Investigation on the Reversibility of Asphaltene Precipitation

observed for Cold Lake bitumen. In both cases, the precipitation could be completely reversed. Temperature-reversibility was also investigated with n-...
0 downloads 0 Views 117KB Size
910

Energy & Fuels 2001, 15, 910-917

Investigation on the Reversibility of Asphaltene Precipitation Subodhsen Peramanu,† Chandresh Singh,‡ Mayur Agrawala,‡ and Harvey W. Yarranton* Department of Chemical & Petroleum Engineering, The University of Calgary, Calgary, Canada - T2N 1N4 Received January 2, 2001. Revised Manuscript Received May 15, 2001

The precipitation and redissolution of asphaltenes upon the addition and removal of solvent were investigated for Athabasca and Cold Lake bitumens using a flow-loop apparatus. The presence of precipitate was detected through an increase in pressure drop across an in-line filter. These solvent-reversibility experiments were conducted at 40 and 60 °C with n-heptane solvent. A significant hysteresis was observed for Athabasca bitumen while little or no hysteresis was observed for Cold Lake bitumen. In both cases, the precipitation could be completely reversed. Temperature-reversibility was also investigated with n-dodecane solvent at temperatures ranging from 40 to 160 °C. A hysteresis was observed for both bitumens and only partial reversibility was achieved. Benchtop solvent-reversibility experiments were also conducted on the two bitumens at room temperature. In this case, the precipitate was recovered by centrifugation. The benchtop results were in good agreement with the flow-loop experimental results. Two heavy oils were also tested and both exhibited hysteresis and complete reversibility. Precipitation and redissolution in n-heptane were measured over time for the Athabasca bitumen. Precipitation increased over time reaching an apparent equilibrium after 8 days. Redissolution experiments reached virtually the same equilibrium position in less than 1 day. The slow kinetics of precipitation suggests rate-limiting nucleation, growth, or flocculation of the asphaltenes. Redissolution experiments with their more rapid kinetics are better suited for obtaining equilibrium solubility data.

Introduction Asphaltenes are high molar mass brown to black solids and consist of a complex mixture of heavy crude constituents. By definition, they are soluble in aromatic solvents (benzene or toluene) and insoluble in paraffinic solvents (n-pentane or n-heptane). Asphaltene deposits can form during petroleum production, transportation, and upgrading processes due to changes in temperature, pressure, and composition. In all these cases, it is desirable to know under what conditions the asphaltenes precipitate and to what extent precipitated asphaltenes can be redissolved. Although there is considerable data available in the literature on asphaltene precipitation, these studies mostly focus on the measurement of the amount of precipitated asphaltene and the measurement of the “critical” or “onset” solvent-to-bitumen ratio (i.e., the ratio of solvent to bitumen at which asphaltenes begin to precipitate). There is limited data available on the effect of the reversal of solvent concentration or temperature on asphaltene precipitation. And yet, data on reversibility of asphaltene precipitation is essential for bitumen or heavy oil processes that involve solvent * Author to whom correspondence should be addressed. Phone: (403) 220-6529. Fax: (403) 282-3945. E-mail: [email protected]. † Present address: VECO Canada Ltd., 1200 401 9th Avenue S.W., Calgary, Alberta, Canada, T2P 3C5. ‡ Present address: Hyprotech Ltd., Suite 800, 707 8th Avenue S.W., Calgary, Alberta, Canada, T2P 1H5.

addition and recovery, such as bitumen transportation, solvent de-asphalting, and upgrading. Furthermore, these data help in investigating the validity of asphaltene precipitation models based on solution thermodynamics1-3 or colloidal theory.4 For example, thermodynamic theories predict that asphaltene precipitation is fully reversible while the colloidal theory predicts irreversible precipitation. Note that if the solution thermodynamics view is correct, asphaltenes dissolve in solution whereas the colloidal model holds that asphaltenes disperse in solution. For simplicity, the term “dissolution” is used throughout this paper without differentiating between models. Reversibility experiments are normally difficult to perform without disturbing the system; for example, by removing and/or reintroducing precipitated material. Andersen5 performed reversibility experiments by the gravimetric method using a mixed solvent (toluene and (1) Hirschberg, A.; deJong, L. N. J.; Schipper, B. A.; Meijer, J. G. Influence of Temperature and Pressure on Asphaltene Flocculation. SPE J. 1984, June, 283-289. (2) Burke, N. E.; Hobbs, R. E.; Kashou, S. F. Measurement and Modeling of Asphaltene Precipitation. J. Pet. Technol. 1990, 11, 14401446. (3) Kawanaka, S.; Park, S. J.; Mansoori, G. A. Organic Deposition from Reservoir Fluids: A Thermodynamic Predictive Technique. SPE Reservoir Engineering. 1991, May 185. (4) Leontaritis, K. J.; Mansoori, G. A. Asphaltene Flocculation During Oil Production and Processing: A Thermodynamic Colloidal Model. SPE Int. Symp. Oilfield Chem., San Antonio, TX, 1987, February 4-6; SPE Paper 16258. (5) Andersen, S. I. Hysteresis in Precipitation and Dissolution of Petroleum Asphaltenes. Fuel Sci. Technol. Intl. 1992, 10, 1743-1749.

10.1021/ef010002k CCC: $20.00 © 2001 American Chemical Society Published on Web 06/22/2001

Reversibility of Asphaltene Precipitation

n-heptane) with toluene contents ranging from 10 to 40 volume percent. He measured the precipitation of asphaltenes from crude oils upon addition of a given toluene/heptane mixture. He then measured the dissolution of asphaltenes extracted from the same crude oil in the same mixtures of toluene and heptane. He observed that the amount of precipitated asphaltenes was higher during the dissolution experiments than the precipitation experiments, producing a hysteresis curve. However, since the precipitation experiments were performed with whole oil and the dissolution experiments were performed with only the asphaltene fraction, the experiments could not be categorized as true reversibility runs. The hysteresis could be caused by maltenes (the non-asphaltenic crude oil fraction) enhancing the solubility of asphaltenes during the precipitation experiments. Clarke and Pruden6 indicated that there is a fundamental difference between attempts to redissolve asphaltenes using aromatic solvents and performing true reversibility studies. With a heat transfer apparatus Clarke and Pruden6-8 performed reversibility experiments at a constant temperature by adding and then removing n-heptane from Cold Lake bitumen. Although the amount of precipitated asphaltenes was not measured directly, the temperature profiles indicated that the asphaltenes redissolved as n-heptane was removed. However, the dissolution did not follow the precipitation path. They speculated that the segregation kinetics is slower than the aggregation kinetics leading to a hysteresis. This means that the redissolution curve would follow the precipitation curve if the time for redissolution were increased sufficiently. Andersen and Stenby9 studied the effect of temperature on asphaltene precipitation/dissolution. They used a mixed solvent (toluene and n-heptane) and performed solvent reversibility runs at 24, 50, and 80 °C. Although the reversibility of precipitation with temperature was not explicitly investigated, the results demonstrated that asphaltenes partially redissolve with an increase in temperature. Mohamed et al.10 and Rassamdana et al.11 investigated the redissolution of precipitated asphaltenes from crude oil upon the removal of the precipitant and upon addition of fresh oil. They investigated a limited set of conditions and observed partial reversibility of the precipitation even after 24 h. They both attributed the partial reversibility to slow redissolution kinetics although neither group investigated the kinetics explicitly. Mohamed et al.10 found that complete reversibility (6) Clarke, P. F.; Pruden, B. B. Heat Transfer Analysis for Detection of Asphaltene Precipitation and Resuspension. 47th Annual Technical Meeting of The Petroleum Society, Calgary, June 10-12, 1996; Paper 96-112. (7) Clarke, P. F.; Pruden, B. B. Asphaltene Precipitation: Detection using Heat Transfer Analysis, and Inhibition using Chemical Additives. Fuel 1997, 76 (7), 607-614. (8) Clarke, P. F.; Pruden, B. B. Asphaltene Precipitation from Cold Lake and Athabasca Bitumens. Petr. Sci. Technol. 1998, 16 (3&4), 287-305. (9) Andersen, S. I.; Stenby, E. H. Thermodynamics of Asphaltene Precipitation and Dissolution Investigation of Temperature and Solvent Effects. Fuel Sci. Technol. Intl. 1996, 14 (1&2), 261-287. (10) Mohamed, R. S.; Loh, W.; Ramos, A. C. S.; Delgado, C. C.; Almeida, V. R. Reversibility and inhibition of Asphaltene Precipitation in Brazilian Crude Oils. Petr. Sci. Technol. 1999, 17, 877-896. (11) Rassamdana, H.; Dabir, B.; Nematy, M.; Farhani, M.; Sahimi, M. Asphalt Flocculation and Deposition: 1. The Onset of Precipitation. AIChE J. 1996, 42, 10-22.

Energy & Fuels, Vol. 15, No. 4, 2001 911

could be obtained within a few hours with the application of ultrasound (sonication). Hammami et al.12 also found that asphaltene precipitation was partially reversible, in their case with a change in pressure in a PVT cell. They also attributed the partial reversibility to slow dissolution kinetics. However, the effect of more turbulent mixing could not be assessed with their apparatus. The purpose of this paper is to investigate the reversibility of asphaltene precipitation under various conditions and to identify kinetic effects on asphaltene precipitation and dissolution. A flow-loop apparatus developed by Peramanu et al.13 was used to perform the reversibility experiments. With this technique, the precipitation onset is identified by an increase in the pressure drop across an in-line filter. The technique is capable of handling opaque samples at high temperature and pressure conditions. Benchtop solvent reversibility experiments were performed to evaluate the flowloop experiments and to study the kinetic effects. Here, the precipitated asphaltenes were recovered by centrifugation. The flow-loop experiments were conducted on two bitumens and the benchtop experiments on the two bitumens and two heavy oils. Experimental Section Chemicals and Materials. Athabasca and Cold Lake bitumens were obtained from Syncrude Canada Ltd, Lloydminster heavy oil was obtained from Husky Oil Ltd., and Peace River heavy oil was obtained from Amoco Canada Petroleum Co. Ltd. The Athabasca bitumen is an oilsands bitumen that has been processed to remove sand and water. The Cold Lake and Peace River samples were recovered by steam injection from an underground reservoir and have also been processed to remove sand and water. The Lloydminster heavy oil is a wellhead sample from a cyclic steam injection project and has not been processed. Both the Lloydminster and Peace River samples contained water. It was removed by diluting the oils with n-heptane to a ratio of 0.8 cm3/g. At this ratio, the density of the oil was reduced sufficiently for the water to settle. The supernatant was decanted and the n-heptane was removed in a rotoevaporator at 70 °C. The water content of the treated bitumen was measured using Karl Fischer titration and was found to be 2 ( 0.1 wt % in both cases. This water content is similar to that of coker feed bitumen.17 The properties and SARA fractionation results for the four oils are given in Table 1. Further details on the properties for Athabasca and Cold Lake bitumens could be obtained from Peramanu et al.14 Toluene, n-heptane, and n-dodecane solvents were obtained from Aldrich Chemical Co. and were 99%+ pure. Nitrogen, (12) Hammami, A.; Phelps, C. H.; Monger-McClure, T.; Little, T. M. Asphaltene Precipitaiton from Live Oils: An Experimental Investigation of Onset Conditions and Reversibility. Energy Fuels 2000, 14, 14-18. (13) Peramanu, S.; Clarke, P. F.; Pruden, B. B. Flow Loop Apparatus to Study the Effect of Solvent, Temperature and Additives on Asphaltene Precipitation. J. Pet. Sci. Eng. 1999, 23, 133-143. (14) Peramanu, S.; Pruden, B. B.; Rahimi, P. Molecular Weight and Specific Gravity Distributions for Athabasca and Cold Lake Bitumens and Their Saturate, Aromatic, Resin, and Asphaltene Fractions. Ind. Eng. Chem. Res. 1999, 38, 3121-3130. (15) Long, Y.; Dabros, T.; Hamza, H.; Power, W. J. Reversibility of Asphaltene Precipitation from Paraffinic Solvent-Diluted Bitumen, Proceeding of the 50th Canadian Chemical Engineering Conference, Montreal, 2000, October 15-18. (16) Yarranton, H. W.; Masliyah, J. H. Molar Mass Distribution and Solubility Modeling of Asphaltenes. AIChE J. 1996, 42, 3533-3543. (17) Long, Y.; Tipman, R. N. Studies on Froth Water and Solids Removal from Froth to Enhance Diluted Bitumen Quality, Internal Report, Syncrude Canada Ltd., November 1994.

912

Energy & Fuels, Vol. 15, No. 4, 2001

Peramanu et al.

Table 1: Properties and SARA Analysis of Athabasca, Cold Lake, Lloydminster, and Peace River Samples on a Water-Free Basis Cold Peace Athabasca Lake Lloydminster River API gravity saturates (wt %) aromatics (wt %) resins (wt %) C5 - asphaltenesa (wt %) C7 - asphaltenesb (wt %) a

8.05 16.3 39.8 26.4 17.5 13.4

10.7 19.4 38.1 26.7 15.8 11.3

12.5 23.1 41.7 20.4 14.8 13.1

11.1 20.8 41.1 22.1 16.0 11.8

Extracted with n-pentane. b Extracted with n-heptane.

Figure 1. Flow-loop experimental apparatus. which was used for blanketing, was 99.5% pure and was obtained from Praxair Inc. Flow-Loop Apparatus. In the flow-loop apparatus,13 the diluted bitumen is continuously circulated through an in-line filter at a constant rate. As the bitumen is diluted, the viscosity decreases and so does the pressure drop through the filter. As precipitation occurs, the pressure drop through the filter increases. Hence, the minimum value on the curve of pressure drop versus solvent-to-bitumen ratio indicates the onset of precipitation. This apparatus was modified so as to enable solvent reversibility and temperature reversibility runs. The experimental apparatus consisted of a 750 cm3 vessel with provisions for stirring, heating, and cooling as shown in Figure 1. The stirring was accomplished using a magnetic drive mixer (Dyna-Mag, Pressure Products Industries, Warminster, PA). Heating clamps were attached around the cooling coils and wired to a temperature controller (Omron, Kyoto, Japan). A positive displacement metering pump (Bran-Luebbe, Buffalo Grove, IL) was used to circulate the vessel contents through a sintered stainless steel in-line filter (Swagelok, Solon, OH) of nominal pore size 60 µm. The pressure drop across the filter was measured by a differential pressure transmitter (Rosemount, Eden Prairie, MN). The bitumen sample was blanketed with nitrogen and a pressure relief valve was installed for the vessel safety. The vessel and tubing were completely insulated to minimize any heat loss. Thermocouples were located in the vessel and close to the in-line filter to monitor the temperature. A small injection pump (Milton Roy, Ivyland, PA) continuously dispenses the solvent from the reservoir into the vessel. The vessel temperature, in-line filter temperature, vessel pressure, and pressure drop signal across the in-line filter were continuously monitored using a data acquisition board (Keithley, Taunton, MA). Procedure for the Solvent Reversibility Flow-Loop Experiments. Solvent reversibility experiments were conducted at 40 and 60 °C for Athabasca and Cold Lake bitumens. For each run, about 100 g of bitumen was mixed thoroughly

with 25 cm3 n-heptane solvent in the vessel. The contents were then blanketed with nitrogen pressure to prevent the mixture from boiling. Next, the sample was heated to a desired temperature. Then, the valve underneath the vessel was opened, the circulating pump was turned on, and the sample was circulated at approximately 200 cm3/min. Initially the flow was through a fully opened bypass valve mounted across the in-line filter. The bypass valve was then closed slowly until the entire sample was passing through the in-line filter. After the bitumen was circulating through the in-line filter for about 15 min, n-heptane (solvent) was injected into the vessel continuously at a rate of 2 cm3/min and the vessel pressure and pressure drop across the filter were recorded at various solvent-to-bitumen ratios. The solvent injection and the recirculation pumps were stopped when the solvent-to-bitumen ratio reached a value of about 3.5 cm3/g, a ratio well above the onset of precipitation. Solvent addition took about 3 h at which time the vessel pressure was approximately 300 kPa. Differential pressure across the filter and vessel pressure were noted at various solvent-to-bitumen ratios during the precipitation (injection) run. To initiate the dissolution (extraction) run the vessel pressure was released slowly while nitrogen with vapors of n-heptane solvent was condensed and carefully collected in a conical flask to measure the mass. When needed, extra nitrogen was flowed through the vessel to remove the desired amount of solvent. Then the vessel was pressurized with nitrogen until the vessel pressure equaled the pressure of the forward run at the given solvent-to-bitumen ratio. In this way, forward and reverse run measurements could always be compared at the same pressure. The recirculation pump was started and, once steady state was achieved, the differential pressure across the filter corresponding to the solvent-to-bitumen ratio was noted. This solvent removal procedure was repeated to collect differential pressure values at a number of solvent-to-bitumen ratios. In the solvent removal run more than 80% of the solvent injected was recovered in about 5 h. Procedure for Bench-Top Experiments. Benchtop experiments were conducted to verify the solvent reversibility results obtained using the flow-loop apparatus. These experiments were based on direct gravimetric measurements of precipitated asphaltenes at various solvent-to-bitumen ratios. Each experiment was started by placing 20 g of bitumen into an Erlenmeyer flask, followed by the addition of n-heptane solvent. For precipitation (solvent addition) runs, 6-8 samples were prepared and the mixture was sonicated for 30 min and then left to settle for at least 18 h. The precipitated asphaltenes were separated from the solution using a centrifuge at 3500 rpm (900g), then dried and measured. For dissolution (solvent extraction) runs, 6-7 samples were prepared having a solvent-to-bitumen ratio of 3.5 cm3/g. The samples were sonicated and allowed to settle as above. Then the solvent was extracted from these samples in a rotary evaporator at 40 °C under vacuum to reach different solvent-to-bitumen ratios. The remaining mixtures were again sonicated and settled, and the asphaltenes were collected after centrifuging and drying. It is possible that oxidation occurred in the benchtop experiments and new asphaltenes were generated. A few experiments were repeated under nitrogen and identical results were obtained. Hence, oxidation is not significant over the duration of these low-temperature experiments.

Results and Discussion Solvent Reversibility. Figures 2 and 3 show flowloop solvent reversibility experiments for Athabasca bitumen at 40 and 60 °C, respectively. As n-heptane is added, the viscosity of the sample decreases and therefore the pressure drop across the filter also decreases. Further addition causes asphaltene precipitation and

Reversibility of Asphaltene Precipitation

Figure 2. Flow-loop solvent reversibility for Athabasca bitumen at 40 °C.

Figure 3. Flow-loop solvent reversibility for Athabasca bitumen at 60 °C.

accumulation on the filter resulting in an increase in the pressure drop. The continuous increase in the pressure drop following the onset of asphaltene precipitation indicates a continuous deposition of asphaltenes on the filter surface. When n-heptane is extracted from the mixture, the precipitated asphaltenes begin to redissolve, resulting in a decreased pressure drop across the filter. Athabasca bitumen exhibited a clear hysteresis at 40 °C and 60 °C with differential pressures during solvent extraction period much higher than those during the solvent injection period. For Cold Lake bitumen at 40 and 60 °C (Figures 4 and 5), the solvent extraction curve was nearly overlapping the solvent addition curve and there was no apparent hysteresis. With both bitumens, the extraction curves and solvent addition curves overlap near the onset solvent-to-bitumen ratio indicating a complete redissolution of asphaltenes. In other words, asphaltene precipitation is reversible upon removal of solvent. The only exception is the Athabasca bitumen run at 40 °C where a small difference in differential pressure is apparent even at low solvent-to-bitumen ratios. However, as will be discussed below, the benchtop experiments demonstrated that there was no precipitate left at these conditions and therefore the small difference in differential pressure is likely experimental

Energy & Fuels, Vol. 15, No. 4, 2001 913

Figure 4. Flow-loop solvent reversibility for Cold Lake bitumen at 40 °C.

Figure 5. Flow-loop solvent reversibility for Cold Lake bitumen at 60 °C.

error. The most probable source of error is plugging of the filter, i.e., physical trapping of a deposit in the filter. Such deposits can introduce random errors of unpredictable magnitude. A large deposit may not dissolve within the duration of the experiment. During the flow-loop experiments, it is possible that the precipitated asphaltenes settle in the vessel or stick to the walls of the tubing. Hence, the amount of precipitate on the filter may not be proportional to the amount of precipitate in the flow-loop apparatus at all times. Since this potential error could affect all the flowloop results, benchtop experiments were conducted to verify flow-loop solvent reversibility results. Figures 6 and 7 give the benchtop experimental results for Athabasca and Cold Lake bitumens, respectively. Note that the precipitation experiment is equivalent to the injection run in the flow-loop apparatus and the dissolution experiment corresponds to the extraction run in the flow-loop apparatus. In these figures, the ratio of the amount of precipitate to the total asphaltene content (fractional asphaltene precipitation) was plotted against solvent-to-bitumen ratio. Similar to the flowloop apparatus results, a significant hysteresis was observed for Athabasca bitumen (Figure 6) and little or no hysteresis was observed for Cold Lake bitumen (Figure 7). Also both bitumens exhibited complete redissolution at the end of the cycle.

914

Energy & Fuels, Vol. 15, No. 4, 2001

Peramanu et al.

Figure 6. Benchtop solvent reversibility for Athabasca bitumen.

Figure 9. Benchtop solvent reversibility for Peace River heavy oil.

Figure 7. Benchtop solvent reversibility for Cold Lake bitumen.

Figure 10. T-x phase diagram for Athabasca bitumen and n-dodecane.

Figure 8. Benchtop solvent reversibility for Lloydminster heavy oil.

The benchtop experiments were also performed for two heavy oils. A large hysteresis was observed for Lloydminster heavy oil (Figure 8) and a relatively smaller hysteresis was found for the Peace River heavy oil (Figure 9). The asphaltenes completely redissolved for both heavy oils at the end of the cycle. Hence, the precipitate from all four oils could be completely redissolved by removing the solvent and three out of four oils exhibited a hyteresis. The only oil that did not exhibit a hysteresis was the Cold Lake bitumen. The

properties of the Cold Lake bitumen (Table 1) are similar to those of the other oils and the reason for the different behavior is not clear. One possibility is the presence of an additive from the field processing. As will be discussed later, the hysteresis appears to be caused by slow precipitation kinetics, possibly the slow flocculation of asphaltenes. Hence, an added flocculent (for emulsion breaking) could reduce the observed hysteresis. Unfortunately, what chemicals, if any, were added to the oils is not known. Note that the Lloydminster sample, which contains no additives, did exhibit a hysteresis. Temperature Reversibility. Temperature-composition phase diagrams for bitumen-solvent systems indicating single-phase (liquid) and two-phase (solidliquid) regions were created using the onset data from the flow-loop apparatus, as shown in Figures 10 and 11 for the Athabasca and Cold Lake bitumens, respectively. In Figures 10 and 11, the region below the curve is single-phase liquid whereas the region above the curve is two-phase (liquid with precipitated asphaltenes). The solvent used here was n-dodecane. In ndodecane, the onset solvent-to-bitumen ratio changed significantly with temperature so that a temperature path through an onset could be found, as shown in Figure 10. The solvent first tested, n-heptane, provided too little change in onset ratio with temperature.13

Reversibility of Asphaltene Precipitation

Figure 11. T-x phase diagram for Cold Lake bitumen and n-dodecane.

Figure 12. Temperature reversibility for Athabasca bitumen at an n-dodecane-to-bitumen ratio of 2.25 cm3/g.

Figures 10 and 11 were used to select appropriate conditions for the temperature reversibility experiments. A 2.25 cm3/g solvent-to-bitumen ratio was selected. At this ratio, the solid phase is expected to appear at temperatures below approximately 60 °C for Athabasca bitumen and 90 °C for Cold Lake bitumen. Both temperatures are well within the operating range of the flow-loop. To begin the temperature reversibility experiments, n-dodecane solvent was mixed with bitumen to a ratio of 2.25 cm3/g at 160 °C (single-phase region) in the flowloop vessel. The sample was then continuously circulated through the in-line filter. The mixture was cooled to 40 °C (two-phase) and then reheated to 160 °C. The rate of cooling and heating was maintained at 0.4-0.5 °C/min. The temperature reversibility path is represented by dashed line in Figures 10 and 11 for Athabasca and Cold Lake bitumens, respectively. The path proceeds from single-phase to a two-phase region and then returns to single-phase region. The changes in differential pressure with temperature for the temperature reversibility runs are given by Figure 12 and Figure 13 for Athabasca and Cold Lake bitumens, respectively. As the temperature is decreased from 160 °C, the differential pressure begins to rise due to an increase in viscosity. After the onset of precipita-

Energy & Fuels, Vol. 15, No. 4, 2001 915

Figure 13. Temperature reversibility for Cold Lake bitumen at an n-dodecane-to-bitumen ratio of 2.25 cm3/g.

tion, the rise in differential pressure results primarily from accumulation of precipitate on the filter. However, to identify the onset point it was necessary to distinguish the pressure drop rise due to viscosity change from the pressure drop rise due to asphaltene precipitation. The change in differential pressure due to viscosity was obtained from the solvent injection experiments on the flow-loop apparatus. For example, consider Figures 2 and 3. The differential pressure at solvent-to-bitumen ratios less than 2.2 cm3/g is simply the pressure required to flow single-phase fluid through the filter at the given conditions. The trend on the plots (below 2.2 cm3/g) was extrapolated to find the differential pressure at a solvent-to-bitumen ratio of 2.25 cm3/g as if the fluid were still single-phase. These single-phase differential pressures are plotted versus temperature in Figures 12 and 13. The point where the temperature reversibility data departs from the single-phase differential pressure curve indicates the onset of asphaltene precipitation. The phase transition temperatures are approximately 65 (( 10) °C for Athabasca bitumen and 85 (( 10) °C for Cold Lake bitumen which are in good agreement with the values obtained from the temperaturecomposition diagrams (Figures 10 and 11). Again a hysteresis was observed. However, unlike the solvent reversibility runs, the precipitated asphaltenes did not appear to redissolve completely, i.e., the differential pressure did not return to its original value. Benchtop Kinetic Experiments. The presence of a hysteresis indicates that equilibrium was not reached in the precipitation and/or redissolution of asphaltenes and there must be a slow kinetic step(s). To identify the kinetic effects on precipitation and dissolution, the amount of precipitation over time was measured for Athabasca bitumen at an n-heptane-to-bitumen ratio of 2.25 cm3/g. Both precipitation and redissolution were measured up to a maximum time of 40 days and 30 days, respectively. The results of kinetic experiments are given in Figure 14. Although the data is quite scattered, some trends are apparent. During the precipitation runs, the fractional asphaltene precipitation increases with time and reaches a constant value at approximately 8 days.

916

Energy & Fuels, Vol. 15, No. 4, 2001

Figure 14. Time dependence of precipitation and redissolution of of Athabasca bitumen and model oil (0.025 g/cm3 Athabasca asphaltenes in toluene) in n-heptane. (n-heptaneto-bitumen ratio ) 2.25 cm3/g; n-heptane-to-model oil ratio ) 1.7 cm3/g)

During the redissolution runs, the fractional precipitation reaches a constant value within 1 day. The precipitation and redissolution curves reach approximately the same fractional precipitation over time suggesting that both systems eventually reach equilibrium. The slow kinetic steps appear to occur in the precipitation but not the dissolution of asphaltenes. As discussed previously, Mohamed et al.10 also observed relatively fast kinetics for redissolution upon solvent removal with sonication. Without sonication, they found the kinetics of dissolution to be slow. A likely explanation is that sonication disintegrates flocculated asphaltenes and increases the surface area exposed to the liquid phase. Increased surface area is expected to increase the rate of dissolution. Hence, without sonication or some form of turbulent mixing, the observed hysteresis would be greater and complete reversibility may not be possible. The kinetics of asphaltene precipitation, on the other hand, are slow even with sonication. Asphaltene precipitation likely consists of a series of steps including nucleation, growth, and flocculation. As with any phase separation, nucleation involves the aggregation of molecules into a stable volume. As more of the second phase forms, new nuclei may form or existing nuclei may grow. These newly formed particles or droplets may then flocculate into larger and larger aggregates. Detectable precipitation may require significant particle growth or flocculation. Hence, slow growth or flocculation could likely lead to a gradual increase in precipitation over time much as was observed in the kinetic experiments. Identification of the exact rate controlling step requires a microscopic study on particle growth and flocculation, which is outside the scope of this paper. However, Long et al.15 have reported slow growth and flocculation kinetics for asphaltene precipitation in a microscopic study. The rate-limiting effects could be caused by intrinsic properties of the asphaltenes, by interactions with other bitumen constituents such as resins, or by both these factors. Therefore, benchtop kinetic experiments were conducted for a model oil consisting of 0.025 g/cm3 asphaltenes dissolved in toluene. The amount of pre-

Peramanu et al.

Figure 15. Benchtop solvent reversibility for model oil (0.025 g/cm3 Athabasca asphaltenes in toluene).

cipitation over time was measured at an n-heptane-tomodel oil ratio of 1.7 cm3/g. Only precipitation experiments were conducted for the model oil to a maximum time of 10 days. The redissolution experiments could not be conducted for the model oil since toluene also evaporates when the n-heptane is removed, changing the model oil composition. The amount of precipitate increases over time for the model oil just as with the bitumen suggesting that the slow kinetics are intrinsic to the asphaltenes and do not depend on maltene components such as resins. The kinetic effect on the precipitation from the model oil indicates that a hysteresis should be obtained if solvent reversibility experiments were conducted. To confirm this hypothesis, a modified benchtop solvent reversibility experiment was conducted for the model oil. Precipitation experiments were conducted by adding n-heptane to the model oil which is a method similar to that used for bitumens. However, for the dissolution experiments, a different method was used since extracting n-heptane without loosing toluene was impractical. Instead, solid asphaltenes were first added to n-heptane and then toluene was added to the mixture. Since the asphaltenes begin as a dry solid with this method rather than a wet precipitate, the starting point of the dissolution is different. Extra driving force may be required to dissolve dry and partially crystallized asphaltenes as opposed to wet amorphous asphaltenes. Hence, the hysteresis may be more severe in this case. The precipitation and dissolution experiments resulted in a hysteresis curve as shown in Figure 15. The results are comparable to the precipitation and dissolution experiments conducted by Yarranton and Masliyah16 at various asphaltene concentrations in heptanetoluene solutions. A small hysteresis is observed at high asphaltene concentrations (low solvent-to-model oil ratio) and a large hysteresis at low asphaltene concentrations (high solvent-to-model oil ratio). While this is not a “true” hysteresis because solid asphaltenes were used to reconstruct the system, the hysteresis is similar to the bitumen systems. Hence, slow precipitation kinetics appears to be intrinsic to the asphaltenes. Of course, as the Cold Lake results demonstrate, other factors (perhaps additives and maltenes) can modify the kinetics.

Reversibility of Asphaltene Precipitation

Conclusions The precipitation and redissolution of asphaltenes upon the addition and removal of n-heptane were investigated for Athabasca and Cold Lake bitumen using a flow-loop apparatus and with benchtop centrifugation experiments. A significant hysteresis was observed for Athabasca bitumen while little or no hysteresis was observed for Cold Lake bitumen. Lloydminster and Peace River heavy oils were tested with the benchtop apparatus and both exhibited a hysteresis. In all cases, the precipitation could be completely reversed. The results confirm that solvent treatments can be an effective method for redissolving asphaltenes as long as there is sufficient turbulence to break up the asphaltene particles. Temperature-reversibility was also investigated with the flow-loop apparatus. Athabasca and Cold Lake bitumens in n-dodecane solvent were tested at temperatures ranging from 40 to 160 °C. A hysteresis was observed for both bitumens and only partial reversibility was achieved. The results suggest that temperature treatments may not be the best method for redissolving asphaltenes.

Energy & Fuels, Vol. 15, No. 4, 2001 917

Precipitation and redissolution in n-heptane were measured over time for the Athabasca bitumen. Precipitation increased over time reaching an apparent equilibrium after 8 days. Redissolution experiments reached virtually the same equilibrium position in less than 1 day. Slow precipitation kinetics and a hysteresis between precipitation and redissolution was also observed for a model oil consisting of asphaltenes in toluene. The slow kinetics of precipitation suggests rate limiting nucleation, growth, or flocculation of the asphaltenes. Redissolution experiments with sonication are better suited for obtaining equilibrium solubility data as the kinetics are relatively fast.

Acknowledgment. We thank Natural Sciences and Engineering Research Council of Canada, Shell Canada, Barry Pruden (Former Industrial Hydrogen Chair), Patrick Clarke, and Hussein Alboudwarej for their support and input in this work. EF010002K