Fuel Processing Technology 64 2000.

253269
www.elsevier.comrlocaterfuproc

A process simulation of the NEDOL coal

liquefaction process
Masaki Onozaki a , Yasuki Namiki a , Hirohito Ishibashi a ,
Masatoshi Kobayashi a , Hidenobu Itoh b, Masataka Hiraide b,
Shigeharu Morooka c,)
a

Nippon Coal Oil Co. Ltd., KS Bldg. 2, Sanban-cho, Chiyoda-ku, Tokyo, 102-0075, Japan
Mitsui Engineering and Shipbuilding Co. Ltd., ST Nishikasai Bld., 8-4-6, Nishikasai, Edogawa-ku,
Tokyo, 104-8489, Japan
c
Department of Materials Physics and Chemistry, Kyushu Uniersity, Fukuoka 812-8581, Japan
Accepted 1 January 2000

Abstract
A coal liquefaction pilot plant of the NEDOL process, supported by the New Energy and
Industrial Technology Development Organization NEDO., was in successful operation for a total
of 269 days at Kashima, Japan. The liquefaction section consisted of three reactors, 1 m in
diameter, connected in series, and processed 150 tons of coal per day. Product yields were
modeled assuming 13 reaction paths, and a steady-state simulator was developed. The performance of the entire process, including distillation and solvent hydrogenation sections, was further
incorporated into the simulation. The design procedure for the liquefaction reactors, which would
possess a diameter of 4 m and process 2500 tons of coal per day, was discussed in sight of the
simulator. q 2000 Elsevier Science B.V. All rights reserved.
Keywords: Coal liquefaction; Reactor; Bubble column; Scale-up

1. Introduction
A coal liquefaction pilot plant, having a capacity of 150 tons of coal per day, has
been successfully operated for the period 19961998 at Kashima, Japan. The pilot plant
)

Corresponding author. Tel.: q81-92-642-3551; fax: q81-92-651-5605.

E-mail address: smorotcf@mbox.nc.kyushu-u.ac.jp S. Morooka..

0378-3820r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 8 - 3 8 2 0 0 0 . 0 0 0 6 9 - 2

254

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

consists of four sections, a coal slurry preparation section including the pulverizing and
drying of coal and mixing with solvent, a liquefaction section including a preheater, a
high-pressure vaporliquid separation, a distillation section with atmospheric and vacuum towers, and a solvent hydrogenation section which is comprised of a solvent
hydrogenation reactor with six NirMo catalyst beds w1,2x. The reactor system of the
liquefaction section consists of three bubble column reactors in series. The goals of this
project were to obtain engineering data, as well as yield data, for the future design of
liquefaction reactors. The design principle of the NEDOL process involves the use of
a fine pyrite catalyst and the hydrogenation of recycle oil at a connected plant, in order
to attain a high oil yield 58 wt.% of the dry and ash-free daf. coal. under mild reaction
conditions pressures 16.818.8 MPa; temperatures 723735 K..
Large-scale liquefaction reactors, which would be constructed in the future, have
been conceptually designed w35x. The proposed EDS process involves a plug flow-type
reactor of 2.9 m in diameter and 47.5 m in height w3x, having a capacity of 6300 tons of
Illinois a6 coal per day. The proposed SRC-II process involves a backmix-type reactor
of 3.35 m in diameter and 20.4 m in height with a capacity of 1500 tons of coal per day
w4x. However, differences in reactor designs and operation conditions exist, as a result of
the philosophy of the processes.
The objectives of the present study are to analyze the performance in the Kashima
pilot plant using Tanitoharum coal, and to develop a design procedure for a large-scale
plant based on the NEDOL process. Careful attention was paid for the design of the
liquefaction reactors, which would have a capacity of 2500 tons of coal per day. The
process design simulator used was validated via the use of reaction kinetics data and
vaporliquid equilibria, which were obtained at the Kashima pilot plant.

2. Pilot plant reactors at Kashima

Table 1 shows the properties of Tanitoharum coal, which was processed in the pilot
plant in Kashima, Japan. The feed was prepared using powdered coal, recycle oil, finely

Table 1
Properties of Tanitoharum coal
Proximate analysis, dry coal basis, wt.%
Volatile matter
Fixed carbon
Ash
Moisture in feed coal

47.0
48.0
5.0
16.2

Ultimate analysis, daf coal basis, wt.%

C
H
N
S
O difference.

76.9
5.8
1.9
0.15
15.25

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

255

divided pyrite powder average diameters 0.7 mm., and hydrogen-rich recycle gas, and
was heated to 660690 K with a slurry heat exchanger and a fired heater. The mixture
was then introduced into the first reactor through a 107-mm i.d. upward nozzle. At the
Kashima pilot plant, three reactors of the same size were connected in series. The outlet
nozzle of the first reactor was conducted to the inlet nozzle of the next reactor, and the
second one to the third one. Fig. 1 shows the dimensions of the reactor. The axial
position was measured from the height of the boundary between the cylindrical and
conical portions at the bottom of the reactor. This line is referred to as the bottom
tangential line. The distance between the bottom and top tangential lines was 11.0 m.
The effective reactor height was 11.8 m, as calculated by considering the volume of the
spherical cap section at the top of the reactor and the conical section at the bottom.
Several thermocouples and pressure taps were installed at different heights in order to
record the radial and axial temperature distributions and pressure-difference in each
reactor. Two downward injection nozzles for the quench gas were installed at different
heights, and an additional nozzle was installed to inject cold oil in case of an emergency,
but was never used.

Fig. 1. Dimensions of the liquefaction reactor at the Kashima pilot plant.

256

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

Table 2 shows the typical operating conditions of Tanitoharum coal, as well as the
data obtained on the pilot plant. The recycle gas was fed into the reactors as the feed
component and the quench gas. The compositions of the recycle gas were slightly
different for case 1 and case 2. The slurry was a mixture of coal, recycle oil, and catalyst
powder. The flow rate of the recycle gas is given as the volume STP. per unit time and
is denoted as Gr . The flow rate of the slurry is given as the mass per unit time and is
denoted as Lf . In the pilot plant, the temperature at the top of each reactor was
maintained at a prescribed value by adjusting the flow rate of cold recycle gas quench
gas., which was introduced into each reactor at different heights. In the first reactor, the
temperature at the bottom was 2030 K lower than that at the top, thus minimizing the
amount of quench gas, as shown in Table 2. In the second and third reactors, however,
the temperature at the bottom was nearly equal to that at the top. One reason for the

Table 2
Operating conditions and data for the Kashima pilot plant

Recycle gas
Fraction of hydrogen, vol%
Makeup slurry
Slurry feed rate, kgrh
Coal concentration in slurry,
dry coal basis, wt.%
Catalyst pyrite powder. in slurry,
dry coal basis, wt.%

Case 1

Case 2

86

85

16 200
40

12 700
48.5

Reaction
Operating pressure, MPa
Operating temperature
Feed of the first reactor, K
Bottom of the first reactor, K
Top of the first reactor, K
Bottom of the second reactor, K
Top of the second reactor, K
Bottom of the third reactor, K
Top of the third reactor, K
Gr r Lf ,a m3 STP.rkg-slurry

16.616.8

16.616.8

673
699
728
727
729
726
728
0.71

664
713
733
731
733
731
733
0.90

Yields, daf coal basis, wt.%

Liquefactionb

Hydrogenationc

Liquefactionb

Hydrogenationc

Gas
Water
Oil C 4 to BP 811 K fraction.
Residue
Total
Hydrogen consumption

17.2
10.2
51.0
26.1
104.5
4.5

0.4
0.7
y0.3
0.0
0.8
0.8

20.4
10.0
55.6
18.8
104.8
4.8

0.7
1.3
y0.7
0.0
1.3
1.3

Volumetric flow rate of total recycle gas fed to the three reactors.rmass flow rate of makeup coal
slurry..
b
In the liquefaction section.
c
In the solvent hydrogenation section.

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

257

temperature difference in the first reactor can be attributed to the feed slurry temperature, which was lower than the temperature in the reactor. The feed slurry temperature is
determined from the following criteria:
i. to decrease the heat required in the preheating section, and
ii. to minimize the temperature distribution in the first reactor, where the axial
dispersion is rather limited.
Table 2 also shows product yields and hydrogen consumption in the liquefaction
section and the solvent hydrogenation section. Since the heavy components are converted to lighter oils, gas, and water via the consumption of hydrogen in the hydrogenation section, the final oil yields are lower than those obtained at the liquefaction step.
3. Large-scale reactors
The liquefaction reactors are bubble columns for gas, liquid, and solid systems. The
gas phase consists largely of hydrogen, which is introduced into the reactor as bubbles,
acts as a reactant in the liquefaction process, and causes the mixing of the slurry phase.
The gas phase also contains light hydrocarbons, which are produced by liquefaction.
Coal is disintegrated into small particles in the preheating section, and the fine particles
of coal, ash, and catalyst are homogeneously suspended in the liquid. This slurry phase
resides in the reactors for approximately 1 h. After a lengthy operation, however, coarse
particles are formed and accumulated in the reactors w6x. It is defined in this article that
the coarse particles are not included in the slurry phase, which thus consists of only the
liquid and fine particles.
Two extreme mixing conditions in a bubble column reactor are possible; backmix
flow and plug flow. Large-scale reactors are assigned to the backmix flow. Since coal
liquefaction is highly exothermic, the heat of reaction should be compensated for by
quenching gas or liquid, which is introduced into the reactors. In a backmix flow, the
temperature of the feed slurry rapidly reaches a prescribed value after the slurry is
introduced into the reactor. Thus, the temperature at the outlet of the preheater can be
maintained at a value, which is lower than the temperature in the reactor and is
determined from a balance with the heat production in the first reactor. A reduction in
the preheater outlet temperature is useful to avoid coking, which is possible at
temperatures over 680 K.
On the other hand, a plug flow reactor minimizes the reactor volume. Direct coal
liquefaction involves multi-step conversions of coal to pre-asphaltenes and asphaltenes,
and then to gas, oil, and residue w7x. A plug flow reactor will bring higher yields of the
intermediate components. However, a considerable temperature distribution may be
formed in each reactor. As a result, a quench is often required to control the heat, which
is generated by the liquefaction reaction. A multi-staged reactor system is a compromise
between both flow types, but the optimum number of reactors, which can be connected
in series is strongly dependent on the design concept of the plant.
If liquefaction yields are controlled by reaction rates in a backmix flow or plug flow
reactor, which is maintained at a prescribed temperature, the mean residence time of the

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

258

slurry phase in the reactor is only the design parameter. Thus, the scale-up of the reactor
is performed via each of the following principles:
i. The aspect ratio of the reactor is maintained, and the slurry velocity is increased.
ii. The reactor diameter is increased, and the slurry velocity is maintained.
When the degree of mixing in the reactor is intermediate, however, gas, and slurry
velocities, as well as reactor dimensions, influence the temperature and the reactant
concentrations along the axial positions in the reactor.
The gas holdup in the reactors at the Kashima pilot plant is estimated to be 0.450.60
w8x. A gas holdup of 0.55 is likely to be the maximum for a homogeneous bubble flow
w9x. Thus, the superficial gas velocity of the large-scale reactors cannot be increased
beyond the value of the Kashima pilot plant, if the homogeneous bubble flow regime is
to be maintained in the large-scale reactors, as in the Kashima pilot plant.
The above discussion suggests that the reactor diameter can be increased in accord
with increasing feed rate. Since the ratio of the recycle gas to the feed slurry, GrrLf , is
determined by the required hydrogen supply and the vaporization of product oil, the
sectional area of the reactor will be in proportion to the amount of coal which is
processed. When the total height of the reactor and the superficial gas velocity are
maintained at the approximate values of the Kashima pilot plant, the mass transfer rate
of gaseous hydrogen to the liquid phase will not be very different for the large-scale
reactors. If the large-scale reactors are constructed using the 3Cr1Mo1r4VTiB
alloy ASME code; SA336-F3V., which was developed by the New Energy and
Industrial Technology Development Organization NEDO., and using a stainless steel
347 overlay, the maximum inner diameter of the reactor will be limited to approximately
5 m. This is dictated by currently used manufacturing techniques. The maximum
capacity per each train of reactors 5 m in diameter. is then 25-fold larger than the
Kashima pilot plant and is capable of processing 4000 tons of coal per day. In the
following part of this study, the hydrodynamic and physical properties in the large-scale
reactors with a diameter of 4 m capacitys 2500 tons of coal per day. will be discussed.

4. Fluid dynamics in the large-scale reactors

4.1. Gas holdup
The most important factor affecting the reactor volume is gas holdup, g . It has been
reported that the gas holdup at elevated pressures is several times larger than the data
measured for airwater systems at ambient conditions w9,10x. Onozaki et al. w8x found
that the gas holdup data obtained in the Kashima pilot plant were in agreement with the
correlation proposed by Tarmy et al. w9x for Ug s 0.050.07 mrs.
Ugrg s Ug q Usl . q Ub 1 y g .

1.

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

259

where Ub s 0.09 mrs and m s 0.65. Koide et al. w11x reported that the effect of
superficial gas velocity on gas holdup in a 5.5-m diameter bubble column approximately 4 m in height. was nearly the same as in small-size bubble columns operated
under ambient conditions. Wilkinson et al. w12x also discussed the design parameters of
bubble columns at elevated pressures and concluded that the gas holdup was virtually
independent of column dimension and sparger layout, provided the following criteria
were fulfilled:
i. diameter ) 0.15 m,
ii. ratio of height to diameter aspect ratio. ) 5, and
iii. sparger hole size ) 12 mm.
All these conditions will be satisfied in large-scale liquefaction reactors.
4.2. Flow regime
The Kashima pilot plant was operated without hydrodynamic instability at superficial
gas velocities of 0.050.085 mrs. The axial dispersion coefficients of the liquid slurry.
phase were determined by Sakai et al. w13x by a neutron absorption tracer technique. The
small values observed suggest that a homogeneous bubble flow regime prevailed at
superficial gas velocities less than 0.075 mrs. In the reactors of a brown coal
liquefaction pilot plant in Victoria, Australia w14x, the gas holdup was approximately
0.68 at superficial gas velocities of 0.100.16 mrs. The axial dispersion coefficient of
the liquid phase increased from 0.13 to 0.25 m2rs with increasing gas velocity. These
data suggest that the flow regime in the brown coal liquefaction reactors is assigned to
the homogeneous bubble flow regime with some disturbances. Letzel et al. w15x
investigated the effect of pressure on flow in a bubble column for the nitrogenwater
system at elevated pressures ranging from 0.1 to 0.9 MPa. The homogeneous bubble
flow was changed to the heterogeneous bubbly flow regime at a superficial gas velocity
of 0.080.10 mrs at 0.9 MPa. From the above data, it can be assumed that the
homogeneous bubble flow regime is maintained in large-scale reactors, provided the
superficial gas velocity is less than 0.085 mrs. However, a certain amount of turbulence
can be expected, depending on the physical properties of the gas and slurry, the aspect
ratio of the reactors, and the temperature profile in the reactors.
4.3. Physical properties of the gas and slurry
The physical properties of coal liquids have been carefully determined in the EDS
process w16x as well as the SRC-II process w17x. The results show that the physical
properties of coal liquids are strongly dependent on liquefaction conditions and coal
samples. As described above, a characteristic of the NEDOL process is the downstream
solvent hydrogenation section. The recycle oil, which was equilibrated after continuous
operation of 1000 h at the Kashima pilot plant, was fractionated into 12 narrow cuts by
means of ASTM D 2829. The physical properties, such as boiling point referred to
herein as BP., vapor pressure, heat of vaporization, specific heat capacity, thermal

260

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

Table 3
Properties of the recycle oil
Case 1

Ultimate analysis, wt.%

C
H
N
S
O difference.
Specific graity (288r 277 K)
PONA analysis, wt.%
Parafin
Orefine
Naphthene
Aroma
Unidentified

Case 2

Liquefactiona

Hydrogenationb

Liquefactiona

Hydrogenationb

87.19
9.53
0.97
0.03
2.28

87.40
9.99
0.82
0.02
1.77

87.91
9.29
1.08
0.02
1.70

88.61
10.04
0.61
0.01
0.73

1.004

20.1
0.0
0.4
58.2
21.3

0.991

20.5
0.0
0.5
57.5
21.5

Distillation (gas chromatographic distillation, ASTM D 2887), wt.%

473 to 493 K
2.3
2.3
493 to 533 K
16.6
16.9
533 to 623 K
44.9
45.3
623 to 723 K
32.5
31.9
723 to 811 K
3.7
3.6

1.008

0.983

19.6
0.0
0.4
56.6
23.5

19.8
0.0
0.6
50.4
29.2

2.2
14.6
45.2
32.4
5.6

2.1
15.5
48.2
29.6
4.6

Feed oil to the hydrogenation section, referred to as liquefied recycle oil in Fig. 4.
Feed oil to the coal slurry preparation section after hydrogenation, referred to as hydrogenated recycle oil
in Fig. 4.
b

conductivity, specific gravity and molecular weight, were then determined for each
fraction. These data were stored in the reaction simulator w18x. Table 3 shows the
properties of the liquefied oil, which was fed to the solvent hydrogenation section, and
the recycle oil, which was used to make the coal slurry. Although the composition of the
recycle oil depends on liquefaction and hydrogenation conditions, the fraction of
components having BPs higher than 623 K reached approximately 35 wt.%, and that of
aromatic compounds was higher than 50 wt.%. A part of the liquefied oil several wt.%.
was discharged to the outside along with off-gas, waste water, and residue.

5. Estimation of liquefaction yields in the large-scale reactors

The simulation model, which was developed by Hiraide et al. w18x, can be applied to
predict the yields of the Kashima pilot plant. The following assumptions are made.
i. The plant processes 150 tons of dry coal per day.
ii. The reactor system consists of three backmix-type vessels connected in series.
There are no temperature differences among the gas, liquid, and solid particles in each

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

261

reactor. The continuous stirred tank reactor CSTR. model is adopted in the present
calculation, since the simulator contains a number of parameters, which should be
decided.
iii. The vapor and liquid phases are in equilibrium.
iv. Liquefaction proceeds only in the slurry phase, in which the pyrite catalyst
resides.
v. The volume of the slurry phase in the reactor is calculated from the gas holdup
given by Eq. 1..

The liquefaction consists of 13 reaction steps from coal to gas, three fractions of oil
oil a1, C 4 to BP s 493 K; oil a2, BP s 493623 K; oil a3, BP s 623811 K., and a
residue including ash and catalyst. The reaction scheme is shown in Fig. 2. A portion of
coal, denoted as CA , is rapidly liquefied to asphaltenes, pre-asphaltenes, heavy oil
BP s higher than 811 K; referred to as PAAO., oil a3 and CO, CO 2 , H 2 O. The
reaction rate for the path from a portion of coal, which is denoted as C B , to PAAO
reaction rate constant, K 1 . is considerably lower than that from CA to PAAO reaction
rate constant, k CA .. Ci represents the non-reactive portion of the coal. The PAAO
fraction is hydrogenated to H 2 O, gas, oil a1 and oil a3, which is a precursor of oil a2.
The hydrogenation from oil a2 to oil a1 is neglected, since most of oil a1 is produced
directly from the PAAO. These reactions are irreversible except for the watergas shift
reaction.
The reaction rate coefficient of each reaction was determined from the yields of the
process supporting unit PSU, capacitys 1 ton of coal per day. w19x, which was

Fig. 2. Assumed reaction paths in the simulation model.

262

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

constructed and operated based on the NEDOL process using Tanitoharum coal. The
physical properties of the coal liquids were obtained using the products of the Kashima
pilot plant as described above. Vaporliquid equilibria were calculated using the SRK
equation w20x and were verified using data obtained from the high-pressure pressure
separator of the PSU. Details of the simulation model including the reaction rate
coefficients will be published separately.
For case 1 of Table 2, the total oil yield is calculated to be 49 wt.%, which is less
than the experimental value of 51 wt.%. This difference may be due to errors in
calculation in the mean residence time of the slurry in the reactor, and due to the lower
temperature at the bottom part of the reactor. However, the correction of vaporliquid
equilibria and temperature profiles are rather complicated. For simplicity, therefore, the

Fig. 3. Correction factors for gas, oil, and residue yields for case 1 a. and case 2 b.. Lines, simulation; B, oil
yield of experiment; ^, gas yield of experiment; `, residue yield of experiment; I, hydrogen consumption
of experiment.

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

263

total calculated oil yield is adjusted to the experimental value using a correction factor
with respect to the reactor length, defined as follows:

effective reactor height. r actual reactor height; 11.8 m . .

2.

As shown in Fig. 3a, the correction factor is 1.08 for case 1. It is noteworthy that the
gas and residue yields, as well as the hydrogen consumption are in agreement with the
data when the reactor height is corrected as above.
On the other hand, for case 2 of Table 2, the correction factor is 0.92 as shown in
Fig. 3b. However, it was found that 18% of the reactor volume of the first reactor was
occupied with the accumulated solids in case 2 w6x. Thus, the net volume fraction of the
slurry phase for case 2 should be reduced by 6% from the volume fraction calculated
from 1 y g .. The correction factor is then 0.98, after the reactor volume is adjusted by
the accumulation of solid particles.
The mean residence time of the slurry phase ascending through the three reactors at
the Kashima pilot plant, as calculated from the simulator, is 4400 s for case 1 and 6000 s
for case 2. The mean residence time of the slurry phase was also determined by the
neutron absorption tracer technique w13x, to be 5100 s for case 1 and 7200 s for case 2.
The experimental data of the mean residence time are longer than the calculated values
for both cases. However, the discrepancies, which may be attributed to errors in the
calculation of vaporliquid equilibria, are sufficiently small to predict the gas and oil
yields in the large-scale reactors.
Table 4 shows the conditions used for the design of the large-scale demonstration
plant with a capacity of 2500 tons of coal per day. The reactor system consists of two or
three CSTRs. The operating conditions of the large-scale reactors are the same as cases
1 and 2 shown in Table 2, except for the production capacity and the residence time of
the slurry in the reactors. Light oil fractions are relatively unreactive with the coal and
pre-asphaltene. Thus, the amount of oil a3 BP s 623811 K. should be maintained at
a certain level in order to preserve the hydrogen donation characteristics of the recycle

Table 4
Design parameters for a large-scale demonstration plant
Case 1

Case 2

Recycle gas
Average molecular weight
Fraction of hydrogen, vol%

5.5
86

5.5
86

Makeup slurry
Coal feed rate, dry coal basis, kgrh
Recycle oil in slurry
Catalyst pyrite powder. in slurry, dry coal basis, wt.%,

104 200
156 300
3

104 200
104 200
3

Reaction
Operating pressure, MPa
Operating temperature, top of the two reactors, K
Gr r Lf ,a m3 STP.rkg-slurry

16.616.8
728
0.70

16.616.8
733
0.90

Volumetric flow rate of total recycle gas fed to the two reactors.rmass flow rate of makeup coal slurry..

264

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

oil w21x. As shown in Fig. 4, most of the oil a3 produced in the liquefaction reactors is
fed to the solvent hydrogenation section, but a small fraction is discharged to the
outside, accompanied by the oil and residue produced. Furthermore, a part of oil a3 is
converted to lighter oil fractions at the catalytic hydrogenation section. Thus, the
required feed rate of oil a3 can be calculated from the following balance:

required yield of oil a3. s a . q b . q c . ,

3.

a. s production rate of oil a2. = fraction of oil a3 to the oil a2.,

b. s production rate of residue. = fraction of oil a3 in the residue., and
c. s reaction rate of oil a3 hydrogenated to lighter fractions..
All quantities in Eq. 3. are expressed in units of mass per unit time and per unit
mass of daf coal in the feed slurry. If the yield of oil a3 produced in the liquefaction
reactors is lower or higher than this requirement, the recycle oil becomes lighter or
heavier during the long-term operation of the liquefaction. Each term in the right-hand
side of Eq. 3. is estimated as follows: a. The fraction of oil a3 contained in the
outflow as oil a2 is assumed to be 5 wt.% from the engineering experience of the
petroleum atmospheric fractionators. b. Although the content of oil fractions in the
residue should be as low as possible to achieve a higher oil recovery yield, the softening
point of the residue should be at least lower than 445 K. Fig. 5 shows the relationship
between the softening temperature determined by a ring-and-ball method, ASME
D36-95. and the hexane soluble content of the residue. In order to assure smooth
handling, the softening point of the residue is assigned to be 435 K, which is 10 K lower
than the limit. Thus, the content of oil a3 in the residue is assumed to be 8 wt.%. c.
The severity of the hydrogenation section should be increased when the slurry is

Fig. 4. Flow chart of the NEDOL liquefaction process. Solid lines, streams containing oil a3; broken lines,
streams with no oil a3.

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

265

Fig. 5. Effect of hexane solubles on the softening point of the residue from Tanitoharum coal.

prepared with a higher coal concentration, as in case 2. Thus, the oil a3 yield, which is
decreased in the hydrogenation section, is increased for the high concentration slurry.
Based on the data in the Kashima pilot plant, the decrease in the oil a3 yield in the
hydrogenation section is assumed to be 4.2 wt.% for case 1 coal concentrations 40
wt.%. and 7.2 wt.% for case 2 coal concentrations 48.5 wt.%..
Fig. 6a shows the effect of the mean residence time of the slurry phase in the reactors
on the oil a3 yield at the outlet of the liquefaction section for case 1. The reactor system
is assumed to consist of two CSTRs of the same volume. As the residence time
increases, the oil a3 yield at the outlet of the liquefaction section gradually decreases.
As shown in Fig. 7, the increase in total oil yield and the decreases in gas and residue
yields are always accompanied by an increase in hydrogen consumption. The oil a3
yield in the recycle oil is also decreased, as shown in Fig. 6a. However, the slope of the
required yield of oil a3 is smaller than that of the oil a3 yield in the total oil at the
outlet of the liquefaction section, since the reaction at the hydrogenation section is not
directly linked with the mean residence time of the slurry phase in the liquefaction
reactors. Thus, the required yield of oil a3 should be balanced with a. q b. q c. as
shown in Eq. 3.. The intersection point in Fig. 6a gives a value for the optimum total
residence time of the slurry phase in the reactors, which is approximately 5100 s for case
1.
Fig. 6b shows the relationship between the total oil yield at the outlet of the reactor
system and the total residence time of the slurry phase. The total oil yield increases with
increasing total residence time of the slurry phase. The total oil yield, which is
calculated for the reactor system with three CSTRs, is also shown in the figure. In this
calculation, the total residence time is equally divided into the three reactors. The total
oil yield in the three CSTR system is higher than that in the two CSTR system, but the
difference is approximately 1 wt.%. This suggests that the effect of the number of
reactors, provided the total residence time of the slurry phase is the same, is not
appreciable for two vs. three CSTRs.
Fig. 8a shows the relationship between the oil a3 yield and the total residence time
of the slurry phase for case 2, shown in Table 4. The oil a3 yields are optimized at a

266

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

Fig. 6. Effect of slurry residence time on oil a3 yield a. and total oil yield b. for case 1.

mean residence time of 7500 s. The total oil yield is 59 wt.% on a daf coal basis, and 5.0
wt.% of the hydrogen on a daf coal basis is consumed in the liquefaction reactors. Fig.

Fig. 7. The relationship between estimated hydrogen consumption and product yields under the operating
conditions of case 1.

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

267

Fig. 8. Effect of slurry residence time on oil a3 yield a. and total oil yield b. for case 2.

8b shows the effect of the number of reactors on the total oil yield for case 2. The result
is similar to case 1. The product yields, which are estimated for cases 1 and 2, are shown
in Table 5.

Table 5
Estimated yields of liquefaction and solvent hydrogenation sections
Case 1
Liquefactiona
Oil yield, daf coal basis, wt.%
C 4 493 K
27.8
493623 K
18.1
623811 K
7.2
Total
53.1

Case 2
Hydrogenationb

Total

Liquefactiona

Hydrogenationb

Total

2.11
0.70
y3.16
y0.35

29.91
18.80
4.04
52.75

26.2
22.7
10.3
59.2

3.83
2.54
y7.17
y0.80

30.03
25.24
3.13
58.40

5.2

5.0

1.3

6.3

Hydrogen consumption, daf coal basis, wt.%

4.4
0.8
a
b

In the liquefaction section.

In the solvent hydrogenation section.

268

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

6. Dimensions of the large-scale reactors

The effective reactor height is determined so as to satisfy the prescribed mean
residence time. When the inner diameter of the reactors is assumed to be 4.0 m in case 1
for the two reactors connected in series, the superficial gas velocities are calculated to be
0.058 mrs in the first reactor and 0.077 mrs in the second reactor. The values of the
gas holdup are 0.52 and 0.59 for the first and second reactors, respectively. The effective
reactor height is calculated to be 23.7 m, and the actual reactor height to be 21.9 m,
using the correction factor, 1.08, defined by Eq. 2.. The length between the top and
bottom tangential lines is calculated to be 20.7 m, after correction for the effective liquid
volume in the semi-spherical sections at the top and bottom of the reactor. The
dimensions of the reactors for the two-reactor system for case 2 are also determined to
be the length between the tangential lines s 20.7 m, providing the diameter of the
reactors is fixed as 4.0 m. These sizes agree with the values for case 1 by chance.
If the required residence time of the slurry phase is divided into the three reactors
connected in series, the length between the top and bottom tangential lines is calculated
to be 14.4 m. As shown in Fig. 6, however, the total oil yield is not greatly different for
the systems with two versus three reactors. Thus, the number of reactors should be
determined from the viewpoints of aspect ratio, maximum size for manufacturing and
transportation, stable introduction of feed and quench, etc. The aspect ratio of each
reactor for the three-reactor systems is as low as 3, which may cause unstable flows of
gas and slurry phases in the reactors. Moreover, the two-reactor system can reduce the
cost for equipment and construction in comparison with the three-reactor system. If the
dimensions of the reactors, which are connected in series, are not limited by local
conditions of the construction site, the two-reactor system is more advantageous than the
three-reactor system.
In order to establish the steady-state operation of liquefaction plants, design items
with respect to thermal stability, such as feed temperature, locations of quench gas
injection, amounts of quench gas and liquid, and thermal dispersion in the reactors, need
to be more fully examined. These will be discussed in a separate article using a
simulator developed by Onozaki et al. w22x.

7. Conclusions
The product yields of the Kashima pilot plant and the large-scale plant were predicted
using a simulator developed for the NEDOL process. The compositions and physical
properties of the recycle oils obtained in the PSU and the Kashima pilot plant were
applied to these calculations. The procedure for designing the reactors of 4 m in
diameter, processing 2500 tons of coal per day dry coal basis., were developed based
on the yields estimated by the simulator. The yield of oil a3 in the recycle oil was a key
determinant in maintaining the hydrogen donor ability of the recycle oil. The total
residence time of the slurry phase in the reactors was determined from this requirement.
The total oil yield was not greatly changed for the reactor systems with two vs. three
CSTRs.

M. Onozaki et al.r Fuel Processing Technology 64 (2000) 253269

269

Acknowledgements
The authors wish to acknowledge the support of NEDO in line with the New
Sunshine Project.

References
w1x K. Hirano, Fuel Process. Technol., in press.
w2x M. Kouzu, K. Koyama, M. Oneyama, T. Aramaki, T. Hayashi, M. Kobayashi, H. Itoh, H. Hattori, Fuel
79 2000. 365.
w3x A.R. Irvine, H.D. Cochran, O.L. Culberson, J.F. Fisher, W.R. Gambill, G.E. Oswald, R. Salmon, Direct
Liquefaction Technology Assessment. Task 1. Technical Readiness of the Developing Plant Functions,
Oak Ridge National Laboratory, ORNLrTM-9181, 1985.
w4x C.F. Irwin, A.J. Sincali, E.W. Wong, Hydrodynamics of Coal-Liquefaction Dissolver Configurations,
Oak Ridge National Laboratory, ORNLrMIT-326, 1981.
w5x Exxon Research and Engineering, EDS Consolidation Programme, Reactor Optimization Design Study,
Phase V, FE-2893-152, 1985.
w6x M. Onozaki, Y. Namiki, T. Aramaki, T. Takagi, M. Kobayashi, S. Morooka, Ind. Eng. Chem. Res.,
submitted for publication.
w7x Y.T. Shah, Reaction Engineering in Direct Coal Liquefaction, Addison-Wesley Publishing, Massachusetts, 1981.
w8x M. Onozaki, H. Ishibashi, Y. Namiki, T. Takagi, M. Kobayashi, in: 63rd Annual Meeting of Chem. Eng.,
Osaka, Japan, 1998, N117.
w9x B.L. Tarmy, M. Chang, C.A. Coulaloglou, P.R. Ponzi, Inst. Chem. Eng., Symp. Ser. 87 1984. 303.
w10x M. Takeshita, M. Takayasu, M. Fukuyama, N. Sakai, S. Ueda, H. Yoshida, K. Yokoyama, T. Hayashi,
Intern. Conf. Coal Sci., Tokyo, 1989. 859.
w11x K. Koide, S. Morooka, K. Ueyama, A. Matsuura, F. Yamashita, S. Iwamoto, Y. Kato, H. Inoue, M.
Shigeta, S. Suzuki, T. Akehata, J. Chem. Eng. Jpn. 12 1979. 98.
w12x P.M. Wilkinson, A.P. Spek, L.L. Dierendonck, AIChE J. 38 1992. 544.
w13x N. Sakai, M. Onozaki, H. Saegusa, H. Ishibashi, T. Hayashi, M. Kobayashi, N. Tachikawa, I. Ishikawa,
S. Morooka, AIChE J., submitted for publication.
w14x Y. Tanaka, M. Tamura, H. Kageyama, K. Clark, in: 2nd JapanrAustralia Joint Technical Meeting on
Coal, Tokyo, 1992, p. 198.
w15x H.M. Letzel, J.C. Schouten, C.M. van den Bleek, R. Krishna, Chem. Eng. Sci. 52 1997. 3733.
w16x C. Tsonopoulos, J.L. Heideman, S.C. Hwang, Thermodynamic and Transport Properties of Coal Liquid,
An Exxon Monograph, Wiley, New York, 1986.
w17x J.A. Gray, C.J. Brady, J.R. Cunningham, J.R. Freeman, G.M. Wilson, Ind. Eng. Chem. Process Des. Dev.
22 1983. 410.
w18x M. Hiraide, H. Itoh, A. Kidoguchi, E. Kaneda, M. Kobayashi, Y. Namiki, Y. Imada, K. Inokuchi, in:
Proc. Int. Symp. Fundamentals for Innovative Coal Utilization, Sapporo, 1999, p. 103.
w19x M. Mochizuki, K. Imada, K. Ikeda, K. Inokuchi, Y. Nogami, T. Takeda, K. Sakawaki, in: 4th
JapanrGerman Symp. On Bubble Columns, Kyoto, 1997, p. 393.
w20x A. Soave, Chem. Eng. Sci. 27 1972. 1197.
w21x M. Kouzu, H. Saegeusa, T. Hayashi, M. Kobayashi, H. Itoh, H. Hattori, Fuel Process. Technol.,
submitted for publication.
w22x M. Onozaki, Y. Namiki, H. Ishibashi, T. Takagi, M. Kobayashi, S. Morooka, Energy and Fuels, in press.