Fuel Processing Technology 178 (2018) 180–188

Contents lists available at ScienceDirect

Fuel Processing Technology

journal homepage: www.elsevier.com/locate/fuproc

Research article

Experimental study and modeling of heavy tar steam reforming T

a a b a,⁎
Qiang Li , Qian Wang , Anchan Kayamori , Jiansheng Zhang
a
Key Laboratory for Thermal Science and Power Engineering of the Ministry of Education, Department of Thermal Engineering, Tsinghua University, Beijing 100084, China
b
TIGAR PJ Group, Business Development Department, Resources, Energy & Environment Business Area, IHI Corporation, Tokyo, 135-8710 Japan

A R T I C LE I N FO A B S T R A C T

Keywords: In this paper, the elimination of heavy tar by steam reforming at high temperature was studied with experiments
Heavy tar and numerical simulation. The experiments were conducted in a tube reactor with five model compounds at
Steam reforming temperatures of 1273–1673 K, steam of 0–40 vol% and sample weight of 3–20 mg with residence time of 2 s. The
Numerical simulation simulation was performed with the plug flow model in CHEMKIN program based on a kinetic model that consists
CHEMKIN
of > 200 chemical species and 2000 elementary step-like reactions. The results of experiments indicate that
Tube reactor
increasing temperature will accelerate the decomposition of heavy tar; moisture could significantly prompt the
decomposition of heavy tars, but not so efficient above 20 vol%; less sample could help the decomposition of
heavy tar; longer residence time can slightly increase the decomposition of heavy tar. At high temperature
(1573 K or above), the heavy tar is mainly cracked into two parts: the one carbon molecules (CO and CO2) and
the soot at the beginning of reactor. Then the soot decomposes into CO, CO2 and H2 by steam reforming. The
conversion and kinetic data of heavy tar can be divided by two temperature sections. The carbon conversion
degrees of different model compounds from 1173 K to1473 K are close with similar trends, which may be re-
presented by a universal kinetic model. Our results provide an insight in the understanding of the heavy tar
decomposition and give the necessary information for the designation and operation of the tar cracker to
eliminate heavy tar.

1. Introduction The main reactions involved in these processes on the model com-
pounds are as follows:
At present, the circulating and bubbling fluidized bed technology Cracking:pCn Hx → qCm Hy + rH2 (1)
has been successfully used in the gasification of coal and biomass with a
high production capacity [1]. However, the major problem of biomass Steam reforming:Cn Hx + mH2 O → (m + (x /2)) H2 + nCO (2)
and coal gasification is the presence of tar in the produced that blocks
the pipeline and causes corrosion, erosion and abrasion of equipment. Methane formation:CO + 3H2 ↔ CH4 + H2 O (3)
Therefore, it is necessary to remove tar contents in the product gas to an Carbon formation:Cn Hx → nC + (x /2) H2 (4)
acceptable value or transform tar into other valuable gas, either in the
gasifier or in the downstream. Water Gas shift: CO + H2 O ↔ CO2 + H2 (5)
Tar is a complex mixture of single-ring to 5-ring aromatic com-
Carbon/soot‐steam reaction: C + H2 O ↔ CO + H2 (6)
pounds with other oxygen-contained hydrocarbons and complex poly-
aromatic hydrocarbons (PAH) [2,3]. There are many techniques to re- The syngas from fluidized bed gasifier contains both heavy tar and
move or transform tar [4,5]. Many experiments have been conducted steam [15–17]. If adding a tar cracker at the outlet of the gasifier to
on the thermal and catalytic cracking of primary tars and secondary increase the temperature of syngas, the steam reforming will occur and
tars, including naphthalene a tertiary tar, based on model components, eliminate heavy tar into H2, CO and CO2. To design the tar cracker and
such as phenol [6], toluene [7–11], naphthalene [12–14]. Also, several reduce the heavy tar in the syngas, it is essential to study the process
numerical simulations on tar reforming with light model compounds, and mechanism of heavy tar steam reforming at high temperature.
such as benzene and toluene, have been performed with reaction ki- Though many researchers have studies the steam reforming of light tar,
netic models, and some simple reaction schemes have been built up less data and studies are found on the heavy tar and at high temperature
[1,2,4,9]. (> 1273 K), especially for the compounds more than two benzene

⁎
Corresponding author.
E-mail address: zhang-jsh@tsinghua.edu.cn (J. Zhang).

https://doi.org/10.1016/j.fuproc.2018.05.020
Received 31 January 2018; Received in revised form 12 May 2018; Accepted 12 May 2018
Available online 01 June 2018
0378-3820/ © 2018 Elsevier B.V. All rights reserved.
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

Fig. 1. Schematic graph of experimental setup (solid feeding of PAHs).

rings. Table 1
In this study, naphthalene, phenanthrene, acenaphthylene, fluor- Reaction conditions.a
anthene and pyrene will be used as model compounds to study the Tar model Sample Temperature (K) Steam Residence
steam reforming of heavy tar at high temperature (as high as 1573 K). compounds weight (vol%) time(s)
Naphthalene represents two-ring compounds; acenaphthylene and (mg)
phenanthrene represent three-ring compounds; pyrene and fluor-
Naphthalene ( ) 3–20 (10) 1273–1673 0–40 2–10 (2)
anthene represent four-ring and higher compounds, which are the (1573) (30)
major contents in their class of heavy tars [17–21]. The effects of the Acenaphthylene 3–20 (10) 1273–1673 0–40 2–10 (2)
reactor temperature profile, operating temperature, steam content, ( ) (1573) (30)
sample weight and residence time on the carbon conversion degree of Phenanthrene 3–20 (10) 1273–1673 0–40 2–10 (2)
heavy tar will be investigated in a tube reactor. The steam reforming (1573) (30)
( )
process will be described and studied by both experiments and simu-
lation. The kinetic data of five model compounds will be achieved. Fluoranthene 3–20 (10) 1273–1673 0–40 2–10 (2)
(1573) (30)
( )

Pyrene ( ) 3–20 (10) 1273–1673 0–40 2–10 (2)

2. Experimental (1573) (30)

a
Standard conditions are in brackets; total pressure is around 1.05 atm; Ar is
2.1. Experimental setup
the balance gas.

The newly designed horizontal tube reactor is shown as the elec-

2.2. Evaluation of experimental data
trical furnace in Fig. 1. This reactor is special modified with large
heating area (800 mm in length, 30 mm in diameter) and long iso-
The carbon conversion degree, Xc, of model compound j with carbon
thermal space (250 mm range from 773 to 1673 K). There is a cor-
number NC,j and hydrogen number NH,j is the ratio of one carbon gas
undum sample injector that can be controlled to put sample in any-
product (CO, CO2, CH4) to model compound j, which describes the
where inside heating area of the tube reactor. The temperature
amount of carbon in model compound converted to one carbon gas:
distribution of the tube reactor was measured by a Pt-Rh thermocouple
(HT1270S, Beijing Aerospace Oriental, China) with accuracy class nCO, out + nCO2, out + nCH 4, out
of ± 0.25% × temperature (e.g. ± 2.5 K at 1273 K). Xc (%) = × 100
nj, in ∙NC, j (7)
Naphthalene (99.6%, Xiya Chemical Industry, China), phenanthrene
(97%, Xiya Chemical Industry, China), acenaphthylene (99%, Xiya H2 yield is defined as the percentage of the stoichiometric potential,
Chemical Industry, China), fluoranthene (98%, J&K Scientific, China) which describes the amount of hydrogen and carbon in model com-
and pyrene (98%, J&K Scientific, China) were used as model compound pound converted to hydrogen gas:
representing heavy tars. The reacting gas is a mixture of Ar
(> 99.999%) and steam from steam generator. nH 2, out
YH 2 (%) =
3–20 mg model compound was put on the sample injector and then nj, in ∙ (2NC, j + NH , j /2) (8)
was moved to the beginning (at 20 cm) of the isothermal space at
certain temperature (1073–1673 K) and flow rate (0.5–10 L/min at The reaction rate rj of each model compound can be described by
s.t.p.). The model compound was vaporized immediately and carried by following equation:
the gas flow through the high temperature isothermal area. The gas
r j = kj ∙c jm ∙c Hn2 O (9)
product after condensation and filtration was collected by a gas storage
bag. When the gas bag was cooled to room temperature, the composi-
tion of the product gas was analyzed by gas chromatograph (490 Micro EA, j ⎞
kj = Aj ∙exp ⎛− ⎜ ⎟

GC, Agilent, United States). Reaction conditions of the experiments are ⎝ RTR⎠ (10)
listed in Table 1.
The residence time t can be calculated by volume or length divided
by gas flow rate Q or gas velocity v as:

181
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

1300
(a) (b) Integrated value (L )=34.9 cm
R
1.2
Integrated value (LR)=28.4 cm

Temperature/K
1200

Integral term
0.8
Integrated value (LR)=20.7 cm

1100 0.4
0 L/min
6.36 L/min Ar with 30% moisture
6.36 L/min Ar
Polynomial fitting line
Polynomial fitting line 0.0 0 L/min
6.36 L/min Ar with 30% moisture
Polynomial fitting line 6.36 L/min Ar
1000
20 30 40 50 60 20 30 40 50 60
Length along the tube reactor/cm Length along the tube reactor/cm
Fig. 2. Temperature profile along the length of the tube reactor (a) and the integration part along the length of the tube reactor (b) at 1273 K.

300 300
Naphthalene (a) (b)
250 Pyrene 250
CH4 content/ mol

CO content/ mol
200 200

150 150

100 100

50 50

0 0
1200 1300 1400 1500 1600 1700 1200 1300 1400 1500 1600 1700
Temperature/K Temperature/K
300 1000
(c) (d)
250 800
CO2 content/ mol

H2 content/ mol

200
600
150
400
100

50 200

0 0
1200 1300 1400 1500 1600 1700 1200 1300 1400 1500 1600 1700
Temperature/K Temperature/K
Fig. 3. Content of CH4, CO, CO2 and H2 at different temperature from the steam reforming experiments (30 vol% steam, 10 mg sample, 2 s residence time).

VR LR L EA, j ⎛ 1 1 ⎞⎤
t=
Q (Tinlet , pinlet )
=
v (Tinlet , pinlet ) LR = ∫0 exp ⎡ ∙ ⎜ − dL ⎟
(11) ⎢ R ⎝ Tx (L x ) TR ⎠ ⎥ (13)
⎣ ⎦
The reference temperature TR is the setting temperature of the tube
reactor. However, the temperature along the tube reactor Tx(L) is not 2.3. Simulation
isothermal as shown in Fig. 2(a). Therefore, the LR is defined as the
effective reaction length that equals to the length at reference tem- The simulations were performed with the plug-flow reactor (PFR)
perature TR, which has an equivalent reaction for the reaction at re- model by employing the CHEMKIN program. The kinetic model to si-
ference temperature rj(TR). The effective reaction length LR can be de- mulate the steam reforming of heavy tars was based on a reaction
fined as: mechanism for hydrocarbon combustion and polycyclic aromatic hy-
drocarbon growth, which was developed by Richter and Howard [23].
L LR
∫0 r j (Tx ) dL = ∫0 r j (TR ) dLR (12)
The mechanism and the thermodynamic data can be found in the re-
ference [24]. This reaction mechanism consisted of 2216 reactions,
The reaction rate rj depends on the concentration of model com- including 257 chemical species from the smallest species (hydrogen
pound and steam, and also on the rate constant kj. At the same reaction radicals) to the largest molecule (coronene). This mechanism success-
concentration, LR can be calculated from the obtained temperature fully predicted the characteristics of partial oxidation of gas emitted
profile by combining Eqs. (9), (10) and (12): [22] from metallurgical coke ovens and the thermal reactions of aromatic
hydrocarbons from pyrolysis of solid fuels [22,25].
The reaction conditions required for the simulations were

182
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

Carbon conversion degree/%

50 50
Naphthalene (a) (b)
Pyrene
40 40

Hydrogen yield/%
30 30

20 20

10 10

0 0
1200 1300 1400 1500 1600 1700 1200 1300 1400 1500 1600 1700
Temperature/K Temperature/K
50 50
(c) (d)
Carbon conversion degree/

40 40

Hydrogen yield/
30 30

20 20

10 10

0 0
0 10 20 30 40 0 10 20 30 40
Moisture/vol% Moisture/vol%
80 80
Carbon conversion degree/

(e) (f)
60 60
Hydrogen yield/

40 40

20 20

0 0
0 5 10 15 20 0 5 10 15 20
Sample weight/mg Sample weight/mg
50 50
Carbon conversion degree/%

(g) (h)
40 40
Hydrogen yield/%

30 30

20 20

10 10

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Residence time/s Residence time/s
Fig. 4. Carbon conversion degree and hydrogen yield of naphthalene and pyrene at different temperature, moisture content, sample weight and residence time
(default condition: 30% steam, 1573 K, 10 mg sample weight, 2 s residence time).

determined from the experimental conditions as shown in Table 1. 3. Results and discussion
When doing simulation, the input data include some experimental data
to narrow the deviation. The major data used are the evaporation time 3.1. Influences of the temperature profile along the reactor length
of the sample and the temperature profile of the tube reactor. The
temperature profile along the tube reactor was given from a polynomial Fig. 2(a) shows the temperature profile at different flow rate and
function fitted to measured temperature profiles, as demonstrated in moisture content, in which the quasi-isothermal temperature (TR, the
Fig. 2(a). reference temperature) is 1273 K. The Tx(L), namely the temperature
profile function along the length of tube reactor, can be fitted by
polynomial function as shown the line in Fig. 2(a). Putting the Tx(L)

183
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

Carbon conversion degree/% Carbon conversion degree/%

Carbon conversion degree/% Carbon conversion degree/%
50 50
simulation 凚 a凛 凚 b凛
40 experimental 40

30 30

20 20

10 10

0 0
1200 1300 1400 1500 1600 1700 0 10 20 30 40 50
Temperature/K Moisture/vol%
80 50
凚 c凛 凚 d凛
70
40
60
50 30
40
30 20
20
10
10
0 0
0 5 10 15 20 25 0 4 8 12
Sample weight/mg Residence time/s

Fig. 5. Comparison of experimental and simulation data.

function into Eq. (13), the integral term of Eq. (13) can be drawn in a shown in Fig. 4(d). The hydrogen yield of naphthalene increases all the
function of length L as shown in Fig. 2(b). By integrating the line in way along with the rising of steam content.
Fig. 2(b), the effective reaction length (LR) can be obtained. In Fig. 2(a), The sample weight has an important influence on the elimination of
the gas flow can reduce the highest temperature and push temperature heavy tar, as the carbon conversion of both naphthalene and pyrene
profile toward outlet, which will reduce the effective reaction length in decreases with the increase of sample weight and then reaches a plateau
Fig. 2(b). This is due to the gas in the flow absorbs the heat near the at 20 mg, as shown in Fig. 4(e). The hydrogen yield shares the similar
inlet, and then releases the heat near the outlet. By adjusting the flow trends with the carbon conversion degree. In the tube reactor, the gas
rate, a precise residence time, calculated by Eq. (11) can be gained. flow rate (5.0 L/min) and steam volume percent (30 vol%) in the gas
flow are constant. The higher sample weight means a larger sample
concentration in the reactor, but the steam content is limited. From the
3.2. Optimization of heavy tar steam reform—the effects of temperature, thermodynamic aspect, with the consumption of steam and the high
moisture, weight and residence time sample concentration, the equilibrium of the reaction will inevitably
toward the reactants, which reduces the carbon conversion degree and
The effects of temperature on the yields of CH4, CO, CO2 and H2 in hydrogen yield.
the steam reforming were shown in Fig. 3. With the increase of tem- Longer residence time will help the decomposition of both naph-
perature, the yield of methane increases first, then decreases to zero thalene and pyrene, as shown in Fig. 4(g). Pyrene is more sensitive with
with peaks at around 1550 K. The total content of methane is much less the residence time, as at the residence time of 10 s, the carbon con-
than CO, CO2 and H2. With the raise of temperature, the content of CO version degree of pyrene increases > 15%, compared with the data at
has no apparent trends, which is influenced with balance of water-gas the residence time of 2 s. In Fig. 4(h), there are no apparent trends for
shift reaction (Eq. (5)), steam reform reaction (Eq. (2)) and other re- the hydrogen yields, which is affected by the equilibrium of water gas
actions. The content of CO2 from naphthalene and pyrene increase shift reaction and other reactions.
steadily then slows down at high temperature. The content of H2 in-
creases with the rising of temperature in whole temperature range.
Generally, increasing temperature can prompt the decomposition of 3.3. Process of heavy tar steam reforming—the soot generation and
heavy tar. decomposition from simulation
The carbon conversion degree, which evaluates the efficiency of tar
elimination by steam reform, is presented in Fig. 4. The hydrogen yield The comparison of experimental and simulation data for carbon
is expressed as the stoichiometric potential of heavy tar converting into conversion degree are presented in Fig. 5. The simulations of carbon
hydrogen. In Fig. 4(a), with the increase of temperature, the carbon conversion degree in the change of temperature, moisture content,
conversion degree of both naphthalene and pyrene increases. At high sample weight and residence time are close to the experimental data,
temperature, such as 1673 K, more naphthalene is converted to small which ensures a reasonable basis for the analysis from data of simula-
molecules (CO, CO2, CH4), compared with pyrene. In Fig. 4(b), the tion.
hydrogen yield has the same trend of the hydrogen content in the Fig. 3, Fig. 6 shows the decomposition of each model compound along the
which increases with the rising of temperature. length of tube reactor at 1373 K and 1573 K from simulation with major
The steam can help eliminate the heavy tar significantly as the components in their carbon mole fraction (components less than 2% are
carbon conversion degree increase at 20 vol% steam, as shown in not drafted in the figure). The model compounds finish the decom-
Fig. 4(c). Naphthalene is more sensitive to the steam than pyrene. position in the beginning of the tube reactor at high temperature
However, when the steam composition above 20 vol%, it only has slight (1573 K), while at lower temperature (1373 K) the model compounds
effect on elimination of heavy tar. The hydrogen yield of pyrene in- decompose along the whole length of the tube reactor with a lower
creases slow at beginning and then accelerates at high steam content, as decomposition rate. With further analysis, the reaction products from

184
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

100 100 naphthalene

(a) naphthalene (b)

Mole fraction/C%

Mole fraction/C%
CO2 benzo[k]fluoranthene
80 indene 80 CO2
phenanthrene CO
perylene coronene
60 benzo[k]fluoranthene 60

40 40

20 20

0 0
20 40 60 20 40 60
Distance/cm (1373 K) Distance/cm (1573 K)
100 100 phenanthrene
(d)
Mole fraction/C%

Mole fraction/C%
(c) pyrene
80 80 cyclopenta[cd]pyrene
phenanthrene CO2
anthracene CO
60 C10H8 60 coronene
indene
flouranthene
40 pyrene 40

20 20

0 0
20 40 60 20 40 60
Distance/cm (1373 K) Distance/cm (1573 K)
100 fluoranthene
100 fluoranthene
(e) (f)
Mole fraction/C%

Mole fraction/C%

aceanthrylene CO
80 acephenanthrylene 80 pyrene
cyclopenta[cd]pyrene
CO2
60 60 acephenanthrylene
aceanthrylene
40 40

20 20

0 0
20 40 60 20 40 60 80
Distance/cm (1373 K) Distance/cm (1573 K)
100 pyrene
100 pyrene
(g) (h)
Mole fraction/C%

Mole fraction/C%

CO2 CO2
80 phenanthrene 80 cyclopenta[cd]pyrene
anthracene CO
CO coronene
60 60

40 40

20 20

0 0
20 40 60 20 40 60
Distance/cm (1373 K) Distance/cm (1573 K)
Fig. 6. Decomposition of model compounds along the length of tube reactor: (a), (b) decomposition of naphthalene at 1373 K and 1573 K; (c), (d) decomposition of
phenathrene at 1373 K and 1573 K; (e), (f) decomposition of fluranthene at 1373 K and 1573 K; (g), (h) decomposition of pyrene at 1373 K and 1573 K (C% means the
mole fraction is in carbon percent).

the decomposition of model compounds are quite different between the reactor. Then the soot is steam reformed to one carbon molecules. In
temperature of 1373 K and 1573 K. this paper, the soot is defined as the aromatic hydrocarbon with carbon
For the decomposition of naphthalene in Fig. 6(b), at high tem- number > 16, including soot precursor, as the largest molecule of the
perature (1573 K) naphthalene mainly decomposed to one carbon kinetic model applied in CHEMKIN is coronene with 26 carbon num-
models (CO and CO2) and soot (benzo[k]fluoranthene and coronene). bers. In Fig. 6(a), at lower temperature (1373 K), the naphthalene de-
The conversion of naphthalene is very fast, which completed at the first composes relative slow with large portion converted to indene in the
5 cm in the length of the tube reactor and the benzo[k]fluoranthene is balance of temperature. Naphthalene is close to indene in their carbon
generated at the very beginning with intermediate products, and then number. Only a small portion are converted to one carbon molecules
the coronene is generated at first 10 cm in the length of the tube and soot.

185
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

Naphthalene (a) Naphthalene (b)

100 Phenanthrene 100 Phenanthrene
Fluoranthene Fluoranthene

decomposition rate/%
Conversion rate/%
Pyrene Pyrene
80 80

60 60

40 40

20 20

0 0

1000 1200 1400 1600 1800 1000 1200 1400 1600 1800
Temperature/K Temperature/K
Fig. 7. Difference between (a) conversion and (b) decomposition (solid line is only for showing the trend).

this paper, pyrene is a kind of soot precursor or soot, and is relative

stable at high temperature, as around 20% remains at the end of the
tube reactor. In Fig. 6(g), at lower temperature (1373 K), pyrene de-
composes relative slow with large portion converted to phenanthrene,
which close to pyrene in their carbon number. Only a small portion are
converted to one carbon molecules and soot.
From the analysis of the model compounds above, at high tem-
perature (1573 K), model compounds are mainly converted to one
carbon models (CO and CO2) and soot. At lower temperature (1373 K),
there is a different conversion style, as the model compounds are more
like to convert to similar aromatic hydrocarbons, rather than decom-
posed to one carbon molecules. For example, in Fig. 6(e), fluoranthene
(C16H10) is converted to aceanthrylene (C16H10) and acephenanthrylene
(C16H10) at 1373 K, rather than decomposed to one carbon molecules
and soot. Therefore, it is necessary to make a distinction between them.
In this paper, we define that the change of model compounds to similar
aromatic hydrocarbon as conversion (e.g. at 1373 K, fluoranthene
C16H10 is mainly converted to aceanthrylene C16H10 and acephenan-
thrylene C16H10). We define the change of model compounds to one
Fig. 8. Carbon conversion degree of five model compounds and a mixture
carbon molecules and soot as decomposition (e.g. at 1573 K. fluor-
(50 wt% of phenanthrene and 50 wt% of fluoranthene; solid line is only for
anthene C16H10 is mainly converted to CO, CO2 and soot).
showing the trend).
Fig. 7 shows the difference between conversion and decomposition
of model compounds from the experimental data. Model compounds
For the decomposition of phenanthrene in Fig. 6(d), at high tem- are easy to convert other compounds at relative lower temperature, but
perature (1573 K) phenanthrene mainly decomposed to one carbon hard to be decomposed, which needs a higher temperature. The stabi-
models (CO and CO2) and soot (pyrene, cyclopenta[k]pyrene and cor- lity rate is pyrene (C16H10) or fluoranthene (C16H10) > phenanthrene
onene). The conversion of naphthalene is very fast, which happened at (C14H10) > naphthalene (C10H8).
the first 10 cm in the length of the tube reactor and the soot is generated
with the decomposition of phenanthrene. In the meanwhile, the soot
and other hydrocarbon decompose to one carbon molecules. In 3.4. Kinetics of model compounds—two temperature sections with different
Fig. 6(c), at lower temperature (1373 K), phenanthrene decomposes kinetics
relative slow with only 20% converted to similar aromatic hydrocarbon
(naphthalene, indene and anthracene), with close carbon number. The kinetic model and data is gained by treating the whole steam
For the decomposition of fluoranthene in Fig. 6(f), at high tem- reforming reactions as a general and empirical process. The steam re-
perature (1573 K) phenanthrene mainly decomposed to one carbon forming process is defined by the decomposition of model compounds
models (CO and CO2) and soot (pyrene and cyclopenta[k]pyrene). to one carbon molecules (CO, CO2, CH4), namely the carbon conversion
However, at the very beginning of the reaction, fluoranthene is con- degree, not simply the conversion of model compounds. Fig. 8 shows
verted to aceanthrylene and acephenanthrylene immediately with an the carbon conversion degree of five model compounds and a mixture
equilibrium. This is equilibrium is shown very clearly at lower tem- (50 wt% phenanthrene mixed with 50 wt% fluoranthene) with the in-
perature (1373 K) in Fig. 6(e). There is little decomposition in the whole creasing of temperature at standard condition showed in Table 1. The
range of the tube reactor, but only converted to aceanthrylene and carbon conversion degrees of different model compounds from 1173 K
acephenanthrylene in the equilibrium of temperature. In this way, the to1473 K are close with similar trends. But at high temperature above
change of fluoranthene may not be defined as decomposition, and it 1473 K, the trends and values are quite different. For example, phe-
should be probably defined as conversion, which will be discussed more nanthrene is much easier to decompose than pyrene at high tempera-
in Fig. 7. ture. At high temperature sections, the mechanisms of the process are
For the decomposition of pyrene in Fig. 6(h), at high temperature probably different and possibly controlled by gas diffusion. When build
(1573 K) pyrene mainly decomposed to one carbon models (CO and the kinetic model, the difference trends of carbon conversion degree at
CO2) and soot (cyclopenta[k]pyrene and coronene). In the definition of different temperature section has to be considered.
From Fig. 4, when the moisture content in the gas flow above 20%,

186
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

0 0
(a) Naphthalene (b) Acenaphthylene
linear fitting line linear fitting line
-1 -1

-2 -2

ln (kj)

ln (kj)
-3 -3

-4 -4

-5 -5
0.0006 0.0007 0.0008 0.0009 0.0006 0.0007 0.0008 0.0009
1/T (1/K) 1/T (1/K)
0 0
(c) Phenanthrene (d) Fluoranthene
linear fitting line linear fitting line
-1 -1

-2 -2
ln (kj)

ln (kj)
-3 -3

-4 -4

-5 -5
0.0006 0.0007 0.0008 0.0009 0.0006 0.0007 0.0008 0.0009
1/T (1/K) 1/T (1/K)
0 0
(e) Pyrene (f) Mix
linear fitting line linear fitting line
-1 -1

-2 -2
ln (kj)

ln (kj)

-3 -3

-4 -4

-5 -5
0.0006 0.0007 0.0008 0.0009 0.0006 0.0007 0.0008 0.0009
1/T (1/K) 1/T (1/K)
Fig. 9. Kinetic data fitting of model compounds (Mix: 50 wt% phenanthrene mixed with 50 wt% fluoranthene).

Table 2 Fig. 8 and the value of t is the residence time. The kinetic data can be
Kinetic data of low temperature section. acquired by combining Eqs. (10) and (14).
Model compound Pre-exponential factor Activation energy (KJ mol−1)
Fig. 9 shows the kinetic data fitting of model compounds and the
fitting line is derived from Eqs. (10) and (14). In a large temperature
Naphthalene 4558.4 125.5 range, the points are not on one line, namely can't use one Arrhenius
Acenaphthylene 1013.7 105.3 equation to fitting. Similar to Fig. 8, data can be divided into two
Phenanthrene 36,930.8 143.5
temperature sections, namely the low temperature section (1173 K to
Fluoranthene 5231.0 124.0
Pyrene 9108.5 130.2 1473 K) and high temperature section (above 1473 K), then fitted by
Mix 24,523.8 140.5 Arrhenius equation. The mechanisms in different temperature sections
are probably different. For the design of tar cracker, the kinetic data of
high temperature is better than the one of whole temperature range.
the moisture has not much influence on the carbon conversion degree. Gained by the equation of the fitting line, the kinetic data of each
Therefore, the value of n in Eq. (9) can be assumed as zero. When as- model compound at two temperature sections are listed in Tables 2 and
suming m = 1, then the value of kj in Eq. (9) can be calculated by: 3. The activation energy at low temperature section is similar, also the
fitting lines of carbon conversion degree are similar. The decomposition
−ln(1 − Xc ) of heavy tar of different model compounds at this temperature rage is
kj =
t (14)
not highly depends on its components, which may be represented by a
universal model. The activation energy at high temperature is relative
The value of Xc, the carbon conversion degree, can be gained from

187
Q. Li et al. Fuel Processing Technology 178 (2018) 180–188

Table 3 [7] R. Zhang, Y. Wang, R.C. Brown, Steam reforming of tar compounds over Ni/olivine
Kinetic data of high temperature section. catalysts doped with CeO2, Energy Convers. Manag. 48 (1) (2007) 68–77.
[8] K. Narayanaswamy, G. Blanquart, H. Pitsch, A consistent chemical mechanism for
Model compound Pre-exponential factor Activation energy (KJ mol−1) oxidation of substituted aromatic species, Combust. Flame 157 (10) (2010)
1879–1898.
Naphthalene 5.3 42.0 [9] A. Fourcault, F. Marias, U. Michon, Modelling of thermal removal of tars in a high
temperature stage fed by a plasma torch, Biomass Bioenergy 34 (9) (2010)
Acenaphthylene 1.5 24.1
1363–1374.
Phenanthrene 81.2 73.8 [10] B. Zhao, X. Zhang, L. Chen, R. Qu, G. Meng, X. Yi, L. Sun, Steam reforming of
Fluoranthene 3.0 33.5 toluene as model compound of biomass pyrolysis tar for hydrogen, Biomass
Pyrene 0.8 20.8 Bioenergy 34 (1) (2010) 140–144.
Mix 10.3 47.1 [11] D. Swierczynski, C. Courson, A. Kiennemann, Study of steam reforming of toluene
used as model compound of tar produced by biomass gasification, Chem. Eng.
Process. 47 (3) (2008) 508–513.
[12] T. Furusawa, A. Tsutsumi, Development of cobalt catalysts for the steam reforming
small, and the reaction rate is not so sensitive with temperature. It is of naphthalene as a model compound of tar derived from biomass gasification,
possible that with the increasing of temperature, the total process may Appl. Catal. A Gen. 278 (2) (2005) 195–205.
[13] D.N. Bangala, N. Abatzoglou, E. Chornet, Steam reforming of naphthalene on Ni-Cr/
change from chemical reaction control to diffusion control, which Al2O3 catalysts doped with MgO, TiO2, and La2O3, AICHE J. 44 (4) (1998) 927–936.
makes the kinetic data into two sections. [14] P. Morf, P. Hasler, T. Nussbaumer, Mechanisms and kinetics of homogeneous sec-
ondary reactions of tar from continuous pyrolysis of wood chips, Fuel 81 (7) (2002)
843–853.
4. Conclusion [15] C. Font Palma, Modelling of tar formation and evolution for biomass gasification: a
review, Appl. Energy 111 (2013) 129–141.
[16] C. Li, K. Suzuki, Tar property, analysis, reforming mechanism and model for bio-
The experiments and simulation of heavy tar steam reforming were mass gasification—an overview, Renew. Sust. Energ. Rev. 13 (3) (2009) 594–604.
successfully conducted. The main results obtained are summarized as [17] G. Taralas, M.G. Kontominas, X. Kakatsios, Modeling the thermal destruction of
follows: toluene (C7H8) as tar-related species for fuel gas cleanup, Energy Fuel 17 (2) (2003)
329–337.
[18] S.A. Wise, D.L. Poster, S.D. Leigh, C.A. Rimmer, S. Mossner, P. Schubert,
(1) To optimize the reforming of heavy tar: increasing temperature will L.C. Sander, M.M. Schantz, Polycyclic aromatic hydrocarbons (PAHs) in a coal tar
accelerate the decomposition of heavy tar; moisture could sig- standard reference material–SRM 1597a updated, Anal. Bioanal. Chem. 398 (2)
(2010) 717–728.
nificantly prompt the decomposition of heavy tars, but not so effi- [19] C. Brage, Q.Z. Yu, G.X. Chen, K. Sjostrom, Use of amino phase adsorbent for biomass
cient above 20 vol%; less sample could help the decomposition of tar sampling and separation, Fuel 76 (2) (1997) 137–142.
[20] E. Kurkela, P. Ståhlberg, P. Simell, J. Leppälahti, Updraft gasification of peat and
heavy tar; longer residence time can slightly increase the decom- biomass, Biomass 19 (1–2) (1989) 37–46.
position of heavy tar. [21] J. Gil, M.P. Aznar, M.A. Caballero, E. Frances, J. Corella, Biomass gasification in
(2) At high temperature (1573 K or above), the heavy tar is mainly fluidized bed at pilot scale with steam-oxygen mixtures. Product distribution for
very different operating conditions, Energy Fuel 11 (6) (1997) 1109–1118.
cracked into two parts: the one carbon molecules (CH4, CO and [22] A. Jess, Mechanisms and kinetics of thermal reactions of aromatic hydrocarbons
CO2) and the soot. It is critical to convert soot into one carbon from pyrolysis of solid fuels, Fuel 75 (12) (1996) 1441–1448.
[23] H. Richter, J.B. Howard, Formation and consumption of single-ring aromatic hy-
molecules. The decomposition of heavy tar is completed at begin-
drocarbons and their precursors in premixed acetylene, ethylene and benzene
ning of the tube reactor, but soot just begins to decompose into CO, flames, Phys. Chem. Chem. Phys. 4 (11) (2002) 2038–2055.
CO2 and H2 by steam reforming. [24] K. Norinaga, Y. Sakurai, R. Sato, J. Hayashi, Numerical simulation of thermal
conversion of aromatic hydrocarbons in the presence of hydrogen and steam using a
(3) The conversion and kinetic data of heavy tar can be divided by two detailed chemical kinetic model, Chem. Eng. J. 178 (2011) 282–290.
temperature sections. The carbon conversion degrees of different [25] K. Norinaga, H. Yatabe, M. Matsuoka, J. Hayashi, Application of an existing de-
model compounds from 1173 K to1473 K are close with similar tailed chemical kinetic model to a practical system of hot coke oven gas reforming
by noncatalytic partial oxidation, Ind. Eng. Chem. Res. 49 (21) (2010)
trends, which may be represented by a universal kinetic model. The 10565–10571.
activation energy at high temperature is relative small, and the
reaction rate is not so sensitive with temperature. Nomenclature

The results of this paper have demonstrated that the steam re- Aj: pre-exponential factor
forming is an effective method to eliminate heavy tar. The experimental c: concentration (mol m−3)
data and simulation results provide an insight in the understanding of EA, j: activation energy (J mol−1)
kj: rate constant of decomposition of the aromatic hydrocarbon
the heavy tar decomposition and give the necessary information for the LR: effective reaction length (m)
three-dimension simulation, designation and operation of the tar Lx: reaction length (m)
cracker. n: mole number (mol)
NC,j: model compound j with carbon number N
NH,j: model compound j with hydrogen number N
Acknowledgements p: pressure (Pa)
Q: flux (mol s−1)
rj: reaction rate (mol m−3 s−1)
This work is financially supported by IHI Corporation. R: universal gas constant (8.314 J mol−1 K−1)
t: residence time (s)
References TR: reference temperature (K)
Tx: temperature along the tube reactor (K)
v: velocity of gas (m s−1)
[1] Y. Fang, J. Huang, Y. Wang, B. Zhang, Experiment and mathematical modeling of a VR: reference volume (m3)
bench-scale circulating fluidized bed gasifier, Fuel Process. Technol. 69 (1) (2001) Xc: carbon conversion degree (%)
29–44. YH2: H2 yield (%)
[2] L. Devi, K.J. Ptasinski, F.J.J.G. Janssen, A review of the primary measures for tar
elimination in biomass gasification processes, Biomass Bioenergy 24 (2) (2003)
125–140. Subscripts
[3] T.A. Milne, N. Abatzoglou, R.J. Evans, Biomass gasifier tars: their nature, formation,
and conversion, National Renewable Energy Laboratory Technical Report, NREL/
TP_570_25357, (1998). c or C: carbon
[4] S. Anis, Z.A. Zainal, Tar reduction in biomass producer gas via mechanical, catalytic H: hydrogen
and thermal methods: a review, Renew. Sust. Energ. Rev. 15 (5) (2011) 2355–2377. in: at inlet
[5] A. Paethanom, S. Nakahara, M. Kobayashi, P. Prawisudha, K. Yoshikawa, j: aromatic hydrocarbon
Performance of tar removal by absorption and adsorption for biomass gasification, out: at outlet
Fuel Process. Technol. 104 (2012) 144–154. R: reference
[6] K. Polychronopoulou, J.L.G. Fierro, A.M. Efstathiou, The phenol steam reforming
reaction over MgO-based supported Rh catalysts, J. Catal. 228 (2) (2004) 417–432.

188