Characterization of a Carbonate Reservoir With Pressure-Transient Pressure-Transient Tests Tests and Production Logs: Tengiz Field, Kazakhstan K.T. Chambers,* SPE, and W.S. Hallager,** SPE, Tengizchevroil; C.S. Kabir, SPE, Chevron Overseas Petroleum Technology Co.; and R.A. Garber, Chevron Petroleum Technology Co.
Summary The combination of pressure-transient and production-log (PL) analyses has proved valuable in characterizing reservoir flow behavior in the giant Tengiz field. Among the important findings is the absence of clear dual-porosity flow. This observation contradicts an earlier interpretation that the reservoir contains a wellconnected, natural fracture network. Fracturing and other secondary porosity mechanisms play a role in enhancing matrix permeability, but their impact is insufficient to cause dual-porosity flow behavior to develop. Flow profiles measured with production logs consistently show several thin (10 to 30 ft) zones dominating well deliverability over the thick (up to 1,040 ft) perforation intervals at Tengiz. A comparison of of PL results and core descriptions descriptions reveals reveals a good correlation between high deliverability zones and probable exposure surfaces in the carbonate reservoir. Contrary to earlier postulations, results obtained from pressuretransient and and PL data at Tengiz Tengiz do not not support rate-sensitive rate-sensitive productivity indices (PI’s). Inclusion of rate variations in reconciling buildup and drawdown test results addressed this issue. We developed wellbore hydraulic models and calibrated them with PL data for extending extending PI results to wells that do not have have measured measured values. A simplified simplified equation-of equation-of-state -state (EOS) fluid description was an important component of the models because the available black-oil fluid correlations do not provide reliable results for the 47°API volatile Tengiz oil. Clear trends in reservoir quality emerge from the PI results. Introduction A plethora of publications publications exists on transient transient testing. However, However, only a few papers address the issue of combining multidisciplinary data to understand reservoir flow behavior (Refs. 1 through 4 are worthy of note). We used a synergistic approach by combining geology, petrophysics, transient tests, PL’s, and wellbore-flow modeling to characterize the reservoir flow behavior in the Tengiz field. Understanding this flow behavior is crucial to formulating guidelines guidelines f or reservoir management. management. Permeability estimation from pressure-transient data is sensitive to the effective reservoir thickness contributing to flow. Unfortunately, difficulties associated with the calibration of old openhole logs, sparse core coverage, and a major diagenetic overprint of solid bitumen combine to limit the identification of an effective reservoir at Tengiz based on openhole log data alone. Consequently, PL’s have been used to identify an effective reservoir in terms of its flow potential. A limitation of production production logs is that they only measure fluid entering the wellbore and are not necessarily indicative of flow in the reservoir away from the well. Pressure data from buildup and drawdown tests, on the other hand, provide insights into flow behavior both near the well and farther into the reservoir. The combination of pressure-transient **Now with Chevron Petroleum Technology Co. **Now with Chevron Canada Resources Copyright © 2001 Society of Petroleum Engineers This paper (SPE 72598) was revised for publication from paper SPE 38657, first presented at the 1997 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, 5–8 October. Original manuscript received received for review 2 December December 1997. Revised manuscript received 2 January 2001. Paper peer approved approved 20 June 2001.
250
analysis using simultaneous downhole pressure and flow-rate data along with measured production profiles provides an opportunity to reconcile near-wellbore and in-situ flow behavior. Expansion of reservoir fluids along with formation compaction provides the current drive mechanism at Tengiz because the reservoir is undersaturated by over 8,000 psia. As the field is produced, reservoir stresses will increase in response to pressure decreases. 5 Increased stresses can significantly reduce permeability if natural fractures provide the primary flow capacity in the reservoir. Wells producing at high drawdowns provide an opportunity to investigate the pressure sensitivity of fractures within the near-wellbore region. Early interpretations of pressure-transient tests at Tengiz uncovered a significant discrepancy between buildup and drawdown permeability, despite efforts to carefully control flow rates during the tests. Drawdown permeabilities typically exceeded the buildup results by 20 to 50%. Although this finding appears counterintuitive to the expectation that drawdowns (that is, higher stresses) would lead to lower permeability, it indicated a possible stress dependence on well deliverability. The method proposed by Kabir 6 to reconcile differences between drawdown and buildup results proved useful in addressing this issue. The opportunities opportunities to collect PL and downhole downhole pressure data at Tengiz are limited by mechanical conditions in some wells and by the requirement to meet the processing capacity of the oil and gas plant. On the other hand, accurate wellhead-pressure and flow-rate data are routinely available. Wellbore hydraulic calculations provide a basis for calculating flowing bottomhole pressures (FBHP’s) with the available surface data. Calculated FBHP’s can be combined with available reservoir pressure data to determine PI’s for wells lacking bottomhole measurements. The ability to compute accurate fluid properties is critical in applying this approach. Unfortunately, the black-oil correlations routinely used in wellbore hydraulic calculations 7–9 do not provide reliable results for the volatile Tengiz oil. We obtained good agreement between laboratory measurements of fluid properties and calculated values using a simplified EOS. 10 Surface and bottomhole data collected collected during PL operations operations provide a basis for validating wellbore hydraulic calculations. Networks of natural fractures can dominate the producing behavior of carbonate reservoirs such as Tengiz. Early identification of fractured reservoir behavior is critical to the successful development of these types of reservoirs. 11 We present an approach for resolving reservoir flow behavior by combining production profiles, pressure-transient tests, and wellbore hydraulic calculations. Furthermore, Furthermore, we discuss discuss the PL procedures developed developed to to allow acquisition of the data required for all three types of analyses in a single logging run. Field examples from Tengiz highlight the usefulness of this approach.
Background The Tengiz field is located on the south side of the Pri-Caspian basin. It is one of several large carbonate banks found at various depths all around the edge of the basin, as shown in Fig. 1. The platform evolved during Devonian and Carboniferous time by recurrent deposition of skeletal fragments and lime mud. Fig. 2 shows that the top of the central, or platform, portion of the reservoir is relatively flat with localized structural highs on the northern and eastern edges. The platform is bounded by faults or lithologic August 2001 SPE Reservoir Evaluation & Engineering
Fig. 1—Cross section of North Caspian basin.
breaks and surrounded by gently sloping flanks of carbonate debris. We define a rim of variable width (0.3 to 1.2 miles) at the transition from the platform interior to the flanks based on core, log, and production results. The Carboniferous reservoir is primarily limestone with minor dolomitization. The Devonian appears much more dolomitized, although the amount of core material upon which this
Fig. 2—Top structure map showing well locations.
observation is based is limited. Porosity varies from less than 3% to over 20%, while matrix permeability varies from less than 0.01 md to over 100 md, as shown in Fig. 3. The reservoir is almost 5,000 ft thick based on the lowest known oil. Initial reservoir pressure was 11,948 psia at a datum of 14,764 ft subsea.
Integrating Flow Profiles and Transient Tests Production-logging operations are a challenge at Tengiz because of the high reservoir (11,600 psia) and wellhead pressures (3,500 to 7,500 psia), in conjunction with a hydrogen sulfide content in excess of 16 mol% in the produced fluids. On the other hand, the calculation of flow profiles from the PL data is usually straightforward because the oil remains as a single phase at downhole conditions. The radius of investigation of the short (1 to 4 hours) buildup and drawdown tests performed while production logging is on the order of 100 to a few hundred feet based on typical reservoir permeability values. Equipment and Procedures. The PL string used at Tengiz includes gamma ray, pressure, temperature, and spinner sensors. A typical logging program includes an initial pass down and then up across the completion interval to investigate crossflow during shutin. This information is useful in identifying differential pressure depletion over a thick reservoir interval. Following calibration of the spinner tool in a blank section of casing above the completion interval, the well is placed on production with the logging tool located above the top perforations. In some wells, the existence of tail pipe below the packer requires placing the logging tool within the completion interval. The logging tool records pressure, temperature, and spinner response while the well is placed on production and allowed to stabilize. A series of down-and-up logging
Fig. 3—No apparent correlation between core permeability and core porosity. August 2001 SPE Reservoir Evaluation & Engineering
251
Oil Rate (STB/D)
Relative Inflow and Porosity
Gamma API 0
20
40
0
13,100
0.1
0.2
0
0.3
13,100
13,100
13,200
13,200
1,000
2,000
3,000
Openhole
13,200 Bashkirian
t f , h t p e
D
Porosity
13,300
13,300
13,300
13,400
13,400
13,400
13,500
13,600
PL
t f , h t p e D
Serpukhovian
13,500
t f , h t p e D
13,600
13,500
13,600 Perfs 13 ,622 –13,947 ft
13,700
13,700
13,800
13,800
13,900
13,900
13,700
Inflow
13,800
13,900 Bottom PL at 13,911 ft
14,000
14,000
14,000
Fig. 4—Production profile in T-112 before acid treatment.
passes across the completion interval are obtained at four different cable velocities after the well stabilizes. Stationary measurements are used to resolve discrepancies in the flow profiles indicated by the flowing passes. Finally, the logging tool is returned to the same depth where the original drawdown data were collected, and a buildup test is performed. Wellhead pressures are measured while logging with a highresolution electronic gauge. Surface flow rates are measured with an automated test separator. Typical Production Profiles. Fig. 4 presents the zonal flow profile for T-112 overlain on the openhole porosity results and next to the cumulative rate profile. The gamma ray correlation logs from the openhole and PL are also shown. Soviet openhole gamma ray logs run in Tengiz do not exhibit the same resolution or absolute readings as western cased-hole gamma ray logs. This discrepancy is apparent in Fig. 4 but does not limit our ability to correlate on depth using the two vintages of gamma ray logs. The dark horizontal lines on the correlation log plot correspond to the marker picks for the Bashkirian (top reservoir) and Serpukhovian stratigraphic units. In general, zones of inflow correspond to the higher porosity intervals in the well. On the other hand, approximately 50% of the production in the well comes from only 10% of the perforations starting at a depth of 13,736 ft. Another interesting result is that log porosity and gamma ray signature alone do not appear to provide a good indication of reservoir permeability. Fig. 5 presents the flow-profile results for T-112 15 months later following an acid treatment. A comparison of Figs. 4 and 5 reveals that the only significant profile changes, induced by the acid treatment, were moderate increases in the relative contributions of the major inflow zone at 13,736 ft and the high porosity zone below 13,890 ft. The acid was pumped above the fracture gradient (0.91 psi/ft) of the reservoir and included leak-off control and diversion additives. Acid diversion was unsuccessful based on the absence of clear improvements in the flow profile. Shut-in passes obtained during production logging in T-112 uncovered evidence for fluid crossflow in the wellbore. The logs showed that fluid was flowing up the wellbore from below the bottom-logged depth of 13,910 ft and exiting to the reservoir opposite the major inflow zones. The rate of crossflow was significantly higher following the acid treatment (410 STB/D) compared to the result obtained 15 months earlier (72 STB/D). This behavior 252
is indicative of differential pressure depletion over the thickreservoir interval owing to the existence of laterally continuous permeability barriers. Fig. 6 presents the flow-profile results for T-8. The existence of several thin, high-permeability layers is evident in this well. Approximately 30 out of 1,040 ft of perforations account for over 70% of the total production. Once again, high porosity appears to be a necessary, but not sufficient, condition for inflow to occur. One of the biggest drivers for running a production log in T-8 is that cores are available over much of the perforated interval. Consequently, we were able to investigate the importance of lithology and diagenesis on reservoir effectiveness with data from this well. Core recovered within the major inflow zone in T-8 from 13,091 to 13,110 ft exhibits grainstone texture, which indicates deposition in a very shallow subtidal-to-shoal environment. The cored interval is bioturbated, heavily leached, and extensively brecciated in places. Thin fractures, often partially or completely filledwithcalcitecementorsolidbitumen,arepresent.Vugsarecommon. Furthermore, thin sections from this core interval uncovered evidence for recrystallization. These characteristics are consistent with subareal exposure and karstification. Cores from within the other two major inflow zones exhibit the same characteristics. A transition in depositional environment occurs in T-8 at 13,248 ft in going from the Bashkirian to the Serpukhovian age deposits. The texture changes from grainstones to packstones/wackestones at this point. Furthermore, the porosity changes from predominantly vuggy and intergranular to microporosity associated with the micritic matrix of the packstone/wackestone carbonates. This distinction in porosity type is consistent with the higher permeability of the Bashkirian grainstones as demonstrated by the PL. Scattered grainstone intervals exist below the Bashkirian in T-8, but packstone/wackestone textures dominate. Fig. 7 presents the flow-profile results for T-104. Although this profile is much more homogeneous than the one for T-8 (Fig. 6), a few thin intervals still dominate deliverability. The rate of inflow is typically low opposite low porosity troughs and high opposite several high porosity lobes in the openhole log data. On the other hand, a high porosity interval from 13,281 to 13,317 ft exhibits almost no inflow. We present evidence in the following section, based on pressure-transient analysis, that formation damage is a likely cause of the low inflow opposite this interval. August 2001 SPE Reservoir Evaluation & Engineering
0
Oil Rate (STB/D)
Relative Inflow and Porosity
Gamma API 20
40
0
13,100
0.1
0.2
0
0.3
13,100
13,100
13,200
13,200
1,000
2,000
3,000
Openhole
13,200 Bashkirian
t f , h t p e D
13,300
13,300
13,400
13,400
13,500
PL
t f , h t p e
D
13,600
Serpukhovian
Porosity
13,400
13,500
t f , h t p e
D
13,600
13,700
13,700
13,800
13,800
13,300
13,500 Perfs 13,622 – 13,947 ft
13,600
13,700
13,800 Inflow
13,900
13,900
13,900
14,000
14,000
14,000
Bottom PL at 13,880 ft
Fig. 5—Production profile in T-112 after acid fracture treatment.
Pressure-Transient Analysis Using Downhole Rate Data. Fig. 8 shows an example of the stationary pressure and rate data collected in T-112. Flow rate is calculated from the spinner data based on the spinner calibration results. The data have been resampled from the initial 1-second frequency to improve the appearance of the plot and to facilitate pressure-transient interpretation. The missing section of data corresponds to the time during which the flow-profile data were collected. Flow conditions had not completely stabilized while collecting the flowing pass data based on a comparison of the stationary data recorded just before and just after this period of time. However, variations in rate and pressure were small. Safety
20
12,800 Openhole
Oil Rate (STB/D)
Relative Inflow and Porosity
Gamma API 0
and field production considerations led us to limit logging operations to daylight hours over a single day. This practice, in turn, limits stabilization times to between 1 and 2 hours. The data in Fig. 8 are typical of the results obtained in other Tengiz wells. Note that the rate decreases by approximately 7% from its maximum value over the course of 3 hours. The log-log results, including rate superposition, for the stationary pressure and rate data from T-112 appear in Fig. 9. The buildup and drawdown data exhibit good agreement in terms of pressure, although the appearance of a shallow trough in the derivative is more pronounced in the former. The sudden increase in the
40
0
0 .1
0 .2
0 .3 0 .4
0
0.5
12,800
PL
12,900
12,900 Bashkirian
1 ,0 00
2 ,0 00
3 ,0 00 4 ,0 00
12,800 12,900
Perfs 12,907 –13,533 ft
Porosity
13,000
13,000
13,000
13,100
13,100
13,100
13,200
13,200
13,200 Inflow
13,300
Serpukhovian
13,300
13,400 t f , h t p e D
13,300
13,400 t f , h t p e D
13,500 13,600
13,500
13,400 t f , h t p e D
13,500
13,600
13,600
13,700
13,700
13,800
13,800
13,800
13,900
13,900
13,900
14,000
14,000
14,000
14,100
14,100
14,100
14,200
14,200
14,200
13,700
Oksky
Perfs 13,917 –14,331 ft
Bottom PL @ 14,072 ft
Fig. 6—Production profile in T-8. August 2001 SPE Reservoir Evaluation & Engineering
253
10
20
Oil Rate (STB/D)
Relative Inflow and Porosity
Gamma API 0
30
40
0
13,100
0.1
0.2
0
0.3
13,100
1 ,0 00
2 ,0 00
3 ,0 00 4 00 0
13,100
Openhole Porosity
Perfs 13,136 –13,399 ft
Bashkirian Inflow
13,200
13,200
PL
t f , h t p e D
t f , h t p e D
13,300
13,300
13,200
t f , h t p e D
13,300
Bottom PL @ 13,356 ft
13,400
13,400
13,400
13,500
13,500
Serpukhovian
13,500
Fig. 7—Production profile in T-104.
drawdown derivative between 0.02 and 0.025 hours of time corresponds to a sharp inflection point in the pressure and rate data caused by rapidly increasing the flow rate in the well by opening the flowline valve. The superposition algorithm experiences difficulty in treating this sudden change. Failure to include rate superposition results in poor agreement between the drawdown and buildup response, as shown in Fig. 10.
Fig. 8—Pressure and instantaneous rate data from the T-112 PL.
Fig. 9—Diagnostic graph for T-112 using instantaneous rate data. 254
Both buildup and drawdown data reveal the onset of radial flow at around 0.6 hours. Semilog analysis of the buildup data, not shown here, indicates a permeability of 1.15 md with a skin of 1.2, based upon perforation thickness of 325 ft. The effective permeability of the best reservoir rock is probably closer to 3.7 md based upon the PL results shown in Fig. 4, where 100 ft of the perforations provide more than 70% of the flow. The small positive skin indicates minor formation damage. Questions arose about the type of reservoir behavior that may be discerned from the derivative response. Both dual-porosity and dual-permeability models exhibit a dip in the derivative signature. The former model is typically associated with naturally fractured reservoirs, whereas the latter represents crossflow in a layeredreservoir system. Naturally fractured reservoirs typically exhibit negative skins and fairly large storage coefficients, owing to the presence of fractures.12 Furthermore, the dip in the derivative is sharper and deeper than that for the dual-permeability model. The shape of the derivative curve for the T-112 buildup data is more consistent with a dual-permeability model. The absence of significant storage effects plus the small positive skin confirm the applicability of the dual-permeability over the dual-porosity model. The log-log results, including rate superposition, for the stationary pressure and rate data from the second production log in T-112 appear in Fig. 11. The buildup and drawdown data are in
Fig. 10—Diagnostic graph for T-112 without variable rate data.
August 2001 SPE Reservoir Evaluation & Engineering
TABLE 1—COMPARISON OF BUILDUP RESULTS IN T-112
Property C s (bbl/psi) k (md) s
Before Acid < 0.001 1.15 1.2
After Acid 0.007 1.60 –4.0
excellent agreement, indicating consistent reservoir properties. Interpretation of test results is complicated by the absence of radial flow. Nonetheless, the development of parallel half-slope lines prior to 0.04 hours provides evidence for infinite-conductivity vertical-fracture behavior. Absence of dual-storage 13 behavior further corroborates our earlier observations on lack of natural fracture development in this reservoir. Fig. 12 presents the results after applying nonlinear regression to the buildup data. The estimated fracture half- length of 44 ft and the storage coefficient of 0.007 bbl/psi are consistent with the generation of an etched-acid fracture. Table 1 summarizes the results of the pressure-transient analyses before and after the acid treatment. Although the acid treatment increased the effective PI by a factor of 3.4, the results are not indicative of dual-porosity behavior. Instead, a few higherpermeability intervals, now with negative skin owing to the acid treatment, dominate well deliverability. Well-test-derived permeability of these intervals appears to be an order of magnitude higher than those of the core, as shown in Fig. 3. However, this contrast is not indicative of a highly fractured system. That is because the permeability contrast between the fracture and matrix typically exceeds 1,000 before distinct dual-porosity behavior develops. 12 The log-log results, including rate superposition, for the stationary pressure and rate data from T-8 appear in Fig. 13. The short duration of the buildup test (10 minutes) is a consequence of having to retrieve the production-logging tools before nightfall. The erratic behavior of the drawdown data near 0.01 hours again reflects the inability of the superposition algorithm to handle a sudden sharp change in flow rate while opening the well to production. Apart from the slight upward bend in the buildup data near 0.1 hours, the pressure and derivative data from both test periods exhibit parallel one-fourth-slope lines. This behavior is representative of flow in a finite-conductivity fracture. We believe this response, previously reported only for hydraulically induced fractures, is a rare occurrence in an acid-stimulated carbonate reser-
voir. We postulate that the stimulated zone sustained damage, leading to the bilinear flow response. The small discrepancy between the buildup and drawdown results is a reflection of a slight change in spinner response during the course of logging the well because of the impact of severe turbulence on the spinner opposite to the major inflow zone. Slightly bent spinner blades, upon inspection of the recovered tool, confirmed our suspicion. We were unable to determine a reliable permeability-thickness value for T-8 because pseudoradial flow did not develop during transient tests. We used a value of 500 md-ft based on local experiences and an effective reservoir thickness of 82 ft based on the PL results in performing the bilinear analysis on the drawdown data, as shown in Fig. 14. The timing of bilinear flow behavior in the measured data is consistent with the dimensionless fracture conductivity, F cD , and fracture half-length, x f , results of the analysis. The T-8 well received a small-volume (260 bbl) acid treatment in 1991. Although details concerning the treatment pressures and acid additives are sketchy, a typical acid treatment of this vintage included a soak for 12 hours before flowing back the spent acid. We conjecture that the conductivity of an acid-induced fracture can be impaired significantly during such a long soak period in the absence of effective sequestering additives. Both pressure-transient and PL data once again do not support the dual-porosity behavior. The log-log results, including rate superposition, for the stationary pressure and rate data from T-104 appear in Fig. 15. The buildup and drawdown pressure data exhibit good agreement beyond 0.01 hours. Shapes of the pressure derivatives are also in good agreement, although the drawdown derivative lags the buildup by 0.01 hours. The data from T-104 is much noisier than the data from T-112 (Figs. 9 and 11) and T-8 (Fig. 13). The explanation for this behavior is that spinner quality degraded when logging wells with 7-in. casing (T-104) compared to wells with 5-in. casing (T-8 and T-112) because the maximum opening of the fullbore spinner used was 4.5 in. Both buildup and drawdown data revealed the onset of radial flow at around 0.2 hours. Semilog analysis of the buildup data indicates a permeability of 23.7 md, a skin of 47.1, and p* of 11,303 psia based on the perforation thickness of 262 ft. The large positive skin reflects formation damage resulting from mud losses while working over the well in 1996. We conjecture that the perforations opposite to the high porosity interval not contributing to production are ineffective owing to mud damage. We used the rectangularhyperbola method 14 to calculate an average reservoir pressure of 11,269 psia. This result is in good agreement with the reservoir pressure measured following an extended shut-in of the well 6 months later.
Fig. 12—Nonlinear regression match to buildup following acid in T-112.
Fig. 13—Diagnostic plot for T-8 indicating finite-conductivity fracture.
Fig. 11—Diagnostic plot for T-112 showing acid-induced fracture behavior.
August 2001 SPE Reservoir Evaluation & Engineering
255
Fig. 14—Bilinear analysis of Tengiz-8 buildup.
Fig. 15—Diagnostic plot for T-104 demonstrating radial flow.
Fig. 16—Extended pressure-transient test in T-21.
Fig. 17—Overlay of first drawdown and final buildup for T-21.
Fig. 18—Simulated rate during drawdown based on buildup results, Well T-21.
Fig. 19—T-21 drawdown and buildup after incorporating rate variations.
Extended Pressure-Transient Test Results Extended drawdown and buildup tests lasting approximately 1 month are necessary to investigate reservoir regions of 500 acres, a typical well spacing at Tengiz. Fig. 16 presents data from an extended pressure-transient test in T-21. The short buildup between the first and second drawdowns corresponds to the shut-in time required to replace the fixed choke in the wellhead. Surface rates were recorded with an automated testing system. Rate data are typically collected over 6-hour test cycles during the flow periods. Problems with the testing system limited the number of valid data points during the second drawdown. The first rate measurement in T-21 was completed approximately 7 hours after initiating the first drawdown. The spinner data collected with PL’s at Tengiz indicate that the majority of the rate variation, upon initiating a drawdown, occurs during the first few hours. Consequently, it is not surprising that the rate data for T-21 does not exhibit a clear decrease from the initial measurement. An overlay of the buildup and drawdown data on a log-log graph reveals a significant discrepancy in permeability, by 36%, as shown in Fig. 17. Constant rates of 3,077 and 4,786 STB/D during the first and second drawdowns formed the basis for these calcula-
tions. As discussed earlier, this behavior can be misconstrued as stress-sensitive reservoir properties. Reconciliation 6 of drawdown and buildup results proved useful at Tengiz. Fig. 18 shows the rate profile generated by applying the reconciliation technique to the first drawdown in T-21. The calculated rate declines approximately 10% from a maximum of 3,429 to 3,064 STB/D over the course of the 5-day drawdown; however, the majority of the decrease occurs within the first 5 hours. Furthermore, the average rate of about 3,100 STB/D is in good agreement with the field measurements, as Fig. 16 testifies. Fig. 19 shows an overlay of the first drawdown and extended buildup data, including the calculated rate history, on a log-log graph. The good agreement between the results of the two tests is apparent from this plot. This example demonstrates the importance of including small, gradually varying rates in the analysis of drawdown data. This approach has been used successfully on several Tengiz wells. Furthermore, the magnitude and timing of rate variations is corroborated by the PL spinner data, as shown in Fig. 8. Both drawdowns at T-21 lasted approximately 120 hours. Because pseudosteady-state flow did not develop during the drawdowns, comparing transient PI’s at the same elapsed time appears
256
August 2001 SPE Reservoir Evaluation & Engineering
appropriate. PI for both drawdowns based on measured average rates of 3,077 and 4,786 STB/D, respectively, is 1.5 STB/D/psia. The average reservoir pressure used in these calculations is 10,308 psia and was determined with the rectangular-hyperbola method. 14 The theoretical pseudosteady-state PI, based on the reservoir properties determined from the buildup data and assuming a typical well spacing of 500 acres, is also 1.5 STB/D/psia. Clearly, the consistency between the drawdown and buildup PI’s does not support the notion of stress-sensitive permeability.
Wellbore Hydraulic Results We developed wellbore hydraulic models to extend PI results to wells that do not have measured bottomhole pressure data. Before estimating PI’s, we used PL data to verify the accuracy of the calculations. Calculated FBHP’s are sensitive to the fluid properties and the flow correlations used in the calculations. The former impacts the hydrostatic and the latter the frictional pressure gradient. We found that the available black-oil correlations 7–9 yielded hydrostatic pressure gradients—that is, fluid densities that are in poor agreement with measured values of around 0.27 psi/ft. A simplified EOS model, 10 on the other hand, provided density results that are typically only 2% higher than the measured values. The agreement between calculated and measured oil viscosity was also significantly better with the EOS model (within 10%) compared to the black-oil correlations, which were off by more than 20%. Wellhead pressures remained above the bubblepoint pressure for all of the Tengiz PL’s. Consequently, it is not possible to choose a preferred flow correlation from these data. A comparison of calculated and measured FBHP’s for the seven Tengiz wells with PL data based on wellhead pressure and surface flow rate measurements appears in Table 2. We found that a pipe roughness factor of 0.0039 in. provided the best agreement between calculated and measured FBHP’s. The calculated FBHP’s should be approximately 75 psi higher than the measured values
because the EOS model overpredicts densities by approximately 2%. Furthermore, the accuracy of the pressure measurements is about ±10 psi for the surface pressure transducer used. Calculated FBHP’s are within 1% of the measured values after accounting for these two effects. We performed flow-after-flow (FAF) deliverability tests upon completing a workover or placing a new well on production. Furthermore, we measured reservoir pressures before flowing the wells. We combined calculated FBHP’s with these reservoir pressure data to determine PI’s. Table 3 summarizes FBHP and PI results from FAF tests completed on four wells. Consistency among PI’s for each well in the table over a wide range of flow rates indicates the absence of either non-Darcy behavior or ratesensitive reservoir properties in the vicinity of the wellbore. Except for T-110, flowing wellhead pressure dropped below the bubblepoint pressure at the highest flow rate for all wells in the table. The Hagedorn and Brown 15 flow correlation appears to provide good estimates of frictional pressure losses in the two-phase flow regime based on the consistency between PI results obtained above and below the bubblepoint pressure.
Trends in Reservoir Quality from PI Results Fig. 20 shows an overlay of relative PI’s for the 37 Tengiz wells currently on production. The area of the circles is proportional to PI. The rim exhibits the best reservoir quality compared to the platform interior or the flanks based on the PI results. A limitation of using PI’s to infer reservoir quality is that they contain a skin component in addition to the permeability-thickness and drainagearea components. The PI comparison was used because permeability-thickness information is unavailable for many of the producing wells. The trends in reservoir quality become even clearer after taking into account the large positive skins (approximately +40) determined with pressure-transient tests in rim wells T-20, 102,
TABLE 2—EVALUATION OF WELLBORE HYDRAULIC RESULTS USING PRODUCTION-LOG DATA
Well
q (STB/D)
p wh measured (psia)
p wf measured* (psia)
p wf calculated* (psia)
T-5K T-8
3,176 3,819
5,182 6,528
9,599 10,424
9,741 10,575
T-102 T-104 T-106 T-112 T-113
2,461 3,636 3,375 2,660 2,954
6,231 5,678 5,172 4,061 5,114
10,262 9,445 9,437 7,752 8,923
10,310 9,572 9,621 7,830 9,006
*Referenced to midperforation depth.
TABLE 3—SUMMARY OF FBHP AND PI RESULTS FROM FLOW-AFTER-FLOW TESTS
Well T-5K
T-72
T-107
T-110
q (STB/D)
p wh (psia)
p wf (psia)
p (psia)
PI (STB/D-psi)
2,191 4,081 5,232 6,598 2,469 4,319 5,859 6,979 1,588 2,715 3,097 3,605 2,557 4,566 5,328 5,876
6,090 4,988 4,220 3,176 5,916 4,684 3,451 2,711 5,174 3,795 3,006 2,320 6,163 5,017 4,655 4,350
10,038 9,071 8,493 7,763 9,809 8,744 7,789 7,350 8,989 7,555 6,730 6,033 10,096 9,183 8,980 8,803
11,052
2.2 2.1 2.0 2.0 1.7 1.7 1.7 1.8 0.7 0.8 0.7 0.7 2.3 2.2 2.4 2.4
August 2001 SPE Reservoir Evaluation & Engineering
11,252
11,135
11,226
257
Wellbore hydraulic calculations provide a basis for extending PI results to wells lacking bottomhole pressure data. Production logs provided the data necessary to validate the accuracy of this approach. Development of a simplified EOS model greatly enhanced the accuracy of the calculated results compared to those obtained with the available black-oil fluid correlations. Calculation of PI’s over a wide range of drawdowns revealed the absence of any significant stress-sensitivity owing to the closure of natural fractures in the vicinity of producing wells. Although fractures appear to play a role in enhancing matrix permeability in Tengiz, they exhibit low compressibility based on the absence of stresssensitive permeability and PI results. The purpose of this paper was to demonstrate how existing multidisciplinary techniques could be applied to understand the performance of a giant, complex field. In particular, we sought synergy among and within various data sets before reaching conclusions.
Fig. 20—PI distribution superimposed on top reservoir contours.
and 104. Platform interior wells T-111, 112, and 113, on the other hand, benefit from negative skins (about -4) resulting from acid fracture treatments. Another trend that emerges from the PI and pressure-transient results is degradation in reservoir quality in the southern platform compared to the northern platform interior. Core analyses and reservoir dip suggest lower energy conditions and a deeper depositional environment in the southern part of the platform. The small number of wells producing from the flank of the reservoir limits the conclusions that can be made regarding reservoir quality there. However, a review of data obtained during short (4 to 12 hours) flow tests completed before suspending delineation wells indicates greater heterogeneity in the flank compared to the platform interior or rim.
Discussion Cores recovered from Tengiz frequently exhibit thin fractures. However, the majority of these fractures are partially or completely filled with either calcite cement or solid bitumen. Although the reservoir at Tengiz does not exhibit a clear relationship between porosity and permeability, as shown in Fig. 3, matrix permeability is typically above 0.1 md for porosity above 10%. Examination of Tengiz production profiles in Figs. 4 to 7 indicates that the major inflow zones typically correspond to intervals characterized by porosity of 10% or higher. Review of permeability results from pressure-transient tests in the same wells, as shown in Figs. 9, 12, 14, and 19, reveals that effective permeability is in the range of a few to a few tens of millidarcies. Although permeabilities in this range have been measured in Tengiz core, as shown in Fig. 3, they comprise a small portion of the total measurements. The discrepancy between typical core and well-test permeability results reflects in part the impact of fractures on enhancing matrix permeability. The small magnitude of permeability enhancement provided by fractures explains the absence of clear dual-porosity flow behavior. Wellbore-storage coefficients determined from pressure-transient testing are typically in good agreement with those calculated from wellbore volume and fluid compressibility data. Absence of storativity associated with natural fractures corroborates the limited fracture connectivity indicated by pressure-transient tests. 258
Conclusions A synergistic approach of combining engineering (pressuretransient, PL, and wellbore hydraulics), geologic, and petrophysical data provided important insights into the reservoir-flow behavior in the Tengiz field. Findings of this study are being used to manage this field. Specific conclusions include the following. 1. The reservoir rock exhibits neither dual-porosity flow behavior nor stress sensitivity. Transient pressure tests provided crucial evidences to both elements. 2. Vertical heterogeneity confines flow contribution to a small fraction of the total interval and promotes reservoir layer crossflow in some cases. 3. Comparison of the major inflow zones in T-8 to core revealed a good correlation between probable karst surfaces and highpermeability zones in the reservoir. 4. Incorporation of transient rate data is important in reconciling drawdown and buildup test results. 5. Computing well PI’s from wellhead data with fluid-flow and fluid PVT models aided the reservoir-wide coverage and understanding of heterogeneity in an areal sense. Nomenclature C s = F cD = h= k = p* = pwf = pwh = q= s= x f = D p / q = D p¢ / q =
wellbore-storage coefficient, bbl/psi dimensionless fracture conductivity formation thickness, ft horizontal permeability, md Horner’s extrapolated pressure, psia flowing bottomhole pressure, psia flowing wellhead pressure, psia flow rate, STB/D steady-state skin factor, dimensionless fracture half-length, ft normalized pressure change, psi/STB/D normalized log superposition-time-derivative of D p / q, psi/STB/D
Acknowledgments We are indebted to the Tengizchevroil Partnership Council and management for permission to publish this work. The views expressed within the paper do not necessarily represent those of the minority partners of Tengizchevroil. References 1. Ayestaran, L.C. et al.: “Well Test Design and Final Interpretation Improved by Integrated Well Testing and Geological Efforts,” paper SPE 17945 presented at the 1989 SPE Middle East Oil Technical Conference and Exhibition, Bahrain, 11–14 March. 2. Massonnat, G.J., Norris, R.J., and Chalmette, J.-C.: “Well Test Interpretation in Geologically Complex Channelized Reservoirs,” paper SPE 26464 presented at the 1993 SPE Annual Technical Conference and Exhibition, Houston, 3–6 October. August 2001 SPE Reservoir Evaluation & Engineering
3. Bourgeois, M.J., Daviau, F.H., and Boutaud de la Combe, J-L.: “Pressure Behavior in Finite Channel-Levee Complexes,” SPEFE (September 1996) 177. 4. Kabir, C.S. et al.: “Characterizing the Greater Burgan Field: Integration of Well-Test, Geologic, and Other Data,” paper SPE 37749 presented at the 1997 SPE Middle East Oil Conference and Exhibition, Bahrain, 17–20 March. 5. Nelson, R.A.: Geologic Analysis of Naturally Fractured Reservoirs, Gulf Publishing Co., Houston (1985) 87. 6. Kabir, C.S.: “Seeking Synergy Between Drawdown and Buildup Analyses,” SPEFE (June 1997) 125. 7. Standing, M.B.: “A Pressure-Volume-Temperature Correlation for Mixtures of California Oils and Gases,” Drill. & Prod. Prac., API (1947) 275. 8. Lasater,J.A.: “BubblePoint Correlation,” Trans., AIME(1958) 213, 379. 9. Vazquez, M., and Beggs, H.D.: “Correlations for Fluid Physical Property Prediction,” JPT (June 1980) 968. 10. Furnival, S.R. and Baillie, J.M.: “Successful Prediction of Condensate Wellbore Behaviour Using an EOS Generated From Black Oil Data,” paper SPE 26683 presented at the 1993 SPE Offshore European Conference, Aberdeen, 7–10 September. 11. Missman, R.A. and Jameson, J.: “An Evolving Description of a Fractured Carbonate Reservoir: The Lisburne Field, Prudhoe Bay, Alaska,” paper SPE 22161 presented at the 1991 SPE International Arctic Technology Conference, Anchorage, 29–31 May. 12. Gringarten, A.C.: “Interpretation of Tests in Fissured and Multilayered Reservoirs With Double-Porosity Behavior: Theory and Practice,” JPT (April 1984) 549. 13. Azari, M. et al.: “A Comprehensive Study of Reservoirs Exhibiting Dual-Storage Effects During Well Testing,” paper SPE 24708 presented at the 1992 SPE Annual Technical Conference and Exhibition, Washington, DC, 4–7 October. 14. Kabir, C.S. and Hasan, A.R.: “Estimating Average Reservoir Pressure Using the Hyperbola Approach: New Algorithm and Field Examples,” paper SPE 36255 presented at the 1996 Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, UAE, 13–16 October. 15. Hagedorn, A.R. and Brown, K.E.: “Experimental Study of Pressure Gradients Occurring During Continuous Two-Phase Flow in Small Diameter Vertical Conduits,” JPT (April 1965) 475; Trans., AIME, 234.
August 2001 SPE Reservoir Evaluation & Engineering
SI Metric Conversion Factors °API bbl ft in. mile psi
´
141.5/(131.5 +°API) ´ 1.589 873 ´ 3.048* ´ 2.54* ´ 1.609 344* ´ 6.894 757
*Conversion factor is exact.
g/cm3 3 =m =m = cm = km = kPa =
E E E E E
-
01 01 + 00 + 00 + 00 -
SPEREE
Kevin Chambers is the geostatistics team leader at Chevron Petroleum Technology Co. in San Ramon, California. He previously worked as both a reservoir and production engineer with Chevron Overseas Petroleum Inc. Chambers holds a BS degree in chemical and petroleum refining engineering from the Colorado School of Mines and an MS degree in chemical engineering from the U. of California, Berkeley. Bill Hallager manages earth science services for Chevron Canada Resources in Calgary. He began his career with Chevron in ore deposits research and worked briefly in structural geology research at Chevron Oil Field Research Co., then moved to oil and gas development projects in Papua New Guinea, Kazakhstan, and Nigeria for Chevron Overseas Petroleum. Hallager holds a BA degree from Dartmouth College and MA and PhD degrees in economic geology from the U. of California, Berkeley. Shah Kabir is a senior adviser of petroleum engineering currently working in the Mid-Africa Business Unit of Chevron Overseas Petroleum Inc. in San Ramon, California. Kabir has more than 24 years of experience in the oil industry, with the last 11 years at Chevron, including Chevron Oil Field Research Co. in La Habra. He holds an MS degree from the U. of Calgary. Kabir has published extensively and has served various SPE committees, including the Editorial Review committees for SPEPF, SPEREE, and SPEJ . Raymond Garber is a staff reservoir geologist currently working in the Mid-Africa Business Unit of Chevron Overseas Petroleum Inc. in Houston. He has worked in exploration, development, and research and specializes in carbonate stratigraphy. Previously, he worked for Gulf Research and Chevron Oil Field Research Co. in La Habra. He has published articles on carbonate evaporite geology and reservoir modeling. Garber holds a BA degree from the U. of Rochester and MS and PhD degrees from Rensselaer Polytechnic Inst., all in geology.
259