gas-phase reaction phenomena; however, elementary step-like reaction models describe the phenomena more realistically, because gas-phase reactions without solids are unaffected by catalysts and pores, although they are significant for primary pyrolysis. Additionally, numerical calculation methods are expected to be further developed in the future, which is beneficial for elementary step-like reaction models; therefore, this model was adapted in this study.

Elementary step-like reaction models involving aromatics have been mainly developed for gaseous or liquid fuels. Because the main reaction path changes depending on the reaction conditions, some cases require the addition of new reactions when the conditions are altered (Thimthong et al. 2015). Therefore, automatic generation methods have been developed to add reactions using the kinetic parameters of reactions, including similar structural species (Miyoshi 2005; Zhu et al. 2022). Determination of the main reaction pathway is difficult because the models have become huge. In addition, current computers require substantial time to consider more than a thousand elementary step-like reactions when applying computational fluid dynamics (CFD) for analysis inside gasifiers. Therefore, several simplifying approaches have been proposed to extract the main reactions from large reaction models. There is another method which calculates multiple conditions in one dimension in advance to create a look-up table, and refers to the table in a three-dimensional calculation (Akaotsu et al. 2020). However, reference cases must be added to the library for calculating new conditions. In addition, there is a complicated thermal distribution in the gasifier, and the thermal history of the library cannot be easily described.

There are three methods for simplifying elementary step-like reactions. The first is the skeletal method, which simply reduces elementary step-like reactions that minimally affect the calculation results. The second is the lumping method, wherein various isomers are combined into a single chemical species. The third method involves a quasi-steady-state approximation, in which the amount of intermediate product is assumed to be constant during the reaction, thereby combining the generation and consumption reactions of intermediate products in chain reactions. In the present study, the skeletal method is primarily considered, because it is often used first for identifying the main reaction path from a large-scale elementary step-like reaction model.

The directed relation graph (DRG) method was proposed as a typical skeletal method (Lu and Law 2005). Niemeyer et al. (2010) then proposed an advanced DRG model that considered error propagation and sensitivity analysis. However, these automatically reducing methods like DRG model typically use only one standard to select reactions and species, whereas there are several important reactions. For example, when a reaction has a high initial reaction rate, another reaction can have a low initial reaction rate, with a high extent of reaction overall. Kawase et al. (2006) applied an elementary step-like reaction model to analyze a process of CVD of carbon from propane, and simplified the reaction to extract 26 reactions as the main reaction path from 651 reactions.

In this study, gas-phase reactions under gasifier conditions were initially calculated using an elementary step-like reaction model; subsequently, the model was analyzed to extract the main reaction paths.

2. Gasifier Condition

The gasifier considered in this study is an air-blown two-stage entrained flow type, as shown in Figure 1. The gasifier consists of two parts: a combustor and a reductor. Coal mainly burns with O₂ in the air, thus generating CO₂, H₂O, and heat in the combustor. Then, in the reductor, coal mainly reacts with CO₂ and H₂O obtained from the combustor. The combustor has an oxidative atmosphere, whereas the reductor has a reductive atmosphere. In this study, both the combustor and reductor conditions were used for analyzing the gas-phase reactions.

Figure 1. Schematic of the gasifier for calculations.

The initial conditions for analyzing the gas-phase reactions were as follows: They were established from public information (Asano 2012). The pressure was 2.7 MPa; the ratio of the reductor input coal to the total input coal (R/T) was 0.5; the ratio of input air to the theoretical amount of air for completely combusting the total input coal (air ratio) was 0.45; the oxygen content in the total input gas (oxygen-enriched air) was 25%. Additionally, the single-pass carbon conversion efficiency was assumed to be 0.8 to determine the amount of char recycled into the combustor. Oxygen-enriched air without a carrier gas for the reductor coal was fed into the combustor. Coal pyrolysis gas data from the Ex-CPD model (Umemoto et al. 2017, 2023) were used for analyzing the reactions. DT coal (Chinese bituminous coal) data was used for the calculation. Its properties are listed in Table 1.

Table 1. Properties of coal samples by ultimate and proximate analyses.

(DT coal is used in Sections 4–6; others are used in Section 6)
Coal	Ultimate analysis (wt%, d.a.f.^a)				Proximate analysis (wt%, d.b.^b)
	C	H	N	O (diff.)	FC^c	VM^d	Ash
DT	83.2	4.3	0.3	12.2	61.5	27.3	11.2
MN	81.1	5.7	2	11.2	51	40.7	8.4
BT	73.4	5.3	1.7	19.7	48.2	45.3	6.4
AD	73.1	5.1	1	20.8	47.7	48.7	3.7

^ad.a.f.: dry and ash free basis.

^bd.b.: dry basis.

^cFC: fixed carbon

^dVM: volatile matter.

The primary pyrolysis and most gas-phase reactions before soot formation can be assumed to be instantaneously completed near the burner. In addition, completely different reactions should be predominant in the combustor and the reductor. Therefore, the combustor and reductor are analyzed separately, and the temperatures are analyzed under constant conditions of 1800 and 1400 °C, which are considered the temperatures at the respective burners.

3. Calculation Method of VM Reaction in Gasifier

The Ex-CPD model was used to calculate the VM composition of DT coal and determine the initial gas phase compositions in the gasifier. The heating rate was assumed to be 10⁴ °C/s based on experiences with CFD (Watanabe and Otaka 2006; Watanabe et al. 2015). Table 2 lists the initial compositions in the combustor and the reductor. The initial composition in the combustor was obtained by mixing the input gas and the VM of the coal charged into the combustor. The initial composition in the reductor was obtained by mixing the equilibrium composition in the combustor at 1800 °C with the input N₂ for the coal carrier and the VM of the coal charged into the combustor. Partially revised Ex-CPD which treated acetylene as VM was used for the combustor. The compositions are not exactly the same as the real primary pyrolysis gas (for example, CH₄ or C₂H₂ composition may be much higher than real primary pyrolysis gas) because these are the model-calculated compositions. However, it is difficult to consider complicated primary pyrolysis product species in elementary reaction models, and Table 2 contains representative chemical species contained in pyrolysis gas.

Table 2. Initial gas compositions for kinetic analyses of the gasifier.

	Unit: % (molar basis)]
	Combustor condition	Reductor condition
N2	69.3	61.5
O2	23.1	0
CO	1.02	17.5
CO2	0.29	12.4
H2	2.39	4.14
H2O	1.66	2.59
CH4	0.37	1.85
C2H2	1.81	0
Benzene	6.2 × 10^−3	2.7 × 10^−2
Naphthalene	1.2 × 10^−3	5.1 × 10^−3
Phenanthrene	1.1 × 10^−2	6.1 × 10^−2

The gas-phase reaction kinetics calculation software CHEMKIN-PRO 15101 was used in this study. The gas-phase reaction model developed by Richter and Howard (2002) was applied. The model consists of 1107 elementary step-like reactions involving 257 chemical species with a molecular weight of a six-ring compound (coronene) or less. A set of reversible reactions (forward and backward) was regarded as one reaction, which could express the reaction behavior of aromatics at high temperatures. The model is not exact elementary step reaction mechanism. Therefore, it and its skeletal models include some highly chemically unstable species. Here, soot formation and soot gasification were not considered. As mentioned in the Introduction, soot gasification is much slower than char gasification. Therefore, PAH (here, hydrocarbons larger than phenanthrene) yield was treated as a surrogate for soot yield.

4. Complete Model Calculation Results

The elementary step-like reaction model without reduction (the complete model) was initially utilized to calculate the combustor and reductor conditions. Figure 2 presents the gas composition profiles of the combustor and the reductor. High-concentration reactive species (CO, CO₂, H₂, H₂O, CH₄, and O₂) and primary tar components (benzene, naphthalene, and phenanthrene) with acetylene (C₂H₂) are indicated in the volume fraction. PAHs are observed in the mass fraction, which corresponds to the mass fraction of soot because the model does not include soot formation reactions.

Figure 2. Fractions of gas species and PAHs calculated using the complete model.

The time required to obtain stable PAH fractions significantly differed between the combustor and reductor conditions. There was a difference between the reaction times of the combustor condition and the reductor condition because their temperatures are different. Therefore, the major reaction paths in the subsequent sections were defined using 0–2 μs and 0–0.8 s for the combustor and the reductor, respectively.

Additionally, in the combustor condition, the mass fraction of PAH was approximately 0.05 g/kg at 2 μs, while that in the reductor condition was approximately 1 g/kg at 0.8 s. The value of 1 g/kg is equivalent to approximately 1% of the input coal on the carbon basis, which indicates that the soot yield of the actual gasifier can be significant, although it depends on the operating conditions.

Under the combustor condition, light gas, primary tar, and PAH fractions exhibited the following behaviors: First, reactive species such as H₂, CH₄, benzene, naphthalene, and phenanthrene monotonically decreased upon oxidation by O₂. Second, the oxidation products H₂O and CO₂ monotonically increased. Third, other species such as CO, C₂H₂, and PAH, initially increased and subsequently decreased. Particularly, inorganic gases and PAHs were generated by the oxidation or pyrolysis of the reactive species. Then, CO and PAHs reacted with O₂ relatively slowly, generating CO₂ and H₂O. The time lag between H₂O and CO₂ formation was attributed to the difference between the reactivities of H₂ and CO. Hydrocarbon oxidation generated CO₂ via CO formation.

Under the reductor condition (Figure 2(b)), at first, benzene, naphthalene and phenanthrene decomposed or become heavier to generate C₂H₂ or PAH. After that, the fractions of C₂H₂ and PAH decreased, while that of CO gradually increased. However, in contrast to the phenomenon in the combustor condition, fast oxidation did not occur owing to the absence of O₂. The reaction behaviors can be qualitatively explained using the calculated results of the complete model. However, the main reaction path from the primary tar to PAHs or soot cannot be easily defined using only complete model calculations. For example, reaction paths can be drawn according to reaction rates using specific software; however, it is difficult to ascertain whether other reactions can be eliminated to describe the target reaction behaviors. Therefore, a method to extract significant reactions is proposed in the following sections. The extracted reaction paths are subsequently used as simplified reaction models to calculate the PAH formation behavior and compared with the complete model.

5. Main Reaction Path Extracted Model

The authors attempted to determine an effective method for reducing reactions and species while maintaining small differences between the calculated PAH concentrations of the reduced and complete models. The calculation time is roughly assumed to be almost proportional to the number of reactions and species. In this study, we aimed to reduce the number of reactions from 1107 to < 100 and the number of species from 257 to < 100, to consequently reduce the calculation time to < 3%. By trial and error, the steps for extracting important reactions were constructed as follows:

Step 0: Grouping of species and reactions, and selection of major species

Step 1: Extraction of reactions according to contributions to the mass balance of major species

Step 2: Addition of fast reactions (average and maximum)

Step 3: Identification of the main reaction path

Details of these steps are described below.

5.1. Grouping of species and reactions, and selection of major species

(1) Grouping of species and reactions

As shown in Figure 2, the mole fractions of the light gas species were considerably larger than those of the tar species. This indicates that most reactions involving only light gas species have higher reaction rates than those of reactions involving tar species. If reactions with higher rates are chosen from all reactions, most of them involve only light gas species. Therefore, four groups of species and reactions were created. The classifications are listed in Table 3. Species were grouped into SG1–SG4 in the order of molecular size, and reactions involving SG1–SG4 were grouped into RG1–RG4; SG1 includes the most species, and RG1 includes the most reactions because the complete model includes several light gas species. Nevertheless, the total number of species in SG2–SG4 (179) is greater than that in SG1 (78), and the total number of reactions in RG2–RG4 (505) is comparable to that in RG1 (602). Thus, 505 reactions must be compared even for extracting important reactions related to tar species. Additionally, the initial mole fraction of naphthalene was less than half that of benzene and phenanthrene. Therefore, tar species and reactions were divided into three groups. In steps 1 to 3, important or primary species and reactions were extracted from these groups.

Table 3. Species and reaction groups (Cn: species containing n carbons).

Species group	Classification	Quantity	Reaction group	Classification	Quantity
SG1	C0–C5 (light gas)	78	RG1	Reactions concerning to C0–C5	602
SG2	C6–C9 (benzene etc.)	41	RG2	Reactions concerning to C6–C9	171
SG3	C10–C13 (naphthalene etc.)	46	RG3	Reactions concerning to C10–C13	89
SG4	C14–C24 (phenanthrene, PAH etc.)	92	RG4	Reactions concerning to C14–C24	245
Total		257	Total		1107

(2) Selection of major species

Ten species, with the highest maximum concentrations in each group, are selected as the major species. The mole fraction profiles of the major species calculated using the complete model are displayed in Figures 3 and 4 for the combustor and reductor conditions, respectively.

Figure 3. Ten major species in each species group (SG) in the combustor condition.

Figure 4. Ten major species of each SG in the reductor condition.

The mole fraction axes are presented in logarithms owing to the substantial differences between the mole fractions, even in the same group. The order of the mole fractions changes over time in all groups and conditions. If the major species are selected by comparing concentrations at a certain time, some species with high concentrations at other times may be excluded from the selection. Therefore, the maximum mole fractions over the entire time were used for selecting the major species.

Figures 5 and 6 exhibit the maximum mole fractions of all species under the combustor and reductor conditions, respectively. The graphs on the left indicate the complete ranges, whereas those on the right depict the magnified portions. In SG2–SG4, under reductor conditions (Figure 6), the primary tar species (benzene, naphthalene, and phenanthrene) have the highest mole fractions in each group. In contrast, in SG3 under the combustor condition (Figure 5(c)), 1-ethylnaphthalene (A₂C₂H_-1) presents the highest mole fraction. The species is not directly generated from naphthalene but from phenanthrene. The top four species in SG3 are mainly formed through phenanthrene decomposition. Thus, the major species were different between the combustor and reductor conditions. Therefore, the reaction pathways for PAH formation are expected to be different.

Figure 5. Maximum mole fractions of chemical species in the combustor condition (0–2 μs) (the right figures are the magnification of the left figures indicated by brackets).

Figure 6. Maximum mole fractions of chemical species in the reductor condition (0–0.8 s) (the right figures are the magnification of the left figures indicated by brackets).

There were reaction paths from a small species in SG2 to a large species in SG3 and SG4. If these species in the three groups were entered into the same group, only species with similar molecular sizes would be extracted, and the reaction paths would be cut off, due to the bias in mole fractions. Therefore, these three groups were created to compare the species and responses within each group effectively.

Major species, particularly, the top 10 species with high mole fractions, were selected for each group. The summation of mole fractions of the 10 major species exceeded 0.9 in any group.

5.2. Extraction of reactions according to contributions to major species mass balance (Step 1)

Most simplifying methods of elementary step-like reaction models are automatically reducing methods using thresholds, such as reaction rates, which are compared at every time interval in the calculation range (Niemeyer et al. 2010). The authors selected a method that uses comprehensible standards and extracts important reaction paths. The contribution of the major species to the mass balance was used as an indicator of the extraction in Step 1. Figure 7 is a schematic of the mass balance contribution. Initially, the extent of a reaction was calculated for 2 μs in the combustor condition and for 0.8 s in the reductor condition. All reactions in the complete model were considered reversible. The difference between the extent of the forward and backward reactions was then calculated as the extent of the reaction. For example, if the forward and backward reactions proceeded at 3 and 2 mol, respectively, then the extent of the reaction was 1 mol (Figure 7(a)). Regarding the contribution to mass balance, the absolute values of the extent of consumption and formation reactions were accumulated as the mass balance, because both consumption and formation reactions were significant for the reaction path. If the extent of reaction r₁ consuming species A is 1 mol, and the extent of reaction r₂ producing and consuming A is 3 mol, the overall contribution of reaction r₁ for A is 0.25. (Figure 7(b)). Table 4 lists the species and reactions after extracting reactions whose contributions to the mass balance of the major species are higher than 0.1. Both species and reactions were fewer than 100 in the (a) combustor and (b) reductor.

Figure 7. Schematic of the extent and the contribution of reactions to the mass balance of species.

Table 4. Numbers of species and reactions after extraction based on contributions to mass balance (Step 1).

Group	Total	SG1/RG1	SG2/RG2	SG3/RG3	SG4/RG4
(a) Combustor condition
Species	61	19	13	12	17
Reaction	82	17	21	13	31
(b) Reductor condition
Species	67	25	16	20	6
Reaction	91	19	23	19	30

Figures 8 and 9 depict the calculated results for the combustor and reductor conditions, respectively. The solid lines indicate the calculated results for the species and reactions listed in Table 4. The dotted lines indicate the results calculated using the complete model. For the reductor condition, magnified figures are shown for 0–0.1 s and 0–0.8 s, specifically because benzene, naphthalene, and phenanthrene fractions significantly changed during the initial stage.

Figure 8. Fractions calculated by the extracted reaction model according to contributions to the mass balance of major species (combustor condition) (solid line: extracted model (Table 4, Step 1); dotted line: complete model).

Figure 9. Fractions calculated by the extracted reaction model according to contributions to the mass balance of major species (reductor condition) (solid line: extracted model (Table 4, Step 1); dotted line: complete model).

Under the combustor condition (Figure 8), the O₂ fraction of the extracted model calculation decreased more gradually than that of the complete model calculation. Accordingly, H₂O and CO₂ increased more gradually, whereas CH₄, C₂H₂, benzene, naphthalene, and phenanthrene gradually decreased. In addition, the PAH mole fraction at the completion of the extracted model calculation was higher than that of the complete model calculation. This is because reactions involving O₂ were eliminated from the extracted model, and non-oxidation reactions dominantly proceed.

Under the reductor condition, the calculated results of the extracted model were similar to those of the complete model, especially during the initial stage. However, in the results of the extracted model, the fraction of PAH initially increased and then maintained, and light gas mole fractions differed from those calculated by the complete model calculation. Additionally, SG4 included only six species in the reductor condition (Table 4). Fewer than ten species are selected as the major species. Major species larger than pyrene did not form in the calculation of the extracted model because the reaction paths were disconnected.

Therefore, extracting species and reactions from one perspective generated a significant error; multiple species and reactions should be adopted to reduce the margin of error.

5.3. Addition of fast reactions (average and maximum) (Step 2)

The previous section demonstrated the difficulty of maintaining a low error and reducing species and reactions from one perspective. Therefore, other perspectives are adopted to extract important reactions in this section. Figure 10 presents the two types of reactions added here.

Figure 10. Schematic of reactions with high reaction rates.

The first type displays an extremely high reaction rate. Active intermediates, such as radicals, exhibit high reactivities at high temperatures. The complete model includes several reactions that involve active intermediates. A reaction from one active intermediate to another proceeds promptly, as shown in Eq. (a-2) in Figure 10. Even if species A and B are major species, intermediates A* and B* are not similarly considered, because they are immediately consumed in reactions a-2 and a-3. Subsequently, reaction a-2 is eliminated (Step 1 in section 5.2). For example, in the formation reaction of Benzyl radical, both the reactant (phenyl radical (C₆H₅)) and the product (C₇H₇) demonstrate high reactivity; consequently, these species were instantaneously consumed, and were not included in the major species. Nevertheless, this reaction is important in the reaction path for benzene to convert to heavier molecules.

The second type demonstrates a high initial reaction rate. The reaction is similar to that of b-4 in Figure 10, wherein an active intermediate (A*) that is directly generated from the primary tar or gas species produces a minor species (C). The total extent of this type of reaction is low. However, the reaction affects the mass balance of A, and eliminating b-4 results in a significant error. For example, in the reaction shown in Figure 10(b), the methyl radical (CH₃) produces the methanol radical (CH₂O). A high reaction rate was observed at the beginning in the combustor condition owing to the high concentration of methane; however, the reaction rate decreased after a while.

According to these perspectives, the reactions demonstrating high extents or maximum reaction rates were added to the extracted model in Step 1 to maintain a little difference from the calculated results of the complete model. Table 5 lists the number of species and reactions, and Figures 11 and 12 display the calculated results for the combustor and reductor conditions, respectively.

Table 5. Numbers of species and reactions after adding the reactions with high reaction rates (Step 2).

Group	Total	SG1/RG1	SG2/RG2	SG3/RG3	SG4/RG4
(a) Combustor condition
Species	91	28	26	13	24
Reaction	158	37	41	29	51
(b) Reductor condition
Group	Total	SG1/RG1	SG2/RG2	SG3/RG3	SG4/RG4
Species	113	34	17	32	30
Reaction	146	29	33	29	55

Figure 11. Fractions calculated by extracted reaction model after adding the reactions with high reaction rates (combustor condition) (solid line: extracted model (Table 5, Step 2); dotted line: complete model)

Figure 12. Fractions calculated by extracted reaction model after adding the reactions with high reaction rates (reductor condition) (solid line: extracted model (Table 5, Step 2); dotted line: complete model).

In the combustor condition, reactions were added only to RG1. This implies that the error in step 1 was generated because oxidation reactions were absent in RG1. The calculated results of the extracted model in Step 2 for all the species shown in Figure 11 were almost similar to those of the complete model.

In contrast, reactions are added to all groups in the reductor condition. This indicates the influence of intermediates, from small to large sizes, on PAH formation, because they can remain in the reductor condition, unlike in the combustor conditions where oxygen is present. The intermediate reactions are possibly reduced to generate global reactions by approximating the quasi-steady states. Global reactions had been previously used for CFD when considering only light-gas reactions (Watanabe and Otaka 2006; Watanabe et al. 2015). To apply such an approximation, the significant reaction path must be extracted from a large reaction model. In the calculated results shown in Figure 12, little differences were observed between the mole fractions of tar species (benzene and naphthalene) calculated with the extracted and complete models, This is because the derivatives (such as C₆H₅CH) which are not strongly correlated with other major species were removed from the model and original species (such as benzene) were likely to remain. However, there was little error in light gas fractions, and the PAH mass fraction was considerably improved compared to Figure 9.

5.4. Identification of the main reaction path (Step 3)

In the previous section, both the species and reactions were reduced to approximately 100, and the calculated results were similar to those obtained using the complete model. Furthermore, species and reactions in each group were reduced to only a few dozen. Therefore, connections between each reaction can be found, and the main reaction path can be adequately identified. Therefore, to extract smaller reaction models, the reaction paths from the species included in the initial conditions to the other major species were determined, and the reactions not included in the paths were removed.

In the combustor, the reaction path from primary tars to PAH involves the following steps: phenanthrene (A3) becomes a phenanthrene radical, which is attacked by acetylene to produce pyrene and acephenanthrene (A3R5). The main decomposition reaction path from phenanthrene proceeds from the formation of ethynyl naphthalene (A2C₂H_-1, A2C₂H_-2) by H₂O and OH to that of the benzene radical by O₂ and O radicals.

In the reductor condition, the tar reactions proceed more extensively than in the combustor condition. The main reaction pathway for PAHs is from phenanthrene to coronene, which is one of the largest molecules in the complete model.

Figures 13 and 14 illustrate the main reaction paths of the PAH formation extracted from the combustor and reductor conditions, respectively. Under the combustor condition, the PAH formation reaction pathway is from phenanthrene to pyrene or acephenanthrene. Species larger than pyrene were rarely formed and were eliminated from the extracted model. The phenanthrene decomposition reaction had a higher extent than reactions to convert heavier. These decomposition reactions proceeded preferentially, thus suppressing reactions to convert heavier. However, in the reductor condition (Figure 14), after the conversion of phenanthrene to pyrene, pyrene converts to heavier molecules and the reaction path finally reaches benzo[ghi]perylene and coronene. The difference between the combustor and reductor is attributed to acetylene, which substantially contributes to HACA (Hydrogen-Abstraction-C₂H₂-Addition) reactions in the reductor condition. However, in the combustor condition, acetylene reacts with O to form the ketenyl radical (HCCO) and weakly contributes to HACA reactions.

Figure 13. Main reaction paths involving phenanthrene in the combustor condition (the values indicate the extent of the reaction calculated by the complete model [mol/kg] (0–2.0 μs)).

Figure 14. Main reaction paths from phenanthrene to coronene in the reductor condition (the values indicate the extent of the reaction calculated by the complete model[mol/kg] (0–0.8 s)).

The number of reactions for the combustor condition exceeds 100. When using for numerical analysis, it is desirable to keep the number of reactions less than 100. Therefore, several sets of reactions were combined to reduce the species and reactions when the intermediate species were only included in these reactions. Two or more reactions with similar extents were lumped together, because the preceding reaction is the rate-determining step and the succeeding reactions proceed promptly; thus, steady-state approximation can be applied. Figure 15 depicts the combinations of lumped reactions.

Figure 15. Lumped reactions for the combustor condition.

The number of species and reactions of the extracted model constructed after Step 3 are listed in Table 6. The calculated results using the main reaction path extracted (MRE) model are shown in Figures 16 and 17. In both the combustor and reductor conditions, the calculated mole fractions of light gas species by the MRE model were similar to those of the complete model, although slight errors were observed between the tar mole fractions calculated by the MRE and the complete models. Although there was some error in the final mass fraction of the PAHs, its increase and decrease were captured by the MRE model. A high accuracy was expected with the significantly reduced number of reactions for calculations.

Table 6. Numbers of species and reactions of MRE models after identification of the main reaction paths.

Group	Total	SG1/RG1	SG2/RG2	SG3/RG3	SG4/RG4
(a) Combustor condition
Species	52	24	12	10	6
Reaction	96	38	26	13	19
(b) Reductor condition
Species	55	21	6	12	16
Reaction	71	22	9	14	26

Figure 16. Fractions calculated by the MRE model (combustor condition) (solid line: MRE model (Table 6); dotted line: complete model).

Steps 1 to 3 were applied to construct MRE models composed of 52 species and 96 reactions for the combustor condition and 55 species and 71 reactions for the reductor condition. The species are still more than 50 and it may be not little enough for CFD of large-size gasifiers. There are several possibilities to reduce the number, such as lumping like Figure 15. However, it is necessary to reduce the number of species carefully because reducing species decrease the accuracy. The main purpose of this study is to extract the main reaction path. Nevertheless, the MRE model may be useful for a kind of CFD. Therefore, the applicability of the MRE model is examined below.

5.5. Effect of temperature on MRE model calculation results

To investigate the effect of temperature on the MRE model calculation results, the reaction behaviors under combustor and reductor conditions were calculated at several temperatures. The temperatures of previous calculations were 1800 °C for the combustor case and 1400 °C for the reductor case. Then, the MRE models derived at these temperatures are used for different temperatures. Figures 18 and 19 show the calculated results of the MRE and complete models at 1700 and 1900 °C with the initial composition in the combustor condition and at 1300 and 1500 °C with the initial composition in the reductor condition, respectively.

Figure 17. Fractions calculated by the MRE model (reductor condition) (solid line: MRE model (Table 6); dotted line: complete model).

Figure 18. Applying the MRE model for different temperature conditions (combustor condition) (a-1∼c-1: 1700 °C, a-2∼c-2: 1900 °C, solid line: MRE model (Table 6); dotted line: complete model).

Figure 19. Applying the MRE model for different temperature conditions (reductor condition). (a-1∼c-1: 1300 °C, a-2∼c-2: 1500 °C, solid line: MRE model (Table 6); dotted line: complete model).

Under both conditions, the time required for stabilizing the PAH mass fraction significantly varied with temperature. Consequently, temperature strongly influenced the formation of PAHs. Despite these marked changes, the light gas concentrations calculated using the MRE models were almost identical to those calculated using the complete model. There were differences between concentrations calculated by the MRE model, however, the production and consumption behaviors of PAHs and tars calculated using the MRE were similar to those calculated using the complete model. The reliability of the model in the case of other temperatures or non-isothermal processes will be considered in future studies.

6. Verification of MRE Models Applying a PDTF Gasification Experiment

Coal gasification tests using a pressurized drop tube furnace (PDTF) were performed using gasifying agents such as CO₂, O₂, or H₂O. The experimental conditions and results are shown in the past literature (Umemoto et al. 2016). The coal properties considered in the experiments are listed in Table 1. The Ex-CPD and char gasification models (Kajitani et al. 2002; Umemoto et al. 2013) were utilized to calculate primary coal devolatilization and char gasification, respectively. In the calculation, the heating rate was assumed at 10³ °C/s and the holding time was 1.1 s which corresponds to the position in which samples were captured in the experiments. The gas-phase reactions were calculated using MRE models with the DVODE solver. For the cases including O₂ in the primary gas components, the MRE model constructed for the combustor conditions was applied. In the other cases, the model for the reductor condition was applied.

Figures 20–23 exhibit the carbon conversion and char and soot (PAH) yields for four types of coal with different initial gasifying agent compositions. The soot yields from the experiments are compared with the calculated PAH yields. The yield [mol/mol] is defined as a ratio of the amount of carbon in PAHs or soot to the amount of carbon in coals. In the experiments, PAHs were not detected because they were converted to soot. The large PAHs are not easy to decompose to light hydrocarbons, then the net formation rate of PAHs is almost the same as that of soot. In the calculation, the conversions from PAHs to soot were not included. The measured yield of soot possibly includes an influence of soot gasification, but soot has lower gasification reactivity than char (Miura et al. 2004) and the reactivity can be assumed to be similar to or lower than large PAHs. Therefore, the calculated yield of PAHs and measured yield of soot were compared. The temperature settings were 1300 °C for the DT and MN coals and 1200 °C for the BT and AD coals based on the experimental results. The calculations for all coal types indicated trends that were similar to the experimental results: gasification by CO₂ demonstrates higher carbon conversion and lower char and soot yields than those obtained via gasification by H₂O. Furthermore, at the same initial gasifying concentration, gasification by H₂O demonstrated higher carbon conversion and lower char and soot yields than those obtained via gasification by O₂.

Figure 20. Comparison of carbon conversion and soot and char yields between the MRE model and pressurized drop tube furnace (PDTF) experimental results of DT coal (1300 °C; 0.5 MPa; N₂ balance. Line: MRE model calculation; points: experiment).

Figure 21. Comparison of carbon conversion and soot and char yields between the MRE model and pressurized drop tube furnace (PDTF) experimental results of MN coal (1300 °C; 0.5 MPa; N₂ balance. Line: MRE model calculation; points: experiment).

Figure 22. Comparison of carbon conversion and soot and char yields between the MRE model and pressurized drop tube furnace (PDTF) experimental results of BT coal (1200 °C; 0.5 MPa; N₂ balance. Line: MRE model calculation; points: experiment).

Figure 23. Comparison of carbon conversion and soot and char yields between the MRE model and pressurized drop tube furnace (PDTF) experimental results of AD coal (1200 °C; 0.5 MPa; N₂ balance. Line: MRE model calculation; points: experiment).

Some conditions reveal differences between the experimental soot yield and the calculated PAH yield. These differences corresponded to the results calculated using the complete model in a previous study (Umemoto et al. 2023), which considered that the differences were generated because the calculated PAH yields did not include soot formation from PAH in the gas-phase reaction. However, the calculated PAH yields were still close to the experimental soot yield. Furthermore, the tendency of soot yields with gasifying agents in the experiments was clearly described by the calculations. A high CO₂ concentration did not decrease the soot (PAH) yield, whereas high H₂O and O₂ concentrations tended to decrease the soot (PAH) yields, and the effect of O₂ was stronger than that of H₂O. Therefore, the reaction behavior of the precursors of PAHs that were decomposed by gasifying agents were appropriately expressed by the MRE models.

7. Conclusions

In this study, gas phase reaction in coal gasification was analyzed to understand soot formation behavior. A method identifying the main reaction paths from an elementary step-like reaction model was proposed to analyze PAH formation during coal gasification. The complete model was composed of 257 species and 1107 reactions, which could describe hydrocarbon pyrolysis and combustion reactions. The main reaction paths were extracted by analyzing the combustor and reductor conditions of a coal gasifier.

Initially, 257 species were classified into non-aromatic species: one-ring (carbon number C6 to C9), two-ring (C10 to C13), and three- or more ring aromatics (C14 and more). Ten species, with high maximum concentrations in each group, were selected as the major species. The following steps were used to extract the reactions. Step 1: Reactions were extracted according to their contributions to the major species mass balance. Step 2: Fast reactions were added (average and maximum). Step 3: The main reaction path was identified.

Accordingly, the main reaction path extracted model (MRE model) for the combustor condition with 52 species and 96 reactions, and that for the reductor condition with 55 species and 71 reactions were constructed. These models clearly outline the main reaction path from the primary tar to PAH, the precursor of soot. Additionally, it has a practical advantage because it is more convenient to use in a numerical analysis than the original elementary step-like model.

As coal gasification experiments in the PDTF were simulated using the MRE models, the tendencies of the experimental results were successfully captured. The soot (PAH) yield negligibly decreased even at high CO₂ concentrations, while the yield decreased at high concentrations of H₂O and O₂; the yield was affected to a greater degree by O₂ than by H₂O.

Acknowledgments

Part of the presented work is based on results obtained from a project, JPNP10016 and JPNP16002, commissioned by the New Energy and Industrial Technology Development Organization (NEDO).

Appendix A. All species in the complete model (Richter and Howard 2002) are listed in Table A1.

 

Table A1. All species considered in the complete model.

H	I-C3H7	C6H3	A2-T2	A2R5YNE1	A3LR5	B[k]Fluor
H2	CH2CHO	Benzyne	C10H7-J1	A2R5YNE3	Pyrene	COR1^−
C	CH3CO	C6H4	C10H7-J2	A2R5YNE4	C17H12	COR1
CH	C3H8	C6H5	C10H8	A2R5YNE5	BenzNapJP	AnthanJS
CH2	CH3CHO	C6H5(L)	A2CH2-1	A2R5R5	BenzNap	B[ghi]PeJS1
HCH	CO2	C6H6	A2CH2-2	A3J1	CpcdPyrJS	Anthan
CH3	CH3OCH2	C6H6-F	A2CH3-1	A3J2	PyC2H-1JP	B[ghi]Per
CH4	CH3OCH3	C6H7	A2CH3-2	A3J4	PyC2H-2JS	InPyr
O	C4H	C6H8-3	C10H7O-1	A3J9	PyC2H-4JS	COR2−
OH	C4H2	C6H8-4	C10H7O-2	A3LJ1	CpcdFltJS	COR2
H2O	HCCHCCH	C5H5CH3	C10H7OH-1	A3LJ2	B[ghi]F-	CpbPer
C2	H2CCCCH	C5H4O	C10H7OH-2	A3LJ9	CpcdPyr	COR3−
C2H	C4H4	C5H5O	A2R5-T	A3 (Phenanthrene)	PyC2H-1	COR3
C2H2	CH2CHCHCH	C5H4OH	A2C2H-1J2	A3L	PyC2H-2	COR4−
C2H3	CH2CHCCH2	C6H5CH	A2C2H-2J1	A3CH2	PyC2H-4	COR4
C2H4	HCCCO	C7H7	A2C2H-2J3	A3CH2R	CpcdFlth	C24H12
CO	C4H6-1	C7H8	A2R5J1	A3CH3	B[ghi]F	Coronen
N2	C4H6-2	C6H5O	A2R5J3	A3C2H-1JP	A4J1
HCO	C4H6-3	C6H4OH	A2R5J4	A3C2H-2JS	A4J4
C2H5	HCCCHO	C6H5OH	A2R5J5	A3LC2H-1P	A4J12
C2H6	I-C4H7	A1C2H-J2	A2C2H-1	A3LC2H-2S	A4LJS
CH2O	CH2CHCO	C8H6	A2C2H-2	A3LC2H-2P	ChrysenJ1
CH3O	CHCHCHO	C6H5CHCH	A2R5	A3C2H-1	ChrysenJ4
CH2OH	C4H8	C8H7-J2	Biphen	A3C2H-2	ChrysenJ5
CH3OH	CH2CHCHO	C8H8	A2VINP	A3CL2H-1	A4
O2	C2H5CO	C6H5CO	C12H9	A3CL2H-2	A4L
HO2	C2H5CHO	C8H9	HA2R5	A2-1C6H4	Chrysen
H2O2	CH3COCH3	C8H10	BiphenH	A2-2C6H4	B[ghi]FRJS
C3H2	C5H2 (L)	C6H5CHO	A2C2H3-2	Flthn-J1	COR-
H2CCCH	C5H3	C6H3O2	C12H10	Flthn-J3	B[ghi]FR
C3H4-CY	C5H3(L)	C6H5CH2OH	A2R5H2	Flthn-J7	COR
C3H4	C5H4	C7H8O	Fluorene	A3R5J7	Dcpcdfg
C3H4-P	C5H4(L)	C6H5OCH3	Benzyl BJ	A3R5J10	B[a]PyrJS
C2O	C5H5	OC6H4O2	Benzyl B	A3LR5JS	BePyrenJS
C3H5	C5H4H	P-C6H4O2	C6H5OC6H5	PyreneJ1	PerylenJS
HCCO	C5H5(L)	Indene-J	A2R5YN1J2	PyreneJ2	B[b]FluorJS
C3H6	C5H6	C6H5C3H2	A2R5YN3J4	PyreneJ4	B[a]Pyr
CH2CO	H2C4O	Indene	A2R5YN4J3	A2C6H5-2	BePyren
HCCOH	C6H	C6H3O3	A2R5YN4J5	Flthn	Perylen
N-C3H2	C6H2	A2-T1	A2R5YN5J4	A3R5	B[b]Fluor

Note: A: benzene ring; Anthan: anthanthracene; B: benzo; Benz: benzyl; COR: corannulene; COR: corannulene radical; COR1: cylcopenta[cd]corannulene.; CpbPer: cylcopentabenzo[ghi]perylene; Cp: cycropenta; -CY: cycro; Dc, Dcp: dicylopenta; Dcpcdfg: dicylopenta [cd,fg] pyrene; F,Fl,Fluor,Flthn: fluoranthene; FR: cyclopenta[cd]fluoranthene; I-:iso; InPyr: indeno[1,2,3cd]pyrene; J: radical; L: linear; N-: normal; NAP: naphthalene; P-: primary; Py,Pyr: pyrene; R₅: five-membered ring; S-: secondary; T: tertiary; VIN: vinyl; YN: ethynyl; 1 ∼ 10: number of atom, or position of substituent or bond.