SciELO - Scientific Electronic Library Online

vol.17 issue3Parameter estimation of thermodynamic models for high-pressure systems employing a stochastic method of global optimizationHydrogenation of diesel aromatic compounds in supercritical solvent environment author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Brazilian Journal of Chemical Engineering

Print version ISSN 0104-6632On-line version ISSN 1678-4383

Braz. J. Chem. Eng. vol.17 n.3 São Paulo Sept. 2000 



S.Waintraub, C.N.Fonseca, G. M.G. Soares and E.A.Campagnolo
PETROBRAS/CENPES, Cidade Universitária, Quadra 7, Ilha do Fundão,
CEP 21949-900, Rio de Janeiro - RJ, Brazil


(Received: February 10, 2000 ; Accepted: May 10, 2000)



Abstract -In order to reduce energy consumption and to increase deasphalted oil yield, an old PETROBRAS deasphalting unit was converted for use in the process of supercritical solvent recovery. In-plant and pilot tests were performed to determine the ideal solvent-to-oil ratio. The optimum conditions for separation of the supercritical solvent from the solvent-plus-oil liquid mixture were determined by experimental tests in PVT cells. These tests also allowed measurement of the dew and bubble points, determination of the retrograde region, observation of supercritical fluid compressibility and as a result construction of a phase equilibrium diagram.
Keywords: deasphalted, process of supercritical solvent, supercritical fluid compressibility.




The solvent deasphalting process has been continuously used and improved due to the economic benefits of upgrading the bottom-of-the-barrel heavy fuel oils into higher value products from vacuum distillation residues.

The five PETROBRAS solvent deasphalting plants were designed for the conventional system of solvent recovery by either single- or double-effect evaporation and for production of FCC feeds or bright stocks lubes. In these techniques, the solvent is recovered after vaporization in a series of decreasing pressure flashes and steam stripping under subcritical conditions. A large fraction of the total operational cost of each unit is spent on this step.

In the 1970’s supercritical fluid technology began to attract special attention, due to growing concern about improving energy utilization. Nelson and Roodman (1985) pointed out that the supercritical solvent recovery process has a much lower utility consumption than conventional systems. Later, Cervi (1989) showed that this reduction in utility requirements was about of 36% when compared with a double-effect solvent recovery in a Livorno’s refinery. In contrast to conventional solvent recovery, the supercritical solvent recovery process is virtually independent of the solvent-to-oil ratio (Hood, 1994). As a consequence of the decreased energy consumption, the plant can operate with higher solvent-to-oil ratios, resulting in an increase of deasphalted oil production and/or quality. Therefore the conversion of a conventional solvent deasphalting unit for operation under supercritical conditions results in decreased utility costs for recovery of the solvent, allowing an increase in deasphalted oil yield, for the same quality.

In this paper we will discuss the steps involved in the conversion design of an old PETROBRAS single-effect deasphalting unit, located at the RLAM refinery (Bahia, Brazil) for the operation under supercritical conditions. This unit consumes more energy than any of all the PETROBRAS deasphalting plants, operating with a 9:1 solvent-to-feed ratio, in order to produce Bright Stock lube.



The commercially available process simulators don’t accurately calculate the properties and phase equilibrium near the critical point, and calculations are worse when the mixture is composed of a very light component (solvent) and a very heavy one (deasphalted oil). In order to better understand the behavior and properties of solvent plus deasphalted oil (DAO) in this region, tests were performed in PVT cells. These cells are normally used for other kinds of studies, such as determination of the phase equilibrium of petroleum-plus-gas mixtures from the reservoirs and the asphaltene onset floculation in oils with gas in solution in a liquid-solid equilibrium.

PVT Cell Description

The experiments to determine the propane-plus-deasphalted-oil mixture phase envelope were conducted in a D.B Robinson’s equipment, called the DBR Solid Detection System.

The PVT cell is a cylinder of a transparent glass. In its lower region there is a magnetic system responsible for the agitation of the fluid inside the cell. In its upper part there is a mobile piston that is responsible for the mixture pressurization and depressurization. The piston displacement is promoted by mineral oil which is connected to a computerized pump. The whole system, consisting of the PVT glass cell and the steel external cell (with two glass windows), is inside a controlled temperature air bath. The maximum volume of the cell is 130 cm3.

There is a laser source that transmits a 2 mW light through an optic fiber cable into the sample and another optic fiber cable located on the opposite side receives the residual light and sends to a receiver. Variation in pressure inside the cell is controlled by a computer and causes a change in light transmittance which can be registered graphically.

During the tests sample real images are sent to a video monitor that enlarges them and they can also be recorded for a subsequent VCR viewing.

Sample Handling

The ratio of propane to deasphalted oil in the sample reproduces the deasphalting tower top when operating with a 12:1 solvent-to-oil ratio.

Propane and deasphalted oil heated were conditioned in metallic cylinders with displacement pistons (sample bottles), both of which were pressurized to 2000 psig . The lower parts of these bottles were connected to displacement pumps, which can control precisely the amount of each fluid transferred to the cell.

Tests Procedure and Results

All the tests were done for a fixed temperature and decreasing pressures, starting with a one-phase high-pressure condition (about 1400 psig). Under these initial conditions the mixture has a yellowish color and looks like a liquid.

For test temperatures above the mixture critical point, as pressure is reduced, there is a point with high sample turbidity and a large decrease in transmittance, corresponding to the formation of the first liquid drop (dew point). Below this point, as the pressure falls, the monitor image shows liquid drops accumulating on the cell base. This indicates retrograde condensation. After a maximum limit of the liquid volume, a reduction in pressure causes liquid vaporization, indicating the end of the retrograde region. There is also a pressure at which the dense gas is sharply clear and transparent, indicating the best region for phase separation and as a consequence for propane recovery.

For test temperatures below the critical point, as pressure is reduced a second liquid phase is formed. A further decrease in pressure causes bubble formation, showing a three-phase equilibrium (liquid+liquid+vapor).

At all times it was possible to measure the volume of each phase formed. Although the supercritical fluid looks like a liquid, we noticed a highly compressible behavior, similar to a vapor behavior. These test results are plotted graphically in Figure 1.




Plant tests were performed to determine the best value of the solvent-to-oil ratio to be used in the conversion to supercritical conditions.

For a Ramsbottom carbon residue constant value, increments of the solvent-to-oil ratio were made and the corresponding deasphalted oil yields were computed. These increases in the solvent-to-oil ratio were achieved by feed reduction due to solvent pump and solvent recovery system limitations.

The tests are described in reference Waintraub et alli (1998) and the results are summarized in Table 1. It was noticed that an increase in the solvent-to-oil ratio causes an increase in DAO yield, for the same quality, up to a limiting value. Above this point the increase in yield is negligible.




The RLAM solvent deasphalting unit was designed by the M. W. Kellogg Company in 1957, in order to process 3600 BPSD (572 m3/d) of a vacuum residue of a Brazilian oil. Currently the unit’s feedstock is 634 m3/d and it operates with a 9:1 solvent-to-oil ratio and a 54% yield of Bright Stock.

The basic design goal was the conversion to propane supercritical recovery with the same capacity, an increase to 12:1 in the solvent-to-oil ratio, an increase to 58% in lube yield and a reduction in energy consumption of 50% (Fonseca et alii, 1994) .

The conventional and the converted simplified flow diagrams are shown in Figures 2 and 3, respectively.




In the original configuration, the deasphalted oil plus propane from the top of the deasphalting tower flows to the low temperature evaporator. Almost all of the solvent is vaporized in this equipment. The remaining solvent in solution is vaporized in a second evaporator, the high temperature evaporator. All the propane vapor recovered from the DAO system is condensed and stored in the propane drum.

In the supercritical process, the solvent plus oil at the top of the deasphalting tower is pumped in order to increase the pressure to above the critical point. The temperature is increased in the supercritical exchanger and afterwards in the steam heater. The supercritical solvent (about 90%) is recovered in the supercritical drum, and cooled in the supercritical exchange by heating the deasphalted oil plus solvent mixture. After cooling with water the solvent is returned by pressure difference directly to the deasphalting tower. As there is no vaporization involved in this step, a large reduction in energy is achieved. The small amount of solvent that isn’t circulating in the supercritical system is recovered in the same way as it is in the conventional process, by condensation and storaged in the solvent drum.



Cervi, G. An Application of Phase Equilibrium Phenomena in the Supercritical Region for Oil Refining at AgipPlas Refinery in Livorno, 1989 European Federation of Chemical Engineers Meeting, May 1989.        [ Links ]

DB Robinson Design & Manufacturing Ltd. DBR Solid Detection System, Operating and Maintenance Manual, Edmonton, Alberta, Canada, February 1997.        [ Links ]

DB Robinson Design & Manufacturing Ltd. Jefri High Pressure PVT Cell Model PVT-0150-100-200-316-155, Operating and Maintenance Manual, Edmonton, Alberta, Canada, March 1997.        [ Links ]

Fonseca, C.N. Estudo da Modernização da Unidade de Desasfaltação da RLAM (U-11), Relatório

Gerencial, RL-GCOQUE-02/94, August 1994.

Hood, R.L. The Importance of Supercritical Solvent Recovery, Hydrocarbon Technology International Quarterly, London, 94( 4), March 1994, pp. 45-47.         [ Links ]

M.W.Kellogg Company. Operating Instructions Manual for 3600 BPSD Propane Deasphalting Unit, 1957.        [ Links ]

Nelson, S.R. and Roodman, R.G. ROSE®: The Energy Efficient, Bottom of the Barrel Alternative, Spring AIChE Meeting, Houston, Texas, March 1985.        [ Links ]

Waintraub, S., Fonseca, C.N., Mainenti, M.R.M. and Ferreira, R.A.; Teste Operacional na Unidade de Desasfaltação da RLAM (U-11) (Operational Test in the RLAM Deasphalting Unit), Comunicação Técnica DIPRIND/SEDET no 05/98, PETROBRAS/CENPES, April 1998.        [ Links ]

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License