Ind. Eng. Chem. Res. 1987,26, 1147-1153 Subscripts
b = bulk solution g = gas phase s = solid phase Registry No. C6H&H0, 100-52-7;C6H5CO3H,93-59-4.
Literature Cited Bamford, C. H.; Tipper, C. F. H. Comprehensive Chemical Kinetics; Elseiver: Amsterdam, 1980;Vol 16,pp 89-124. Caloyannis, A. G.; Graydon, W. F. J. Catal. 1971,22,287. Chou, T. C.; Lee, C. C. Znd. Chem. Fundam. 1985,24(1),32. Chou, T. C.; Lin, F. S. Can. J . Chem. 1983,61, 1295. Gould, E.S.;Rado, M. J. Catal. 1969,13, 238. Hendrih, C. F.;Van Beek, H. C. A.; Heertjes, P. M. Znd. Eng. Chem.
Prod. Res. Deu. 1977,16(4),270. Hendrih, C. F.; Van Beek, H. C. A.; Heertjes, P. M. Ind. Eng. Chem.
Prod. Res. Dev. 1978,17(3), 260. Hodge, P.; Sherrington, D. C. Polymer-supported Reactions in Organic Synthesis; Wiley: New York, 1980.
1147
Hwang, B. J.; Chou, T. C. Ind. Eng. Chem. Res. 1987,preceding paper in this issue. Kuo, M. C.; Chou, T. C. Ind. Ehg. Chem. Res. 1987,26,277. Maslov, S. A.; Blyumberg, E. A. Russ. Chem. Rev. 1976,45, 155. Meyer, C.; Clement, G.; Balaceanu, J. C. Proc. Int. Congr. Catal. 3rd,
1964 1965,I, 184. Neuberg, H. J.; Basset, J. M.; Graydon, W. F. J. Catal. 1972,25,425. Sadana, A. Ind. Eng. Chem. Process Des. Dev. 1979,18(1),50. Sadana, A. Ind. Eng. Chem. Process Des. Dev. 1981, 20(2), 397. Sadana, A,; Katzer, J. R. J. Catal. 1974a,35, 140. Sadana, A.;Katzer, J. R. Ind. Eng. Chem. Fundam. 1974b, 13(2),
127. Sheldon, R. A.; Kochi, J. K. Metal Catalyzed Oxidations of Organic Compounds; Academic: New York, 1981. Taylor, W. F. J. Catal. 1970,16, 20. Wang, C.K. Master thesis, National Cheng Kung University, Tainan, Taiwan, R.O.C., 1984.
Received for reuiew February 6 , 1986 Revised manuscript received January 5, 1987 Accepted February 24, 1987
Measurement and Prediction of Swelling Factors and Bubble Points for Paraffinic Crude Oils in the Presence of C 0 2 and Contaminant Gases Teresa G. Monger Department of Petroleum Engineering, Louisiana State University, Baton Rouge, Louisiana 70803
Effects on the phase equilibria and volumes of C02/crude oil mixtures of adding N2, CHI, H2S, or SO2 contaminants are reported. Phase equilibria data were measured at temperatures above and below the C 0 2 critical temperature and pressures up to 240 atm. Visual observations indicated conditions which promoted solid-phase formation. The data show that contaminant gases can have a pronounced effect on COz solubility and oil swelling. These effects must be considered in C 0 2 source selection and recycling for enhanced oil recovery. Measured swelling factors and bubble points are compared with the predictions of an empirical correlation, and a simple Peng-Robinson equation of state which treats the mixture as a binary with the crude oil as a single pseudocomponent. Swelling factor predictions using either method are in good agreement with the measured values. Equation of state predictions of bubble-point pressure were less successful. Introduction Carbon dioxide is gaining popularity as an enhanced oil recovery fluid and as the preferred solvent for several industrial extraction processes (Holm, 1982). Use of COz is usually attractive because it is relatively available, inexpensive, and environmentally safe. As more applications for COz arise, users will begin to consider less pure sources. For enhanced oil recovery specifically, operating costs can be significantly reduced with the use of extensive recycling. Overall consumption of the available C02supply will be optimized if complicated and expensive clean-up procedures are not required. What is needed is the assurance that some lower limit of contaminant gas(es) will not adversely affect the process performance. One of the aims of this study is to provide experimental data on how four contaminant gases, N2, CHI, H2S,and SOz, affect the phase behavior of C02/crude oil mixtures. Natural deposits of C02 commonly have associated Nz or CHI gas. The byproduct COz from ammonia and cement plants will likely contain N2. The COz supplied by acid gas separation plants may have appreciable levels of H a , and SOz is often present in COz generated by coal-fired power plants. Experimental data for C02/crude oil mixtures containing impure COz are scarce, particularly for paraffinic crude oils. 0888-5885/87/2626-1147$01.50/0
In enhanced oil recovery applications, the COz property most germane to the oil displacement mechanism is its ability to become miscible with reservoir oil through in situ multiple contacts. Two types of mass transfer occur (Holm and Josendal, 1974; Metcalfe and Yarborough, 1979). Carbon dioxide dissolves in and subsequently swells the oil, and in some cases, the composition of the displacing COz-richphase is altered by hydrocarbon enrichment to the point that it is first contact miscible with the original in place oil. Presently, there is much research activity directed toward predicting how C02 interacts with reservoir oil (Grigg and Lingane, 1983). One approach is to use generalized pressure-volume-temperature correlations like those of Simon and Graue (1965)for pure COz. In this case, there is no theoretical basis for including contaminant gas effects. This paper presents results which show that the Simon-Graue correlation for oil swelling can be used when small amounts of N2,CHI, or H2Sare present in the CO,; however, SO2 effects may be inadequately described. Another approach to predicting C02/reservoir oil phase behavior is to use a computer-implemented equatiofi of state. In this case, modeling the effects of contaminant gases is facilitated; however, characterization of the heavy oil components is difficult because rigorous compositional 0 1987 American Chemical Society
1148 Ind. Eng. Chem. Res., Vol. 26, No. 6, 1987 r---7 r---
I
A - Positive Displacement Pump B - Pressure Goqe C - P-V.T Cell D - A i r Bath
E - Cothelometer F - Mercury Reservoir
TC
I -7
-----~
G-
2 50
i
Mercury Level lndicotor
H - Line Filter
-
I TC-
-
Storoqe R e s e r v o i r s Temperature Controller To Vent, Voccum, or Compressed Gos
Figure 1. Schematic of experimental apparatus.
simulation requires enormous amounts of computer time. Mulliken and Sandler (1980) showed that it was possible to develop an analytical equation of state which used a single pseudocomponent characterization of the oil and gave COz solubility and oil swelling predictions of comparable accuracy to the empirical Simon-Graue results. An additional aim of this study is to test whether the Mulliken-Sandler method can be extended to impure C02 systems. The results presented for paraffinic crude oils show that the Mulliken-Sandler method can be easily adapted to accurately account for volumetric effects, but changes in the bubble-point pressure are difficult to describe. Experimental Section The apparatus used in this study is schematized in Figure 1. The Ruska P-V-T setup incorporated a viewable 484-cm3windowed condensate cell, housed in a temperature-controlled air bath, and manifolded to a 1000-cm3 positive displacement pump filled with mercury. The cell and pump were also manifolded to storage reservoirs. Temperature was controlled and monitored to 3Z0.3 "C by using Omega equipment employing numerous Pt-resistance thermometers. Pressure was measured with a Heise Bourdon tube gauge to rt0.5 atm at the maximum pressure tested, based upon calibration vs. an accurate dead weight tester. Volumes were calculated from pump displacements measured to 0.01 cm3and cathetometer readings measured to 3ZO.l mm. The overall accuracy of the P-V-T apparatus was checked vs. the pressure-volume isotherm for the binary system n-decane/C02 a t 71.1 "C and was found to agree within 1.0% of the data reported by Reamer and Sage (1963). In a typical experiment, gases and crude oil were volumetrically charged from pressurized storage reservoirs into the P-V-T cell. The cell's contents were mechanically mixed, and all but 6 cm3 of the cell volume could be observed with an external cathetometer view site calibrated to 0.1 cm3. Because the upper 6 cm3of the P-V-T cell was not visible through the cell windows, data were not collected at an observed bubble point. Bubble-point pressures were instead determined from linear and quadratic least-squares curve fitting of the pressure-volume dependence of the total sample mixture. Phase equilibria were measured as a function of increasing pressure by injecting mercury into the cell. Additional steps were also taken to minimize the contamination of manifold mercury. Corrections were made for cell expansion and thermal expansion of mercury. The excellent reproducibility of the P-V-T data and a typical curve fit are illustrated by the pressure-volume isotherm shown in Figure 2. The data points plotted in Figure 2 were obtained from two inde-
01
200
300 VOLUME ( c c )
400
Figure 2. Pressure-volume behavior in repeat runs determined for Hilly Upland crude oil (mixture 1). Curve fit shown is for open circles. Table I. Properties of Auoalachian Crudes property molecular weight, demo1 density, g/cm3 viscosity, cSt composition, %
method freezing pt depression Watson K factor equation of state densitometer 15.56 "C, 1 atm 21.11 OC, 1 atm kinematic 37.78 "C, 1 atm 98.89 O C , 1 atm gas chromatographs - -
Wt c5-c,,
I3C NMR saturated C aromatic C
Hilly Upland 224 290 324.21
Bath 193 277 312.53
0.8176 0.8141
0.8198 0.8155
4.69 1.75
4.26 1.68
66
78
94.3 5.7
89.8 10.2
pendent runs, one performed at the initiation of this study and a repeat run recorded at the completion of the experimental work. The COz and contaminant gases probed in this study were the highest purity provided by the manufacturer and were used without further purification, except for flashing from main storage cylinders prior to liquefication. The COz used in this study had a stated minimum mole purity of 99.99%. The contaminant gases were typically 99+% pure with the exception of CH4 which was technical grade (98+% pure). The paraffinic crude oils examined were from the Hilly Upland field in northcentral West Virginia and the Bath field in northcentral Ohio. The colors of the two stock tank oils contrast sharply; Hilly Upland crude is reddish brown, while Bath crude is nearly black. Additional properties of the crude oils are presented in Table I. The molecular weight of each crude oil was determined experimentally by the Beckmann method of freezing point depression and empirically from the K factor chart of Watson et al. (1935). Crude oil densities were measured by using a Mettler-Paar DMA 45 densitometer. Viscosity measurements followed standard ASTM procedures. Gas chromatography results were obtained by using a Hewlett-Packard 5880A gas chromatograph with an OV-101 column. These measurements indicated that both crude oils were rich in high molecular weight components with 34 wt % Hilly Upland
Ind. Eng. Chem. Res., Vol. 26, No. 6, 1987 1149
PARAFFIN
"c
NMR
HILLY U U N D CRUDE OIL
l3C W
R
BATH CRUDE OIL
I F i g u r e 3. Natural abundance NMR spectrum and integration of Hilly Upland crude oil in deuteriochloroform containing about 0.05 M CrAcAc. The spectrum is plotted from 0 to 170 ppm downfield from tetramethylsilane.
Figure 4. Natural abundance I3C NMR spectrum and integration of Bath crude oil in deuteriochloroform containing about 0.05 M CrAcAc. The spectrum is plotted from 0 to 170 ppm downfield from tetramethylsilane.
crude and 22 wt % Bath crude heavier than C3& Several experimental spectroscopic techniques and an empirical method were tried to analyze the aromatic contents of the crude oils. The best results were obtained by using natural abundance carbon-13 nuclear magnetic resonance spectrometry, as described in method 3 of Shoolery and Budde (1976). The CMR method provided an accurate and time-effective measurement, scoring aromatic and saturated carbon contents on a straightforward atomic basis, with no sample degradation. The natural abundance 13C NMR spectra and integrations obtained for Hilly Upland and Bath crude oils shown in Figures 3 and 4 c o n f i i that both of these crude oils are highly paraffinic.
Results and Discussion The raw experimental values of pressure and volume from the isotherms of the examined CO2/oil mixtures are presented in Table I1 for Hilly Upland crude and Table I11 for Bath crude. For each crude oil, the condition of no contaminant gas present was examined above the C02 critical temperature (mixture 1)and at room temperature (mixture 2). The effects of added N,, CH4,H2S, or SO, were then determined at room temperature (mixtures 3, 4,6, and 8, respectively). The effects of added H2S or SO, were also recorded at the higher temperature (mixtures 5 and 7 , respectively). Visual observations during isotherm measurements revealed precipitate formation for some mixtures. No attempt was made to quantitate formation of the black, tar-like material since it adhered strongly to the cell windows. Precipitation was generally more evident with Bath crude and most extreme when HzS or SOpwas present. In contrast to what has been previously reported for a Colorado reservoir oil (Graue and Zana, 1981), the addition of Nz alone did not enhance precipitation. Table IV summarizes the results obtained on the swelling of both oils. The Simon and Graue (1965) definition of a dimensionless swelling factor was used to calculate a value from experimental data oil volume at P and T swelling factor = (1) oil volume at 1 atm and T where P is the bubble-point pressure and T is the run temperature, Swelling can be measured at pressures other than the saturation pressure by using eq 1. Typically, for all the mixtures examined in this study, the swelling factor increased with increasing pressure until the vapor phase vanished. Above the bubble point, the swelling factor decreased slightly with increasing pressure, as live oil is more compressible than dead oil. For any given mixture, swelling was maximum at the bubble point. Comparisons of swelling factors determined when the same sample was run at different temperatures (Hilly Upland mixtures 7 and 8, Bath mixtures 1and 2,5 and 6, and 7 and 8) suggest that within the estimated accuracy of the data, swelling is temperature independent. As the absence of temperature in the Simon-Graue correlation suggests, this result is known for oil mixed with pure COz. In Table IV, swelling factors calculated from experimental data are compared with values given by the Si-
Predicting Method The equation of state proposed by Peng and Robinson (1976) was used to predict bubble-point pressures and volumes for C02/crude oil mixtures. All sample mixtures were treated as gas/oil binary systems. When impure COz was used, the critical properties and accentric factor of the gaseous pseudocomponent were determined from the physical properties of the pure components by molefraction mixing. The crude oil pseudocomponent was characterized by using the equations suggested by Kesler and Lee (1976) and the empirical correlation of Mulliken and Sandler (1980). Sensitivity tests were run on each of the input parameters, with special attention paid to variables derived from experimental data. It was found that predictions were most sensitive to the choice of oil molecular weight (mole fractions). This is perhaps not surprising since crudes are colloidal suspensions having no true molecular weight, and the average molecular weight determined is highly dependent upon the method used (See Table I). It was thus decided to tune the predicting method to a base case for each crude oil and generate the oil molecular weight which gave the best match of the experimental values of bubble-point pressure and saturation volume at run temperature. By analogy to the earlier work (Simon and Graue, 1965; Mulliken and Sandler, 1980), the phase equilibria data measured for crude mixed with pure COPabove its critical temperature were selected as the base case and are referred to as mixture 1. The oil molecular weights generated by the equation of state are listed in Table I and were used as input parameters for all bubble-point pressure and oil swelling predictions.
1150 Ind. Eng. Chem. Res., Vol. 26, No. 6, 1987 Table 11. Measured Values of Pressure and Volume for Isotherms of CO,/Hilly Upland Oil Mixtures pressure, phase total vapor, liquid, pressure, phase total vapor, liquid, atm region vol, cm3 vol % vol % atm region vol, cm3 vol % vol % Mixture 5: 98.35 g of oil + 19.06 g of COz + 3.31 g of HzS, Mixture 1: 142.84 g of oil + 28.09 g of CO?, 39.1 "C 28.2 L+V 484.44 60.56 39.46 39.0 "C 30.9 L+V 324.79 58.08 41.92 384.37 34.1 L+V 49.33 50.67 40.1 42.2 L+V 29.67 70.33 38.37 61.63 L+V 230.51 287.63 6.59 93.41 53.0 L+V 163.07 50.9 L+V 9.68 90.32 222.79 L 148.78 0 100 0 100 76.7 208.74 88.8 L 0 100 114.2 L 148.08 0 100 120.2 L 207.99 L 161.4 L 0 100 0 100 157.7 147.37 207.02 0 100 198.2 L 146.71 206.13 0 100 202.0 L L 146.28 0 100 0 100 228.5 205.38 239.0 L L 0 100 238.2 146.14 Mixture 2: 73.16 g of oil + 14.94 g of COz, 21.7 "C Mixture 6: 151.54 g of oil + 29.76 g of COz + 5.13 g of 27.6 L+V 202.25 HzS, 24.2 "C 37.2 L+V 131.05 20.52 79.48 37.2 L+V 131.00 20.31 79.69 27.9 L+V 484.13 57.16 42.84 31.6 L+V 414.75 38.1 L+V 129.41 19.64 80.35 49.14 50.86 37.4 L+V 302.16 27.38 72.62 41.5 L+V 112.09 5.33 94.66 44.1 L+V 241.48 7.18 92.82 41.2 L+V 111.58 4.82 95.18 L 224.44 0 100 80.6 64.0 L 106.15 0 100 114.4 L 223.54 0 100 76.3 L 105.90 0 100 155.1 L 222.57 0 100 93.9 L 105.74 0 100 195.8 L 221.59 0 100 0 100 110.8 L 105.56 236.0 L 220.70 0 100 L 105.40 0 100 128.5 0 100 138.8 L 105.29 Mixture 7: 74.49 g of oil + 14.52 g of COz + 5.06 g of SOz, 0 100 163.3 L 105.06 38.4 "C L 104.87 0 100 179.8 20.9 L+V 385.25 66.37 33.63 0 100 198.5 L 104.72 24.3 L+V 296.08 62.02 37.98 0 100 216.2 L 104.55 29.0 L+V 205.94 42.67 57.33 239.5 L 104.39 0 100 34.2 L+V 140.86 13.84 86.16 0 100 232.9 L 104.26 0 L 122.00 100 69.7 111.4 0 L 121.37 100 Mixture 3: 50.20 g of oil + 10.24 g of COP+ 0.70 g of Nz, 0 100 157.7 L 120.71 21.7 "C 0 L 120.15 153.92 54.55 45.45 100 199.0 34.2 L+V L 0 100 43.07 56.93 242.1 119.63 123.82 L+V 43.8 74.27 95.93 25.73 L+V 60.6 Mixture 8: 74.49 g of oil + 14.52 g of COz + 5.06 g of SO2, 82.28 11.61 88.39 80.2 L+V 23.6 "C .~ 77.16 90.9 L+V 15.3 L+V 483.81 77.66 22.34 100 112.8 0 75.04 L L+V 350.09 68.15 31.85 18.5 0 100 74.13 L 159.4 L+V 208.22 43.67 23.3 56.33 0 100 210.5 73.70 L 27.0 L+V 138.31 13.25 86.75 0 100 243.1 73.45 L 0 L 100 65.8 120.00 L 108.7 0 100 119.37 Mixture 4: 142.69 g of oil + 29.05 g of COP + 1.09 g of 0 147.0 L 118.85 CHI, 22.8 "C 100 0 L 60.69 39.31 100 189.3 118.29 L+V 483.80 28.8 0 100 232.6 L 117.75 48.31 51.69 L+V 375.57 35.8 25.80 74.20 L+V 273.73 45.0 5.65 94.35 L+V 220.90 54.4 0 100 L 208.66 113.5 L 208.00 0 100 144.2 100 207.36 0 L 176.0 L 206.67 0 100 210.5 0 100 L 206.05 243.8 ~
mon-Graue correlation. The assumption used was that the total gas present behaved like COz. Results are tabulated for two estimates of oil molecular weight. The agreement between experimental and Simon-Graue values is better using oil molecular weights determined by freezing point depression. A comparison of the results for the two crudes suggests that the Simon-Graue correlation is more accurate for the less paraffinic oil (Bath crude). The Bath crude results have an average error ("Simon-Graue correlationcncompared to "exptl value" in Table IV) of 0.5% and a maximum error of 1.4%, which are within the estimated accuracy of the data. The Hilly Upland crude , results have an average error of 2.9% and a maximum error of 8.9% which exceed the estimated accuracy of the data. For both crudes, errors are greatest for mixtures containing SO2 (mixtures 7 and 8). The results in Table IV suggest that the Simon-Graue correlation can be used to estimate oil swelling for impure COPsources containing 5-10 mol % contaminant gas. Best results are expected for less
paraffinic oils and contaminant gases other than SOz. The experimental swelling values are also compared to equation of state predictions in Table IV. Results are tabulated for two approaches. One approach (designated "pred by EOSbnin Table IV) predicted sample volumes for input pressures of the experimentally determined bubble point and 1 atm. Better results were usually obtained by using the other approach (designated "pred by EOS"" in Table IV)which divided the experimental value of sample volume at the bubble point by the predicted sample volume at 1 atm. With this approach, swelling factors predicted by the equation of state usually matched the experimental values, as well as observed with the Simon-Graue correlation. The Bath crude results have an average error ("pred by EOS"" or "pred by EOSb") of 1.1% or 1.2% and a maximum error of 1.7% or 1.8%. The Hilly Upland crude results have an average error of 0.9% or 2.3% and a maximum error of 1.1% or 5.7%. The equation of state method appears to have an advantage over
Ind. Eng. Chem. Res., Vol. 26, No. 6, 1987 1151 Table 111. Measured Values of Pressure and Volume for Isotherms of C02/Bath Field Oil Mixtures pressure, phase total vapor, liquid, pressure, phase total vapor, liquid, atm region vol, cm3 vol % vol % atm region vol, cm3 vol % vol % Mixture 5: 118.38 g of oil + 24.08 g of COz + 1.96 g of Mixture 1: 139.02 g of oil + 28.43 g of COP, 38.3 "C 35.1 L+V 370.08 48.72 51.28 H,S, 38.9 "C 31.0 L+V 35.2 L+V 369.78 48.70 51.30 391.66 58.75 41.25 29.21 L+V L+V 278.18 45.2 70.79 332.38 51.26 48.74 36.3 7.21 L+V 218.82 92.79 256.04 34.34 45.4 L+V 51.0 65.66 7.97 190.49 53.1 L+V L 203.60 0 100 101.0 92.03 175.91 L L 202.81 0 0 100 100 134.7 99.6 0 0 L 202.03 136.0 100 100 169.2 175.16 L 0 170.8 174.48 L L 201.27 0 100 100 205.0 L 200.63 0 100 0 100 204.0 236.2 173.86 L 237.6 L 173.25 0 100 Mixture 2: 139.02 g of oil + 28.43 g of CO,, 22.8 "C 32.6 L+V 319.98 40.01 59.99 Mixture 6: 118.38 g of oil + 24.08 g of CO, + 1.96 g of 35.5 L+V 276.08 28.98 71.02 HZS, 23.3 "C 40.0 L+V 220.16 9.03 90.97 26.7 L+V 390.60 58.90 41.10 67.6 L 201.09 L+V 331.02 0 100 51.22 48.78 30.1 105.1 L 200.29 0 100 L+V 253.14 35.6 33.65 66.35 199.66 0 100 42.0 136.2 L L+V 190.46 9.35 90.65 71.4 170.6 L 198.95 0 100 L 173.53 0 100 205.9 L 198.34 0 100 104.8 L 172.95 0 100 L 237.8 L 197.72 172.37 0 0 100 100 138.7 170.3 L 171.84 0 100 Mixture 3: 121.68 g of oil + 24.78 g of COP + 1.66 g of N,, 206.0 L 171.28 0 100 25.0 "C 239.0 L 170.78 0 100 27.0 L+V 484.09 67.09 32.91 38.0 L+V 343.47 52.54 47.46 Mixture 7: 75.50 g of oil + 15.15 g of CO, + 5.16 g of SOz, 79.1 L+V 191.23 7.77 92.23 39.0 "C 119.3 L 178.64 0 100 21.3 L+V 388.35 73.73 26.27 0 100 26.5 L+V 289.29 64.89 35.11 155.8 L 177.69 0 100 43.5 L+V 134.23 14.88 85.12 195.3 L 176.88 236.2 L 176.10 0 100 49.0 L 117.09 0 100 0 100 89.1 L 115.96 Mixture 4: 139.02 g of oil + 28.43 g of CO, + 1.09 g of 0 100 136.5 L 115.33 CH,,23.3 "C 0 100 185.4 L 114.71 34.2 L+V 372.45 48.95 51.05 232.6 L 114.15 0 100 34.9 L+V 371.31 48.80 51.20 43.9 L+V 275.34 28.39 71.61 Mixture 8: 75.50 g of oil + 15.15 g of Cz + 5.16 g of SOz, 91.12 8.88 21.8 "C 52.2 L+V 222.20 0 100 18.4 L+V 387.41 70.7 L 204.34 75.19 24.81 L 0 100 22.5 103.1 203.64 63.27 36.73 L+V 283.56 L 0 100 33.5 L+V 142.79 136.0 202.94 13.87 86.13 0 100 L 54.4 L 114.31 169.6 202.25 0 100 0 100 L L 201.58 0 100 97.0 113.82 204.7 141.8 L 0 100 100 L 113.34 238.0 200.99 0 191.3 L 112.84 0 100 0 238.2 100 L 112.40 Table IV. Swelling Factor Results
mixture
contaminant gas
exptl value 1.182 1.182 1.219 1.196 1.221 1.208 1.325 1.315
none none
1.183 1.181 1.198 1.198 1.199 1.196 1.243 1.236
pred by EOS" pred by EOSb Hilly Upland Crude 1.195 1.195 1.191 1.194 1.229 1.208 1.214 1.205 1.235 1.235 1.218 1.229 1.340 1.250 1.326 1.240 Bath Crude 1.201 1.190 1.209 1.208 1.218 1.206 1.264 1.245
1.201 1.194 1.209 1.215 1.221 1.213 1.252 1.242
SimonGraue correlationC
SimonGraue correlationd
1.168 1.175 1.193 1.191 1.203 1.205 1.207 1.207
1.162 1.169 1.188 1.185 1.197 1.199 1.200 1.200
1.180 1.180 1.204 1.205 1.202 1.202 1.226 1.226
1.167 1.167 1.182 1.183 1.182 1.182 1.197 1.197
OEOS predicted crude volume a t 1 atm. bEOS predicted mixture volume at measured bubble point, and crude volume a t 1 atm. 'Oil molecular weight by freezing point depression. dOil molecular weight by Watson K Factor.
the Simon-Graue correlation in predicting the effects of SOz, especially for the more paraffinic oil (Hilly Upland
crude). This effect of SO, is desirable from an enhanced oil recovery standpoint.
1152 Ind. Eng. Chem. Res., Vol. 26, No. 6, 1987 Table V. Contaminant Gas Effects on Bubble Point bubble-pt pressure change." atm/mol % eas gas Hilly Upland Bath critical temp, "C NZ +9.8 +9.1 -147.2 CH4 +2.9 +3.3 -82.8 CO2 +0.9 +1.0 31.1 H2S +0.7 +1.0 100.6 so2 -1.1 -0.1 157.2 "Change expected for each mole percentage of additional gas mixed with an equimolar C02/crude oil sample.
Table V summarizes the measured effects of contaminant gas on the bubble point. The values in Table V quantitate how the bubble-point pressure of a nearly equimolar COz/crude oil sample changes for each mole percentage of additional gas introduced. The changes appear to correlate inversely with the gas critical temperature. The most adverse effect on COz solubility was observed when Nzwas present. The increase in bubblepoint pressure caused by an additional mole percentage of gas is significantly greater for Nzthan COz and is somewhat more severe for the more paraffinic oil (Hilly Upland). Compared to COz, CHI also increased the bubble-point pressure, although to a lesser degree than Nz. The results for Nzand CH4are in excellent agreement with reports in the literature (Peterson, 1978; Orr et al., 1981). The results shown in Table V for HzS seem surprising in light of published slim tube displacement experiments which show that 25-50 mol 9'0 additions of HzS can effectively lower COP minimum miscibility pressure requirements on the order of 15-30% (Metcalfe, 1982). Although displacement results might not be directly comparable with P-V-T data, a more likely explanation for the absence of a significant HzS effect in this work is suggested by an analysis of the phase equilibria properties of COZ/H2Smixtures (Bierlein and Kay, 1953). None of the critical properties for the COZ/HzSsystem obeys an additive law with respect to composition. Instead, strong intermolecular forces create abrupt curvatures near the 100% HzS and 100% C02compositional end points; hence, as long as the HzS content is below a threshold level of about 20 mol % , the mixture essentially behaves as if it were pure COz. In the reported studies the HzS content was below this threshold level; consequently, bubble-point pressures (and swelling factors) typical for crude oil mixed with pure COz were observed. The results shown in Table V for SOz are desirable from an enhanced oil recovery standpoint. Compared to COz, SOz decreased the bubble-point pressure for both crudes. This effect is more pronounced for the more paraffinic oil (Hilly Upland), perhaps because SOz promoted extensive precipitation in the Bath sample mixture. The critical properties for the COz/SOz system also do not obey an additive law with respect to composition; however, unlike the COz/H2S system, abrupt curvatures in the critical properties do not occur (Caubet, 1900). In addition, the experimental work of Sayegh et al. (1981) has demonstrated that COz/S02 mixtures have the potential of being more soluble with crude oil than either component alone. The results obtained in predicting bubble-point properties are summarized in Table VI. Overall the performance of the equation of state in estimating bubble-point pressures was disappointing. A close match to the experimental value was achieved for only two mixtures, Hilly Upland mixture 5 and Bath mixture 3. These results may not reflect a shortcoming of the predicting method. The sample mixtures demonstrated rapid pressure changes with volume above the bubble point (Figure 2). For many
Table VI. Bubble-Point Results ~ ~ _ _ _ _ _ _ _ _ _ _ exptl EOS predictions contami- pressure, bubble-pt pressure: mixture nant gas atm vol, cm3 atm v01,~cm3 Hilly Upland Crude 1 none 52.5 209.52 c C 2 none 42.6 106.26 58.2 106.53 3 NZ 91.7 75.20 d 73.94 4 CH4 55.7 209.79 100.6 211.33 5 H2S 56.3 149.06 54.4 149.00 6 H2S 45.4 225.24 100.7 227.34 7 so2 35.7 122.43 d 114.24 8 so2 27.8 120.48 d 112.74 ~
1 2 3 4
5 6 7 8
none none N2 CH4 HZS HZS SO2 SO2
Bath Crude 52.3 204.64 41.7 201.56 83.6 179.35 55.5 204.60 55.0 176.74 43.8 173.96 46.1 116.91 36.4 114.47
~
C
C
61.6 84.9 88.4 67.9 79.1 5.9 20.3
202.19 179.38 205.71 177.18 174.96 115.86 114.13
Input sample volume measured a t bubble-point pressure. *Input experimental value of pressure at bubble-point. Base case. "Negative pressure.
mixtures, small adjustments in the input sample volume yielded predictions which matched the experimental data. These adjustments were within the estimated accuracy of the data. The predicting method was more successful at approximating the sample volume a t the bubble point. These results for Bath crude are within the estimated accuracy of the data. The bubble-point volumes predicted for Bath crude have an average error (compared to the experimental value) of 0.4% and a maximum error of 0.9%. For Hilly Upland crude the average and maximum errors in predicted bubble-point volumes were 2.4% and 6.7%. For both crudes, errors were largest for the mixtures containing SOz. To test whether these errors grew out of using the simple mixing rule to determine the critical properties and accentric factor of the gaseous pseudocomponent, measured values for the critical properties and accentric factor were tried (Caubet, 1900; Bierlein and Kay, 1953). Despite the nonadditive behavior of critical properties for the COZ/HzSand COz/SOz systems, this modification did not improve the results. Conclusions The experimental data reported here establish that contaminant gases present at the level of 5-10 mol % affect the swelling and bubble-point behavior of paraffinic crude oils in the presence of COz. The modeling results show that two relatively simple approaches can be used to predict how impure COPinteracts with reservoir oil. The following conclusions are provided by the experimental data: 1. Oil is swelled by impure CO, to about the extent observed for pure COz unless the contaminant gas is SOz and the crude oil is highly paraffinic. The increased swelling by SO2is desirable from an enhanced oil recovery standpoint. 2. Contaminant gases do not alter the observation that oil swelling by COPis temperature insensitive. 3. The presence of small levels of HzS or SOz stimulates precipitate formation in COz/crude oil mixtures, especially for less paraffinic oil. 4. Measured changes in bubble-point pressure induced by a contaminant gas correlate inversely with the gas critical temperatures. From an enhanced oil recovery standpoint, the large pressure increase caused by N2 and
Ind. Eng. Chem. Res. 1987,26, 1153-1162 the smaller increase caused by CH4 are adverse, while the pressure decreases caused by the acid gases, especially SOz, are beneficial. Additional conclusions are suggested by the modeling efforts: 1. Swelling factors can be adequately estimated for COz sources contaminated with small levels of Nz, CHI, HzS, or SOz by using either a generalized P-V-T correlation designed for pure C 0 2 or a simple equation of state. Swelling factors predided by the Simon-Graue correlation were usually closer to the measured values; however, extension of the Mulliken-Sandler approach to the PengRobinson equation of state was more successful when SOz was present and less sensitive to the chemical nature of the crude oil. 2. The simple equation of state gave unsatisfactory predictions of bubble-point pressure but adequately estimated sample volumes at the bubble point except for mixtures of the'more paraffinic oil containing SOz. Acknowledgment
The research summarized in this report was supported by funds from the US Department of Energy. Summer salary support for Dr. Monger was provided by the Exxon Education Foundation Faculty Assistance Program. Some of the experiments were performed at the US DOE Morgantown Energy Technology Center in Morgantown, WV. Nomenclature L = liquid phase P = pressure
1153
T = temperature V = vapor phase V = volume Registry NO. C H ~74-82-8; , coz,124-38-9; N, 17778-880; soz, 7446-09-5; HZS, 7783-06-4.
Literature Cited Bierlein, J. A,; Kay, W. B. Znd. Eng. Chem. 1953, 45, 618. Caubet, F. Comp. Red. 1900,130,828. Graue, D.J.; Zana, E. J . Pet. Technol. 1981, 33, 1312. Grigg, R. B.; Lingane, P. J. Presented at the 58th Annual Fall Technical Conference of the Society of Petroleum Engineers, San Francisco, CA, Oct 1983; paper SPE 11960. Holm, L. W. J. Pet. Technol. 1982, 34, 2739. Holm, L. W.; Josendal, V. A. J. Pet. Technol. 1974, 16, 1427. Kesler, M. G.; Lee, B. I. Hydrocarbon Process. 1976, 55(3), 153. Metcalfe, R. S. SOC.Pet. Eng. J . 1982, 22, 219. Metcalfe, R. S.; Yarborough, L. SOC.Pet. Eng. J. 1979, 19, 242. Mulliken, C. A.; Sandler, S. I. Znd. Eng. Chem. Process Des. Deu. 1980, 19, 709. Orr, F. M.; Yu,A. D.;Lien, C. L. SOC.Pet. Eng. J . 1981, 21, 480. Peng, D.; Robinson, D.B. Znd. Eng. Chem. Fundam. 1976,15, 59. Peterson, A. V. Pet. Eng. 1978, 50, 40. Reamer, H. H.; Sage, B. H. J. Chem. Eng. Data 1963,8, 508. Sayegh, S. G.; Najman, J.; Hlavacek, B. Presented at the 32nd Annual Technical Meeting of the Petroleum Society of the Canadisn Institute of Mining and Metallurgy, Calgary, Alberta, May 1981; paper 81-32-14. Shoolery, J. N.; Budde, W. L. Anal. Chem. 1976, 48, 1458. Simon, R.; Graue, D.J. J . Pet. Technol. 1965, 13, 102. Watson, K. M.; Nelson, E. F.; Murphy, G. B. Znd. Eng. Chem. 1935, 27, 1460.
Received for review January 22, 1985 Accepted November 11, 1986
Solid-Liquid Phase Relations of Some Normal Long-chain Fatty Acids in Selected Organic One- and Two-Component Solvents Urszula Domafiska Department of Physical Chemistry, Warsaw Technical University, 00664 Warsaw, Poland
Solubilities of octadecanoic acid and docosanoic acid were determined from 290 to 340 K in several pure solvents and in several binary solvent mixtures. The effects of temperature changes on the observed solubilities were in close agreement with the general principles of thermodynamics, using the Wilson equation to represent the activities in the solutions. Use of the Scatchard-Hildebrand regular-solution theory was less successful, even though the chemical association of the acids was allowed for. A knowledge of the solubility characteristics of longchain fatty acids in organic solvents is important to fats and oil technology and research. Extensive solubility data have already been published (Singleton, 1960), but new processes with two-component solvents require these data to be constantly supplemented. Ralston and Hoerr (1942, 1945), Hoerr and Ralston (1944), and Hoerr et al. (1946) have measured the solubility of stearic acid in more than 20 solvents of different chemical properties. Preckshot and Nouri (1957) has dealt with halogen derivatives of hydrocarbons, whereas Kolb (1959) has used various hydrocarbons at low temperatures. Harris et al. (1968) has worked with N,N-dimethylformamide, and N,N-dimethylacetamide. Bailey et al. (1969) have made comprehensive investigations of the solubility of octadecanoic acid and docosanoic acid based on the literature data and 0888-5885/ 871 2626-1 153$01.50/0
their own measurements. The literature data concern solubilities of these two acids in a series of more than 50 one- and two-component solvents (Brandreth and Johnson, 1971). However, the discrepancy of solubility measurements, obtained by various investigators, is observed, due to the differences ih purity of used components, to .be the best solubility of stearic acid in chloroform, N,N-dimethylacetamide, and N,N-dimethylformamide. According to the literature data, aliphatic halogen derivatives, especially asymmetric compounds with a great number of chlorine or fluorine atoms in molecules, appear to be the best solvents at a temperature range of 290-300 K. Pure alcohols are not attractive solvents, but the increment of the solubility of acids with an increase in the chain length of alcohol molecules (C&) is observed. The azeotropic mixtures of alcohols with halogen derivatives of hydro0 1987 American Chemical Society