You are on page 1of 17

Third SPE Comparative Solution

Project: Gas Cycling of Retrograde
Condensate Reservoirs
Douglas E. Kenyon, SPE, Marathon Oil Co.
O. Aida Behie, SPE, Dynamic Reservoir Systems

Summary. Nine companies participated in this artificial modeling study of gas cycling in a rich retrogradegas-condensate reservoir. Surface oil rate predictions differ in the early years of cycling but agree better late
in cycling. The amount of condensate precipitated near the production well and its rate of evaporation varied
widely among participants. The explanation appears to be in K-value techniques used. Precomputed tables for
K values produced rapid and thorough removal of condensate during later years of cycling. Equation-of-state
(EOS) methods produced a stabilized condensate saturation sufficient to flow liquid during the greater part of
cycling, and the condensate never completely revaporized. We do not know which prediction is more nearly
correct because our PVT data did not cover the range of compositions that exists in this area of the reservoir
model.

Introduction
SPE conducted two earlier solution projects, 1,2 both designed to measure the state-of-the-art simulation capability
for challenging and timely modeling problems. The first
project involved a three-layer black-oil simulation with
gas injection into the top layer. 1 Both constant and variable bubblepoint pressure assumptions were used. Model
predictions were in fair agreement. No simulator performance data (run times, timestep size, etc.) were given.
Seven companies participated in the project. The second
project was a study of water and gas coning with a radial
grid and 15 layers.2 Authors of the project felt that unusual well rate variations and a high assumed solution
GOR contributed to the difficulty of the problem. Some
significant discrepancies in oil rate and pressure were obtained. Eleven companies joined in the project.
For the third comparative solution project, the Committee for the Numerical Simulation Symposium sought
a compositional modeling problem. Numerical comparisons of the PVT data match were considered important.
Speed of the simulators was not to be of major interest.
The problem we designed is the outcome of this fairly
general request. Some features of interest in current production practice of pressure maintenance by gas injection
are included. The results confirm the well-known tradeoff between the timing of gas sales and the amount of condensate recovered. Several features of interest in a more
complete examination of production from gas-condensate
reservoirs are ignored. These include the effects of nearwell liquid saturation buildup on well productivity and of
water encroachment and water production on hydrocarbon productivity. We did not address the role of numerical dispersion. In addition, the surface process is
simplified and not representative of economical liquid
recovery in typical offshore operations. We simplified·the
surface process to attract a larger number of participants
Copyright 1987 Society of Petroleum Engineers

Journal of Petroleum Technology, August 1987

because not all companies had facilities for simulating gas
plant processing with gas recycling in their compositional simulators.
Nine companies responded to the invitation for participation. Table 1 is a list of the participants in this project.
Participant responses were well prepared and required a
minimum of discussion. We invited all the companies to
use as many components as necessary for the accurate
match of the PVT data and for the simulation of gas cycling. Companies were asked to give components actually
used in the reservoir model, how these components were
characterized, and the match to the PVT data obtained
with the components.
We first outline the problem specifications, including
sufficient data for others who may wish to try the problem. the pertinent PVT data are given. We show each
participant's components, the properties of these components, and the basic PVT match obtained. In many cases,
EOS methods were used exclusively, but in others, a combination of methods was applied. The results of the reservoir simulation are given and comparisons are shown
between companies for both cycling-strategy cases. Finally, some facts regarding simulator performance are
given, although this information was voluntary.

Problem Statement
The two major parts to a compositional model study are
the PVT data and the reservoir grid. For the PVT data,
participants were supplied with a companion set of fluid
analysis reports. The specification of the reservoir model
is given in Tables 2 and 3 and the grid is shown in Fig.
1. Note that the grid is 9x9x4 and symmetrical, indicating that it would be possible to simulate half the indicated
grid. Most participants chose to model the full grid. Note
also that the layers are homogeneous and of constant
porosity, but that permeability and thickness vary among
layers.
981

00 0.O.450 Saturation Data Phase Saturation Marathon Oil Co. CO 80160-0269 McCord-Lewis Energy Services P.040 0.900 1.00 0. explicit saturation (IMPES) models.250 0.76 0.00 0.058 0.. P.500 0.3 1.2 7. TX 75245 Petek.92 0.0 9.. psi Initial pressure at contact.13 7360 II 7400 II 733011 7450 It 0.060 0. August 1987 . TX 75247 Computer Modelling Group (CMG) 3512-33 Street N.1 PV compressibility.119 0.72 0. The producer is not in the very corner of the grid.550 63.005 0.400 7.36 0. In this area. Most of the area behind the producer undergoes pressure depletion only because it is not swept by injection gas.3011 20md.DATUM = 7500 II (subsurface) INJECTION COMPLETIONS PRODUCTION COMPLETIONS Fig.00 0.002 0.260 0.88 0. NZ=4 DX = DY = 293. Box 2819 Dallas.7 0.330 0.52 0.500 1. 130md.082 0. psi .O.00 0.3 0.620 0.513 0.5 0.0 0. P. Calgary.126 0.187 0.24 0. CO 80202-2699 7.84 0.W.680 0.400 0. P.800 GaslWater Capillary Pressure (psi) 0.60 0.00 0. 50 II 150md.O. 1-Third comparative solution project 9 x 9 x 4 reser· voir model grid.00 0. Ibm/ft 3 compressibility.00 0.00 ~ ~ ~ 0. Fourth Floor Denver.00 o 3.650 0. Canada T2L 2A6 Soc.20 0.026 0.505 0. Alta. retrograde condensation occurs without significant evaporation by recycle gas to simulate areas of minimal sweep in a real reservoir.620 0.040 0.227 0.50ll. Box 45307 Dallas.7 2.562 0. N-7034 Trondheim NTH Norway • Now Scientific Software-Intercomp. TX 75221 Grid Data NX=NY=9.7 1.00 0.0 3.300 0.277 0. Avenue des Lilas 64018 Pau Cedex France Intercomp' 1801 California S!.049 0.68 0. ft Water saturation at contact Capillary pressure at contact.196 0. Journal of Petroleum Technology.00 0.066 0.00 0.08 0.00 0.64 0. however.2 3.156 0.012 0.6 Layer Horizontal Permeability Vertical Permeability Thickness (ft) Depth to Center (ft) 1 2 3 4 130 40 20 150 13 4 2 15 30 30 50 50 7.48 0.450 0.04 0.O.005 0.462 0.12 0.O.013 0.010 0.150 0. The very small compressibility and volume of water.0x10.40 0.186 0.1 1. The grid size selected represents a reasonable grid for certain offshore applications but is somewhat too refined for a full-field simulation.315 0-400 0.000 >50 >50 >50 >50 50 32 21 15. 982 The grid size sets the value of numerical dispersion in these implicit pressure.090 0. CA 90631 Core Laboratories Inc.112 0.348 0.6 4. psia Water properties density at contact.56 0. The initial conditions for the location of the gas/water contact and the capillary pressure data generate a water/ gas transition zone extending into the pay layers.1 0.44 0.00 0.80 0.311 4> E~~~~~I~~ = 0.32 0.28 0.0 5.540 0. H293.3 4.TABLE 1-COMPANIES PARTICIPATING IN THIRD SPE COMPARATIVE SOLUTION PROJECT TABLE 2-RESERVOIR GRID AND SATURATION INPUT DATA ~es!ilrvoir Arco Oil and Gas Co.033 0.4 0. 3011 40md.360 7.00 0. Natl.390 0. Box 269 Littleton.078 0.0 Capillary pressure for gas/oil is assumed to be zero.150 0. The Petroleum Technology Research Ins!. psi -1 Chevron Oil Field Research Co.96 1.100 0. .0x10. Box 47547 Dallas.5 12. 7600 Carpenter Freeway P.3 ft Datum (subsurface). ft Porosity (at initial reservoir pressure) Gas/water contact.024 0.740 0.020 0.710 0. Box 446 La Habra.2 0.13 7. Elf Aquitaine 26.330 7.222 0.4 2.16 0.800 0.

86 100.74 to 0. Layer 1 is a high-permeability layer (130 md) with rapid movement of injected gas. Reservoir pressure falls rapidly during the last years of cycling in Case 2 and surface liquid falls accordingly.81 8. reduces available sales gas volume.71 9.403 4. 2. 4 (bottom layers) Radius. which is completed in Layers 3 and 4.39 0.7737 140 0. psi Sales Rate Case 1 (constant sales rate to blowdown) 0<t<10 years: 1. liquid is random condensate sample.871 1.39 1. We speculate that buoyancy. August 1987 983 . Injected gas is taken from the combined vapor streams of the three-stage separation.476 2.94 0.119 1." The path of migration of injected gas is along Layer 1 with a turn downward only in a small zone around the producer.00 8. Production is controlled by a specified separator-gas rate.000) =0. then switch to primary separator at 315 psia. but our review of saturation array data revealed that most of the injected gas that reaches the producer in Layer 4 has come across Layer 1 and turned downward as it approaches the producer.63 0.200 500 Injection Well Data Location I =J = 1 Perforations K = 1.000-psi [6. and the models were run to 15 years or 1.78 8. which- TABLE 4-HYDROCARBON ANALYSES OF SEPARATOR PRODUCTS AND CALCULATED WELL STREAM Component Separator Gas' Well Stream Separator Liquid (mol %) (gallscfx 10 3 ) (mol %) (gallscfx 10 3 ) (mol %) 0.736 Calculated gross heating value for separator gas = 1. Case 1 uses a constant recycle-gas rate (4. Layer 4 is also a highpermeability layer (150 md).34 1.37 0. and some extra water in Layer 4 explain the favored flow of dry gas through Layer 1.777 0.04 1. More gas is recycled in the critical early years in Case 2.03 8.774 140 Calculated separator-gas gravity (air = 1.00 2.000 Case 2 (deferred sales) 0< t< 5 years: 500 MscfID 5 < t < 10 years: 2.59 3.13 1. rw' ft Rate (separator gas rate).9-MPa] average reservoir pressure.27 nil nil 6.000 psia. and Sales Data Production Well Data Location 1= J = 7 Perforations K = 3.500 MscfID t> 10 years: all produced gas to sales 4.40 as condensate builds to this saturation with irreducible water present. 2 (top layers) Radius.363 5.734 1. MscfID Minimum bottomhole pressure.000 psia 'Gas synthetically prepared in 1he laboratory.52 0. the two cases provide for exactly the same amount of recycle gas to be injected over the duration of the cycling period (10 years).00 Carbon dioxide Nitrogen Methane Ethane Propane Isobutane n-Butane Isopentane n-Pentane Hexanes Heptanes plus Total Properties of heptanes plus API gravity at 60°F Specific gravity at 60/60°F Molecular weight 51.4 0.812 scf/bbl at 72°F. This promotes pressure maintenance and increases surface liquid yield (less condensation in the reservoir) but TABLE 3-WELL AND SEPARATOR INPUT DATA Production.21 1.500 MscfID t> 10 years: all produced gas to sales Separator Pressures and Temperatures Separator Primary' Primary' Second stage Stock tank Pressure (psia) 815 315 65 14. Note that condensate is immobile up to 24% saturation and that krg is reduced from 0. Injection.99 12.21 0:52 0.189 1.488 2. Journal of Petroleum Technology. gas and liquid not in equilibrium at 2. Case 2 uses a somewhat higher rate (5.700 Mscf/D [105 X 10 3 std m 3 / d)) for the last 5 years of cycling.57 0. Volumetrically.7 Temperature (OF) 80 80 80 60 * Primary separator at 815 psia until reservoir pressure (at datum) falls below 2.65 psia and 60°F ' Primary separator-gas/separator-liquid ratio 4.571 6.98 10. Two cases were requested and differ by the recycle-gas rate assumed. Liquid production by multistage separation is the unknown to be predicted.745 100.756 35. high vertical permeability.617 6.69 6.50 0.110 0.729 100.33 65.33 0.700 MscflD [133x 10 3 std m 3 /d)) for the entire cycling period.39 1. The produced gas becomes a mix of reservoir gas with "dry gas. Relative permeability data were based on the simplistic assumption that the relative permeability of any phase depends only on its saturation.91 4. psi 1 6. rw' ft Rate (separator-gas rate minus sales-gas rate) Maximum bottom hole pressure.216 Btulft 3 of dry gas at 14. Blowdown (all gas to sales) starts at the end of the 10th year of cycling.700 MscflD [161 X 10 3 std m 3 /d)) for the first 5 years of cycling and a somewhat lower rate (3. The primary separator pressure depends on reservoir pressure as given in Table 3.12 0.500 psia.make water rather insignificant for this problem.67 0.55 78.23 2.

400 3.1 15. At least one participant chose to predict surface volumes without recourse to the data in Table 9 because such data are calculated.000 2. the less severe swelling and dewpoint pressure excursions in the reservoir model should be adequately covered.69 5.08 0. These fluids were physically recombined at a gas/liquid ratio of 4. However. Most participants chose to use these data to match surface volumes produced by reservoir gas processed in the multistage separators.400 3. none of the companies used this many components for the PVT match. We will see later that all companies matched the relative volume in expansion accurately but that there were some minor differences in calculated Z factors.428 3. PVT Data Measured PVT data are given in Tables 4 through 15.0043 1.129 1.2 10.000 1.7 8.1 15.803* (dewpoint) • Gas expansion factor = 1 .50 1.2 17.49 0. This peculiarity was spelled out in the cover letter of the fluid-analysis report sent to all potential participants.0997 1. and swelling data of four mixtures of reservoir gas with lean gas.000 5.2 3.933 0. Retrograde condensate observed during constantvolume depletion of the original mixture is shown in Table 7.12 1. ever occurred first.26 0. including calculated Z factors at and above the dewpoint pressure.400 2.8856 0.812 scf/STB [857 std m 3 /stock-tank m 3 ].05 0.200 700 o • First depletion level.0142 1.15 100.94 65.78 1.34 0.11 0. Because gas and liquid samples used for recombination are not in equilibrium.9 2. Swelling tests with the reservoir gas and a synthetically prepared lean gas were performed.295 Mscf/bbl.200 3.600 3.443 psia [24 MPa]. Models were initialized at pressures about 100 psi [690 kPa] above the dew point pressure of 3.9745 1. This contrasts with the separator gas used as recycle gas in the reservoir problem.0 0. and swelling tests.3412 1.0* 19.73 0. TABLE 6-PRESSUREIVOLUME RELATIONS OF RESERVOIR FLUID AT 200°F (Constant-Composition Expansion) Pressure (psig) Relative Volume Deviation Factor. not measured. the well stream will not flash to the gas and liquid compositions of Table 4 at the indicated pressure and temperature.350 3.500 5.6113 2. Unlike most fluid analyses.822 0.998 0.300 1.0412 2. The reservoir model operates Journal of Petroleum Technology. August 1987 .05 0. constantcomposition expansion data.800 1.200 3.030 836 0. Furthermore.99 8.9 19.350 3.2925 4. however.600 1. The data include hydrocarbon sample analyses. Because participants matched the swelling data for the lean gas (with varied success).000 4.91 2.869 0.21 1.57 1.13 0. Note that the lean gas is virtually free from C H fractions.0000 1.06 0.39 2.00 Total • Assumed molecular weight = 325. Thus the relevance of matching the swelling data is in question for the problem at hand.400 1.20 0.1393 1. Z 6. which has approximately 10% C H . Compositions of equilibrium gas are given in Table 8.000* 2. and the calculated yields of separator and gas-plant products are given in Table 9.1644 1. Table 5 gives more detail on the distribution of components in the synthetic reservoir fluid.8530 0. The lean-gas composition is given in Table 10. Table 6 gives constant-composition expansion data. the separator liquid is a random condensate sample.0468 1.063 0. however. The resultant well-stream composition is correctly given in Table 4.8045 0. Liquid condensation data are given for each of the expansions.428 3. Tables 11 through 15 give pressure/volume data for expansions at 200°F [93°C] for four mixtures of lean gas with reservoir gas. Table 4 gives compositions of liquid and gas used to create a reservoir well-stream composition for depletion 984 TABLE 7-RETROGRADE CONDENSATION DURING GAS DEPLETION AT 200°F (Constant-Volume Depletion) Pressure (psig) Retrograde Liquid Volume (% hydrocarbon pore space) 0.800 2.000 3.8268 0.TABLE 5-HYDROCARBON ANALYSIS OF RESERVOIR FLUID SAMPLE Component Mol % Carbon dioxide Nitrogen Methane Ethane Propane Isobutane n-Butane Isopentane n-Pentane Hexanes Heptanes Octanes Nonanes Decanes Undecanes Dodecanes Tridecanes Tetradecanes Pentadecanes Hexadecanes Heptadecanes Octadecanes Nonadecanes Eicosanes plus * 1.500 4. the separator-gas composition was prepared in the laboratory with pure components and not collected in the field.44 1. constant-volume depletion data.9284 0.81 1.5542 3.

67 2.54 1.14 530.729 6.507 5.400 1.02 0.830 0.35 74.687 74.45 9.535 Pentanes plus 5.78 1.60 1.02 14.31 2.665 2.70 20.22 23.94 65.752 0.732 0.761 0.24 2.287 'Equilibrium liquid phase representing 10.464 3.27 2.49 1.76 3.762% of original well stream.428 3.730 0. "Primary separator at 800 psig and 80°F.95 247.800 1. Journal of Petroleum Technology.50 1.21 1.69 29.06 0.12 0.407 2.58 3.464 0 0 0 141 137 321 374 352 678 629 580 973 900 821 1.740 0.96 4.84 1.44 1.81 1.49 0.03 71.872 1.40 0.36 9. Mscf Total plant products in primary-separator gas.23 0.01 2.06 1.702 4.748 0.57 1.63 5. second stage at 50 psig and 80°F.00 1.79 5.09 9.88 0.802 0.69 0.18 1.07 0.20 0.924 0.44 0.64 31.428 3.04 100. Mscf Second-stage gas.60 0. gal Propane Butanes (total) Pentanes plus Total plant products in well stream.27 2.66 0.06 0.14 12.019 4.80 5.99 8.76 7.16 0.22 0.798 0.146 1.200 psig.40 0.91 2.000 9.20 2.000 2.75 7.46 8.43 369.19 131.06 100.19 32.25 2.00 1. TABLE 9-CALCULATED CUMULATIVE RECOVERY DURING DEPLETION' Reservoir Pressure (psig) Well stream.27 7.745 0.803 3 Gal/scf x 10 from smooth compositions 8.08 0.46 5.200 700 700' Carbon dioxide Nitrogen Methane Ethane Propane Isobutane n-Butane Isopentane n-Pentane Hexanes Heptanes Octanes Nonanes Decanes Undecanes Dodecanes plus Total 1. stock tank at o psig and 80°F.01 2.488 'Cumulative recovery per MMscf of original fluid in place.400 1.96 1.000 2.20 73.92 11.89 16.98 8. reduced to 300 psig and 80°F for pressures below 1.33 2.74 8.68 9. bbl Primary separator gas.095 42.00 1.03 0.485 3.877 0.26 0.43 1.18 0.00 1.73 0.44 5.13 100.07 4.12 4.13 69.39 0.774 0.200 700 1.93 2.66 6.12 0.78 8.617 1.00 1.32 2.02 420.702 0.50 100.598 Propane plus 5.73 0.46 2.00 0.27 73. Z Equilibrium gas Two-phase flow 0.72 7.96 12.648 5.043 3.159 3.05 0.26 596.34 0.08 0.02 0.112 Butanes plus 5.013 1.87 740.55 0.66 5.40 2.TABLE 8-DEPLETION STUDY AT 200°F Hydrocarbon Analyses of Produced Well Stream (mol %) Reservoir Pressure (psig) Component 3.46 107. Mscf Stock-tank gas. August 1987 985 . gal Propane Butanes (total) Pentanes plus Initial 3.83 0.703 0.61 2.049 1.240 1.38 1.27 1.12 5.25 0 0 0 0 7.739 Deviation factor.803 0.328 1.026 59.800 1.31 1.24 '72.39 2.000 0 90.83 211.38 100.44 0.12 1.25 0.67 801 492 206 0 0 0 85 54 22 249 163 67 443 295 120 654 440 176 876 617 255 496 394 164 0 0 0 35 30 12 80 69 29 119 106 45 161 146 62 168 153 65 1.02 100.09 0. gal Propane Butanes (total) Pentanes plus Total plant products in second-stage gas.43 0.774 0. 106 105 148 0.69 5.73 0.15 25.642 Cumulative initial well stream produced (%) 24.82 1.05 27.72 8.25 666.72 0.00 750.09 0.00 118 111 .781 Molecular weight of heptanes plus 140 127 Specific gravity of heptanes plus 0.02 100. Mscf Normal temperature separation" Stock-tank liquid.046 7.

8827 1.401 0.1 0.428 psig and 200°F.500 4.2918 Liquid Volume< (% saturated volume) Pressure (psig) Relative Volume< Liquid Volume< (% saturated volume) 6. :t: Original reservoir fluid.950 3.1116 1.1 6.300 4.800 3.0135 1.1224 1 1.65 psia and 60°F per barrel of original reservoir fluid at 3.300 5.05 nil 100.600 5.800 3.000 1.9248 3 1.9043 1.1969 1..650 3.3667 1.9719 1.00 Gal/scf X 10 3 1.428 psig and 200°F.9115 0.200 3.610 4.523 0.467 0.9248 1.500 3.580 1.3385 1. 986 TABLE 14-PRESSURE/VOLUME RELATIONS OF MIXTURE 3 AT 200°F (Constant-Composition Expansion) at and below the dewpoint pressure during cycling. but the reservoir calJournal of Petroleum Technology.415 Calculated gas gravity (air = 1.428 psig and 200°F.7413 1.000 5.8620 1.0687 1. and swelling volume and dewpoint pressure during swelling of reservoir gas with lean gas.7884 1.2162 1.9387 0.500 5.TABLE 10-HYDROCARBON ANALYSIS OF LEAN-GAS SAMPLE' Component Mol % Carbon dioxide Nitrogen Methane Ethane Propane Butanes plus Total nil nil 94. t Barrels of indicated mixture at its dewpoint pressure and 20QoF per barrel of original reservoir fluid at 3. PVT Matches to the PVT Data We asked for matches of total volume in constantcomposition expansion.800 4.1378 2.1224 1.000 1.000 4.100 4.0000 190 0. Btu/ft 3 0. TABLE 12-PRESSUREIVOLUME RELATIONS OF MIXTURE 1 AT 200°F (Constant-Composition Expansion) Pressure (psig) Relative Volume< 6.500 4.1508 1.0 0.3 0.950 4. Two companies (Elf Aquitaine and Petek) chose to match phase volumes in the swelling test only for pressures in the range expected to occur during cycling.880 'Cumulative cubic feet of injection gas at 14.000 5.428 psig and 200°F. August 1987 .400 3.048 'Synthetically prepared in the laboratory. The number of components used ranged from a low of 5 components (Chevron and Core Laboratories) to highs of 12 (Marathon) and 13 (Petek).7 6.014 nil 1.428 3.900 3.428 psig and 200°F.3479 1.6 2. We also asked companies to describe techniques used for Kvalues.0 *Relative volumes and liquid volume percents are based on the original hydrocarbon PV at 3.1298 1. We believe this to be a valid approach but do not know how this affects the cycling problem.8422 1.502 5.050 4.6709 0.1 0.610 4. ·Cumulative moles of injection gas per total moles of indicated mixture.6865 1.1686 1.700 4.500 4. 0. Relative volumes and liquid volume percents are based on the original hydrocarbon PV at 3.3992 1.3542 1.000 0.000 5.1294 1.8 12.000) Calculated gross heating value of dry gas at 14.1271 1.100 5. A special model based on partial densities (McCord-Lewis) used 16 components to obtain the density data needed.600 3.015 4.0 (dewpoint) 0.0 (dewpoint) 0.8519 1.2767 1. TABLE 11-S0LUBILITY AND SWELLING TEST AT 200°F (Injection Gas/Lean Gas) Swollen Mixture Cumulative Gas Injected Number (scf/bbl) < (mol fraction) << Volume t --0*1.700 3.65 psia at 60°F.5 9.635 3. liquid dropout and equilibrium gas yield in constant-volume depletion.68 5.300 3.000 3.3064 1.500 3.5 4.3 1. phase densities and viscosities.8 .27 0. .635 4.4 * Relative volumes and liquid volume percents are based on the original hydrocarbon PV at 3.5115 1.015 3.5043 4 2.0000 o 0. TABLE 13-PRESSUREIVOLUME RELATIONS OF MIXTURE 2 AT 200°F (Constant-Composition Expansion) Pressure (psig) Relative Volume< Liquid Volume' (% saturated volume) 6.000 4.9512 2.000 4.8233 1. and EOS parameters used for the PVT match.6538 Oewpoint Pressure (psig) 3.3542 2 572 0.5384 2.3193 2.3046 1.0965 1..1203 1.900 4.6348 0.0360 2.

600 4. ARCO.5976 0.0 (dewpoint) Trace 0. Actually. CHEVRON. 2-Relative total volume in constant-composition expansion at 200°F (see Table 6). Fig. CORE LAB.6 volumes and liquid volume percents are based on the original hydrocarbon PV at 3.400 psi [17. August 1987 987 .1 0.428 psig and 200°F. INTERCOMP. Separator conditions for the problem differ slightly from the separator conditions in the laboratory reports distributed to the participants in two areas: (1) the primary separator pressure is switched from 815 TABLE 16-COMPONENT GROUPINGS Component CO 2 N2 Cl C2 C3 C4 C5 Cs C7 Cs Cg C lO C ll Hl H2 H3 H4 H5 Total number of components Total C s+ components Total C 7 + components Arco -- Chevron --- Core Laboratories CMG -- Elf Aquitaine Intercomp Marathon I ~ * t * ~+ t *t 1 t X X X X X X X X X X X X t t X X X X * X X X X X X X X X X X I X X X McCord-Lewis X X X X X XX XX X X X X X X X X X X X X X X Petek -X X X X X X X X X X X X X 9 5 5 10 6 8 12 16 13 H 2 H H 2 H 6 7 H 3 H 2 4 H 5 H H 5 Journal of Petroleum Technology.2 and 23. Table 16 indicates the component groups selected by each participant.000 3.500 psi [17.2 MPa] with peak liquid volume varying between about 18 and 22% of the initial (dewpoint) gas volume. Pressure (psig) Relative Volume- Liquid Volume(% saturated volume) 6.000 2.3454 2. Liquid yield by multistage surface separation of equilibrium gas produced during constant-volume depletion is given in Fig. The greatest discrepancies occur in the neighbor- hood of 2.4 CONSTANT COMPOSITION EXPANSION CM' CMG o ' ELF.LEWIS • = CORE LAB PVT DATA 5000 Fig.500 and 3. Tables 17 through 25 give summary data for each company's representation of component properties and the basic PVT match obtained with this set of components. 3 shows liquid dropout in constant-volume depletion. 2 shows pressure/volume data in constantcomposition expansion of the reservoir gas at 200°F [93°C]. between 2.500 5. the reservoir models predict liquid volumes higher than this value in the vicinity of the production well because of convection of heavy end products into this low-pressure area and subsequent deposition.1570 3. TABLE 15-PRESSUREIVOLUME RELATIONS OF MIXTURE 4 AT 200°F (Constant-Composition Expansion) PETEK 3 ML = McCORD .500 3.5288 2. Fig.880 4.5946 2.4 MPa].2435 2.800 4. 2 through 6. which we attribute to the absence of flash data for such compositions.5043 2. MARATHON. there is rather good agreement in the pressure range in which most of the gas cycling takes place. More detailed matches of PVT data are included in Figs. 4. The increased heptanes-plus content leads to compositions and flash behavior not available in the laboratory data provided.000 4.4704 2.400 4.000 5.4 3. While there are some minor discrepancies at the lowest pressures shown.3 0. Results given later show disagreement in the predicted liquid buildup in this area.7 1. ~Re[ative culations do not perform material balance on all 16 components.8478 3.6709 2.

00600 0 0 0 0 C1 C2 C 3-5 C 6-1O C 11 + Acentric Factor Tc Pc Component (atm) 45.4 28.07 54.6 N2 C1 C2 C3 C 4.02 0.00385 0. 3-Relative liquid volume in constant-volume depletion at 200°F (see Table 7).010 0. CONSTANT VOLUME DEPLETION 30 LIQUID SATURATION vs PRESSURE 20 ~ (/) > .8 18.1 0.00600 0.036 0.00630 0.1 0 0 0 0.2 572.0130 0. 4 Basic PVT Match Dewpoint pressure. whereas the lean-gas Journal of Petroleum Technology.0869 0.05 0.1 N2 C1 C2 C3 C 4 -6 C7 P1 C7P2 C7 P3 0 0. 4 Initialization Results Initial wet gas in place.1277 0. Two concerns were expressed. CORE LAB.1 0.811 Simulator Description Standard IMPES compositional reservoir simulator with Gauss 04 5 linear equation solution technique.10 0.LEWIS P = PETEK X = ARCO.6 to 2_2 MPa] at average reservoir pressure of 2.10 0.0044 C 6 .373 Simulator Description Chevron's finite-difference simulator was used for all reservoir calculations.135 0.0153 0.5 45. CHEVRON.2 862.1241 -0.0591 0. psi a Dewpoint Z factor Initialization Results Initial wet gas in place. pSia Dewpoint Z factor 3480 0.5 191.0967 0. 5 and 6.. The relative volumes of reservoir gas blends with increasing amounts of lean gas and the dewpoint pressures of these various blends are shown in Figs.495 0.1 0.5449 16. and this should be kept in mind when these figures are evaluated..9 33. PSIA Fig.0034 Interaction Coefficients CO 2 0 -0.225 0.1392 o o 0 000 0.9 170.833 44.66 PVT Methods Peng-Robinson EOS 3 for vapor/liquid equilibrium and densities.10 67..6 305.7 305..3 3.7538 to 315 psia [5. and in two cases rematches were obtained_ The reservoir model is significantly affected by the match of Fig.152 0.3080 0..234 0.0121 0.5 630.5 25.1 0.0 33.. ELF.200 psig [8_3 MPa].1002 0. Bscf Initial separator gas in place.2765 0.7 8.04 30.0997 -0. MMSTB 26. 6 C7 P1 C 7 P2 C7 P3 Acentric Factor Molecular Mole Weight Fraction 0. Bscf Initial stock-tank oil in place.13 0. MARATHON • = CORE LAB PVT DATA 1000 2000 3000 4000 PRESSURE.58 24.J :::.09281 0 0 0. Several participants expressed skepticism regarding the need to match this part of the PVT data for reservoir modeling purposes.. surface liquid density correlations. Viscosity for gas and liquid by Lohrenz et al. Participants whose data differed significantly from the average were offered opportunities to review their results in light of the trends.00630 0 0 0 0. 988 3501 0. the greatest difference in yield by these two sets of separator conditions is at the dewpoint pressure and is a result of the colder stock-tank temperature used in the reservoir model.2 37.098 0.500 psia [17.2 42. and (2) the stock-tank separator temperature is taken as 60°F [16°C] for the reservoir model.6 16. INTERCOMP.332 0.1 0. Bscf Initial stock-tank oil in place. These data were not distributed to the participants but are shown in Fig_ 4.8 48.0 Interaction Coefficients C1 0 C2 0 C 3_5 -0. August 1987 .1 (K) 190.040 0.00385 0 0 0.6599 0.08 0.01 28.1355 C 11 + PVT Methods Peng-Robinson EOS 3 densities.9 396. MMSTB 28.2 126. Participants were asked to match surface yield for the laboratory separator conditions_ Core Laboratories provided computations for multistage separator products with experimentally determined equilibrium gas compositions in Table 8 for the separator conditions specified in the model problem.50 304.0986 0. Viscosity for 0 -0.0194 0.2 MPa] in the reservoir model.01 16. These data are influenced by K values during depletion. ARCO TABLE 18-CHARACTERIZATION DATA AND PVT MATCH.4 409.010 0 0 0 for vaporlliquid equilibrium and gas and liquid by Lohrenz et a/.2 Molecular Mole Weight Fraction 0. 1_ The actual injection gas derived from the models has a molecular weight of about 22. As seen in Fig_ 4.2492 0 0.1825 0.9 25.85 103. and surface separator K values. CHEVRON Component Characterization Data Component Characterization Data Tc Pc Component (atm) ~ CO 2 72.6 48.4 369.9 282. Gauss with 04 ordering 5 for pressure solution was used.07 44. 10 .2 568.0472 0. Bscf Initial separator gas in place.013 0.06 3.28 110. whereas the laboratory assumed a separator pressure switch at a pressure of 1. 4.TABLE 17-CHARACTERIZATION DATA AND PVT MATCH. Basic PVT Match Dewpoint pressure.04 30. whereas the laboratory data were based on an 80°F [2rC] stock-tank temperature.-0 o o CM = CMG ML = McCORD .2 186.

however.18 C 12•14 19... Gas Z factor was obtained with Yarborough-Hall fit of Standing-Katz charts. Bscf Initial stock-tank oil in place.LEWIS CORE LAB PVT DATA CORE LAB PVT DATA (ADJUSTED FOR SEPAR· ATOR CONDITIONS IN THIS PROBLEM) 60 I~ 40 CL 20 00 E-- CM/.25 C lO. with K values prepared as discussed above. CMG TABLE 19-CHARACTERIZATION DATA AND PVT MATCH. Bscf Initial separator gas in place.112 1. Slight heavy-component K-value adjustment was used to match dewpoint pressure. psia 3443 0. Stiel-Thodos viscosity correlations were used for oil and gas.90 3..773 Interaction Coefficients Not available PVT Methods A 16-component PVT simulator was used to prepare Kvalue data by convergence pressure techniques. MISIM3. K values were fit with methods suitable for the kind of pseudocomponent (lights. the laboratory data on the various lean-gas/reservoir-fluid mixtures are of little use in developing the properties of mixtures of bulked separator gas and reservoir fluid. Initialization Results Initial wet gas in place.024 1.3613 0. Therefore. and depletion-gas compositions.04 3.777 0. MA • ML • " • • • al ~ en 80 CHEVRON CMG CORE LAB ELF INTERCOMP MARATHON McCORD .39 3443 0.8 253.803 Dewpoint Z factor Simulator Description Core Laboratories' compositional model uses up to six components.79 685.689 Basic PVT Match Dewpoint pressure. 4-Three-stage separator yield in constant-volume depletion at 200°F (see Table 9).33 0. August 1987 Component (atm) C l . CORE LABORATORIES Component Characterization Data Pc Component Characterization Data Pc Component (psi) Cl Ca Cg 397 361 332 304 232 C 10 C 11 + Cl 973 1.55 ~ Tc Acentric Factor 573. results from the 16-component PVT simulator were used as the basis for tables of input data to the compositional model. liquid volumes. Once a satisfactory match was obtained..814 0.6 0.37 22. Preconditioned conjugate gradients are used to solve the diagonally dominant matrix equations.7 uses a quasi-Newtonian method of solution called ONSS B developed at CMG. Bscf Initial separator gas in place. Viscosities for gas and liquid by Jossi-Stiel-Thodos.766 0.250 + 688 756 822 886 Molecular Weight Density (g/cm 3 ) 94 110 117 137 213. and nonhydrocarbons). Journal of Petroleum Technology.. psia Dewpoint Z factor 26. Liquid density was obtained with modified Standing correlations. (They do. provide a comparison of calculated and measured gas deviation factors.8 177. Note offered by Core Laboratories: The injection gas of bulked separators contains 22% CO 2 and heavier. U 120 ~ ~ 100 en . a gas considerably heavier than the injection gas used in the laboratory PVT studies.5339 0.8030 Simulator Description CMG's IMPES simulator.0122 0. with properties of the model components representing groups of components between CO 2 and C 10 computed as functions of pressure during the laboratory data match.6 140 0.45 637.99 C 1S + 12.711 0. MMSTB Basic PVT Match Dewpoint pressure. MMSTB 26.9 26.3 Pseudocomponent selection is based on unpublished methods developed at CMG.0080 0.739 0. ONSS was also used to solve the flash obtained from the Peng-Robinson EOS.TABLE 20-CHARACTERIZATION DATA AND PVT MATCH.4501 0.. 11 23. 3-STAGE SEPARATOR YIELD vs PRESSURE CH ~ CM ~ CL ~ E • I . K values obtained were fitted with convergence pressure as the parameter for composition dependence for mixtures of this nature.) Core Laboratories thus used the Peng-Robinson EOS 3 to estimate dewpoints of mixtures of reservoir fluid with separator gas expected in the model.37 23. 989 .4 144.. Core Laboratories has a version of its PVT simulator that uses the Peng-Robinson EOS 3 but it was not used in this problem. The compositional model used five pseudocomponents.0058 Interaction Coefficients Not available PVT Methods Peng-Robinson EOS 3 for vaporlliquid equilibrium and densities. 6 Initialization Results Initial wet gas in place.75 748. Bscf Initial stock-tank oil in place.0399 0.7244 Molecular Mole Weight Fraction 114. heavies.J.070 1.. Five components were used for the present problem."CH 4000 Fig. CONSTANT VOLUME DEPLETION 180 160 140 I.

069. In this study.29 3. Hence the match was based on the liquid deposit at 3. MMSTB 26. 13 is used to overcome timestep size limitations inherent in IMPES models.1763 0.72 C 3 -C s 545.099 0. A special technique. psia Dewpoint Z factor SWELLING WITH LEAN GAS 3. Bscf Initial stock-tank oil in place. a correlation. Bscf Initial stock-tank oil in place. 12 It is a modified IMPES simulator with generalized cubic EOS calculations for phase equilibrium and phase density calculations.75 211. The cumulative surface recoveries from the constant-volume expansion presented in the PVT report were calculated with published equilibrium ratios.0 ML P = ELF = INTERCOMP = McCORD .90 196. '"o 2.7383 0. 11 followed by component lumping to a total of eight components.27 C s -ClO 363. Stock-tank oil rates for constant gas sales rate and for deferred early gas sales are shown in Figs.068 0. No adjustments of the EOS parameters were made to represent the fluid behavior at surface conditions.0115 0.LEWIS = P[TEK !ci p >'" 15 500 1000 1500 0. Note offered by Intercomp: The Peng-Robinson EOS used for the comparative solution project was only calibrated vs.37348 0. MMSTB 26.07 54. VSAT I VSAT• ORIG vs SCF/BBL CL :: CORE LAB ii: F7 Fa F9 FlO F 11 Basic PVT Match Dewpoint pressure.42 Basic PVT Match Dewpoint pressure.93 729. August 1987 . 4 Note offered by Elf Aquitaine: Results of the saturation pressure match are poor.071 0. a very detailed composition analysis (up to C 30 +) would be necessary to match such results adequately.443 0.66 959.56 C 1 +N2 667.01764 0.0908 0. psia Dewpoint Z factor 3. 10 Pseudocomponents were developed by a special version of Whitson's split-out procedure.76 3. From Elf Aquitaine's experience.4 and all phase and equilibrium data were derived from the EOS.07779 0.000 psig. 5-Relative total volume in swelling of reservoir gas with lean hydrocarbon gas at 200°F (see Table 11).000 psig. K-value tables based on component C 1 global mole fraction were used based on precalculations with the PengRobinson EOS. measured data from tests performed at reservoir conditions. The Peng-Robinson EOS3 was used for this problem. Differences between the two methods were small in this problem. No attempts were made to match these data because that would involve calibrating the EOS vs.22 353. the pressure at recombination is considered too far removed from the separator pressures used in the performance simulation to render meaningful calibration of the EOS for surface conditions.01 16.53 23.and nine-point differences were used.08141 0..'s9 formulation of the IMPES equation is used. but the constant composition expansion data (of total volume and liquid drop out) agreed fairly well with the calculations for each mixture. Regression methods are used for the PVT data match. INTERCOMP Component Characterization Data Component Characterization Data Component Pc Tc (psi) (OR) CO 2 1.00 343.034 0.443 0.08 C2 708.07779 0.00370 0. Reservoir Model Performance Table 26 gives the initial surface fluids in place with multistage separation.8452 0. Bscf Initial separator gas in place. molecular weight is about 17. In the absence of measured surface yields.7917 Simulator Description Intercomp's COMP-II was used for the reservoir model. The corresponding cumulative liquid pro- 990 Journal of Petroleum Technology.52310 0.68 254. Both five. no conclusions can be drawn regarding the validity of the EOS or the K-value correlation. The separator compositions and recombination ratio presented in the PVT report are considered not to be representative of a vapor/liquid equilibrium state at 72°F and 2.8112 0. Differences in the swelling characteristics obtained with these gases would be expected.24 0. Initialization Results Initial wet gas in place. The reservoir pressure falls continuously with time in both cases of interest. Thus volumetric behavior at pressures above the initial reservoir pressure is unimportant in the context of the model.50 23.35 151.00721 0.7790 44.08525 0.2250 0.45723 0. The reasons for omitting the EOS match of surface conditions can be summarized as follows.7787 0.04 30. Matrix solution is by Gauss elimination on 04 ordering.67 Interaction Coefficients Not available Acentric Molecular Vc Factor Weight (ft 3/Ibm) 0.3760 0. Initialization Results Initial wet gas in place. the stabilized IMPES method.84 108.3 Phase densities were also obtained from the EOS. with a very small slope of liquid deposit curve at dewpoint.45624 0.5 A = AReo CH :: CHEVRON eM:: CMG E I ~ 2:'" 2. but results based on the nine-point solution are shown. Bscf Initial separator gas in place. ELF AQUITAINE TABLE 22-CHARACTERIZATION DATA AND PVT MATCH.8027 Simulator Description Elf Aquitaine's MULTIKIT compositional model was used.06329 0.0 Acentric Acentric Specific Molecular Mole Component Factor a Factor b Gravity Weight Fraction 2000 2500 SCF/BBL: OF DEW POINT GAS Fig.079 0.18 549. 2. Kazemi et al.301. Viscosity was based on Lohrenz et al.52 547.00062 Interaction Coefficients Not available PVT Methods Intercomp's PVT package is equipped with four choices of EOS's.05 3. or EOS (Peng-Robinson) K values.78 0.TABLE 21-CHARACTERIZATION DATA AND PVT MATCH. 7 and 8.45723 0.67 C 11 + 223.03672 0.139. It allows for either K-value tables (algebraic) convergence pressure relations. Even if the data do represent equilibrium. Viscosity correlation used was Lohrenz et al.8907 108.066 PVT Methods Elf Aquitaine's EQLV PVT package based on the PengRobinson EOS 3 was used for the PVT match.

Oil viscosity was obtained from the correlation of Little and Kennedy 14 and gas viscosity by the Lee 15 correlation. MMSTB SWELLING WITH LEAN GAS CL E I ML 3400 3200 P 3000 0 :: PETEK • :: CORE LAB PVT OATA 1000 500 2000 1500 2500 SCF / BBL OF DEW POINT GAS Fig.3 33. We believe the predicted pressure in early years of cycling is also important.. 991 .3240 0. OJ 500 (j") 400 300 200 100 0 3 0 4 5 6 7 S 9 10 11 12 13 14 15 YEARS OF PRODUCTION duction for these cases is given in Figs. 9 pressure equation and all PVT data are entered as tables.81 533. 5 can be used to find molar volumes of reservoir gas saturated with additions of lean gas. and gas mixtures.98 3.. ~ '" 0 4200 4000 ____=-=::::::::=-----P 3800 Cl... In the present problem.73 Basic PVT Match 3.000 psi [20. K values were functions of pressure only.00 4400 <t iii Cl. For the reservoir model. more pertinent data are molar volumes (Z factors) of mixtures at typical cycling pressures.0 Interaction Coefficients Not available Cs C7 Cs Cg Gas Oil 0. STOCK TANK OIL RATE ..8035 Dewpoint Z factor Simulator Description Marathon's IMPES simulator is based on the Kazemi et al. Initialization Results Initial wet gas in place. For most participants.6 31.2778 0. All yearly production data were connected with straight-line segments in Figs. the pressure decline is affected by Z factors for reservoir gas.2977 0. During the critical early years. Phase densities are also obtained byadjustments to EOS values in such a manner as to allow a match to observed volumetric data and to reported Z factors in depletion experiments. MARATHON 4800 Component Characterization Data Acentric Tc Factor Component (atm) ~ Pc 34. injection gas. August 1987 Fig. because this determines average reservoir pressure for a given excess of production over injection. STOCK TANK OIL RATE ..443 Dewpoint pressure. 9 and 10. Case 2.0105 0. :: CORE LAB = ELF :: INTERCOMP = McCORO .0138 275.LEWIS 600 P . Some limited mixture volume data were available from the laboratory reports at 3.4035 0. the predicted surface oil rate is closely correlated with the liquid yield predictions shown in Fig. 7 through 10.39 22.07 550.00 104. but matches to these data were not requested.7 MPa] and above (Tables 12 through 15). 4. Case 1. The rate of gas recycled for Journal of Petroleum Technology.53 504. primary separator switchout occurred late in the cycling phase (10 years).4 25. 6-Dewpoint pressure during swelling of reservoir gas with lean hydrocarbon gas at 200°F (see Table 11). A 3600 :: AReo -E CH = CHEVRON eM:: CMG PVT Methods Marathon used the Peng-Robinson 3 EOS as a starting point for K-value tables.6000 SCF/BBL P OEW POINT vs 4600 Molecular Mole Weight Fraction 81. a-Reservoir model stock-tank oil rate.5000 TABLE 23-CHARACTERIZATION DATA AND PVT MATCH.50 520.0122 130.CASE 2 A = AReo CH = CHEVRON eM = CMG CL = CORE LAB E "ELF I = INTERCOM? MA = MARATHON ML = McCORD-LEWIS P • PETEK 300 200 100 0 0 3 4 5 6 7 8 9 10 11 12 13 14 15 YEARS OF PRODUCTION Fig. = PETEK .. Most models were already below the dewpoint pressure at 1 year of production.97 598.2 20. In most cases.LEWIS A = AReo CH = CHEVRON eM =CMG CL = CORE LAB E = ELF I = INTERCOMP 700 MA "MARATHON o ML = McCORD .00 693. .0181 0. and surface liquid rate had already dropped below initial rate.00 0. as well as by the rates of wet-gas produced and separator-gas recycled. Bscf Initial separator gas in place. but phase densities and viscosities were adjusted to match both depletion data and estimated bulkedseparator gas properties. this is not the sole explanation for the discrepancies in early oil rate seen in both cycling cases.2592 0.0150 95. Bscf Initial stock-tank oil in place.00 0. psia 0. Hand methods and adjustments of K values generally allow a more precise description of equilibrium data in the two-phase region than by unmodified K values.0144 88.00 0. Swelling data matches in Fig. 7-Reservoir model stock-tank oil rate. However.CASE 1 26.00 0.z- <5 Il..3 30.

09 617.1524 44. but it is important to note that it doesn't require K values per se.2 Interaction Coefficients (with methane) C1 C2 C3 iC 4 nC 4 iC s nC s C6 C7 C8 C9 C lO C 11 C 12 + CO 2 N2 Specific Gravity 0. 3000 3000 STOCK TANK OIL PRODUCED . psia Dewpoint Z factor 3.03918 0.80 1. McCORD-LEWIS Component Characterization Data Acentric Molecular Factor Weight 16. The detailed discussion of the model is not possible here.443 0.0 790.5 845.4946 210.2 969.00 348. The basic assumption is that each reservoir phase can be viewed as a binary mixture of its surface products. 19 Specific gravities and methane binary interaction coefficients of the heavy ends were estimated from the Watson K factor.2 416.064.4800 0. with some minor changes.6 0.12 529. STOCK TANK OIL PRODUCED .8380 0. A = ARea CH = CHEVRON 1200 A 3 4 5 6 7 8 9 10 11 eM = CMG CL = CORE LAB 900 =ELF INTERCOMP = MARATHON ~ 12 13 = ARGO CH = CHEVRON 1200 = CMG : CORE LAB 600 Ml = McCORD-LEWIS P = PETEK 14 E = ELF I = INTERCOMP MA ML s P 1S YEARS OF PRODUCTION Fig.6 ~ ~ ~ ~ ~ ~ ~ ~ ~ 0.3495 114.6 0.5631 0.15 489. termed partial densities.011.3220 1.1 910.04 673.1000 PVT Methods McCord-Lewis used Watson 16 characterization factors for each fraction with correlations of Whitson and Haaland.0400 28.01 0.00 101.6 0.3912 128.0 470..60 419.20 432.17.2010 58.3250 0..6 607. when forming any reservoir phase.7 0.7100 0.TABLE 24-CHARACTERIZATION DATA A~D PVT MATCH. came from Katz and Firoozabadi. 11.0986 30.7 139.071.03596 0.7570 0.02 492.8 541.. 9-Cumulative reservoir model stock-tank oil produced.0 1. August 1987 .7 765.118.1 84. Case 1.0 350.7 391.803 Simulator Description The McCord-Lewis simulator is based on a partial density model that includes condensation from the reservoir gas phase to the reservoir liquid phase. 1500 :::.8 0.30 365.4800 Tc Pc ~ ~ Component Tb (OF) 201.3 227.CASE 1 I 2700 2400 2100 1800 1800 (D t:.h Dewpoint pressure.00 359.04240 0.'CASE 2 2700 J- 4 5 6 7 8 9 10 11 MARATHON = McCORD-LEWIS = PETEK 12 13 14 15 YEARS OF PRODUCTION Fig.18 Initial separator gas in place.4200 0.8010 0.5 828.2 1. Initialization Results 26.18 Boiling points.8 58.4 0.4 665.6 0.4568 155.154. 992 1 ML 2400 2100 (D t.2223 483.219. Case 2.0 692.6248 0.3 549.1 343.6 1.9 0.2806 457. MMSTB Basic PVT Matc.10000 0.8 0.7 0.0 1.6312 0.0 657.5844 0. Journal of Petroleum Technology.56 Initial stock-tank oil in place.04534 0.0 740.7 72. Bscf 3. eM CL E I MA 900 600 1500 :::.6 490.0 875.7340 0.8 556.12860 0.3 547.7800 0.3 0.5077 0.1848 0.03260 0.0 332.2539 72. Bscf 23. 10-Cumulative reservoir model stock-tank oil produced.1 734.0130 0.12 550. Lee_Kesler 2o .21 correlations were used for critical pressure and temperature and molecular weight.0 820.02813 0.15 0.52 Initial wet gas in place.2250 44.6781 0.4354 142.07 708.7 0.

42 3.58 28.5 26.50 26.00092 HC 40 1. Viscosities were calculated from the Lohrenz et al.500 630. 11-lncremental reservoir model stock-tank 011 produced by gas-sales deferral (Case 2 minus Case 1).032080 1.0 500. We did not request predicted surface-liquid molar density from the PVT matches.160840 1.98 23.73 3.0 0.7532 Simulator Description The Petek simulator is IMPES and uses one of several cubic EOS choices for PVT calculations.70 Initial separator gas in place.29 22.000 510.05 23.52 28.76 3.3498 100.57 INCREMENTAL CUMULATIVE OIL PRODUCTION BY GAS SALES DEFERRAL (CASE 1 vs CASE 2) 320 280 240 200 A : ARCO CH : CHEVRON eM .0 HC 10 0.200 960.373 3. 4 and surface-liquid molar density. 993 . The wet-gas rate thus depends on the match to yield data in Fig.00000 HC 10 0.53 26.00000 HC 20 0.76 as the liquid content of the produced 'gas approaches zero.0083 0. Petek made no attempt to match the solubility data of lean-gas injections with reservoir gas as it lacked importance in the context of the model.TABLE 25-CHARACTERIZATION DATA AND PVT MATCH.8000 247. SIp22 was used in the current problem to solve the pressure equations.0104 0. the injection-gas Z factor is approximately 6% higher than the reservoir-gas Z factor.06 25.eM MA ~ 160 en ::.0002 Acentric Factor b Acentric Factor a Component 1.39 3.100 700.CMG CL = CORE LAB E =ELF I = INTERCOMP MA = MARATHON Ml = McCORD· LEWIS P = PETEK .000. Initialization Results Initial wet gas in place. The ratio is 0. liquid relative volumes from the constant-composition expansions of the various mixtures were carefully matched in the range of reservoir pressures. The first version of a joint project venture has recently been released.2 420.37 26.28 24. psia 3.5015 157.56 3.04 22. PETEK Component Characterization Data Molecular Acentric Vc Tc Pc (MPa) Factor Weight (cm 3 /mol) Component ~ 3.0 HC so 1. Bscf Initial stock-tank oil in place..6 1.032080 1.0 Parameter Matching Process (Weight Factors on Acentric Factors) Mole Fraction 0.0 0.18 24.90 23.689 3.7 510.050. August 1987 Fig.3 23.00000 1.7 660. The initial molar rate of separator-gas recycled is approximately 0.37 26.39 26. 120 80 40 O~~~~ o __ 3 ~~~ 4 5 6 __ L-~-L~ 7 8 9 __~-L~__L-~ 10 11 12 13 14 15 YEARS OF PRODUCTION each cycling case is fixed in this problem.452 Dewpoint Z factor 0.00092 HC so Interaction Coefficients Not available PVT Methods Petek used the Peng-Robinson EOS 3 for K value and density data. allowing for sales gas.0 HC 40 2.3846 120.28 Dry Gas (Bscf) 24.0 0.0326 0.57 Basic PVT Match Dewpoint pressure. and Z factors of mixtures of these gases should be someJournal of Petroleum Technology.0 0.993215 1.9000 1.0 HC 30 2. Automatic timestep selection was used and found to be helpful in the problem. Instead.67 times the rate of wet-gas production in Case 1.000 560. Bscf 28. MMSTB 3.791887 HC 30 1. The wet gas produced depends on the surface separator efficiency because the separator-gas rate is specified. TABLE 26-INITIAL FLUIDS IN PLACE Company Arco Chevron Core Laboratories CMG Elf Aquitaine lntercomp Marathon McCord-Lewis Petek Wet Gas (Bscf) 26.7 Stock-Tank Oil (MMSTB) 3.0 HC 20 3. At dewpoint pressure.66 3. This was felt to have a larger effect on correct prediction of reservoir performance. 4 model.0144 0.

55 8.45 0.TABLE 27-AVERAGE OIL SATURATION. the condensate around the producer is partly revaporized. The results for liquid recovery are 55 to 74% of the initial oil (condensate) in place.2 4.51· 0.69 0.08 1.10 0.46 2.91 4.99 10. liquid dropout in the low-pressure region strips liquid from the gas stream.34 3.0 10.57 5.60 10.40 1.89 2.26 6.36 4. Fig.10 6.44 0.10 2. What is perhaps surprising is the widely different predictions for oil saturation at advanced depletion levels in the models.9 4.90 McCord-Lewis 0.76 0. Any model with a high rate of decline in pressure produces a rapid loss in surface liquid yield.44 2. which tends to smear out the initial molecular weight and Z-factor contrast between injection and production gases.0 4.94 6. and surface-liquid density) and assumed gas Z factors in early years of cycling.36 3.0 2.01 1.57 6.20 0.32 4.0 0.5 1. 4) at pressures far below dewpoint pressure explain the near-parallel oil production rates in the advanced stages of cycling and blowdown. We believe that this can be explained by the K values used.43 3.20 5.71 Arco 1 5 10 15 0.02 4.56 1.82 6.10 2.48 7.78 2.89 6.94 4.01 11.16 6.08 1.4) is shown in Figs.1 3. 4.65 3. ranging from 0 to more than 22%.31 1.20 3.40 11.94 9. Conversely.5 7.22 10.16 0.56 3. The stabilized yield value depends on the mixing of injection gas with reservoir gas around the producer and the contribution of the depleted area behind the producer to production.40 4.9 CMG 1 5 10 15 2. A partial compensation for this sensitivity to injection and production-gas Z factors is numerical dispersion.66 0.04 10.06 3.74 Elf Aquitaine 1 5 10 15 0.8 8.60 Elf Aquitaine 1 5 10 15 3.25 Marathon 1 5 10 15 1.33 1.55 5.35 McCord-Lewis 1 5 10 15 1 5 10 15 1.72 6.1 5.38 0.71 1.90 11.0 2. 12 and 13.72 5.0 3.73 1.66 3. 11.04 1.32 5.55 5.94 4.48 4.10 1.18 1.2 0 5.40 0.6 7. For other times.70 9.01 3. This and the merging of the participants' depletion matches (Fig.04 3.56 4.52 - 0.20 9.80 6.54 4.73 5.23 0.89 11.98 5.74 1.37 6.91 4. Layer average oil saturations for selected times are given in Tables 27 and 28.60 7. and reservoir oil ceases to flow.80 1.8 0 3.55 Marathon 1 5 10 15 4. these show 994 Petek - - - - common trends of relatively uniform saturations for the first year.03 3.87 3. These exhibit a considerable rangefrom 3 to 8 % of the initial condensate in place.26 1.61 4.73 4.12 10.75 1 5 10 15 1 5 10 15 0.95 5.68 2.5 0.28 3.96 11.02 5.34 2.70 8.78 8.9 4.06 7.63 6.01 8.93 7.71 4.95 5.62 5.32 2. saturation because little injected gas sweeps this layer.55 2. This indicates a condition of reservoir condensate flow in this area.57 6.81 8. We made two supplementary runs with COMPIII 23 for Case 2 to demonstrate the importance of the K-value techJournal of Petroleum Technology.37 11.23 CMG 1 5 10 15 0.6 10.03 8. Discrepancies in pressure are affected by both wet-gas rate (determined by yield.00 3.36 4.30 10.61 Petek - where between the two.72 5.58 1.97 1.29 5.79 6. during cycling.89 7.35 1.10 4.8 9.24 1. This reduces reservoir voidage and tends to lessen subsequent pressure decline.24 0.15 11.83 3.40 1.76 2.30 9.83 4.75 8.45 8. CASE 2 (%) Company Year Layer 1 Layer 2 Layer 3 Layer 4 Arco 1 5 10 15 3.82 5.79 1.82 6.87 2.76 1.1 1. Layers 2 and 4 are intermediate in sweep efficiency.08 11.0 1.60 13.55 1.98 6. Before this time.27 8.86 3.51 5.24 Intercomp 1 5 10 15 0.73 1.89 - 0.6 Core Laboratories 1 5 10 15 0 0.45 0.27 2.76 3.9 1.97 9.94 1. Another factor compensating for discrepancies in Z factors and separator factors is the reservoir response to falling pressure. Layer 1 (a high-permeability layer) shows almost no liquid.55 1.18 5.62 6.28 4.58 Chevron 1 5 10 15 0.78 1.13 0.55 0. This continues until a small liquid flow begins and the surface yield stabilizes.76 0.92 0.01 0.10 11.33 0. Condensate saturation in Node (7.44 3.20 Chevron 1 5 10 15 1.75 0.26 1.77 0.43 11.63 10.7. CASE 1 (%) Company Year Layer 1 Layer 2 Layer 3 Layer 4 TABLE 28-AVERAGE OIL SATURATION.87 8. Not surprisingly.77 Core Laboratories 1 5 10 15 1.94 9.01 5.84 8.0 5.0 4.0 3. Actual recovery efficiencies achieved by the models are atypical of field values in view of the homogeneous nature of the model grid.22 4.76 0.6 0. Most of the models achieve a fairly stable saturation of slightly more than irreducible oil saturation (24 %).07 9.81 6.59 10.05 3.20 7.30 8.08 6.15 9.42 9.37 5.39 3.51 3.0 7.07 5.91 8.30 5.76 9.40 7.03 8. Liquid yield is partly sustained as some heavyend fractions continue to vaporize and are produced.70 7.55 2.76 0.01 1. August 1987 .58 5.18 4.95 0.62 4.01 5.52 0.28 2.71 6.66 0. The incremental production achieved by gas-sales deferral is shown in Fig.9 3.98 0.31 3.24 4.89 0.57 2.72 1.9 0. Layer 3 (the tight layer) shows high.76 3.0 5. Later.78 8.0 5.39 3.43 5.43 0.28 0.8 0 2.34 3.12 1.42 - Intercomp 1 5 10 15 0.9 0.78 7.1 7.87 6.

4).7 185. PETEK 24 21 ~ J? .0 7.8 383 244 194 14.CASE 2 WITH IN-LINE FUGACITY-BASED K-VALUES 33 OIL SATURATION IN BLOCK (7.8 66.7.4)-CASE 2 ~ 18 o (f) 15 21 18 15 12 12 9 9 _____c..6x10.36 OIL SATURATION IN BLOCK (7.3 5.0 x 10.CASE 2 WITH PRE-COMPUTED K-VALUE TABLES OIL SATURATION IN BLOCK (7.3 8. CMG MA ~ MARATHON CL ~ CORE LAB Ml & McCORD-LEWIS & PETEK P 30 27 --RESPONSE ENVELOPE FOR 4 PARTICIPANTS USING PRECOMPUTED K-VALUE METHODS • eM 24 21 ~ ~ ~ o (f) SUPPLEMENTAL RUN WITH AN 11 COMPONENT MODEL AND A SINGLE PRE-COMPUTED K-VALUE TABLE ~ 18 o (f) 15 15 12 12 9 6 2 3 4 5 6 7 8 9 10 11 345678 YEARS OF PRODUCTION 12 13 14 15 YEARS OF PRODUCTION Fig.7..5 5.7.7 27.7 1 2..4 Average CPU per Timestep 1 2 22.5 1x 10.6 Marathon IMPES Burroughs B7900 365 347 14.4 29.4 44.7 15..1x10. Case 2.4 67.9 27.0 x 10- 5 3..3 Material Balance Error at 15 years 103 6.7x10. CHEVRON 30 eM .0 60. 13-Reservoir model condensate saturation in Block (7. 12-Reservoir model condensate saturation in Block (7.3 163 4.6 21.RESPONSE ENVELOPE FOR 5 PARTICIPANTS USING IN-LINE FUGACITY METHODS • SUPPLEMENTAL RUN WITH AN 11 COMPONENT MODEL AND IN-LINE FUGACITY METHODS 30 E "ELF I " INTERCOMP MA : MARATHON eM 24 OIL SATURATION IN BLOCK (7.19 509 10.4) .5 x 10- 1.4 6.5x10- 4 6.4 7. Case 2..7.!-___ _ 3 4 5 6 7 8 9 10 11 12 13 14 15 2 YEARS OF PRODUCTION 3 4 5 6 7 8 9 10 11 12 13 14 15 YEARS OF PRODUCTION Fig.4).2 116 6.5 10. August 1987 2.2 2 3 13.4).4 8. 36 33 A . 15-Reservoir model condensate saturation in Block (7.6x10.7.4) with in-line fugacity-based K values. AReo CH . McCORD-LEWIS P .0x10.4 2.7.CASE 1 33 A • AReo E ' ELF CH • CHEVRON I : INTERCOMP eM .1 5.. Fig.7. Fig.1x10.4 Elf Aquitaine Intercomp IMPES IMPES IBM 3081 Harris 800 185 128 199 114 39. Case 1.3 28.9 4 McCord-Lewis IMPES VAX-11/780 91 91 60. TABLE 29-RESERVOIR MODEL PERFORMANCE Company Numerical Solution Method Computer Used Number of Timesteps 1 Arco IMPES IBM 4341 325 Chevron IMPES VAX-11/780 375 Core Laboratories CMG IMPES IMPES CDC 6600 Honeywell DPS 68 251 200 Average Timestep Size (days) 2 2 16. CMG CL 27 z CORE LAB 27 ML . Case 2.0 x 10- 121 14.3 995 .7.4) with precomputed K-value tables.4).9 168 192 7.7 Journal of Petroleum Technology.4 13.0 Petek IMPES ND-560 5. 14-Reservoir model condensate saturation in Block (7.5x 10.6x10.

but there is more scatter when companies attempt to use Kvalue tables with no data on which to tune." JPT (Oct.: The Properties of Gases and Liquids." JPT (March 1986) 345-53. Fong. 10. Acknowledgments We thank Marathon Oil Co. C. 14 includes the response envelope of companies (Core Laboratories. McGraw-Hill Book Co. D. Fig.S % of the initial condensate. K. Temperature and Composition. M.B. for allowing us to use the PVT data essential to this problem.Y. C. Bray. The gas used for recycling in the reservoir model was considerably richer in C 3 + than the lean gas used for the swelling tests. but all show rapid condensate evaporation in the late stages of cycling. we used a single K-value table with a constantvolume depletion of the reservoir gas and assumed K values were independent of composition. G. 231. It shows a considerable scatter in predicted condensate.F." Ind." JPT (Aug. Coats. D.. It suggests that differences in pressure caused by physical property errors (Z factors) and/or surface-separator molar split errors may also be responsible.. 996 4. J. and Kennedy. Watson. especially in the late stages of cycling as the gas formation factor changes.7. A. Whitson. Elf Aquitaine. Peng. L. AIME." SPEJ (Oct. 26-29. 6-9. Case 1). This was unavoidable because not all companies had gas-plant capability in the reservoir simulator. 4) does not account for this much error.D." paper EUR 183 presented at the 1980 SPE European Offshore Petroleum Conference. 15. K. Oct. Nghiem. 5. Gonzalez.. and Sherwood.4) for the five companies that used EOS methods. Lohrenz. these data are unavailable and true evaporation rate is unknown at this time. C. (1976) 59-64.H. there was no obvious correlation between the number of components for the heptanes-plus fractions and the predicted rates of evaporation in Node (7. and Computer Modelling Group for permission to publish this paper and for providing the necessary time and support needed to conduct the project. and Aziz. 1966) 997-1000.7. This compositional simulator permits K values to be entered as a table or as calculated in line in the normal manner with an EOS. Figs.T.: "A New Stabilized Method for Use in 1M PESType Numerical Reservoir Simulators.. AIME. Houston. 3.E. Reservoir model performance is indicated in Table 29.. Trans. 243.T.." SPEJ (June 1974) 295-308.M. K. Again. H. O.R. Last. Price. " paper SPE 11197 presented at the 1982 SPE Annual Technical Conference and Exhibition. Journal of Petroleum Technology. we are indebted to Core Laboratories Inc." paper SPE 12242 presented at the 1983 SPE Symposium on Reservoir Simulation. J. The nature of IMPES models restricted the timestep size to a value generally less than 30 days. Reid.4x 10 3 stock-tank m 3 ]. J.M. R. 11. 12. or about 4. Prausnitz." JPT (Jan.: "Application of a Regression-Based EOS PVT Program to Laboratory Data. Nghiem. Unfortunately. The median value was 160 MSTB [2S. Nov. 7..: "A Correlation of the Viscosity of Hydrocarbon Systems With Pressure. AIME.: "Second Comparative Project: A Three-Phase Coning Study. This may be valid.E. 16. New York City (1977) 426..S. A.. 1981) 13-25. References 1. third edition. There is considerable disagreement about condensate saturation in the producing node. 237. 6. and Coats.. B.nique. Chern. Trans . Fig.X. 13. Fund. H. S.C. H.B. and Robinson.S. companies using precalculated K values found wide differences in the amount of condensate formed and its rate of evaporation compared with companies that used EOS methods.. 21-24. (1935)27." SPEJ (Oct. Nelson. Data of this kind have been measured in previous compositional simulation studies 24 -26 and are needed here to decide which answers are correct.: "Compositional Modeling With An Equation of State. but we do not know how it affected results.A. it casts doubt on the usefulness of the swelling data for the problem." paper SPE 5247 presented at the 1974 SPE Annual Technical Conference and Exhibition. 6.: "A New Approach to Quasi-Newton Methods with Application to Compositional Modeling. Several companies chose not to match the high pressure range of the swelling data.H. and only the raw data are given.H. The supplementary runs made were identical in all respects except for the treatment of K values. 1980) 363-76. The response envelope for all companies who used similar in-line K values is also shown. Sept.: "A New Two-Constant Equation of State. The project does not establish which method gave better answers in this case. and Clark. 3. 687-98. Surprisingly. Chappelear. Oct. G. IS shows condensate saturation for the supplemental run with 11 components and in-line K values with an EOS.4). 14 and IS show condensate saturation for the supplemental runs. the range was 3 to 8 % of initial condensate in place. New Orleans. K. Nonetheless.: "Comparison of Solutions to a Three-Dimensional Black-Oil Simulation Problem." SPEJ (Dec. Meijerink.. and Eng. 2. and Shank.L. Marathon. Liquid yield in simple pressure depletion (Fig. and Eakin.K.." SPEJ (June 1968) 157-62. and Eng. This is probably because K values are used as tables or as calculated in line with an EOS.G. Conclusions 1. These show somewhat better agreement with each other and a slower evaporation in the late stages of gas cycling. 8. We thank the participants for their cooperation and well-written responses to the problem. 14 shows a clear indication of high evaporation rate of condensate obtained with a K-value table. K. 9.K. G. 2. B. Trans. Kazemi. Coats.. Results for Case 1 are qualitatively the same as for Case 2. but many other factors are involved. and Nolen. 1981) 21. We were unable to provide these data for this problem. Depletion data and lean-gas swelling data for the retrograde gas condensate are matched well by all companies..: "Characterization of Petroleum Fractions. 1978) 355-68. 1964) 1171-76.X.H.: "Direct Methods in Reservoir Simulation. T. San Francisco. AIME. the surface oil rates disagree by about 20%. Little. J. For the run with K values as tables. Large discrepancies were observed in incremental oil obtained by gas-sales deferral (Case 2 vs. and Smart. Node (7.S. and Murphy. Chern. L. In-line EOS methods seem to increase run times. London. 1460-64. August 1987 . 15-18. J. Vestal.: "Characterizing Hydrocarbon Plus Fractions..R. E.. 257.: "The Viscosity of Natural Gases. Lee.: "An Equation-of-State Compositional Model." Ind. Fig.: "Calculating Viscosities of Reservoir Fluids From Their Compositions. This problem would benefit from PVT data that include some equilibrium flash data for feed compositions that exist in the enrichment zone near the producer. 14. 4. Machine speed differences were not factored into the comparisons.: "An Efficient Multicomponent Numerical Simulator.E. Odeh. J. The pressure range for the swelling data was beyond what is needed for cycling. H. and McCord-Lewis) that used precomputed K-value tables. Trans. In early years of cycling with partial pressure maintenance.

S. 22.67)/1.013 250* E+05 E-Ol bbl x 1.725 895 E+Ol E-Ol fi x 3. August 1987 51 Metric Conversion Factors atm x 1. Revised manuscript received April 23.L. " SPEl (April 1972) 96-102. Trans.831 685 E-02 fi 3 /lbm x 6. H..450 377 E-Ol OR °R/1.801 175 * Conversion factor is exact.048* fi3 x 2. and Spencer.: ''Characterization of North Sea Crude Oils and Petroleum Fractions. Cook.G. C. 997 . Attra. 265. Fund. 26. COMPIII. D. Stone. AlME. Norwegian Inst. B. C. Walker. (March 1976) 153-58. "Equation of State Compositional Reservoir Simulator. Journal of Petroleum Technology. 18.I. and Sandler. Trans.. 1983.I.: "Realistic K Values of C 7 + Hydrocarbons for Calculating Oil Vaporization During Gas Cycling at High Pressures. Chern." Ind.D.8 E-03 gal x 3. and Eng. AIME.B." Scientific Software-Intercomp Corp. and Firoozabadi. 222.D.: "Nonequilibrium Gas Displacement Calculations.8 OF (OF +459. 25. Anal. 1978) 1649-55.: "Improve Prediction of Enthalpy of Fractions.G. M. 1984) 685-96. 24.894757 E+OO psi -I x 1. Kesler.8 E-Ol scf/bbl x 1." JPT(July 1969) 901-15.J. and Lee. S. Whitson.785412 E+Ol lbm/fi 3 x 1. D. M.H.5.589 873 Btu x 1. of Techno!.601 846 psi x 6. 49-54.. A. B.242796 E+Ol OF (OF-32)/1.." Hydrocarbon Proc..2. Paper (SPE 12278) first presented at the 1983 SPE Reservoir Simulation Symposium held in San Francisco. and Yarborough." SPEl (Dec. H. 19.: "Effect of C H Properties on Equation-of-State Predictions. of Trondheim.: "A Third Parameter for Use in General Thermodynamic Correlations. Norway. Kesler. Lee." JPT (Nov. 1987. Release 1. 21. L. 23. Nurner.: "The Effect of Phase Data on Liquid Recovery During Cycling of a Gas Condensate Reservoir. U. Haaland. 1961) 130-36..: "Predicting Phase Behavior of Condensate/Crude-Oil Systems Using Methane Interaction Coefficients. AIME. Fussell. Nov. 24. A. Pa m3 kJ kJ/m 3 m m3 dm 3 /kg °C K m3 kg/m3 kPa kPa. L.1 K std m 3 /m 3 JPT Original manuscript received in the Society of Petroleum Engineers office Oct. (1968) No." SIAM J. G. 1986). Katz.: "Iterative Solution of Implicit Approximations of MultiDimensional Partial Differential Equations. (March 3. Paper accepted for publication Oct. Trans.055056 E+OO Btu/fi 3 x 3.I.." MS thesis. 20. 246. 15-18.17." SPEl (Sept.. 530-58. 24. 1986. (1979) 18.B.