Asphaltene removal from crude oil by means of ceramic membranes

Asphaltene removal from crude oil by means of ceramic membranes

Journal of Petroleum Science and Engineering 82–83 (2012) 44–49 Contents lists available at SciVerse ScienceDirect Journal of Petroleum Science and ...

672KB Sizes 7 Downloads 71 Views

Journal of Petroleum Science and Engineering 82–83 (2012) 44–49

Contents lists available at SciVerse ScienceDirect

Journal of Petroleum Science and Engineering journal homepage:

Asphaltene removal from crude oil by means of ceramic membranes M. Ashtari a, S.N. Ashrafizadeh a,⁎, M. Bayat b a b

Research Lab for Advanced Separation Processes, Department of Chemical Engineering, Iran University of Science and Technology, Narmak, Tehran 16846–13114, Iran Research Institute of Petroleum Industry (RIPI), West Blvd., Azadi Sports Complex, Tehran, Iran

a r t i c l e

i n f o

Article history: Received 31 July 2010 Accepted 3 January 2012 Available online 10 January 2012 Keywords: crude oil asphaltene ceramic membrane separation filtration

a b s t r a c t Separation of asphaltenes from three types of Iranian crude oils of different asphaltene contents, using asymmetric, ceramic monolith membranes with the pore sizes of 0.2 μm and 50 nm was investigated. The experiments were conducted in a batch filtration unit at a differential pressure of 200 kPa and a temperature of 120 °C. Two important operating parameters including flux of permeate and asphaltene rejection were monitored as a function of time. A decline in the flux of permeate and an increase in the asphaltene rejection occurred as a result of fouling the membrane. The flux of permeate did not exhibit a strong function of pore size and this phenomenon confirmed that fouling of the membrane was according to the formation of gel layer. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Asphaltenes are the most polar and heaviest compounds of petroleum which are insoluble in low normal alkanes (nC5–nC7) and soluble in aromatic solvents such as benzene and toluene. In crude oils, asphaltenes, resins and alkanes compose a dynamic stable system, similar to a colloidal system, in which the alkanes act as solvents, the asphaltenes act as micelles and the resins behave as stabilizers (JamiAlahmadi et al., 2009). Asphaltenes have been also identified as a major factor which causes difficulties in the crude oil refining. They accumulate in the separators and other fluid processing units during the crude oil transportation (Zewen and Ansong, 2000). Regarding the serious operational problems encountered in the presence of asphaltenes, separation of this complex and heavy species from crude oil seems to be essential. The conventional separation methods involve the application of paraffinic solvents. A number of researches have been conducted on stabilizing the asphaltene suspensions in the crude oil using some anti scaling agents. Such a procedure would prevent the precipitation of asphaltenes. Although using these methods could eliminate the asphaltene precipitation during transportation and refining of the crude oil but the asphaltene decomposition reactions at high temperatures which cause coke formation in heat exchangers and furnaces, remain as a problem. Separation of asphaltenes not only removes the problem of asphaltene precipitation but also improves the crude oil specifications

⁎ Corresponding author. Tel.: + 98 21 77240496; fax: + 98 21 77240495. E-mail address: (S.N. Ashrafizadeh). 0920-4105/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.petrol.2012.01.001

including increase in API gravity and lowering viscosity. The latter have essential impacts on the price of crude oil. Separation of asphaltenes from vacuum residues is made using SDA 1 process in which asphaltenes become separated from vacuum residues using paraffinic solvents such as propane. Using this process would precipitate the asphaltenes while the used solvent is recovered from the crude oil. The solvent recovery is not though possible by a distillation process due to the wide range of boiling points of crude oil components. On the other hand, another disadvantage of utilizing these techniques involves the consumption of considerable amounts of energy which is relatively expensive. Furthermore, the quality of the deasphalted oil widely decreases as the efficiency of the process increases. Meanwhile, the large amount of waste solvent is also another drawback of these methods. In recent years, alternative processes which are based on membrane technology have been suggested. The membranes have been implemented for two different applications, i.e. for the solvent recovery from initial extraction, phase separation or dilution steps, and for the direct separation of asphaltenes by means of membrane separation (Lai and Smith, 2001). Most of the researches conducted so far have used polymeric membranes for separating asphaltenes from crude oil. For instance, Kutowy et al. (1985) reported a process which involved dilution of the Cold Lake crude oil with 34% naphtha and subsequent separation of asphaltenes via application a 3–30 nm pore size polymeric membrane. By operating the membrane separation at 45 °C and a pressure of 3 MPa, they could reduce the oil viscosity from 90 to 2 cP and the


Solvent De-Asphaltening Process.

M. Ashtari et al. / Journal of Petroleum Science and Engineering 82–83 (2012) 44–49

vanadium and nickel contents from 125 and 41 ppm to 5 and 0.6 ppm, respectively. In another work, Ching et al. (2010) used Gore-Tex polymer membranes with nominal pore sizes of 30, 50, 100 and 200 nm for separating asphaltene from conventional crude oil at ambient pressure and temperatures of 22, 50, and 80 °C. They observed negligible loss of asphaltene due to significant asphaltene flocculation. Their finding could be resulted due to working at low temperatures and also lower effective size of the asphaltene clusters in the conventional crude oil. However, the polymeric membranes are limited to applications of low temperatures (b100 °C) in the case a relatively long life is required. They are also subject to chemical degradation when exposed to light hydrocarbons (Smith, 1998). According to the recent findings, ceramic membranes have strong stability at high temperatures and pressures; the crude oil viscosity is reduced at relatively high temperatures (Bishop et al., 2004). Therefore, application of ceramic membranes at higher temperatures may not only eliminate the addition of paraffin solvents required for dilution (Tsuru et al., 2001) but also may cause formation of larger effective size of asphaltene clusters, which in turn would result in more effective separations. Arod et al. (1993) described the ultrafiltration of a vacuum residue at high temperature (330 °C) using a ceramic membrane with an average pore diameter of 10 nm. At a cross flow velocity of 5.6 ms − 1 and a differential pressure of 500 kPa, the asphaltene content of the vacuum residue was reduced from 6.3 wt.% in the feed to 4.1 wt.% in the permeate. The vanadium content was also reduced from 195 to 90 ppm while the flux of permeate was 667 L/(m The effect of operating duration on the membrane fouling was not reported in this research. Duong et al. (1997) reported the ultrafiltration of Cold Lake heavy oil at temperatures in the range of 80–160 °C using a single-tube ceramic membrane with an average pore diameter of 0.02–0.1 μm. At cross flow velocities in the range of 2–10 ms − 1 and a differential pressure of 600 kPa, the ultrafiltration resulted in rapid fouling of the membrane; the latter significantly reduced the flux of permeate but increased the retention of asphaltenes. In their experiments, the flux of permeate was decreased from 660 to 60 kg/(m after 6 h of operation. At the same time, the asphaltene retention was increased from b1% to 80% and the volume of permeated oil through the single tube membrane was too low during the same period. Lai and Smith (2001) reported the microfiltration of Cold Lake heavy oil at temperatures up to 190 °C, using ceramic monolith membranes with an average pore size in the range of 0.1–1.4 μm. Although the volume of permeate was much more in their experiments, in comparison with that of single tube membrane, but the other results were analogous to those reported by Duong et al. (1997), i.e. the flux of permeate was reduced while the retention of asphaltenes increased with time. During their experiments, the asphaltene rejection increased to 74% after 6 h. Moreover, the membrane fouling resistance varied in the vicinity of 15– 20 × 10 12 m − 1 for different membrane pore sizes. They concluded that both pore restriction and gel layer formation had roles in membrane fouling. In this paper, the microfiltration and ultrafiltration of asphaltenes using a 19-channel ceramic membrane with a surface area of 0.24 m 2 was studied. Membranes of two different pore sizes were examined while a partial recycle was implemented to improve the separation efficiency. The flux of permeate and asphaltene rejection were measured versus time. The objective of this research was to quantify the membrane performance in terms of flux and separation efficiency at different pore sizes and asphaltene contents. This research is unique in monitoring and explaining the mechanism of asphaltene separation by means of membrane.


Table 1 Specifications of used crude oils. Crude oil

Place in Iran

LabSefid Khuzestan Province Foruzan Khark Island Nowruz Khark Island

Asphaltene Density Viscosity at Viscosity at content (wt.%) (g/cm3) 10 °C (cP) 40 °C (cP) 1.50




4.36 9.51

0.876 0.933

26.36 599.40

8.71 103.60

2. Experimental 2.1. Materials Three different Iranian crude oil samples with different asphaltene contents were used in this study. The properties of the crude oil samples are summarized in Table 1. 2.2. Membrane Two different cylindrical 19-channel ceramic monolith membranes made from alumina with pore diameters of 0.2 μm and 50 nm were used. The dimensions of the membrane were: length = 1016 mm, O.D. = 30 mm, and I.D. of each channel = 4 mm. As such, the cylindrical membrane had an efficient surface area of 0.24 m 2. The membrane was secured in a stainless steel housing and was sealed by two special o-rings which were resistant to organic solvents and high temperatures. A picture of the cylindrical membrane cross section along with the schematic diagram of the membrane housing is provided in Fig. 1. 2.3. Apparatus The batch filtration unit included the membrane housing, a feed pump and two pressure gauges. Feed was entered the housing from the tube side and permeate was collected from the shell side through a ball valve located adjacent the bottom of the membrane housing. The crude oil, was entered the circulating loop from a tank using the feed pump. A portion of the crude oil was always recycled to the feed tank in order to adjust the desired feed flow rate. A schematic diagram of the filtration unit is shown in Fig. 2. The crude oil feed tank incorporated a heating element which would maintain the heavy oil at an elevated temperature by means of electric power. The feed pump had a capacity of 10 L/min at a maximum differential pressure of 10 bars and was driven by a 0.75 kW AC motor. The pump was also equipped with a graphite mechanical seal which could withstand high temperatures up to 200 °C in the presence of organic solvents. The flow of the feed stream was controlled through by-passing some portions of the stream to the feed tank. The over pressure of the system was controlled by means of a pressure safety valve placed on top of the feed tank. Thermocouples and pressure gauges were used to detect the temperature and pressure, respectively. The steel pipes and membrane housing were thermally insulated. During each run, the flux of permeate was measured every 2 min for a period of up to 4–5 h. 2.4. Experimental procedures Prior to each experiment, 5 l of heavy crude oil was fed to the feed tank and then the heating devices were turned on. The system was purged with nitrogen to completely remove any trapped air. The feed pump was then turned on and the feed was circulated in the loop line. The V-6 valve was used to regulate the pressure of filtration


M. Ashtari et al. / Journal of Petroleum Science and Engineering 82–83 (2012) 44–49

Fig. 1. The schematic diagram of membrane housing and the membrane cross section.

system and a full open condition of V-6 valve allowed the maximum flow of the fluid to circulate. Some light hydrocarbons which may evaporate after depressurizing the shell side of the filter housing, were condensed by passing through the condenser E-4. Once the system was stabilized, usually within 1.5 h, samples of permeate were periodically collected from the sample line and weighed to determine the flux of permeate. 2.5. Analytical methods

2002). The crude oil density was measured after each run and API was calculated. Furthermore, the crude oil viscosity was measured using glass capillary viscometers in a water bath at 10 and 40 °C, according to ASTM 445–06 standard (D 445–06, 1990). In calculating the crude oil viscosity, both of the parameters of asphaltene contents and temperature should be considered. Therefore, the Beal correlation (Eq. (1)) was used. In this correlation, API varies for different asphaltene contents. ! 1:8ñ107 360 a Þ ð T þ 200 API4:53

2.5.1. Asphaltene analysis The asphaltene contents in the feed and permeate, defined as nheptane insolubles, were determined using a full automated asphaltene analyzer, model APD-500A, which operates based on the measurement of absorbance of two wavelengths of asphaltene particles. This measurement was conducted according to ASTM D-6560 and IP 143/01 methods (, 2009).

a ¼ log

2.5.2. Density and viscosity The crude oil density was measured by a digital density analyzer, model FP-6000, according to ASTM 4052 standard (D 4052–96,

In this correlation, T is the crude oil temperature (F) and μ is the crude oil viscosity (cP) (Sattarin et al., 2007).

μ od ¼ 0:32 þ



  8:33 0:43 þ API

Fig. 2. The schematic diagram of batch filtration system.

M. Ashtari et al. / Journal of Petroleum Science and Engineering 82–83 (2012) 44–49


2.6. Membrane performance Membrane performance was evaluated as the flux of permeate and asphaltene rejection. The flux of permeate was calculated through dividing the mass of permeated sample by the interval of time. The asphaltene rejection was also calculated using Eq. (2).   R ¼ 1−Cp =Cf  100


Where Cp and Cf are the asphaltene contents in permeate and feed streams, respectively. As a matter of fact, the asphaltene rejection is the selectivity of the separation process. Flux decline can be caused by some phenomena, such as gel layer formation and plugging of the pores. These phenomena induce resistances to the transport of feed across the membrane. Because the membrane retains the solutes, there will be accumulation of retained molecules next to the membrane surface. The accumulation of solute molecules may become so high in microfiltration and ultrafiltration that a gel layer can be formed. Such layer would exert a gel-layer resistance. Furthermore, some solutes would penetrate into the membrane pores and block the pores, causing another resistance, so called pore-blocking resistance. The flux of permeate with regard to these resistances can be calculated through Eq. (4). Rov ¼ Rf þ Rm


J0 ¼ Δp=ðμ:Rov Þ


Where J0 is the volumetric flux of the permeate, m 3/(m 2.s), μ is the dynamic viscosity of the crude oil under operating conditions, Pa.s, Δp is the pressure gradient across the ceramic membrane, Rm is the clean membrane resistance, Rf is the fouled membrane resistance, which is representative of gel-layer or pore-blocking resistances, and Rov is the sum of these resistances. J0 can be also calculated through dividing the measured flux of permeate (Jm) by the density of crude oil (Mulder, 1997). The fouled membrane resistance, Rf, is also calculated using Eq. (5): Rf ¼ Δp=ðμ:J0 þ Rm Þ


Rm ¼ Δp=ðμ w :Jw Þ


Fig. 4. Resistance of fouling versus time (Δp = 200 kPa; T = 120 °C; Cf = 9.51 wt.%).

3. Results and discussions 3.1. Flux of permeate and asphaltene rejection Flux of permeate and asphaltene rejection for the membrane with pore size of 50 nm are shown in Fig. 3. The operating conditions were temperature of 120 °C, pressure of 200 kPa, and asphaltene content of the crude oil of 4.36 wt.%. The experiment was continued for about 4 h. According to most studies, the colloidal size of the asphaltene particles is between 3 and 10 nm, depending upon the asphaltene nature and thermodynamic conditions (Mullins, 2010). By heating the crude oil up to particular elevated temperatures, nanometer particles of asphaltene form aggregates of micrometer size which become smoothly separated by membranes. During this period, the flux of permeate reached to one-third of its initial value and became constant while the rejection increased from 20 to 80 wt.%. The observed reduction in the flux can be attributed to the fouling of the membrane due to gel layer or pore blocking formation. According to gel model, asphaltene particles are retained by the membrane. Subsequently, the flux of crude oil through the membrane increases with pressure. By accumulation of asphaltene particles on the membrane surface, the gel layer gets thicker and becomes the limiting factor in the flux of permeate. According to pore blocking model, the asphaltene particles deposit within the membrane pores and block the pores. Therefore, the flux of permeate decreases remarkably and pore blocking becomes the limiting factor in the flux of permeate. Furthermore, increasing the asphaltene rejection can be explained by decreasing the pore size of the membrane and as a result of that, decreasing the passage of fine asphaltene particles through the membrane.

where Jw is the flux of pure water through the clean ceramic membrane and μw is the dynamic viscosity of pure water (Mulder, 1997).

Fig. 3. Flux of permeate and asphaltene rejection versus time (pore size = 50 nm; ΔP = 200 kPa; T = 125 °C).

Fig. 5. Resistance of fouling versus time (Δp = 200 kPa; T = 120 °C; Cf = 4.36 wt.%).


M. Ashtari et al. / Journal of Petroleum Science and Engineering 82–83 (2012) 44–49

Fig. 6. Resistance of fouling versus time (Δp = 200 kPa; T = 120 °C; Cf = 1.5 wt.%).

Fig. 8. The flux of permeate versus time (Δp = 200 kPa; T = 120 °C; membrane pore size = 0.2 μm).

3.2. Effect of pore size and feed asphaltene content on the membrane fouling resistance

the membrane pores and as a result of that a higher differential pressure in smaller pore size membrane is obtained.

Figs. 4, 5, and 6 show the membrane fouling resistance versus time for crude oils of different viscosities and asphaltene contents, and membranes with pore sizes of 0.05 and 0.2 μm. As shown, the fouling resistance of membrane increases with time due to increase in the thickness of gel layer on the membrane surface. Furthermore, as shown in Figs. 5 and 6, after passing some hours, the fouling resistance of membrane becomes equal for membranes of different pore sizes. In fact, according to Eq. (2), the overall membrane resistance is sum of two resistances, i.e. fouling resistance and clean resistance. Fouling resistance might be a function of fouling material specification if gel layer is formed, while it could be a function of both membrane specification and fouling material size if pore blocking is occurred. In addition, clean membrane resistance is a function of membrane specification. Also in both gel layer and pore blocking models, permeate viscosity has a strong effect on resistances. Overall membrane resistance may be limited by clean or fouling resistances. If overall membrane resistance for two different membrane pore sizes becomes equal, it would have two important messages, i.e. clean resistance is not the limiting resistance, and gel layer has been formed on the membrane surface. On the other hand, if overall membrane resistance for two different pore sizes did not become equal, different possibilities might be considered. The first possibility is that clean membrane resistance is limiting resistance, and the second possibility is that pore blocking has been occurred. Similarity of overall resistances on membranes of two different pore sizes, according to Figs. 5 and 6, reveals that the dominant mechanism for asphaltene filtration is gel layer formation. Furthermore, the gel layer model is also well applicable to Fig. 4. However, high viscosity of Nowruz crude oil causes a higher fluid resistance during the crude oil passage through

Fig. 7. The flux of permeate versus time (Δp = 200 kPa; T = 120 °C; Cf = 9.51%).

3.3. Effect of pore size and feed asphaltene content on flux of permeate The flux of permeate versus time for membranes of two pore sizes is shown in Fig. 7. As shown, the flux of permeate decreases with time due to the membrane fouling during the separation. After some hours, the flux of permeate becomes constant and reaches an almost equal value for both pore sizes. According to Fig. 7, the initial flux of permeate was 18 kg/(m 2.h) which declined to approximately onethird of its initial quantity after 6 h. The flux of permeate for three crude oils with different asphaltene contents is shown in Fig. 8. The membrane pore size was 0.2 μm. As it is obvious from Fig. 8, as the asphaltene content in the crude oil increased, the flux of permeate decreased. This phenomenon again confirmed the formation of gel layer for asphaltene separation, because with increasing the asphaltene content, the thickness of the gel layer will be increased and higher fouling resistances as well as lower fluxes of permeate would be resulted. On the other hand, according to pore blocking model, even low contents of particles may block the pores and sudden decrease in the flux of permeate may occur for different contents of fouling materials. In fact, with increasing asphaltene particles in the crude oil, the interactions among asphaltene molecules are increased and larger asphaltene aggregates are formed. These phenomena led to more asphaltene precipitation in the crude oil. The precipitated asphaltene particles form a thick gel layer on the membrane surface and as a result of that, a higher drop in the flux of permeate is observed for heavier crude oils. Accordingly, the flux of permeate somehow depends to the feed asphaltene content.

Fig. 9. Asphaltene rejection versus asphaltene content (Δp = 200 kPa).

M. Ashtari et al. / Journal of Petroleum Science and Engineering 82–83 (2012) 44–49


increasing the asphaltene content as a result of thicker gel layer formation on the membrane surface.

Fig. 10. Resistance of fouling versus time (Δp = 200 kPa; pore size = 0.2 μm).

Nomenclature I.D Inside diameter (mm) O.D Outside diameter (mm) Cf Feed asphaltene content (wt.%) Cp Permeate asphaltene content (wt.%) J0 Permeate volume flux (m 3/m 2.s) Jm Mass permeate flux (kg/m 2.h) Jw Pure water flux (m 3/m 2.h) Δp Pressure difference (bar) R Asphaltene rejection (wt.%) Rf Fouled membrane resistance (m − 1) Rm Clean membrane resistance (m − 1) T Temperature (°C)

3.4. The effect of pore size and feed asphaltene content on asphaltene rejection The asphaltene rejection for different asphaltene contents at optimum temperatures was investigated. The results which are shown in Fig. 9 reveal that the asphaltene rejection for membrane with a pore size of 0.05 μm was in the vicinity of 60 to 87% while the same for the membrane with a pore size of 0.2 μm varied between 44 and 63%. This means that asphaltene particles are so fine which can partially pass from 0.05 μm pores and if the gel layer does not form on the membrane surface, a majority of the asphaltene particles will pass from 0.2 μm pores. Another observation is that the highest rejection occurred for higher asphaltene contents which are related to thicker gel layers and more effective separations. 3.5. The effect of temperature on the resistance of fouling The parameter Rf for the membrane with a pore size of 0.2 μm versus time at different temperatures is shown in Fig. 10. The experiments were conducted in the vicinity of 100 to 200 °C. Obviously, increasing the temperature caused a reduction in the crude oil viscosity. Therefore, the crude oil could permeate faster from the membrane and thus separation increased. Similarly, the crude oil density decreased by heating the system. According to these results, the resistance of fouling was increased with temperature. The increase in Rf, could be as a result of asphaltene precipitation on the membrane surface as well as increasing the gel layer on the surface of the membrane. 4. Conclusions Crude oil filtration was conducted with ceramic monolith membranes. The results show that dominant mechanism in asphaltene separation is formation of a gel layer on the surface of the membrane. Although asphaltene particle sizes are very smaller than the membrane pore sizes, the asphaltene particles get adsorbed on the membrane surface and through gradual aggregation a gel layer is formed on the surface of the membrane which effectively separate asphaltene particles from crude oil. Furthermore, it appeared that flux of permeate and membrane overall resistance are not strong functions of the membrane pore size. This behavior can be well explained by gel layer model. The asphaltene rejection also was increased by

Greek letters μ Viscosity (cP)

Acknowledgements Financial support of Iran University of Science and Technology as well as the technical support of Iran Research Institute of Petroleum Industry are gratefully acknowledged. References Arod, J., Bartoli, B., Bergez, P., Biedermann, J., Caminade, P., Martinet, J., Maurin, J., Rossarie, J., 1993. Process for the treatment of a hydrocarbon charge by high temperature ultrafiltration, US Patent No 4,411,790. Bishop, B., Goldsmith, R., Schucker, R., Rawlins, K., 2004. Ceramic membrane process for upgrading vacuum residual oil. AIChE Conference Proceedings, 89d. Ching, M.-J.T.M., Pomerantz, A.E., Ballard Andrews, A., Dryden, P., Schroeder, R., Mullins, O.C., Harrison, C., 2010. On the nanofiltration of asphaltene solutions, crude oils, and emulsions. Energy Fuel 24, 5028–5037. D 4052–96, 2002. Standard test method for density and relative density of liquids by digital density meter, an American National Standard. D 445–06, 1990. Standard test method for kinematic viscosity of transparent and opaque liquids, an American National Standard. Duong, A., Chattopadhyaya, G., Kwok, W.Y., Smith, K.J., 1997. An experimental study of heavy oil ultrafiltration using ceramic membranes. Fuel 76 (9), 821–828. JamiAlahmadi, M., Soltani, B., Müller-Steinhagen, H., Rashtchian, D., 2009. Measurement and prediction of the rate of deposition of flocculated asphaltene particles from oil. J. Heat Mass Transfer 52 (19–20), 4624–4634. Kutowy, O., Guerin, P., Tweddele, T., Woods, J., 1985. Use of membranes for oil upgrading. 35th Canadian Chem. Eng. Conf., 1, pp. 26–30. Lai, W.C., Smith, K.J., 2001. Heavy oil microfiltration using ceramic monolith membranes. Fuel 80, 1121–1130. Mulder, M., 1997. Basic principles of membrane technology. Kluwer Academic Publishers, Dordrecht, the Netherlands. Mullins, O.C., 2010. The modified Yen model. Energy Fuel 24, 2179–2207. Sattarin, M., Modarresi, H., Bayat, M., Teymori, M., 2007. New viscosity correlations for dead crude oils. Petrol. Coal 49 (2), 33–39. Smith, K.J., 1998. Upgrading heavy oil by ultrafiltration using ceramic membrane, US Patent No 5,785,860. Tsuru, T., Sudoh, T., Yoshioka, T., Asaeda, M., 2001. Nanofiltration in non-aqueous solutions by porous silica–zirconia membranes. J. Membrane Sci. 185 (2), 253–261. Zewen, L., Ansong, G., 2000. Asphaltenes in oil reservoir recovery. Chinese Sci. Bull. 45 (8), 682–688.