Evaluating Constant-Volume Depletion Data Curtis H. Whitson, Rogaland Distt C Stein B. Torp,* Norwegian Ist. of Technology Abstract This paper presents methods for evaluating constant- volume depletion (CVD) data obtained from experimen- tal analyses of gas condensates and volatile oils. ‘Theoretical and practical developments are supported by ‘experimental data from a North Sea gas-condensate Aid, ‘The three major contributions of the work are: (1) presentation of material-balance equations to calculate fluid properties from measured CVD data, (2) a simple ‘method for calculating black-oil formation volume fac tors and solution GOR’s using material-balance results| and a separator flash program, and (3) investigation of the Peng-Robinson (PR) equation of state (EOS) as a tool for matching measured PVT data and studying vapor/liquid phase behavior during CVD. Introduction CVD experiments are performed on gas condensates and volatile ols to simulate reservoir depletion performance ‘and compositional variation. Measured data can be used in a variety of reservoir engineering calculations, among, the mast useful being material-balance calculations, ‘generating black-oil PVT properties and, more recently, the tuning of empirical EOS. All these applications are addressed in this paper. Few engineers are aware of useful fuid properties that ‘can be derived from CVD data—e-g., liquid composition (and therefrom K-values), density, and molecular mass of the C;~ fraction, vapor density, and ‘oral system molecular mass. Only experimental CVD data and sim- pile material-balance equations are used to calculate these data. A procedure outlining these calculations was first presented by Reudelhuber and Hinds.! Their descrip- coy ns Saoey of Peoun Engrs of AME 610 tion, however, is somewhat difficult to follow and not extensively known or used by petroleum engineers. The rmaterial-balance relations are presented here in equation form using current SPE nomenclature. Based on material-balance-derived properties, a method is proposed for calculating black-oil PVT prop- erties — i.e., FVF's and solution GOR’s used in two- phase flow equations and reservoir material balances. ‘The method was first suggested by Dodson er al.? in 1953 for solution-gas/crude-cil systems. Their method, however, requires expensive and time-consuming liquid sample removals and experimental flash separations ‘The proposed method follows the Dodson ef al. pro- cedure but uses experimentally determined vapor com- positions and material-balance- = —0.02, CO, — hydrocarbons = 0.15, Nz ~ hydrocarbons = 0.12, and C, Cy, m = 6, JOURNAL OF PETROLEUM TECHNOLOGYTABLE 2—COMPOSITIONAL AND PROPERTIES DATA OF FLUID NS-1B SAMPLED FROM A WELL OFFSETTING 'NS-1 COMPARED WITH CALCULATED DATA GENERATED USING THE METHOD PRESENTED IN REF. 11 Measured Calculated” Single Boiling Upper Garon Mole Molal Speciic —Kygp Point’ Mole Molal Molal Number Percent Mass Graviy. _Facior (CR) Percent Mass Mass 7 094 95 07158 1205 6417 0935 951 996 8 084 104 07355 1193-6783 0898 1054 111.9, 8 ord 118 07557 11.90 727.3 0739 1188 1268 10 060 132 07639 ©1199-7684 0600 1942 1424 tbat 144 07723 1204 © BOLD 0410 1487 1555 42 034 154 O7e14 1204 832.7 0340 161.7 1684 13031 167 07999 1203 B72 0310 1750 1822 14 026 180 0.053 12.02 © 80700 0.260 198.7 1957 18022 197 08098 1213 947.1 0.220 2021 200.1 18 019-212 08152 1230 1008.1 0.190 D154 220.4 17 O47 226 0.8265 1219 1018.0 0.170 2289 296.1 18 015 298 0.8903 1220 1039.4 0.150 2428 250.1 48 043 250 OMGs1 1225 1069.4 0.190 2568 264.1 2 O11 262 0.8400 12.28 © 10976 0.110 2708 2778 21 008 277 08477 1229 © 1190.8 0.080 2830 2800 22 007 292 0.8591 12.92 1161.0 0.070 294.1 209.9 23-006 808 08577 12.96 11914 0.060 3048 3103, 24 006-329 0.8666 12.98 12349 0.070 3167 3280 25+ 051 471 0.8826 1287 1465.6 0.508 499.1 = ei 17 0.061 12.25 6.190 1770 12.02)" he gana detitan ff 1) ma oad ara gpa aha 0.72 was ns (rn Toco. which were estimated using a linear fit of the Katz and Firoozabadi"® data (their Table 2), 14-74 = 0.0668 2 ‘The interaction coefficient between methane and the heaviest component then was adjusted until a match of the measured dewpoint pressure was obtained. Fluid Description: Rich Gas Condensate NS-1 NS-I is a rich gas condensate first tested at a GOR of 5,500 sef/STB (980 std m*/stock-tank m?) from an in itial reservoir pressure of 7,300 psia (50 MPa) and temperature of 280°F (138°C). Stock-tank oil gravity was 44°API (0.8055 g/em?). Separator samples were taken while flowing the well at 16.3 MMscfiD (46010? std m3/d) and a flowing bottomhole pressure ‘of 7,260 psia (SO MPa). Table 1 presents constant- volume depletion data measured on the recombined reservoir fluid Heptanes-Plus Characterization Extended compositional data of the C7~ fraction were rot available for the NS-I fluid, only molecular mass and specific gravity. Complete true-boiling-point (TBP) data were, however, available from an offset well. These data were adapted (0 the NS-1 fluid using the methods presented in Ref. 11 Molar distribution (mole fraction vs. molecular mass) of the offset well fluid was fit using the gamma distribu- ‘ion parameter alpha and variable upper-boundary MARCH 1983 cso Oper mca macs ee rd by ng be eared ‘molecular masses. The optimal alpha was 0.712 for eta (minimum molecular mass in the C7 fraction) of 92. Table 2 gives results of the match. Molar distribution of the NS-I C7» fraction then was calculated using a = 0.712, 9 = 92, and Mz = 184 (as compared to 177 for the NS-Ib fluid). Results are presented in Table 3, ‘Tuning the Peng-Robinson Equation of State ‘Single carbon number groups were combined into five multiple carbon number (MCN) groups: C7 through Co, Cio through C13, Cig through Cy, Cys through C4, and Ca5-+, as suggested in Ref. 11. Group properties ‘were calculated using Kay's mixing rule, Methane in- teraction coefficients were estimated using MCN specific gravities and Eq. 22. Using these data, the PR EOS yielded a dewpoint pressure much lower than measured, The Cy through Cos interaction coefficient then was increased until dewpoint pressure was matched. ‘The CVD program was run using the MCN properties (Table 3). The overall match was good to excellent, ex- cept for liquid volumes, which were much too high (32% simulated maximum vs. 22% measured maximum). To check whether measured volumes were low, we com- pared material-balance liquid densities with Alani- Kennedy densities (using material-balance composi- tions and molecular masses). Table 4 shows results of the comparison, indicating that measured volumes are consistent except for perhaps small errors in the first two volume measurements (On the basis of these results, we decided to lower the isTABLE 3—PHYSICAL PROPERTIES OF THE C,- SINGLE AND MULTIPLE CARBON NUMBER GROUPS USED IN THE PENG-ROBINSON EQUATION OF STATE "TO DESCRIBE RESERVOIR FLUID BEHAVIOR OF THE NS-1 FLUID Critical Single Boling Temper Methane Carbon Mole Molal Specifie Point ature Pressure centric Interaction Number Percent Mass Gravity _(°R)__(°RI__(psia)__Factor_ Gostficient 7 12196 953 07177 e468 O71 57S 0.2742 8 $1730 1065 0.7409 6905 1021.4 ©4234 0.2086 9 0.8600 1207 0.7599 7094 10734 «3832 (O34 1006672 1347 O.7682 7818 11129 3459 0.9861 1 05681 1487 0.7781 8222 115213168 (0.4251 420478 1627 0.7908 0631 11923 2940 0.4602 13 04074 3767 0.8034 9028 128122749 0.4088 14 03496 1907 08153 9412 12683 Ded 0242 45 0.3021 2047 08168 9726 12942 240.4 OSTEO 18 02621 2187 OMD10 10298 19215 257 OBTa7 17 02282 2327 O8G10 10905 13540 2146 0.8485 18 01992 2456 ea T0720 13838 2043685 1901744 260.6 0.8492 11047 14087 1840 0.7246 2 © 01529 2748 esa T1918 14384 ©1851 0.7634 2101343 2886 0.8554 T1619 14610 1774 O.8OTE 22 1182 9026 0.8602 11902 14849 ©1700 0.8416 2% © Q4D41 3168 0.8699 12174 15070 1629 0.8890 2 © —«G.0918 390.5 0.8690 1245.1 15902 1868 0.9241 25+ 0.7054 962.3 0.9292 19880 17346 © 91.41.0590 52100 188.0 0.8160 ‘Multiple Carbon Number Properties Used in the Final CVD Simulation Tiough® 3.2466 1061 0.7365 6833 10165 425.5 O.s0e¢ 0003659" sO through 19 21410 1527 07697 8350 11636 3131 0.4948 0.04292" ‘4tnvough 17 1.1421 2092 0.8205 9840 13085 © 237.4 0.5856 0.08807" TWtrough 24 0.9749 2515 0.8524 11469 14460 182.7 0.7832 0.05254" 25+ 0.7054 4623 09192 12761" 15642" 1688" 0.8819°" 0.18400T ‘32100 1840 o8t60 ‘sts es oesmsng a Mg cor 140 The cl vale cespond 24 Kng tr 1242 nd oe gon above Goce, {higian save veda mach be eased deport ase ‘TABLE 4—CALCULATED LIQUID DENSITIES AS A FUNCTION ‘OF PRESSURE FORNS-1 FLUID Calculated Liquid Densities (a/em) Teasured CVO Data ‘Simulated CVD Data ‘AlaniKennedy Density AlaniKennedy Material (Material Balance Density Pressure Balance Liquid PR (PR Liquid -"Ipsia) Density Properties)" __Density’* Properties)" 35147 0670 0.608 0541 0570 43147 0.680 0.549 0.584 0595 31147 0.688 0870 0.580 0.632 21187 0.700 0.682 0.608, O66 4247071 0.700 0.696 0.692 47 0.722 om 0.653 0707 “he AiKetnadycentyconlaon request crpostone a ud molecu {Src Tose ea were avaie fom sir wstorsouee eaeuarare oP “gta tng te Sl ae Fe ire even wor EO Gennes ak 1a ge tones er on om te een ve JOURNAL OF PETROLEUM TECHNOLOGYTABLE 5—SIMULATED CONSTANT-VOLUME DEPLETION DATA FOR THE NS-1 FLUID AT 280°F USING THE PENG. jOBINSON EQUATION OF STATE (psia) Compositions Eauilbaam Equilibrium Vapor 7 Laue Expert Caleu menial Component o7es7 S547 40147 9147 BN47 1247 747 _ TAT NAT Gabon cioxice 2370 2409 2.467 2497 2541 2576 2583 0.580 0.595 Nitrogen 6310 0923 0.396 034 0943 05% 0921 0.020 0.029 Motnane 73.180 753.549 7684 79.105 79712 7222 77772 12420 11.939 Ethane ye00 "7779 "7753 "773 “057 8372 B71 3360 3.623 Propane 3550 3474 3405 3363 S444 3660 3989 2900 3.183 Isobutane O70 0686 0800 OGu OGa7 0691 O78 0810 967 ‘routane S480 $590 1.906 $267 1282 137 1567 2000 2.14 isopentane 0640 0608 0564 0590 0516 0548 0698 1.400 1.509 mmpetane 0.500 0633 OSe2 0850 0.832 0863 0.659 1.600 1.770 Hoxanes ‘om 0996 0888 0769 0727 o7a4 0877 3.880 4.228 Heptanes-plus S210 6457 4543 2969 2196 11895 2.105 71.010 69.897 Totals 100,000 100.000 100,000 100.000 100.000 100.000 100.060 100.000 100.000 Me, 1840 161.0 427 1295 1212 N64 1145 2180 2081 a dare 0709 0769 "0770 0762 0758 0.757 0833 0.843 z 1209 4037 0997 0890 0.886 0911 0.938 ‘np. % 01000 9637 22.581 39.492 56.106 72.413 81.595 St 000 1955 26.11 2665 25.11 23.00 21.58 : i teeter | 5 z A TINetneasuaes bata TS Mo 3 Asie (ACTUALY THE RUA WAVE BEEN a ] 5 oraces a apsusieo AccoweiNe ae ees Pe 20g eT E g ont 2 Fee c i 3 . 2 fe z —— a $s ote Ss ye eof wos’ ‘eMEASURED Cr) Pao F $365 "ido0 Foo 05a fan StS 9000 Fig. 6—Caleulated and measured vapor compositions vs. pressure for NS-1 at 280°F, PR liquid volumes by adjusting the characterization fac- tor of the Cs fraction. Lowering the factor from 12.24 to 11,80 resulted in an 8% decrease of the liquid volumes for the maximum drop-out (from 32% to 26%). ‘The adjustment had litle effect on other estimated data. To have lowered the Ky factor more would have created a physically untealistic system. Adjusted physical properties for the C,s~ fraction are found in ‘Table 3, as is the methane interaction coefficient used to MARCH 1983 Fig. 7—Calculated heptanes-plus molecular masses vs. pressure for NS-t at 280°F. adjust dewpoint pressure. Complete results of the CVD simulation are presented in Table 5. PR liquid densities are compared with Alani-Kennedy estimates in Table 4. Many other adjustments of the C7+ characterization procedure were tried. None of these were particularly helpful, though some are worth mentioning: (1) extend- ing the C; + split to Czp~ such that the last component was very heavy, (2) increasing the number of MCN ‘groups used to nine, C2s+ inclusive, (3) splitting the orPRessuat o MATERIAL BALANCE aESULTS Sebs z NEB sug 1 7 1 e = + 3 a ‘2000 "3000 De ae onPonENT CHARKETERIZATION FACTOR, F20(1/Tq = 1/79 Fig NS-1 Kealues al 280°F calculated using the materia balance approach, Fig, 10—Siopes and intercepts otlog Kp vs. Fplois vs. pressure for NS at 280 and 340°F. Fig. 9—NS- Kwvalues at 280°F ealeulated using the PR EOS. ois 0 ws=T —COsTANT COMPOSITION 280° CONSTANT VOLUME DEPLETION 10 10" 10" EQUILIBRIUM CONSTANT, K = y/x | i i i z 5x10" 10 io PRESSURE, psia Fig, 11—PR Kevalues for NS-1 at 280°F representing two deple tion processes. JOURNAL OF PETROLEUM TECHNOLOGYC7 + fraction into eight SCN groups and a C5 fraction, and (4) using TBP Kop factors instead of those estimated from Eq. 23. Results and Discussion Fig. 6 compares measured (or more correctly. smoothed”) vapor compositions with those simulated using the PR EOS, The match is excellent, showing only slight deviation for the C7 = and Cg components. Devia- tion of the hexane component is probably because of its incorrect characterization as n-hexane, Fig. 7 presents heptanes-plus molecular masses of liq- uid and vapor phases and the total system. Simulated and ma lance-derived values match well. Our ex perience has shown that a good match of C; - molecular mass using the PR EOS is usually difficult and very dependent on proper characterization of the plus fraction Calculated equilibrium constants were correlated using the Hoffman et al.® method. Fig. 8 presents NS-l_ K- values calculated using material-palance relations, The log Kp vs. F plots are linear and appear to approach a ‘common point. The convergence point can give an estimate of the apparent convergence pressure. Actually, the most accurate value is obtained by extrapolating the slope vs. pressure curve to zero, as done in Fig. 10. The resulting estimate of pg is 8,000 psia (55 MPa. Fig. 9 presents NS-1 K-values calculated from the EOS. Once again linear plots of log Kp vs. Fconverge to ‘2 point. From the extrapolation of slope to zero in Fig. 10, px = 7,500 psia (52 MPa). Another interesting feature shown in Fig. 9is that heavy components are bet ter correlated using the log Kp vs. F methods at higher pressures ‘Temperature effects on the log Kp vs. F plots of NS-1 ‘were investigated by running the EOS simulator at 340°F (171.1°C), about 60°F (15.5°C) higher than reservoir temperature, Results indicate that temperature influence is (1) largest for heavy components at large pressures, (2) negligible at low pressures (as was found in Ref. 8), (3) relatively small compared to the influence of pressure, and (4) not significant in changing the ap- parent convergence pressure of the system. For lighter systems, there does not always appear such, ‘unique convergence point for log Kp vs. F plots. We thought that this perhaps resulted from a change in the total composition of the system or from alteration in the heptanes-plus properties. We investigated these possibilities by running a constant-composition simula- tion of NS-1, Resulting K-values were compared with CVD K-values and are presented as log K vs. log p plots in Fig. 11. This plot indicates that compositional change during constant volume depletion is not significant ‘enough to influence K-values or convergence pressure. Conclusions 1. Measured CVD data for a gas condensate were analyzed using simple material balances and the PR EOS. 2. A simple method was proposed for calculating black-oil PVT properties (FVF’s and solution GOR’s) of gas condensates and volatile oils 3, Simulated constant-composition and CVD studies [MARCH 1983 of lean and rich gas condensates using the EOS indicate that K-values are independent of the depletion process. 4. Equations of state usually overestimate liquid volumes of petroleum reservoir fluids. The problem was improved by reducing the Watson characterization factor of the heaviest component. Nomenclature Aor A(p) = slope of logkp vs. F plot 'b = slope of the straight fine connecting the critical point and atmospheric boiling point on a log vapor pressure vs. I/T plot, cycle-°R (cycle-K) B = formation volume factor, bbl/STB (m3/stock-tank m3) F or F(T) = component characterization factor. cycle -quilibrium constant (K-value) Warson characterization factor mass, bm (kg) molecular mass, Ibm/lbm-mol (kg/kg-mol) noles, lbm-mol (kg-mol), exponent in K-value correlation pressure, psia (KPa) niversal gas constant, 10.732 psia-cu fmol-°R (9.3143 J/mol-K) Rg, = vapor solution GOR, scf/STB (td m?/stock-tank m3) Rio = liquid solution GOR, scf/STB (std m?/stock-tank m3) S'= saturation, fraction or % T = absolute temperature, °R (K) V = volume, cu ft (m?) x = liquid composition, fraction or % y = vapor composition, fraction or % Yor ¥(p) = intercept of log Kp vs. F plot ‘2 = total system composition Z = vapor compressibility factor ‘y= specific gravity relative to air or water (60160) 4 = interaction coefficient A= incremental p= density, Ibmicu ft (kg/m?) = acentric factor Subscripts a = atmospheric b = bubble point (7) or boiling (T») ¢ = critical cell = cell, pertaining to PVT cell volume d = dewpoint a= Bas i = index for summation | = component identifier k = depletion stage K = convergence owL = liquid phase n= carbon number o=oil p = produced = reduced saturated (bubble- or dewpoint) standard conditions stock-tank oil total (two-phase) ‘vapor phase Acknowledgments We thank H. Norvik, H. Asheim, D. Murphy, V. Dalen, and G. Nielsen for useful comments concerning. this paper. We also acknowledge computer time and facilities provided by Rogaland Dist. C., Norwegian Inst. of Technology, and Continental Shelf Inst. We thank Phillips Petroleum Co, Norway for contributing needed fluid data to the petroleum literature. Suppor from Frying Pan Publications Ine. is appreciated. References 1. Readelhuber, FO. and Hinds, R.F-: "A Compositional Matera Balance Method for Prediction of Recovery from Volatile Oil Depletion Drive Reser,” Trans, AIME (1957) 210, 19-26 2. Detson, CR. Goodwill, D., and Mayer, E.: “Application of Taboratony PVT, Data to Reseroit Engineering. Problems, Trans, AIME (1953) 198, 287-98. 3, Frocabadi, A, Hekim, Y-, and Katz, D.L. “Reservoie Deple- tion Calelation for Gas Condensats Using Extended Analyses in the Peng Robinson Equation of tate," Cat. J. Chem. Eng. (Oct. 1978) $6, 610-15. 4, Conrad, P.G. and Graver, 1. ‘Sate Checks Vaidiy of PVT Experiments (pei 21, 1980) 77-86, “Peng-Robinson Equation of ‘Oil and Gas J. 20 5. Stnding, M.B.: Volumetric and Phase Behavior of Oil Field Hydrocarbon Syutems, SPE, Dallas (197?) 6, Hoffmann, A.E.. Cramp. JS. and Hoeot, C.R: “Equa Constants for a Gas-Condensate System,” Trans. AIME (1953) 198, 1-10. 7. Glaso, 0, and Whitson, CHL: "The Accuracy of PVT Parameters Caleuited from Computer Flash Separation at Pressures Less ‘Than 1,000 psa," J. Pet. Tech. (Aug. 1982) 1811-13. 8. Standing, MiB. "A Set of Equation for Computing Bgiibium Ratioe of a Cride Oil/Natral Gas System at Pressures Below 1,000 pia." J. Pet. Tech. (Sep. 1979) 1193-95, 9, Sinding, M.B. and Katz, D:L-: “VapoeLiquid Equilibria of Natura Gas-Crude Oil Systems,” Tran, AIME (1948) 188, 232-45. 10. Alani, G.H. and Kenedy, H-T.: “Volumes of Liquid Hydroear- bons a High Temperatures and Pressures,” Trans.» AIME (1960) 219, 288-92 11 Whitson, CH: “Chanacterring Hydrocarbon Plus Fractions," paper EUR 183 presented at the 1980 European Offshore Petroleum Conference and Exhibition, Londoa, Ost. 21-24 12. Katz, DL. snd Piroozabadi,A.: "Predicang Phise Behavior of| Condensat/Crade-Oil Systems Using Methane Interaction Coe cients,” Trans, AIME (1978) 268, 1649-85 SI Metric Conversion Factors APL 141,5/(131.5+°APD bbl x 1.589873 E-01 = m> cu ft X 2.831 685 E-02 °F CRF~32)/1.8 psi X 6.894.757 E+00 sof x 2.863640 E-02 tyra man noms Saco peu Eon i 4 "HE Foe fare coo tes pcre ar fet'SE anna acme Cor enc td Eanonon hls San io Ont 57 JOURNAL OF PETROLEUM TECHNOLOGY