Mild cracking solvent deasphalting: a new

method for upgrading petroleum residue

Sheng-Li Chen, Sheng-Sheng Jia, Yun-Hua Luo and Suo-Qi Zhao

National Heavy Oil laboratory, University of Petroleum, Dongying Shandong 257062,
Peoples Republic of China
(Received 18 December 1992; revised 20 day $993)

A new process-mild cracking solvent deasphalting (MCSDj-is described for upgrading heavy oil.
Continuous bench experiments were carried out on Shengh and Gudao vacuum residues. The effects of
cracking temperature and time on the yield, Conradson carbon residue, nickel content of deasphatted oil
and properties of ragmate asphalt were examined. The deasphalted oil (DAO) obtained by MCSD was
superior in yield and quality to that from solvent deasphalting alone. At the same yield of DAO, the
softening noint of the raffinate asnhalt was lower and the penetration and the ductility were greater than
those fo; ;he solvent deasphalting process.

(Keywords: heavy oils; residue; upgrading)

The demand for light oil products has continually been
increasing and that for heavy residual fuel oil decreasing
whereas the crude oil available is becoming heavier. The
economic upgrading of residues has therefore aroused
interest throughout the petroleum refining industry.
Many processes have been proposed for upgrading
residue’,‘, but no one process can upgrade all heavy oils
economically3. Solvent deasphalting is an important
method, in which much improvement has been made
since the early 1970s. Kerr-McGee’s ROSE and UOP’s
DEMEX processes are examples with lower energy
consumption and investment cost and higher yield of
deasphalted oil (DAO) compared with conventional
propane deasphalting4-6. Supercritical fluid extraction,
developed and being scaled up by the Heavy Oil
Laboratory of University of Petroleum (China), is even
more economic for processing residue7. It is known that
the heavier the deasphalting solvent, the higher the yield
of DAO, but the lower its quality’. Because heavier
solvents (i.e. C,, C, or a mixture of them) are used in
ROSE, DEMEX and supercritical fluid extraction, the
quality of the DA0 is lower than that from conventional
propane deasphalting. How to increase both yield and
quality of DA0 is an important question in solvent
deasphalting.
The quality of DA0 depends not only on the solvent
used but also on the concentrations of impurities (e.g.
metals and asphaltenes) and their distribution in the
heavy oil. The accumulation of Ni in the light resin of
Shengli vacuum residue accounts for the high Ni content
of the DA0 obtained from this residueg. Solvent
deasphalting is a purely physical separation process and
cannot modify the impurity distribution in heavy oil. In
order to obtain more DA0 with a lower metals content
and carbon residue, chemical reaction must be introduced
into the solvent deasphalting process. The simplest and
most economic chemical reaction process is mild thermal
cracking, which can move the impurity distribution
towards heavy fractions (heavy resin and asphaltenes).
The HSC-ROSE process, developed by Toyo Engineer-
ing Corporation, Mitsui Coke Company and Kerr-McGee
Corporation, is a combination of solvent deasphalting
with thermal crackinglo,“. It is reported that the
HSC-ROSE liquid product, with lower metals content
and carbon residue, is obtained in greater yield than from
a delayed coker. In addition, the capital cost is lower.
The HSC-ROSE process is a simple combination
of thermal cracking with solvent deasphalting, using
steam as stripping agent in the thermal cracking process.
The process proposed here, mild cracking solvent
deasphalting (MCSD), properly combines solvent de-
asphalting, under supercritical or subcritical conditions,
with mild thermal cracking, using the extraction solvent
as stripping and agitating agent. MCSD, simple in its
flow scheme, can produce more DAO, the quality
of which is higher, with lower hydrogen loss and no coke.
EXPERIMENTAL
Materials
Two kinds of solvent were used: pentane (CP); and a
C,i-C, mixture, of composition (mol%) C, 5.7, n-C, 8.3,
i-C, 85.4, i-C, 0.6,
Gudao and Shengli vacuum residues were used as
feedstocks; their characteristics are presented in Table 1.
Procedure
Figure I shows a simplified flow diagram of the
continuous bench test unit. A first stream of solvent and
the feedstock are pumped out of their tanks in a given
mass ratio, preheated to a desired temperature and then
mixed and charged to the bottom of a thermal cracking
reactor kept at a given temperature. In the course of mild
thermal cracking, some components decompose into
lower-molecular components and others condense into
higher-molecular components which cannot be extracted
by the solvent in downstream deasphalting. In addition
to the moderate severity of thermal cracking, the reacting

~16-2361~94~03/~39~
G 1994 Butterworth-Heinemann Ltd. Fuel 1994 Volume 73 Number 3 439
Mild cracking solvent daas~haf~j~g: S.-L. Chen et al.

liquid is maintained in vigorous turbulence by agitation
by solvent vapour, so coking does not take place
significantly. In consequence, stable and continuous
operation with a simple flow scheme becomes possible.
The solvent also acts as a stripping agent, so that the
residence time of the cracked oil is as short as possible. As
a result, further decomposition and condensation of
cracked oil are minimized.
The thermally cracked residue and first stream of
solvent leaving the top of the reactor mix with a second
stream of cold solvent so that the solvent/residue mass
ratio is 4:l. The mixture is maintained above or below
the critical temperature of the solvent and at a pressure
above the vapour pressure of the solvent at the
corresponding temperature. Under these conditions,
extraction equilibrium can be reached instantaneously.
This mixture is then transferred to the separation column,
in which the temperature and pressure are similar to
those in the mixer. In the separation column, phase
separation is completed rapidly, owing to the large
density difference as well as the low viscosity of the light
fluid phase. The heavy phase is withdrawn from the
bottom of the column, for recovery of raffinate asphalt
after removal of the solvent.
The light phase from the top of the separation column,
after being depressurized and heated above the boiling

Table 1 Feedstock characteristics

Vacuum residue Shengli Cudao

Density, 20°C (g cm - 3, 0.977 0.991

Sulfur (wt %) 0.9 1.7
Conradson carbon residue (wt%) 14.2 16.2
Nickel (ppm) 46.5 34.8
Vanadium (ppm) 2.7
Softening point (“C) 50 42
Penetration, 25°C (0.1 mm) 119 203
Ductility, 25°C (cm) 10 67
Composition (wt%)
Alkanes 21.4 16.5 Figure 1 Flow diagram of continuous bench test unit: 1, solvent tank
Aromatics 34.7 37.1 2, feed pumps; 3, feedstock tank; 4, electronic balance; 5, heaters; 6,
Resins 43.5 42.5 mixer; 7, thermal cracking reactor; 8, separation column; 9, pressure
Asphaltenes 0.4 3.8 gauge; 10, cooler; 11, pressure regulating valve; 12, solvent recovery
column; 13, cracked gas separator; 14, gas collecting bottle

Table 2 Results for Shengli vacuum residue as feedstock (solvent: C&-C, mixture)

DA0 properties
Cracking Cracking Extraction Extraction DA0
temp. time pressure temp. yield CCR Ni
Run no.’ (“C) (h) WPa) (“C) (wt%f (wt%) (ppm)

DE-01 395 1.0 6.0 137 44.0 2.1 4.72

DE-02 395 1.0 6.0 137 43.1 2.1 4.97
DE-03 401 1.0 6.0 137 48.1 2.8 5.54
DE-04 401 1.0 6.0 137 49.0 2.9 5.54
DE-05 410 1.0 6.0 138 60.0 2.9 6.42
DE-06 410 1.0 6.0 140 59.0 2.7 5.56
SL-27 _ 5.5 140 27.5 2.4 3.9
SL-28 - _ 7.0 140 39.3 3.1 6.8
SL-29 8.5 140 46.6 3.5 7.7
SL-30 - 10.0 140 52.7 4.8 10.7
SL-31 _ 11.5 140 59.4 5.4 13.2

“SL runs: solvent deasphalting only

Table 3 Results for Gudao vacuum residue as feedstock solvent: (solvent: C,-I’-C, mixture)

DA0 properties
Cracking Cracking Extraction Extraction DA0
temp. time pressure temp. yield CCR Ni
Run no.* (“C) (h) (MPa) (“C) (wt%) (wt%) (ppm)

GDE-01 400 1.0 6.0 130 52.4 3.7 3.85

GDE-02 400 1.0 6.0 132 50.0 3.2 3.69
GDE-03 410 1.0 6.0 132 54.4 3.0 3.16
GDE-04 410 1.0 6.0 132 52.8 3.0 2.79
GD2-01 _ 4.5 135 20.0 1.9 0.78
GD2-02 - _ 5.5 135 28.9 2.6 1.50
GDt-03 - 6.5 135 34.4 3.3 2.30
GD2-04 - _ 7.5 135 38.7 3.9 3.80
GD2-05 _ 8.5 135 42.7 4.2 4.30

“GD2 runs: solvent deasphalting only

440 Fuel 1994 Volume 73 Number 3

 Mild cracking solvent deasphalting: S.-L. Chen et al.

point of solvent, is separated in a solvent recovery column

into a solvent vapour phase and a heavy phase. The
solvent vapour is condensed and recycled to a solvent
tank. The heavy phase is withdrawn from the bottom of
the column, for recovery of DA0 after removal of the
solvent.

RESULTS AND DISCUSSION

Comparison between MCSD and solvent deasphalting
Tables 2 and 3 show experimental data obtained in
upgrading Shengli and Gudao vacuum residues in the
test unit. The following conclusions may be drawn:
the yield of DA0 is much higher for MCSD than for 0.0 1 , I I I
f

solvent deasphalting alone, under the same extraction 0 20 40 60 8C; DA0yield ( b?cX1

conditions; Figure 2 Effect of crackingtem~rature on Conradson carbon residue

the higher the cracking temperature, the higher the of DA0 from Gudao vacuum residue: l , solvent deasphalting only;
yield of DAO, if other operating conditions are the 0, cracking at 400°C; & cracking at 420°C
same;
demetallization and decarbonization are substantially
tti (PPD) t
improved in comparison with solvent deasphalting.
For instance, the extraction conditions of experiment
SL-27 were similar to those of experiment DE-06 but
the DA0 yield of SL-27 was only about half of that of
DE-06 lJnder the same extraction conditions, the DA0
yields were 44, 49 and 60 wt% at cracking temperatures
of 395, 401 and 410°C respectively. The DA0 yield of
experiment DE-06 was the same as that of experiment
SL-31, but the Conradson carbon residue (CCR) and Ni
content of the DA0 from DE-06 were respectively 50%
and 58% lower than from SL-31. The experimental
results indicate that heavy-metal-containing compounds
in the residue tend to accumulate during thermal cracking
in higher-molecular components, insoluble in the solvent
0 20 40 GO 80 DM yield ( WL% 1
under the extraction conditions. Kikuchi” came to a
similar conclusion when thermally cracking a residue in Figure 3 Effect of cracking temperature on nickel content of DA0
a batch cracker at 410°C followed by precipitation of the from Gudao vacuum residue: 0, solvent deasphalting only; 0, cracking
at 400°C; x , cracking at 410°C; A, cracking at 420°C
product with hexane. Polycyclic aromatic hydrocarbons
with long alkyl chains decompose into alkanes soluble
in the solvent under the extraction conditions and with
a lower CCR, and polycyclic aromatics with a higher
CCR. Some relatively low-molecular aromatics, which
could be extracted by the solvent, condense into higher-
molecular components insoluble in the solvent under the
extraction conditions. Because of this, hydrogen loss
accompanying raffinate asphalt is reduced significantly.

Effect of cracking temperature on DA0 properties

To study the effect of cracking temperature on the
CCR and nickel content of DAO, experiments were
performed on Gudao vacuum residue using pentane as
solvent, with a cracking time of 1.0 h. The results are
shown in Figures 2 and S. It is seen that the CCR and
nickel content decrease as the cracking temperature 0 20 40 GD 80 DM Yield ( Wt% )
increases. This indicates that the higher the severity of
thermal cracking, the deeper the desired reactions occur. Figure 4 Effect of cracking time on Conradson carbon residue of
However, it is better to keep the temperature >42o”C DA0 from Gudao vacuum residue: 0, solvent deasphalting only; .&
cracking time 60 min; x , cracking time 75 min; 0, cracking time 120 min
when the cracking time is 1 h, to avoid coking problems.
Effect of cracking time on DA0 properties in the reactor was adopted. Figures 4 and 5 show the
To study the effect of cracking time, experiments were relation between DA0 properties and the cracking time.
performed using Gudao vacuum residue as feedstock and It is apparent that the CCR and nickel content decrease
pentane as solvent at 400°C cracking temperature. It is as the cracking time increases. In other words, ionger
difficult to estimate the true cracking time, so the average cracking times result in greater demetallization and
residence time of cold liquid feed (residue and solvent) decarbonization. This also indicates that the higher the

Fuel 1994 Volume 73 Number 3 441

Mild cracking solvent deasphalting: S.-L. Chen et al.

lower than that of SL-19 raffinate asphalt, the ductility

Bi ( ppm 1t was 61 greater, and the penetration 26 greater. The
raffinate asphalts from experiments DE-01 and DE-04
could be used as no. 60 bitumen (a Chinese standard
bitumen), whereas that from SL-03-SL-16 could not, and
neither could the feedstock, because of their lower
ductility. This result is of significance to petroleum
refining engineers.

Cracked gas
Since thermal cracking was performed at moderate
severity and the extraction solvent rapidly stripped the
cracked oil, the yield of cracked gas was as low as
l-2wt% of the feedstock, lower than that of the
HSC-ROSE process lo . In consequence, the hydrogen
loss in the form of gaseous by-products is reduced. The
composition (mol%) of the cracked gas was C, 28.6, C,
2.0 I_ -,
48.2, C, 22.3.
0 20 40 GO SO DKI yield ( wt?i)
Figure 5 Effect of cracking time on nickel content of DA0 from Gudao CONCLUSIONS
vacuum residue. Symbols as in Figure 4
MCSD, which properly combines solvent deasphalting
with mild thermal cracking using the extracting solvent
as stripping and agitating agent, is a good method of
Table 4 Raffinate asphalt properties upgrading extra-heavy oils. With a simple flow scheme,
it can produce more DA0 with lower metals content and
Cracking DA0 Softening
temp. yield point Penetration Ductility CCR than solvent deasphalting can.
Run no.’ (“C) (wt%) (“C) (0.1 mm) (cm) Demetallization and decarbonization are greater in
MCSD than in solvent deasphalting, and increase as the
DE-01 395 44.0 55 41 69 cracking temperature and cracking time increase. The
DE-04 401 46.0 55 41 69
DE-06 410 60.0 58 20
hydrogen losses associated with cracked gas and raffinate
SL-03 - 38.1 63.5 15 16 asphalt are reduced significantly.
SL-19 - 42.6 65 15 _ At the same yield of DAO, the softening point of the
SL-46 - 45.2 70 13.5 8.5 raffinate asphalt is lower and the penetration and the
SL-16 ~ 58.0 83 4.5
ductility are greater than in solvent deasphalting.
“SL runs: solvent deasphalting only A no. 60 bitumen and 45 wt% DA0 can be produced
by MCSD from Shengli vacuum residue, which cannot
otherwise be used as road bitumen.

REFERENCES
severity of thermal cracking, the deeper the desired 1 Schuetze, B. and Hofmann, H. Hydrocarb. Process. 1984, 63(2),
reactions occur. However, it is better to keep the reaction 75
time k 2 h when the cracking temperature is 4OO”C,or 2 Chen, S. L. and Jia, S. S. Oil Gas Process (China) 1991, (2), 7;
coke will be formed. 1992, (l), 15; 1992, (2), 23; 1992 (3), 16
Maier, C. E. NPRA Annual Meeting, 1988, Paper AM-88-76
Gearhart, J. A. and Garwin, L. Oil Gas J. 1976, 74(24), 63
RafJinate asphalt Newcomer, R. W. and Soltau, R. C. Oil Gas J. 1982,80(28), 108
The properties of raffinate asphalt from MCSD vary Penning, R. T. et al. Hydrocarb. Process. 1982, 61(5), 145
not only with the cracking conditions but also with the Yang, G. H. et al. in Proceedings of the International Conference
on Petroleum Refining and Petrochemical Processing, Vol. 1,
yield of DAO, as in solvent deasphalting. Table 4 shows
Beijing, 1991, pp. 201-207
the properties of raffinate asphalt obtained by MCSD Speight, J. G. Fuel 1973, 52, 149
and solvent deasphalting. At the same DA0 yield, the Long, J. PhD Thesis, University of Petroleum, Dongying
softening point of the raffinate asphalt is lower and the Shandong, 1990
penetration and ductility are greater for MCSD than for Ishizuka, M. and Ohtsubo, T. Chem. Econ. Eng. Rev. 1986,18(3),
27
solvent deasphalting. For example, although the DA0 Waskimi, K. et al. in Proceedings, 3rd Int. Conf. on Heavy Crude
yield from run DE-04 was similar to that from SL-19, and Tar Sands, Vol. 7, 1985, pp. 1495-1509
the softening point of DE-04 raffinate asphalt was 15°C Kikuchi, K. Sekiyu Gakkaishi 1986, 29, 475

442 Fuel 1994 Volume 73 Number 3