Sie sind auf Seite 1von 6

Consistency Check and Reconciliation of PVT Data From Samples Obtained With Formation Testers Using EOS Models

Jitendra Kikani, SPE, Chevron Petroleum Technology Co., and John Ratulowski, SPE, Shell E&P Technology Co.
Summary

Accurate description of hydrocarbon fluids is critical to field development planning including facilities design, recovery efficiencies, secondary recovery processes, and remedial procedures for intervention. Despite this large impact of fluid characteristics on field development, generally a small number of downhole fluid samples are collected. Often they are not representative of the reservoir fluid for a variety of reasons. These include phase separation caused by drawdown during sampling and leakage during recovery and transport. In this paper we show the utility of a routinely reported piece of data, namely, the tool opening pressure, in a quantitative manner to check the quality of the sampled fluid. The condition of the fluid is modeled by performing an isochoric flash experiment that mimics the actual conditions. A significant discrepancy in modeled and real opening pressure data often suggests leakage. When this method is used, the fluid samples may be approximately reconstructed, multiple fluid samples from the same reservoir may be reconciled, and a consistent phase behavior of the fluid may be determined.
Introduction

Nonrepresentative downhole fluid sample collection can have significant impact. For example, when a two-phase sample leaks, the resulting loss of hydrocarbons (usually gas) causes the fluid to appear denser than the actual reservoir fluid. If these fluids are used to characterize the actual field production, the gas/oil ratio (GOR) would be underestimated or the condensate/gas ratio would be higher than the actual value. This error can translate to undersized gas facilities and/or constrained oil rates from wells. Often the only sources of pressure and pressure/volume/temperature (PVT) data in a new field are samples obtained with a wireline formation-testing tool. Surface recombined samples are available only for wells that are production tested. Wireline formation-testing tools (referred to in a generic sense) are designed for multiple pressure measurements to establish fluid gradients and to collect single or multiple fluid samples for fluid characterization. Some of the tools that belong to this class are repeat formation testers (RFTs),* modular dynamic testers,1,** and reservoir characterization instruments).2 General sampling problems include seal failures and probe plugging that can occur in highly laminated or otherwise heterogeneous formations. Friable and unconsolidated fine sands add to the misery.* In addition to sampling problems, fluid leakage and/or high formation drawdown resulting in two-phase flow may affect the use of a nonrepresentative sample in laboratory measurements. Two-phase flow into the sample chamber typically results in GORs greater than or equal to the actual reservoir fluid GOR. If the fluid in the chamber is in a two-phase nonequilibrium state when the sample chamber is closed downhole, the opening pressure could be slightly depressed. As gas dissolves into the oil, pressure decreases
*J.J. Smolen, RFT Pressure Interpretation, Schlumberger Document, May 1977. **MDT 1995, Schlumberger Internal Document MH-269979. Copyright 1998 Society of Petroleum Engineers This paper (SPE 50979) was revised for publication from paper SPE 36743, first presented at the 1996 SPE Annual Technical Conference and Exhibition held in Denver, Colorado, 69 October. Original manuscript received for review 8 October 1996. Revised manuscript received 13 May 1998. Paper peer approved 19 May 1998.

in the chamber. However, leakage after the fluid has cooled and broken into two phases results in lower GORs and a low tool opening pressure. We have observed this behavior several times recently in the Gulf of Mexico. Low opening pressure is a warning flag for a nonrepresentative sample. Often other data can help in determining the cause of a low value. If it were possible to determine the amount and overall composition of the gas leaked from the tool during transport from reservoir to surface conditions, one could mathematically add the leaked fluid back to the sample in a compositional model. Obviously, the composition of the leaked gas is unknown, but a fair assumption for the light ends can be made. The simple approach described in this paper for estimating in-situ fluid compositions from leaked downhole samples compliments other techniques for reconstructing in-situ reservoir fluids.3 One example showing this procedure for an oil well and a subsequent check with a recombined fluid sample from the gas cap (Prospect Tahoe) illustrates the procedure. This method has also been used to evaluate the quality of two samples obtained at the same depth from the same well but with different GORs (Prospect B). In a third example, fluids collected from the oil rim and the gas cap within the same well are characterized and reconciled with gradient information (Prospect C). For this prospect and at South Pass 62 (SP62), the integrity of the samples collected is verified with this method.
Fluid Sampling With Formation Tester Tools

The wireline formation tester tools are usually run after a logging and a cleaning trip. After the tool is positioned, a hydraulically retractable probe embedded in a rubber packer is forced through the mudcake to make a seal with the formation. Two opposing backup pistons on the other side of the tool push the probe against the formation and help maintain a seal. The backup pistons also help center the tool body in the well, reducing the risk of differential sticking. After a hydraulic connection is established, a set amount of fluid sample is allowed to enter into the pretest chamber.1 Based on the pretest results, one may elect to take samples. There are a number of variations to this theme in different tools and their modules. Some tool combinations have a dual-packer mechanism (used for carrying out mini-drillstem tests or used in areas where a probe seal is difficult to achieve), whereas others have pretest fluid chambers. During fluid sampling, the reservoir fluid is usually preceded by mud and/or mud filtrate. Modern tools look for resistivity contrast or optical transmissivity and reflectivity (with an optical sensor)4 to detect unwanted fluids that may be pumped through the tool and discarded back in the well. Circulation of fluid through the tool is not possible with older-generation tools. Despite all the advances, there are a number of reasons why representative or sufficient sample volumes may not be obtained with these tools. Some of these reasons are packer seal leakage because of high differential pressure, inadequate probe contact with the formation, sand lodging causing leakage, mud filtrate collected in the sample chamber (particularly difficult with oil-based muds), sand lodging in the probe or flowline plugging causing excessive drawdown (flash gas collected in sample chamber), and high drawdown in a low permeability or highly laminated formation. After the samples are collected, the tool is brought back to the surface. The sample chambers are disconnected from the tool and sent to the laboratory for testing or the fluid is transferred from the
SPE Reservoir Evaluation & Engineering, August 1998

348

tool to a laboratory test vessel on site. The first things recorded are the tool opening pressure, color of the fluids, and the amount of bottom sediments and water (BS&W). The tool opening pressure is important when quantifying any leakage that may have occurred. Note that the tool opening pressures are used along with an estimate of the closing pressure and temperature of the tool downhole. This is obtained from the log.
General Methodology

TABLE 1PROSPECT TAHOE EXPERIMENTAL AND EOSMATCHED DATA Matching Parameters Psat (@203F), psia sat, gm/mL C7 Mol. Wt. API gravity Psat, psia Flash GOR (scf/bbl) Bo @BPP (rb/stb) C7 Mol. Wt. EOS Matched 4,879.88 0.2374 138.42 33.14 3,871.9 790.00 1.430 219.70

Well Tahoe 4ST2

Experimental 4,880.0 0.2400 143.50 33.5 3,872.0 782.00 1.420 211.30

Often the formation tester tool opening pressure is ignored. If it is looked at, only a qualitative judgment based on the value is made. If the reported values are lower than a gut feel number, it is assumed that the sample is poor. This, however, could be quite misleading. The procedure proposed here requires the characterization of the imperfect fluid sample with an equation of state (EOS) model tuned to the measured laboratory data. When the characterization is complete, the opening pressure is determined from an isochoric (constant volume) flash of the modeled fluid from reservoir conditions to the surface temperature. This flash is represented as a series of constant composition flashes at the surface temperature and various pressures. A plot of the ratio of total fluid volume at surface to reservoir volume vs. pressure is made. The pressure at Vt /Vres 1 gives the tool opening pressure at the surface. This representation of the isochoric flash allows one to determine the sensitivity of the volume ratio to pressure. If the slope of Vt /Vres is small, the calculated opening pressure is effected to a greater degree by uncertainties in the fluid composition and in the EOS model. The calculated opening pressure when compared with the measured opening pressure provides a diagnostic tool for the sample validity. In case the sample has leaked, an estimate of the moles of light ends that are lost can be made from the measured and ideal opening pressure values. Assuming a composition of the lost gas, the fluid phase behavior can be reconstructed. It should be recognized that the calculated opening pressures are somewhat approximate because the composition, in general, will not be exact because of leakage and imperfect reconstruction. In general, a small amount of an aqueous phase in the sample chamber does not affect the opening pressure calculation because of the small compressibility of the aqueous phase compared to the hydrocarbon phase. However, if the aqueous phase volume is large and the chamber pressure is high, both the aqueous and the hydrocarbon phases must be modeled. This generic procedure, with slight variations, was applied to a number of fields in the Gulf of Mexico. Four field examples of the procedure with slight variations are given next.
Prospect Tahoe

Tahoe 3OH

dewpoint densities, heavy-end molecular weight, and liquid dropout curve. The tuning parameters used were the methane binary interaction parameters and the C7 characterization (usually not determined very accurately). Table 1 shows the parameter match values. The liquid dropout curve, which determines the liquid/gas ratio as a function of pressure depletion, was matched well. Fig. 1 shows both the measured and predicted liquid dropout curve for the production test fluid sample. The tuned EOS model was used to perform compositional grading experiments. These experiments suggested the possibility of the presence of an oil rim. The graded fluid was flashed to the estimated GOC depth and the composition was noted.
Characterization of Tahoe-3OH Fluid

A 2.75-gallon RFT tool caught an oil sample at 10,379 ft. The tool opened on the surface at 1750 psig pressure. After 690 mL of a subsample of the fluid was removed, the tool was bled down to yield 10.8 scf of gas and 2.2 L of oil; 18 mL of BS&W was left in the sample chamber. The reported bubblepoint pressure (BPP) at 87F was 2,590 psig. The fluid was characterized in a manner similar to the Tahoe-4ST2 sample. Binary interaction parameters and C7 characterizations were adjusted to yield the correct API gravity and the BPP.
Discussion of Consistency Check

Viosca Knoll 783 (Prospect Tahoe) is a thinly laminated, relatively lean gas condensate reservoir with an associated thin oil rim. The Prospect is located in approximately 1,500 ft of water in the Gulf of Mexico. Tahoe-4ST2 was drilled in the gas cap, and production was tested. The details of the test design and implementation are given in Ref. 5. A surface recombination sample of the reservoir fluid was collected from the partially perforated sand at the end of an extended flow period. The characterized surface sample from Tahoe-4ST2 with the reservoir fluid gradient and bottomhole pressure data was used to confirm the estimated gas/oil contact (GOC) depth with a compositional grading model. The modeled fluid was also used to obtain composition information at the GOC. The use of this information to confirm sample validity of the Tahoe-3OH RFT sample is discussed later.
Phase Behavior of Tahoe-4ST2

Based on the static data from the permanent downhole pressure gauge in Tahoe-4ST2, the initial reservoir pressure is 4,865 psi at 9,480 ft. The pressure at the GOC at 10,250 ft, after correcting for the gas gradient, is 4,930 psi. The measured dewpoint pressure of Tahoe-4ST2 gas is 4,885 psia, which is consistent with the pressure at the gas/oil contact (4,930 psi). However, the saturation pressure of Tahoe-3OH fluid is 3,872 psi, suggesting a highly undersaturated fluid at reservoir conditions (an unlikely situation for a reservoir in the vicinity of a gas cap).

A good and stable surface separator sample was collected from the well at the end of an extended flow test (150 hours). The experimental data from the sample collected at 970 psig separator pressure were described with an in-house EOS package. The separator fluid was recombined to bottomhole conditions. The model was tuned to provide the experimental dewpoint pressure,
SPE Reservoir Evaluation & Engineering, August 1998

Fig. 1Liquid dropout curves for Tahoe-ST2. 349

To confirm this, isochoric (constant volume) flash experiments were run with the Tahoe-3OH fluid at surface temperature. The characterized fluid is flashed at different pressures and the ratio of total fluid volume to original volume at reservoir conditions is plotted against pressure. The pressure corresponding to a unit value of the above ratio gives the opening pressure. This is shown in Fig. 2. The opening pressure for the fluid should be 2,750 psia compared with the measured opening pressure of 1,765 psia. The following experiments were performed with the two tuned PVT data sets corresponding to Tahoe-4ST2 and Tahoe-3OH. Add C1 to C4 components (with the same composition as the original fluid) to the Tahoe-3OH fluid to increase the BPP to the GOC pressure level. This requires the addition of 0.21 moles of the gas. Next, flash the new reservoir fluid to 10 psi below the GOC pressure to yield two phases. The gas phase composition was compared with the Tahoe-4ST2 gas at the GOC. The results were remarkably similar, not only suggesting a connected reservoir but also confirming fluid leakage from the RFT tool. Table 2 gives the gas composition of the two fluids. Some of the differences in composition can be accounted for by the temperature difference between the two samples. To check whether the gas volumes added in Step 1 are realistic, the Tahoe-3OH original fluid was flashed from the sampling pressure (4,686 psia) and temperature (197F) to the opening pressure (1,765 psia) and temperature (87F) of the RFT tool (isochoric flash). Assuming that the mud and filtrate volumes in the tool stay constant and the tool volume itself does not change when brought to the surface, the initial charge of oil and gas in the tool should be constant. Thus, Vt @1765 psi 4.2476 1.1834, Vo @4686 psi 3.5894 . . . . . . . . . . . . . . . . . . . (1)

TABLE 2COMPARISON OF THE GAS COMPOSITIONS OF GAS AND OIL SAMPLES AT THE GOC FOR PROSPECT TAHOE Recombined Sample 0.00817 0.88508 0.03910 0.01755 0.00399 0.00658 0.00279 0.00269 0.00429 0.02590 203F Reconstructed Sample 0.00447 0.87746 0.04600 0.02316 0.00338 0.00726 0.00289 0.00325 0.00576 0.02637 197F

Composition N2 C1 C2 C3 i-C4 n-C4 i-C5 n-C5 C6 C7 Sample temperature

Prospect B

which is an 18.34% excess, and this is the amount of lighter phase that has been presumably lost in transport. This volume is equivalent to (4.2476 3.5894) 0.6582 ft3. The number of moles of lost gas then is moles of lost gas Vlost gas 0.243 moles. . . . . . . . . . . (2) Mol. wt g

This set of samples from a deepwater Gulf of Mexico sand illustrates the situation in which a sample chamber clearly leaked. Significantly different composition and fluid property data were measured for samples of a reservoir fluid from the two chambers filled at the same depth. In addition, the opening pressures of the two vessels were significantly different. The low GOR vessel (lower 234-gallon chamber) opened at 1,500 psig, and the highGOR vessel (upper 1-gallon chamber) opened at 6,000 psig. Although there was significant drawdown during sampling, the lowest measured pressure was 3,000 psia greater than the highest measured saturation pressure. Fluid models were constructed with compositional data from both samples. The cooling process was simulated by an isochoric (constant volume) flash from the initial reservoir condition to the final state of 61F. Fluid models for both fluids and the results of the flash are summarized below.
EOS Fluid Models

The initial feed volume was 1.6 moles. Thus, moles of gas lost/ lb-mole of feed 0.243/1.6 0.152 moles. To get the correct BPP of the fluid at the GOC, 0.21 moles of gas had to be added to the RFT fluid. This is within the limits of the assumptions made regarding the leaked gas composition. If a heavier gas were to be added to the RFT fluid, the moles of gas required would be lower and close to the 0.152 moles. This confirms that the Tahoe-3OH fluid after gas addition is a representative sample, and the characterization thus performed may be adequate for reservoir-modeling work.

EOS fluid models were created with the laboratory compositional data. The models were tuned to match the saturation pressures at several temperatures, live oil densities, and single and multistage GOR and API gravity. The experimental and EOS-matched values are shown in Table 3 for both models.

TABLE 3PROSPECT B EXPERIMENTAL AND EOSMATCHED DATA Matching Parameters Psat (@180F), psia sat (@180F), gm/mL Psat (@130F), psia sat (@130F), gm/mL Flash GOR, scf/bbl Flash API gravity MS GOR, scf/bbl MS API gravity Psat (@180F), psia sat (@180F), gm/cc Flash GOR, scf/bbl Flash API gravity MS GOR, scf/bbl MS API gravity EOS Matched 8,257.0 0.6166 8,539.0 0.6368 2,272 30.5 1,971 34.5 4,632.0 0.6918 1,022.0 30.1 900.0 32.6

Well B, high-GOR fluid

Experimental 8,258.0 0.6081 8,525.0 0.6281 2,247 30.5 1,944 34.5 4,635.0 0.6927 1,031.0 30.0 851.0 33.5

B, low-GOR fluid

Fig. 2Isochoric flash for Tahoe-3OH. 350

SPE Reservoir Evaluation & Engineering, August 1998

Reconciliation (Results of Isochoric Flash)

An isochoric flash of the low GOR fluid indicated an opening pressure of 3,700 psia. To obtain an opening pressure of 1,500 psia, the hydrocarbon phase would have to expand 1.4 times its volume at reservoir conditions (Fig. 3b). Approximately 1,120 mL of an aqueous phase was drained from the chamber after blowdown. If the aqueous phase is treated as pure water, the change from reservoir to surface conditions would result in a less than 1% volume change of the hydrocarbon phase. The expansion of the water phase caused by depressurization partially counters the shrinkage caused by cooling and degassing. A 40% change in the hydrocarbon phase volume cannot be explained by changes in the volume of the aqueous phase or vessel. An isochoric flash of the high-GOR fluid (Fig. 3a) indicates an opening pressure of 5,840 psia. The reported opening pressure was 6,000 psia. This is certainly within the error of the calculation. Again, the aqueous phase volume change resulted in a less than 1% change in the hydrocarbon phase volume. These two flashes suggest the possibility of a leak in the lower vessel. To check this hypothesis, the high-GOR fluid (good sample) was flashed to 1,500 psia and 61F. Enough of the upper phase was removed to return the system to the initial volume at reservoir conditions. This process crudely approximates the loss of the leaked gas. The properties and compositions of the remaining fluid were compared to the low GOR fluid (Table 4). If the vessel did leak, gas would have been removed at higher temperatures and pressures as the vessel was returned to the surface and processed. This could have caused some of the small differences in composition observed in Table 4. However, the fluid properties and compositions are close enough to confirm a leak in the lower chamber.
Prospect C

TABLE 4COMPARISON OF FLASH RESULT WITH LOWGOR FLUID AT PROSPECT B Composition C1 C2 C3 i-C4 n-C4 i-C5 n-C5 C6 C7 Flash Result 0.504 0.053 0.049 0.0102 0.028 0.013 0.016 0.023 0.300 Low-GOR Fluid 0.520 0.044 0.040 0.008 0.023 0.011 0.015 0.031 0.304

This reservoir contains a large gas cap and a small associated oil rim. Well 1 showed gas to the base of pay, and an RFT fluid sample was collected at a true vertical depth of 11,667 ft. The sample opened at 3,865 psia and 88F. Experimental data were available at both 200 and 130F. The model was tuned to data at both temperatures. Table 5 shows the experimental and model characteristics of the fluid. A close match between the measured and experimental data was obtained at both temperatures. The tuned model was then used to simulate an isochoric flash experiment. For the gas cap sample, Fig. 4a shows a plot of Vt /Vres as a function of pressure. The isochoric flash conditions corresponding to Vt /Vres 1 results in an opening pressure of 4,000 psia compared to the measured value of 3,865 psi. These values are in close agreement considering that the expansion of nonhydrocarbon liquids in the RFT chamber have been neglected. This suggests that the collected gas sample was good. The well was sidetracked and oil was encountered in well 1ST. An oil sample was collected from a true vertical depth of 12,004 ft. The opening pressure recorded was 515 psia and 86F, a surprisingly low value. The experimental and EOS-matched parameters are shown in Table 5. The plot in Fig. 4b suggests that the

opening pressure should be 1,250 psi, a 535-psi difference in the opening pressure. This suggests sample leakage because this difference is too high to explain with tool and mud expansion. The number of moles of lost gas was estimated from this difference in opening pressure, and a synthetic gas comprising C1 to C4 components with the same overall molar composition as the reservoir fluid was added to the characterized fluid to come up with a reconstructed fluid. Good reservoir fluid gradient data are available in this reservoir, which established fluid contacts quite accurately (Fig. 5). The reconstructed oil sample was flashed to the pressure just above the GOC, and the gas composition was compared with the gas cap fluid flashed to the GOC pressure. The two compositions agreed within the limits of the assumptions of the lost gas and the experimental error. This acts as a confirmation of the leak in the oil sample.
South Pass 62

A downhole sample was obtained at South Pass 62 in a 1-gallon chamber. An EOS model was constructed with the reported reservoir fluid compositions. The model was tuned to match surface and reservoir densities, saturation pressure at several temperatures, and multi- and single-stage GORs. The EOS model was compared with measured data in Table 6. An isochoric flash calculation from the reservoir pressure and temperature to the opening temperature of 86F indicated an opening pressure of 4,515 psia (Fig. 6). Although 1,085 psia appears to be a large discrepancy in pressure, at 4,515 psia the hydrocarbon volume has only increased by 6%. A less than 10% error in the hydrocarbon volume is not a clear indication of leakage in the absence of other data. In this case, the 3.4% error in the predicted saturated fluid density at 276F and the lack of density data at 130F causes some concern about the accuracy of the fluid model. If no other data indicating a higher GOR are available, one could assume the sample is representative. When the method described in the previous examples was used, an approximate

Fig. 3Isochoric flash for Prospect B high-GOR (a) and low-GOR (b) fluid. SPE Reservoir Evaluation & Engineering, August 1998 351

TABLE 5PROSPECT C EXPERIMENTAL AND EOS-MATCHED DATA Well C, RFT gas sample Matching Parameters Psat (@200F), psia sat, gm/mL g, cp Psat (@130F), psia sat, gm/mL g (@DPP), cp Condensate gravity (API) LGR (bbl/mmscf) API gravity Psat (@200F), psia sat, gm/mL Oil viscosity, cp Flash GOR, scf/STB Gas viscosity (@1,500 psia), cp Experimental 7,5250 0.3730 0.0545 7,336.0 0.4022 0.0625 443 75.9 33.0 1,695.0 0.7390 0.7730 415.0 0.0154 EOS Matched 7,5270 0.3691 0.0525 7,338.0 0.4025 0.0624 44.5 787 33.0 1,679.0 0.7310 0.7530 410.0 0.01525

C, RFT oil sample

Fig. 4 Isochoric flash for Prospect C gas sample (a) and oil sample (b).

TABLE 6EXPERIMENTAL AND EOS-MATCHED DATA FOR SOUTH PASS 62 Matching Parameters Psat @276F, psia sat @276F, gm/mL Psat @130F, psia sat @130F, gm/mL Flash GOR, scf/bbl Flash API gravity MS GOR, csf/bbl MS API gravity
* Not applicable.

Experimental 7,480.0 0.4909 8,200 N/A* 3,050.0 35.1 2,7530 38.3

EOS Matched 7,469.0 0.5076 8,200 0.5867 3,053.0 35.0 2,700.0 384

Fig. 5Prospect C reservoir fluid gradient.

calculation of the lost gas indicated an increase of only 117 scf/bbl (3.8%) in the single-stage flash GOR.
Discussion

Note that the composition of gas added back to the characterized fluid is an approximation. The composition of the lost gas is unknown and a mixture of the original overall composition has to
352

be added to reconstruct the original fluid. A mixture of C1 to C4 or C1 to C3 with the original composition seems to work well. At times, the BS&W, which is primarily the mud fluid, is present in significant volumes in the formation tester tool. This makes the tool fluid quite incompressible (if the volume/compressibility product of the hydrocarbons is small). This could mean a large change in tool pressure that is undergoing an isochoric expansion while coming to the surface. A low pressure in this case does not necessarily translate to leaked hydrocarbons. One must, in such a case, account for the presence of the mud fluid by means of an iterative process.
SPE Reservoir Evaluation & Engineering, August 1998

References
1. Zimmerman, T. et al.: MDT Tool: A Wireline Testing Breakthrough, Oilfield Review (April 1992). 2. Michaels, J., Moody, M., Shwe, T.: Wireline Fluid Sampling, paper SPE 30610 presented at the 1995 SPE Annual Technical Conference and Exhibition, Dallas, 2225 October. 3. Fevang, O., Whitson, C.H.: Accurate In-Situ Compositions in Petroleum Reservoirs, paper SPE 28829 presented at the 1994 SPE Petroleum Conference, London, 2527 October. 4. Badry, R. et al.:Downhole Optical Analysis of Formation Fluids, Oilfield Review (January 1994). 5. White, C.D. et al.: Reservoir Potential of Thin-Bedded Turbidites: Prospect Tahoe, paper SPE 28975 presented at the 1992 SPE Annual Technical Conference and Exhibition, Washington, DC, 47 October.

Fig. 6 Isochoric flash for South Pass 62 fluid.

SI Metric Conversion Factors

Recent advances in sampling techniques allow one to pump the reservoir fluid into the sample chamber and possibly perform downhole PVT. With these tools the closing pressure and temperature can be measured quite accurately. This should offer an improved quality control check.
Conclusions

bbl 1.589 873 cp 1.000 000 ft3 2.831 685 F (F32)/1.8 F (F459.67)/1.8 gal 3.785 412 lb/ft3 1.601 846 lbm 4.535 924 psi 6.894 757
*Conversion factor is exact.

E01 m3 E03 Pa/s E02 m3 C K E03 m3 E01 kg/m3 E01 kg E00 kPa SPEREE

Often we do not have multiple samples from a single reservoir sand. In such cases, consistency between multiple sets of data should be checked. These sets include estimated in-situ fluid density and sample composition, sampling and opening pressure, saturation pressure, and the proximity of gas/oil contact. All of this information should fit together to form a consistent reservoir fluid description. When this is not the case, it is important to recognize that an EOS simulation package can be used to reconstruct possible fluid compositions based on the procedure described in this paper. The method has been illustrated by a series of four examples in which tool opening pressures and reservoir conditions were used to reconstruct a consistent reservoir fluid. Although this method is not a substitute for good fluid data, it allows one to make reasonable estimates for fluid property data with incomplete information.
Nomenclature Psat saturation pressure, psia Vo formation volume factor, rb/stb Vres bottomhole or sample volume, ft3 Vt total surface volume (liquid gas), ft3 g gas viscosity, cp g gas density, lb/ft3 sat saturation density, lb/ft3

Jitendra Kikani is a staff research scientist at Chevron Petroleum Technology Co. in La Habra, California. Previously, he was a senior research engineer with Shell E&P Technology Co. in Houston, and a senior development engineer with Intera Information Technologies Inc. in Denver, Colorado. His research interests include numerical reservoir modeling, pressure-transient analysis, and reservoir characterization. Kikani holds a Btech degree in petroleum engineering from Indian School of Mines; an MS degree in mechanical engineering from U. of California, Berkeley; and an MS degree in mathematics and a PhD degree in petroleum engineering, both from Stanford U. He is the past chairman of the Pressure Transient Annual Technical Committee. He currently serves on the Editorial Review Committee and Program Coordinating Committee for Annual Meeting. John Ratulowski is a senior research engineer with Shell E&P Technology Co. in Houston. His current focus is on organic deposition and thermodynamic modeling for subsea systems and surface facilities. During his 10 years of experience at Shell, he has also worked in the areas of thermal recovery and reservoir phase behavior. He holds a BS degree from Purdue U. and a PhD degree from the U. of Houston, both in chemical engineering. He currently serves on the Editorial Review Committee and on an Annual Meeting Technical Committee.

SPE Reservoir Evaluation & Engineering, August 1998

353

Das könnte Ihnen auch gefallen