Flocculation Onset Titration of Petroleum Asphaltenes - Energy

---

Energy & Fuels 1999, 13, 315-322

315

Flocculation Onset Titration of Petroleum Asphaltenes Simon Ivar Andersen* IVC-SEP, Department of Chemical Engineering, Building 229, Technical University of Denmark, DK-2800 Lyngby, Denmark Received October 5, 1998. Revised Manuscript Received January 12, 1999

The abundantly used technique of flocculation onset titration (FT) of asphaltenes is investigated, and it is shown how additional information on asphaltene and oil properties in terms of solubility parameters may be derived from the data obtained. The classical ways of interpreting the results are outlined. Analysis of the effect of oil concentration on FT is shown to add information on the stability of the asphaltenes in the oil. Application to various areas of asphaltene stability related problems is shown. This includes emulsion stability by asphaltenes, change in stability by conversion, oil field deposition, segregation, and asphaltene dispersant evaluation. Recommendations are given on the performance of this type of experiment, such as on titration rates. The analyses of more than 25 different oils are reported in the present work.

Introduction Asphaltenes are defined as the phase separating from crude oil or petroleum products on addition of an excess of a hydrocarbon solvent such as heptane. However, when a crude oil or a solution of a heavier petroleum product is titrated with the asphaltene precipitant, the asphaltenes will begin to drop out at a specific magnitude of added precipitant. This magnitude has been recognized as being an indicator of the stability of the oil in terms of asphaltene precipitation during a process for a long time.1 However, the onset is related both to the oil properties and to the solute, the asphaltenes. The appearance of an asphaltene phase in petroleum systems is known to cause immense problems in recovery processes as well as in refinery processes.2 This is not caused by the total asphaltene content of the oil as much as to the accumulation of material during the operation. As an example, severe problems occurred in a German oil field due to accumulation of less than 50 ppm of asphaltenic material,3 leading to field shut-in and regular solvent stimulation operations. Several detection principles have been proposed in the literature that are basically directed toward a solution of the product titrated with a nonsolvent until a solid phase is observed: the Oliensis spot test1 using examination of drops applied on filter paper, microscopic examination of solutions,4,5 optical transmission and light scattering by particles,6-10 conductivity measure* To whom correspondence should be addressed. E-mail: sia@kt. dtu.dk. (1) Taxler, R. N. Asphaltsits Composition, Properties and Uses; Reinhold Publishing Corp.: New York, 1961. (2) Speight, J. G. The Chemistry and Technology of Petroleum, 2nd ed.; Marcel Dekker Inc.: New York, 1991. (3) Kleinitz, W.; Andersen, S. I. Paper presented at the 8th Symposium on Oil Field Chemistry, Geilo, Norway March, 1997. (4) Heithaus, J. J. J. Inst. Pet. 1962, 48, 45. (5) Buckley, J. S. Fuel Sci. Technol. Int. 1996, 14(1/2), 55. (6) Hotier, G.; Robin, M. Rev. Inst. Fr. Pet. 1983, 38(1), 101. (7) Reichert, C.; Fuhr, B. J.; Klein, L.L. J. Can. Pet. Technol. 1986, 25 (5), 33.

ments,11,12 viscosimetry,13,14 fluorescence spectroscopy,12 particle size analysis,15 heat transfer analysis.16 The main problem encountered in the optical detection is the opacity of the oil. Most spectroscopic techniques can only be used on light oils or dilute solutions and, hence, are difficult to apply in investigations of precipitation from samples such as live oils. However, lighter live oils with high GOR and low asphaltene content may show high light transmission properties. It is very important to select the right wavelength, one where the total absorbance of the oil is at a minimum. When low transmission is encountered, methods such a viscometry, conductivity, heat transfer, and filtration and similar procedures are the only options. However, as it will be shown in this paper, the application of techniques using dilute oil samples and solvent/nonsolvent pairs can be applied in monitoring different problems in the oil industry related to asphaltene phase separation. This is only possible if several initial dilution ratios of oil in solvent is used. Comparison of the single-point titration of different products, is often observed in the literature, may unfortunately give poor results as the single-point flocculation onset contains several different contributions from oil, solvent, and solute. Comparison of the qualitative effects of dispersants may be performed by single-point measurements, as can the titration of (8) Lambert, D. C.; Holder, K. A. 4th UNITAR/UNDP International Conference on Heavy Crude and Tar Sands, Edmonton, Canada, August 7-12, 1988 Paper No. 229. (9) Andersen, S. I. Ph.D. Thesis, Technical University of Denmark, 1990. (10) Thomas, F. B.; Bennion, D. B.; Bennion, D. W.; Hunter, B. E. J. Can. Pet. Technol. 1992, 22. (11) Fotland, P.; Arnfridsen, H.; Fadness, F. H. Proceedings of the 6th FPECPD Cortina, July 12-24, 1992. (12) MacMillan, D. J.; Tackett, J. E., Jr.; Jessee, M. A.; MongerMcCluer, T. G. J. Pet. Technol. 1995, 788. (13) Jacoby, R. H. In Heavy Crude Oil Recovery; Okandan, E., Ed.; NATO ASI Series 76, 1984. (14) Escobedo, J.; Mansoori, G. A. SPE Prod. Facil. 1995, 115. (15) Bouts, M. N.; Wiersma, R. J.; Muijs, H. M.; Samuel, A. J. J. Pet. Technol. 1995, 782. (16) Clarke, P.; Pruden, B. Fuel Sci. Technol. Int. 1996, 14(1/2), 117.

10.1021/ef980211d CCC: $18.00 © 1999 American Chemical Society Published on Web 02/20/1999

316 Energy & Fuels, Vol. 13, No. 2, 1999

Figure 1. Analysis of flocculation data according to the procedure of Van Keervort et al., see Heithaus.4

Figure 2. Analysis of flocculation data according to a modified procedure of Mertens.17

“purified” asphaltenes when the concentration is the same and the titrant is the asphaltene precipitant. Interpretation of the results from titrations of solutions of varying concentration are normally performed in two different ways: (1) The flocculation ratio (FR) is the minimum amount of solvent necessary to keep the asphaltenes in solution in a precipitantssolvent mixture.4 In other words, FR is the volume fraction of solvent at the onset. The dilution ratio X is defined as the ratio of the volume of diluent (precipitant) to the volume or mass of residue. If FR is plotted versus the inverse dilution ratio (1/X), a straight line is obtained, Figure 1. The intercept on the FR axis is termed FRmax and is related to the peptizability of the asphaltenes (pa ) 1 - FRmax). The intercept with the x axis is called Xmin and represents the amount of precipitant needed to precipitate the asphaltenes from the neat oil. From these values the peptizing power of the oil (po) is calculated as po ) FRmax(Xmin + 1). An additional parameter used in the interpretation is the “state of peptization” P ) Xmin - 1. Both FRmax and Xmin are changed by a change in solvent type. As this procedure is not further applied herein, see ref 4 for details. (2) The data are analyzed by plotting the volume of precipitant/mass of sample versus the volume of initial solvent/mass of sample, Figure 2. The sample may be oil or asphaltene. The plot reveals for all types of samples a linear relation. The intercept on the y-axis is taken as the relative solvent power of neat oil toward asphaltene, and the slope is related to the power of the solvent-titrant mixture. Originally the analysis was performed using plots of solvent/mass vs precipitant/

Andersen

mass.17 Bichard18 used the angle θ of the line with the x-axis (precipitant/mass) to describe the solvent power as cot θ. This is equal to the slope in the precipitant vs solvent plot. Qualitatively, a relative decrease in slope from one solvent to another indicates the decrease of solvent power. The intercept at the y-axis has been argued in several works to be related to the neat undiluted oil.The x-axis intercept will be constant for a given solvent and independent of the precipitant. Likewise the y-axis intercept is constant for a given precipitant and independent of the solvent. A positive y-axis intercept indicates that the oil is stable, and hence, precipitant is needed to cause instability. A negative y-axis intercept on the other hand indicates that the oil is instable and solid particles were probably present before dilution with the solvent. As reported below, it is possible to get reproducible results with samples in equilibrium with a solid phase if the entire sample is shaken before subsampling and the solids dissolved together with the oil. These oils, of course, yield negative y-axis intercepts. Mertens17 further related the slope of the line to the critical Hildebrand solubility parameter of the solvent/nonsolvent mixture at the onset of asphaltene precipitation. Hotier and Robin6 used plots similar to Figure 2 to evaluate solvent powers relative to heptane. Cimino et al.19 modeled this type of experiment using the Flory-Huggins equation. Of the two methods, the most prevailing in the current literature is analysis B (i.e., ref 20), although method A is used in data analysis in a number of industrial laboratories related to the refining industry using the xylene equivalent test.21 In the present work, spectroscopic titration data from several oils giving rise to very different production problems are reported in combination with the analysis outlined above. However, to avoid the empirical nature of the interpretation based on the original analysis, additional thermodynamic information is obtained from the experimental data as shown below, leading to the deduction of a cirical solvent solubility parameter characterizing the asphaltenes. The effect of concentration is additionally analyzed. The latter may be used in further interpretation of the stability conditions of the oil. The composition and other physical properties of the oils are not revealed in the present work due to the proprietary nature of this information. The aim of the present work is to show the versatility of the flocculation onset titration procedure through examples and to show some of the pitfalls present, as well as to report on the use of the data in obtaining important thermodynamic input for the modeling of asphaltene precipitation using a regular solution theory based approach. (17) Mertens, E. W. ASTM Bull. 1960, 40 (TP 218). (18) Bichard, J. A. Paper presented at the 19th Canadian Chemical Engineering Conference, Edmonton, Alberta, Canada, Oct. 19-22, 1969. (19) Cimino, R.; Correra, S.; Del Bianco, A.; Lockhart, T. P. In AsphaltenessFundamentals and Applications; Sheu, E. Y., Mullins, O. C., Eds.; Plenum Press: New York, 1995. (20) Wallace, D.; Henry, D.; Miandonye, A.; Puttagunta, V. R. Fuel Sci. Technol. Int. 1996, 14, 465. (21) Tort, F. Personal Communication, Elf Solaize sarl., France, 1995.

Titration of Petroleum Asphaltenes

Figure 3. Schematic description of the flocculation titration system.

Experimental Section The onset of precipitation or incipient flocculation threshold was obtained through automatic titration of dilute toluene solutions of either asphaltenes or crude oil. The titration system consists of a Radiometer ABU93 autoburet, a Pharmacia P1 peristaltic pump, and a Beckmann spectrometer model B. The wavelength was 740 nm, and flow cuvettes were either 1 or 10 mm in path length. Data acquisition and system control was performed using a PC. The equipment is outlined in Figure 3. The titration rate was found to have a significant effect, and in order to minimize overtitration and local precipitation, the rate was kept at 30 mL/30 s based on experiments with asphaltene solutions. This rate is significantly lower than titration rates used by others in the literature, which will be in the range of 1 mL/min. Solutions in the range from 0.5 to 1.5 g (to the neareas 0.1 mg) of oil in 5.00 mL of toluene were analyzed. For three oils, tetrahydrofuran (THF) was used as the solvent as the curde oil contained toluene insolubles. Asphaltenes were precipitated by n-heptane at a solventto-oil ratio of 30 mL/g. Asphaltenes were allowed to settle for 16 h and separated by filtration on sintered glass filters by vacuum filtration. The asphaltenes were extracted with toluene, and the solvent was evaporated using rotary evaporation followed by drying at 60 °C under a stream of nitrogen. The asphaltenes so obtained were further washed with aliquots of n-heptane until the supernatant liquid was colorless. This procedure is believed to remove most coprecipitated nonasphaltene components. When testing the asphaltene dispersant, this was added to the precipitant such as to obtain the right oil-to-dispersant ratio before mixing the precipitant and oil. The C10+, C15+, and C20+ residues of oils A-95, L, and L1 were obtained using a Fischer true boiling-point distillation unit. The cut was made 0.5 K after the boiling point of n-C9, n-C14, and n-C19. This was done to increase the asphaltene concentration in the oil to enhance the detection of precipitation. All solvents used were HPLC grade from Rathburn. Oils were obtained from various sources.

Results and Discussion In Figure 4 a typical titration trace of light intensity versus added n-heptane titrant is given for slow titration (30 mL/min). In the first part of the curve, the asphaltenes remain in solution while the oil is being diluted, leading to an increase in the light transmission as the concentration of the oil (c) decreases with the addition of n-heptane (volume Vt)

Energy & Fuels, Vol. 13, No. 2, 1999 317

Figure 4. Flocculation titration trace of oil P2 including the effect of titration rate.

ci ) (V0/(V0 + Vt))c0

(1)

where c0 is the initial oil concentration and V0 is the initial volume of sample. At the onset of flocculation, particles will scatter light and the light transmission decreases. The use of the peak maximum as an indicator of the onset is basically not accurate. The initial part of the titration curve represents dilution due to the addition of n-heptane. As the concentration of chromophores per volume is reduced, the light intensity of the transmitted light I increases. When the peak maximum is reached, sufficient particles are present to scatter the light such that the dilution is counteracted. Hence, the actual onset is to be found on the front of the peak where deviation from dilution takes place. The relation between oil concentration and intensity is described by a curve related to the dilution and the intensity of the incident light Io:I ) Io10kc, where k ) ab (a is the absorbance coefficient and b the path length) and c is the concentration. If a detector is used which only provides an intensity such as a laser with a potential output or as in this case a rebuilt spectrometer, the two key parameters k and Io are determined from the initial part of the curve where no precipitation is expected to occur. Using these parameters and Beer’s law, the dilution line is constructed and the deviation of the titration line from the dilution line then determines the actual precipitation point. With some detectors this theoretical approach has not been found to give realistic dilution curves as the values of k and Io were changing depending on the dilution. This is probably due to the fact that Beer’s law is not suitable for transmissions (I/Io100) below 1%, which is the case observed in the present work. Another way of decreasing the error in the determination is using plots of absorbance/concentration (A/ci) versus mL of titrant. This gives a better reading as A/c will have a constant value up to the onset point where the absorbance increases abrupt.9 This was not possible with the present equipment, so the results given herein are related to the peak maximum reading. The two methods outlined above would give rise to lower values of the precipitant necessary to promote onset as the detection is enhanced. Another important factor affecting the results of this method is the titration rate. In the literature, titration rates as high as 2 mL/min are quoted.22 In the present

318 Energy & Fuels, Vol. 13, No. 2, 1999

Andersen

work, an optimization of the titration rate was performed. The result indicates that a titration rate that is too high may lead to severe local precipitation, leading to an erroneous result with an onset lower than at equilibrium conditions. For the present system, the titration speed selected to minimize this was the addition of aliquots of 0.030 mL every 30 s. Increasing this to 0.125 mL/30 s gave, as seen in Figure 4, a significant effect. Theoretical Treatment of Data. Although asphaltene solutions are recognized as being dominated in part by self-association and to some extent being colloidal in nature (i.e., ref 23), for simplicity we will assume that asphaltenes in oils are in true molecular solution neglecting all associations. As with polymer solutions, we may assume that asphaltenes start to precipitate at a given critical precipitant-solvent-oil mixture. Applying the regular solution theory in simple terms, one may assume that the asphaltene precipitation commence at a certain critical solubility parameter (δcr) of the dilute mixture as the actual concentration of the least-soluble species is very low in the solution. δcr can be estimated from the known volume fractions (φi) of oil (o), solvent (s), and precipitant (p) and the solubility parameter (δi) of each of the groups, assuming that the molar volumes of oil, precipitant, and solvent are equal

δcr ) φoδo + φpδp + φs δs and

∑

δcr,app ) φoδo + φpδp and φs ) 0 φi ) 1

(4)

(2)

The oil is treated as a pure pseudocomponent. For each point, all volumes (Vi) and solubility parameters of the pure solvents are known. Assuming that eq 2 fully describes the onset, a plot of Vp/Vo vs Vs/Vo is a linear function, where the slope s ) (δs - δcr)/(δcr - δp) and the intercept Iy ) (δo - δcr)/(δcr - δp). In the normal analysis, Vo is substituted by mo. Solubility parameters for pure substances referenced in the literature have an apparent inaccuracy of about 0.1 MPa1/2,24giving a fairly large impact on the calculated results. Critical Solubility Parameter Evaluation. It has been observed that the apparent linearity in the plot of precipitant/mass vs solvent/mass persists up to very high dilution ratios (200 mL of toluene/g of oil), where φo , φs + φp. Hence, the critical solubility parameter at infinite dilution can be expressed as

δcr,dil ) φsδs + φpδp

the intercept may be used in evaluating the properties and stability of the oil. Whereas δa of the least-soluble asphaltenes represents a discrete fraction of the oil, δo represents the entire mixture of all components present in the oil. Acknowledging the fact that resins show specific interactions with the asphaltenes, the basic assumption of the regular solution theory (RST) is violated. Nevertheless, a number of applications of RST to asphaltene phase behavior have been reported.25-28 Hirschberg et al.26 generated δo by calculation of the isothermal energy of vaporization (∆Uv) of the oil from the a and b parameters obtained from characterization of the oil and tuning the SRK EoS to VLE data. Molar volumes were obtained from the BWR-EoS. Application of this procedure in our research indicates that pure component properties are reproduced with good accuracy, whereas the oil properties were also affected by the characterization procedure for the oil. Assuming that the linear relation observed can be extrapolated, the intercept with the y-axis represents the mixing of only oil and precipitant. A positive y-axis intercept indicates a stable oil, whereas a negative intercept indicates that solids may already have separated in the neat oil. Equation 2 may now be written

(3)

The slope (s) of the linear relation between precipitant and solvent is related to the volumes of solvent (Vs) and precipitant (Vp): s ) Vp/Vs. Rearranging to φs ) 1/(1 + s) and assuming φp ) 1 - φs, then δcr,dil can be derived from eq 3. Oil Solubility Parameter Evaluation. The intercept with the y-axis represents the mixing of oil and precipitant. In many cases, titration of the neat oil is impossible due to either opacity or viscosity, and hence, (22) Fuhr, B. J.; Klein, L. L.; Komishke, B. D.; Reichert, C.; Ridley, R. K. Proceedings of the 4th UNITAR/UNDP International Conference on Heavy Crude and Tar Sands, Edmonton, Canada, August 7-12, 1988, Paper No. 75. (23) Andersen, S. I.; Birdi, K. S. J. Coll. Interface Sci. 1991, 141, 497. (24) Barton, A. F. M. Handbook of Solubility Parameters and other Cohesive Parameters; CRC Press: Boca Raton, FL, 1983.

Assuming δcr,dil (from the dilute region) ) δcr,app one may directly estimate δo when the oil density Fo is known. The volume fractions are obtained from the y-axis intercept Iy ) Vp/mo and Vo ) mo/Fo. Thus

δo ) IyFo (δcr - δp) + δcr

(5)

The larger the magnitude of Iy is, the more sensitive the calculation will be to the value of the density as well as to the accuracy of the value of δcr. The method could provide an easy estimate of δo, however, as seen in Table 1 the values for δo in some cases get to be either very low or very high. Oil VBH+ has a δo value of 29.45 MPa1/2. This is equal to methanol, which is highly unrealistic. In most cases, appropriate and realistic values are obtained. The reason for this has not yet been revealed. If a positive x-axis slope is observed for the analysis of solid samples such as fractionated asphaltenes, well deposits, or sludge, the solubility of this material in the solvent is the reciprocal intercept given in g/cm3 at the temperature of the experiment. Asphaltene Solubility Parameter. From the calculated critical solubility parameter δcr, one may estimate the solubility parameter of the asphaltene solute δa. This is based on considerations of the difference in solubility parameter ∆δ between solvent and solute. Empirically, in polymer chemistry this is often set equal to 4 MPa1/2: δa ) δcr + 4 MPa1/2.28 The values (19-22 MPa1/2) obtained by this procedure are in agreement (25) Kawanaka, S.; Park, S. J.; Mansoori, G. A. SPE Reservoir Eng. 1991, 6, 185. (26) Hirschberg, A. W.; deJong, L. N. J.; Schipper, B. A.; Miejer, J. G. Soc. Pet. Eng. J. 1984, 24, 283. (27) Andersen, S. I.; Stenby, E. H. Fuel Sci. Technol. Int. 1996, 14, 261. (28) Laux, H. Erdo¨ l Erdgas Kohle 1992, 108, 227.

Titration of Petroleum Asphaltenes

Energy & Fuels, Vol. 13, No. 2, 1999 319

Table 1. Slopes and Intercepts from Flocculation Analysis of Various Oils and Estimated Parameters sample oil Ha A-95 VBH+ (5) VBH- (11) VBH (10) oil La F (H-crack feed) P1 (H-crack prod) P2 (H-crack prod) oil 1 (emulsion) oil 2 (severe emulsion) oil X oil N1 oil N2 (THF)a oil L1 (THF) oil L2 (THF) oil CAa oil Ma

slope intercept 4.96 2.50 0.94 1.01 1.36 1.62 2.42 3.22 3.37 3.77 5.27 1.02 1.23 5.65 2.53 2.10 9.16 7.90

-5.02 1.18 8.30 1.89 0.19 0.38 3.29 -5.13 -2.88 1.82 -6.46 1.68 1.58 -6.38 2.81 2.97 -22.14 -9.06

δcr 15.80 16.16 16.85 16.79 16.57 16.45 16.18 16.01 15.98 15.93 15.77 16.79 16.64 15.80 16.23 16.36 15.60 15.63

δo C7 (Fo ) 1) asph % 13.29 17.17 29.45 20.07 16.81 16.88 19.08 12.37 14.02 17.08 12.73 19.29 18.71 12.80 18.84 18.51 8.96 12.64

18.4 6.5 4.1* 32.7* 13.7* 6.6 8.9 2.4 2.4 1.4 1.1 7.9 0.5 8.5 8.3

a Oils with known well deposition history. Asterisk (/) indicates n-C6 asphaltenes.

with solubility parameters of asphaltenes reported in the literature [i.e., refs 15, 25-28). Solvent Characterization. The above equations may be used as well for the characterization of unknown solvents such as well-stimulation solvents, i.e., refinery cuts. In this type of analysis, the asphaltene precipitating at the onset may be used as a probe of known critical solubility parameter and from eq 4 the solubility parameter of the unknown solvent can be estimated relative to a standard solvent pair such as toluene/ heptane. Assumptions. The approach outlined in all cases relies on the existence of a thermodynamic equilibrium. Hence, it is assumed that the material precipitating is in equilibrium with the solution. Also, the assumption is made that in all cases independent of conditions, the same components or type of components will precipitate at the onset. This has been confirmed by extensive studies of asphaltene precipitation and dissolution in mixtures of heptane and toluene as a function of temperature [i.e., ref 29]. Applications. In our laboratories the procedure has been applied in the evaluation of instability problems in various petroleum industry related problems, such as emulsion stability, conversion leading to coke and sludge formation, segregation, and solid organic deposition in oil wells. In the following a number of examples are given. Most of the data was generated based on titration of four different concentrations. Repeatability of individual concentrations was found to be within 2% in most cases. The linearity given by the regression coefficient r2 was in all cases better than 0.95 and in most cases better than 0.98. The analysis based on the peak maximum was found to have some bias related to the person reading the curves, especially when the peak maximum is not sharp. This gave deviations in the range of 3% for the slope equivalent to a standard deviation of about 0.013 MPa1/2 on the critical solubility parameter. However, the deviation caused by operator bias was up to about 30% in the intercept, basically due to the extrapolation procedure involved in the latter. Concentration Effect. The linear behavior observed in the data analysis of precipitant/mass vs solvent/mass

Figure 5. Relation between initial oil concentration and flocculation onset of four different oils. Note the large difference in slope. See text for details.

often leads to the erroneous conclusion that this is an indication of the lack of influence of the maltene fraction or asphaltene concentration.19 In the case of examination of redissolved solid asphaltenes such as n-heptane asphaltenes, this may be true as the dilution ratio is high and the cosolvency effects may indeed be low in these solutions. However, for titrations of oils, more knowledge is obtained from examination of the effect of concentration, which may be important for the evaluation of the solid-phase behavior. In Figure 5 an example of four different oils are examined by plotting the total amount of heptane at onset versus the concentration of oil in the toluene (g of oil/mL of toluene). As can be seen, additional information is obtained from such a plot. The oils VBH+ (+, top) and VBH- (-, bottom) are from the same reservoir but at different depths. The reservoir fluid is heavily segregated, and the VBH- sample has a higher asphaltene content. As can be seen, the relation between precipitant and oil concentration of VBH- has a positive and relatively larger slope than VBH+, indicating that the solubility power of the oil/toluene mixture increases with increasing oil concentration. In other cases, such as the H-oil, the increased oil level leads to a negative or constant slope, indicating a poor solubility power of the oil. Hence, this gives additional information regarding the solvent power of the oil medium. This may be generalized in that an oil solvency less than the pure solvent added will lead to a negative slope and a better solvency will lead to a positive slope. The increased slope may be explained from a cosolvency effect due to increased specific interactions and from the colloidal stabilization theory on adsorption of resins onto dispersed asphaltene particles. Emulsion Stability. The procedure has successfully been applied in pinpointing the cause of severe water/ oil emulsion formation during oil recovery from specific wells. It is well-known that asphaltenes may stabilize water-in-oil emulsions, especially when energetics favors the adsorption of the asphaltenes at the oil-water interface.30 This is the case if the solubility power of the (29) Andersen, S. I.; Keul A.; Stenby E. H. Pet. Sci. Technol. 1997, 15 (7/8), 611. (30) Isaacs, E. E.; Chow, R. S. In Emulsions-Fundamentals and Applications in the petroleum Industry; Schramm, L. L., Ed.; Advances in Chemistry Series 231; American Chemical Society: Washington, DC.

320 Energy & Fuels, Vol. 13, No. 2, 1999

Figure 6. Analysis of oil from the same reservoir but different wells with and without emulsion problems. The oil analyzed was taken above the emulsion phase after centrifugation.

oil is low or if asphaltenes are already precipitating. A study was undertaken on oil from different wells of the same reservoir. One of these wells had suffered from severe emulsion formation and sludge deposition. Direct titration of the emulsions are hampered by the presence of water, which causes irregular scattering and plugging of lines, but the oil above the emulsion after centrifugation (5000 rpm/15 min) could be expected to be in equilibrium with the oil and the asphaltenes in the emulsion phase. Hence, the oils taken above emulsions were analyzed. In Figure 6 the results are given for two different wells of the same reservoir. As can be seen, the oil (oil 1) having the heavy emulsion problem has a negative y-axis intercept, indicating a very low solvent power of the oil toward the asphaltenes, whereas the oil (oil 2) without problems had a positive y-axis intercept. Oils from wells with less emulsion formation had positive intercepts or intercepts close to zero. The slopes are different, which may be related to the fact that in oil 1 the heavy asphaltenes of high-solubility power are precipitated into the emulsion phase but are still present in oil 2. In this interpretation it is assumed that the asphaltene composition and nature is similar throughout the reservoir. It should be mentioned here that production from the problematic well had not experienced problems related directly to asphaltene precipitation. This indicates that although asphaltenes are not observed as solid deposits in the well-stream, the near critical solubility power of the oil may lead to problems. Hydrotreatment. During hydrotreatment, the stability of the product may change significantly, leading to solid precipitation in reactors and increased coke formation. This may affect catalyst activity as well as the long-term stability of the products. In the present case, three samples of feed and products taken at severe conditions were investigated for alterations of stability. In Figure 7 the flocculation plot is given. There is an obvious change in the solubility power of the oil toward the asphaltenes, as observed from the change in the y-axis intercept. At the same time, the slope is changing. However, more information is obtained from Figure 8 as the slope of the FT vs concentration lines change dramatically, indicating that sample P1 is less stable than both samples F and P2. Hence, a minimum in stability is experienced. In Table 1 the magnitudes of solublity parameters calculated from the above equations are given. Results of sample P1 were dublicated

Andersen

Figure 7. Effect of hydrotreating on flocculation plot of feed (F) and products (P1 and P2).

Figure 8. Relation between initial oil concentration and flocculation onset for hydrotreated oils.

(δcr: 16.01 and 16.04 MPa1/2) on a sample taken approximately 10 h later, indicating a high repeatability of the method. Obviously the solvent power is decreased by the hydrotreating and the cracking of shorter alkyl chains from large molecules as seen from the lowering of the y-axis intercept. At the same time, the asphaltenes solubility parameter decrease from about 20.2 to 20.0 MPa1/2 probably due to both hydrogenation and cracking.31 Instable Crude Oils. In many cases asphaltene deposition has been reported during recovery processes. Here the titration may be used to evaluate the solvent power of the oil in the dead oil, which may be used as an initial guess in modeling. Moreover, important knowledge can be obtained on the asphaltene critical solubility. The asphaltenes in the oil may, in these cases, often differ from those precipitating at elevated pressure as the latter are already removed from the oil and left as a solid phase. With unstable oils, one may either dilute the oil with a solvent, redissolving the precipitate, or after centrifugation and removal of solid examine the oil phase alone at various dilutions. In the latter case, it is assumed that the oil is saturated with asphaltenes and an equilibrium exists between dissolved and precipitated material. Removal of the solids may in some cases, where very low amounts of asphaltenes are present, lead to difficulties in onset detection due to low concentrations of total asphaltenes and hence a low degree of light scattering. In our work, we have found (31) Andersen, S. I.; Bartholdy, J. Paper to be presented at the AIChE Spring Meeting, Houston, March 1999.

Titration of Petroleum Asphaltenes

Figure 9. Effect of commercial asphaltene inhibitor/dispersant on flocculation plots of oil L. The dispersant is successfully applied in the field.

that heating the sample followed by vigorous shaking of the sample in order to disperse the solids immediately before subsampling gives good results and again an excellent repeatability of the titration. In Table 1, the oils H, L, N1, C, and M all are in equilibrium with a solid phase and have caused problems due to well deposition. For some of these oils, the very negative y-axis intercept indicates that the oil is a very poor solvent. On the other hand, the critical solubility parameter is still in agreement with asphaltene solubility parameters in the range of 19.6 MPa1/2, in agreement with Hirschberg et al.26 For these oils, δcr > δo, indicating that the oil is a poor solvent for the asphaltenes. An attempt was made to investigate the effect of an asphaltene inhibitor on the flocculation titration. The inhibitor is successfully used in treating the reservoir from which the oil comes from, so it is known that there is a positive interaction between the inhibitor system and the asphaltenes in the oil. The oil was sampled at a point where no inhibitor was added. The effect of inhibitor concentration of 50, 100, and 500 ppm relative to the oil concentration was tested. The oil was centrifuged to remove water and solids prior to the titration. In Figure 9 the titration plots are given for the neat oil and the oil with inhibitor. As observed, the inhibitor has no apparent large effect on the slope nor on the intercept. Comparing this to successful field experience, one may come up with two answers to the result: (i) The asphaltenes interacting with the dispersant during the well-stream treatment are no longer present and the remaining asphaltenes have little interaction with the dispersant. (ii) The action of the dispersant is not to keep the asphaltenes from flocculating, but to disperse particles and keep these from further flocculation. However, precipitating asphaltenes by heptane with the addition of the inhibitor did affect the yield of solids if sufficient inhibitor was added. Figure 10. This does add evidence to the to antiflocculation mechanism but still does not give a clear answer to this. It is well-known from inhibitor evaluation that the inhibitor concentrations should be much higher than the concentrations needed in the well stimulation [i.e., ref 15]. For the 6 mL of toluene/g of oil data point, in this case one would observe a decrease in stability upon addition of the inhibitor from a single-point experiments (one concentration). The latter emphasizes the danger in overinterpretation of single-point experiments.

Energy & Fuels, Vol. 13, No. 2, 1999 321

Figure 10. Effect of commercial asphaltene inhibitor on asphaltene precipitation from oil L. Table 2. Results of Data Analysis of Neat Crude Oils and Distillation Residues C10+, C15+, and C20+ crude oil C10+ C15+ C20+

a

A-95

Oil L

2.50a 1.18b (4; 0.98)c 3.54 -2.13 (5; 0.98) 2.76 1.54 (5; 0.95) 2.50 2.72 (5; 0.99)

1.61 0.38 (7; 0.98) 1.19 2.47 (4;0.98) 2.15 -1.50 (4; 0.99) 1.81 -2.30 (4; 0.99)

Oil L1 2.53 2.81 2.70 0.44 (5; 0.97) 2.08 3.06 (5; 0.98) 2.31 1.61 (4; 0.98)

Slope. b Intercept. c Statistical parameters n and r2.

An initial study of the possibility of enhancing the detection by increasing the concentration of asphaltenes by distillation was performed. In many cases of instable crude oils leading to production problems, the actual content of asphaltenes in the oil is below 1-0.5% of the oil. As a certain amount of material is necessary for the detection of the solid phase, the onset may be passed undetected for these oils. The data are given in Table 2, and as observed, distillation does not give good results. This may be due to a procedure where no action was taken against possible oxidation. The residue flask was opened after reaching the final cut of either C10 or C15 in order to sample the necessary amount of material, and the distillation was continued. A new series of experiments are planned including purging the system with nitrogen before and during the sampling. However, as performed herein, the method is not good and the differences in asphaltene properties given by the slope are not usable. This may also be an indication of the significance of the presence of some of the lower components in the stability of the asphaltenes. Conclusions In the present work, we have demonstrated the versatility of the flocculation onset titration procedure in the application to very different problems related to asphaltene instability. Normally the evaluation of problems based on this procedure is done on a qualitative and empirical basis. It was demonstrated how the critical solubility parameter of the solvent mixture in which the asphaltene will start to drop out can be

322 Energy & Fuels, Vol. 13, No. 2, 1999

evaluated through simple relations. As the solubility parameter of the oil obtained from the y-axis interecept is very sensitive to the regression analysis, unrealistic results may be obtained by this procedure. Additionally, it has been shown that plots of the concentration dependence of the flocculation onset may give more information about the oil in terms of solvent power. It is important to emphasize that single-point flocculation titration data has very little information if the oilasphaltene system is altered such as during conversion.

Andersen

Acknowledgment. The author thanks Mr. Thoung Dang and Mr. Zacarias Tecle for the excellent help with the experimental work. We are grateful to the companies and people donating the samples for this work: Mærsk Olie & Gas A/S (Denmark), OMV AG (Austria), Elf Production (France), Haldor Topsoe A/S (Denmark), Intevep S.A. (Venezuela), Anticor Chimie sarl (France), Chevron Petroleum Technology Company (USA), and Jill Buckley (New Mexico Tech, USA). EF980211D