Copyright 2006, Society of Petroleum Engineers This paper was prepared for presentation at the 2006 SPE Annual Technical Conference and Exhibition held in San Antonio, Texas, U.S.A., 24–27 September 2006. This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract

Estimation of effective permeability at the reservoir scale has been a long standing challenge in carbonate fields. The carbonate depositional and diagenetic history can be quite complex, and this can lead to a permeability field which is quite difficult to characterize. Permeability in vuggy or fractured intervals can be dramatically different from the matrix permeability measured in core plugs. However realistic estimates of oil recovery, and optimized reservoir management requires good estimates of the reservoir permeability.

In the Tengiz field, a giant carbonate reservoir in western Kazakhstan, a method has recently been developed to calculate apparent permeability (APERM) based on flow rate from production (PLT) logs. Incorporation of this flow calibrated apparent permeability into the static geologic earth model offers an elegant solution to the long-standing problem of how to best incorporate dynamic PLT data into a reservoir model. A reservoir model recently built using APERM resulted in a step change improvement over previous methods where only static log based permeability transforms were used to populate the earth model.

Conventional log based permeability transforms are designed to characterize matrix permeability but not the excess permeability due to fractures & vuggy porosity common in carbonate reservoirs. The APERM method is used for both accurately characterizing total permeability (matrix + excess), and for identifying inaccurate permeability predictions in older wells with poor log quality or limited log data. The log based permeability predictions are more accurate in recent wells with modern logs, but hot streak identification and quantitative permeability estimation from static well logs is still problematic.

The apparent permeability is calculated by solving Darcy’s law on an interval basis, using as input our knowledge of flowing and static pressures, plus well, reservoir, and fluid properties. The method makes several simplifying

assumptions, but the resulting errors are second order in nature, and the method offers improvements over using conventional static log based transform permeability. Application of the method is enhanced by the derivation of coarse scale zonal layer pressures with multi-rate PLTs. Accurate zonal layer pressures improve the accuracy of the permeability derivation. The apparent permeability from PLT is then used as a benchmark to adjust the transform permeability derived from static well logs using a variable multiplier. This technique has the advantage of preserving the original fine scale heterogeneities of the wireline logs, while calibrating their magnitudes. It also has the advantage of identifying higher permeability intervals from rapid changes in well inflow profile, that may not have been characterized by conventional (core plug and wireline) estimates.

Recently a full field reservoir model for Tengiz Field has been constructed using APERM data and petrophysical rock types. Preliminary history match results from the model show that use of APERM has increased confidence in the modeled permeability field before history match. Fewer changes are required to calibrate the full field reservoir model to well pressure data (SGS). Permeability-Height (KH) estimates from the model are a much better match to well test KH, than models derived from core transforms. The models also have significantly more heterogeneity, as shown by Dykstra-Parsons calculations.

A similar improvement in history match was recently observed in a high resolution model constructed using APERM to monitor gas movements in the Tengiz platforms area. History match pulse tests have shown that this model has a much better prediction of the inter-well connectivity than models without APERM. Models with the APERM data have provided higher confidence estimates of the future movement of gas injection in the Tengiz platform.

Future plans are to investigate a correspondence between APERM and Rock Types as well as statistical transforms with open-hole logs. The log based transforms can then be used in wells or intervals without PLT data, improving accuracy of permeability population into the reservoir model.

Introduction

The fundamental building blocks of a reservoir model are porosity, permeability and fluid saturation. Porosity and fluid saturation can be computed from wireline logs with reasonable confidence, but the same often cannot be said for permeability. Permeability can be measured directly from core material or from well tests via pressure transient testing (PTT), pulse-testing between wells, and wireline formation tests. This

SPE 102894

Permeability from Production Logs—Method and Application M.J. Sullivan, D.L. Belanger, M.T. Skalinski, S.D. Jenkins, and P. Dunn, Chevron

2 SPE 102894

paper shows the application of production logging tests (PLT) to provide improved well scale permeability for use in reservoir models.

Scale differences in these sources of permeability data complicate the permeability estimation in a reservoir model3,4. Core permeability measurements are used to develop wireline permeability estimates, and pressure transient tests are used to calibrate these estimates. Determining permeability from well logs is complicated by differences in scale, measurement environment, and measurement physics. Well logs have a vertical resolution of typically 0.5 to 1 meter and provide in-situ measurements, compared to 5 cm diameter core plugs that are measured on the lab bench

In most reservoirs, the amount of core available for direct permeability measurement is limited, and permeability estimates are usually made by proxy using wireline log responses. The science of permeability prediction has evolved to a high degree of sophistication, with each additional level resulting in an incremental gain in predictive accuracy1

Even with sophisticated predictive algorithms, accurate assessment of permeability in carbonate reservoirs is often still problematic because of significant variations in rock types and the nature of the pore geometry. When permeability is matrix dominated and a modern suite of high quality wireline logs is available, the predictive accuracy can be good. However carbonate reservoirs can have many different pore structures2 which have widely varying por/perm relationships making accurate prediction from wireline logs difficult or impossible. Vuggy porosity and fractures, if they are connected, can result in extremely high layer flow rates or very low rates if not connected, for the same value of porosity.

In many carbonate reservoirs the primary source of flow capacity are fractures. Recognizing the existence of fractures, and accounting for them in a reservoir model, may be difficult or impossible in wells with older log suites without image logs or modern vintage sonic logs. Even with high quality image logs, recognizing whether fractures are open or closed can be difficult. Even if open fractures are identified in the borehole, there may be a great deal of uncertainty whether these fractures are laterally continuous, and to what degree they are fed by the rock matrix. These factors increase uncertainty in productivity prediction and greatly complicate reservoir characterization efforts.

Several authors have shown how the accuracy of the reservoir characterization can be improved significantly by incorporating dynamic data, such as pressure transient tests, periodic pressure measurements, and production profiles, into the reservoir characterization process,5,6,7. However, the industry has lacked a robust work process for flow calibrating log derived permeability and distributing it between wells in a multi-layer reservoir model. This obstacle can be overcome with the use of apparent permeability derived from PLT logs.

This paper describes a new method for deriving apparent permeability from production log flow profiles. The method provides a direct integration of static core and wireline log data with dynamic pressure transient and PLT data. It calibrates the wireline scale permeability to “ground truth” pressure transient results, and retains the character of the log data. This improves the accuracy of the permeability used in the model and inherently integrates dynamic data into the

static model. The method overcomes the difficulties created by the differences in scale of various data types mentioned above. The reservoir characterization process is improved because calibration to dynamic flow data is built in at the start.

Two reservoir models have been successfully constructed for the Tengiz reservoir using the APERM technique. These models include a full-field model used for infill drilling and reservoir management, and a high resolution model of the Tengiz platform used to optimize a planned sour gas injection project. Reservoir models constructed using APERM have provided a significant improvement in ability to predict well tests (SGS, pulse tests) with minimal history matching effort. Improved confidence in the history match of these models has provided additional confidence in reservoir management activities such as gas injection and infill drilling of the Tengiz Field. The Tengiz Reservoir

Tengiz is a large carbonate reservoir in western Kazakhstan. Tengiz field (Fig. 1) is located on the south side of the 500,000 square kilometer (km2) Pri-Caspian basin on the northeastern edge of the present-day Caspian Sea. Other giant oil and gas fields located in similar settings include Karachaganak and Orenberg. The reservoir has a very large areal extent. Tengiz is over 110 km2 in area at its top and 400 km2 at its base. The oil column is quite thick, as top of the reservoir is located at approximately −3,850 m subsea and the approximate oil water contact is at −5,450 m subsea.

Fig. 1 – Location of Tengiz Field and the Caspian Sea

The Tengiz reservoir is an isolated carbonate buildup

(“platform”) with a mesa-like geometry (flat-topped and steep-sided). The platform has a slight regional tilt today of less than one degree to the south. The sides of the platform have much greater dip (approximately 25 degrees) on the flanks. The present-day geometry of the Tengiz structure is that of a topographically high-rim area surrounding a depressed central platform. It is flanked by a dipping platform-to-basin

TENGI

10 Km

Tengiz

Korolev

SPE 102894 3

transitional facies along the flank, as shown in Fig. 2. The Tengiz reservoir formed during the Late Devonian and Lower to Middle Carboniferous time by the deposition of calcareous skeletal fragments and lime mud. The cyclicity of these sediments was in response to the interplay of sea-level rise and basin subsidence. The platform appears to have drowned in late Bashkirian or early Moscovian time.

Fig. 2 – The Tengiz Platform: Rim and Flank regions.

The central platform rock type is dominated by the matrix porosity and permeability. Previous methods used core porosity-permeability transforms for converting well log porosity to permeability. This method works well only with matrix-dominated permeability rocks and does not work well with fractured and vug-dominated rocks.

Conversly, the productivity of the rim and flank regions is dominated by fracture permeability that cannot be easily characterized from conventional well logs. Permeabililty estimates from the rim and flank region need to include first order influence of an effective fracture network. Specialized software (FRACSIM) is used to simulate the fracture density, and estimate the effective permeability in the rim and flank when fractures are accounted for. The permeability height (kh) is much higher in the flank than the matrix kh when the effect of fractures is considered (Fig. 3).

Fig. 3 –Map Showing Increased Permeability in Tengiz Flank.

Tengiz has a great deal of remaining development potential, and consequently there is a large focus on reservoir characterization. Pressure differences have been noted in various reservoir intervals, and are related to differences in reservoir porosity, permeability, and voidage.

Surveillance activities in Tengiz include production logging surveys, inter-well pulse testing, and pressure buildup tests.

Most of the wells in the Tengiz reservoir are vertical or near vertical and all are flowing single phase oil with flowing bottom-hole pressure above the bubble point. From this perspective, the conditions for PLT profiling are ideal. Differential depletion of various reservoir intervals has occurred in Tengiz, and this is accounted for by the APERM method described in the next section. APERM Method

A typical production log in Tengiz includes profiles acquired under both stabilized shut-in and then stabilized flowing conditions. After the flowing survey is completed, the well is shut-in for a pressure buildup test until the infinite acting radial flow period is reached8. The profiles are then analyzed with detailed inflow points defining the regions over which permeability will be calculated. The derivative with respect to depth of cumulative flow profile is calculated. This derivative yields the volumetric inflow on a meter by meter basis (Note that the unit of depth in Tengiz is meters, and so will form the convention for this discussion - the technique obviously will work equally as well with depth in feet). This derivative is equivalent to the traditional interval contribution commonly seen on PLT analyses divided by the length of the zonal contribution.

Given the flow contribution per meter from derivative of cumulative flow, for each meter (h=1), the permeability is calculated using the familiar Darcy’s law radial flow equation:

]')([)(

** SRRLn

PPBUQC

w

e

wfe

ooipltk +

−= (1)

PPllaattffoorrmm

RRiimm

FFllaannkk

Cell Permeability

4 SPE 102894

where C=141.2 if depth is in feet, or 43.07 if depth is in meters (Tengiz convention), Qi is flow rate in stbopd, Uo is viscosity in Cp, pressure in psi, and Re/Rw in consistent units.

PLT permeability (kplt) represents average permeability over that specific inflow zone. The kplt data is accumulated over the entire interval to compute khplt. This value is then normalized to the pressure buildup khbu by multiplying each kplt value by khbu/khplt.

kplt_norm = kplt (khbu/khplt) (2) This normalized PLT permeability (kplt_norm) is used as a

baseline reference to which the wireline transform permeability from open hole logs is adjusted using a multiplicative factor (which we call “perm_boost”). This process is repeated, inflow by inflow, from the bottom of the zone upward in a stepwise fashion (Fig. 4). This adjusted transform permeability curve is called APERM for apparent permeability. After each “boost” APERM is accumulated from the bottom up and compared to the accumulated kplt_norm. By matching the increases in cumulative KH between kplt_norm and APERM at each inflow step the final result is normalized to well test permeability. The technique of “boosting” the wireline transform permeability (instead of simply using kplt_norm) offers the significant advantage of preserving the vertical heterogeneity visible in the open-hole wireline logs, while calibrating their magnitudes.

Step 1 Step 2

Fig. 4 – Stepwise calibration to PLT_perm.

A typical Tengiz well with 400m interval and 20 inflow points can be completed in about 30 minutes using this iterative technique. We envision this process could be automated with a purpose built computer code.

Discussion of Details on the Method.

Handling Differential Depletion. The Tengiz reservoir comprises 3 main flow units that behave as different pressure compartments (Fig. 5). The upper-most layer (zone 1) is generally the most prolific so is generally more depleted than the lower two layers (zones 2 & 3). This differential depletion

results in cross flow into zone 1 when the well is shut-in at surface (Fig. 6, 7).

Fig. 5 – Pressure gradient plot showing differential depletion.

The differential depletion causes each layer to flow at a different percentage drawdown for a given Pwf. These different individual layer pressures must be accounted for in the calculation of permeability using the APERM method (Pe in the denominator of equation 1). We routinely measure individual layer pressures using the “Selective Inflow Performance” (SIP) technique9. This technique involves acquisition of production profiles at multiple rates and flowing bottom hole pressures.

Fig. 6-– Typical shut-in PLT profile.

SPE 102894 5

Fig. 7 – Flowing profile from the same well.

Pressure is plotted against rate for each reservoir flow unit

(Fig. 8). For single phase liquid flow, when the bottom hole pressures are above the bubble point pressure, the points will all plot on a straight line. That pressure for which the layer flow rate would be zero defines the layer pressure. By using this technique, no zone ever has to be static, i.e. a rate of zero. Tengiz is an ideal environment for this technique, because all flow is single phase oil above the bubble point pressure. When we initially applied this technique in Tengiz we acquired four PLT profiles; shut-in and 3 different stable flowing rates. Pressure versus rate data plotted on a straight line for each zone as predicted by theory for single phase flow. Modeling and field testing confirmed that the intermediate flow rates have a minor influence on the computed layer pressure when a stabilized shut-in profile with crossflow has been recorded. A practical justification for only conducting two flow rates (one stable shut-in and one stable flowing) is that two longer duration and more stable profiles provide better results than 4 flow rates of shorter duration where each rate suffers from transient flow effects and is not yet stable.

Fig. 8 – SIP Plot of Zonal Rates & Pressures.

Tengiz wells are typically shut-in for several days before logging so a stabilized shut-in profile is available in most

wells. The leftmost point (negative flow rate) on the zone 1 curve in Fig. 8 is the stabilized rate of crossflow into that zone.

Simplified Approach. If there is no differential depletion in the reservoir, and the assumption of uniform skin distribution can be made, a simplified approach may be taken. With no differential depletion, all flowing bottom hole pressures will result in the same drawdown being applied to each layer. The accumulated permeability may simply be normalized to well test KH, and distributed vertically according to interval contribution. This simplified approach has proven successful in other areas10.

Significance of Errors

The accuracy of the APERM calculation can be assessed by examining the certainty factor of the input variables in Eq. 1. As discussed below, uncertainty for many of the input variables result in an impact on permeability that is insignificant for our purposes. In the calculation of Apparent Permeability, we make several simplifying assumptions, but the resulting errors are second order in nature, and do not have a significant impact on the result.

To put the errors in perspective, we first must determine how large of an error is significant. For our purposes, a value within a factor of two of the “true value” of permeability is adequate for reservoir characterization purposes and is a significant improvement over a permeability field based on only on log to core transforms. The most important feature of the APERM method is that it achieves the correct vertical distribution of permeability, helping to correctly characterize the flow units in the reservoir.

To illustrate, Fig. 9 is a crossplot of porosity vs. permeability from 6 cored wells in Unit 1. Rock types are depicted by color scale. There is approximately 1.5 log cycles of scatter in the raw data (50x). A difference of a factor of two is essentially “within the noise”.

Fig. 9 – Core porosity -permeability crossplot.

6 SPE 102894

To further illustrate the impact of a being off by a factor of two, Fig. 10 is a plot of the transform permeability both multiplied and divided by two (track 2). The rightmost track shows the core permeability data (blue dots) plotted on the transform perm (black curve) that is based on a multivariate fit between open hole log data and core permeability for specific rock types. The scatter in core perm data around the transform permeability is approximately equal to the factor two error illustrated in track 2 above. The red blocky curve in this track is the “Apparent Perm” from PLT. The APERM curve is in excellent agreement with core data.

Fig. 10 – Permeability shifted to illustrate impact of twofold error, natural scatter in raw data, and good fit of calculated APERM data.

Discussion of Potential Uncertainties and their Impact on Apparent Permeability

1) Fluid Properties: The product of Uo & Bo changes

little over the typical range of pressures we operate in, and is usually known to within 5%. This does not have a significant impact on calculated permeability (i.e. less than a line width on Fig. 10 above).

2) Pressures: An error in drawdown will cause a proportional error in calculated permeability. We have found in most cases we have been able to assess stabilized drawdown within 10% uncertainty.

A potential source of inaccuracy in the SIP technique is lack of stability (multiphase flow would also be a source of inaccuracy in other reservoirs). We are normally able to assess the layer pressure using the SIP technique to within

about 2% of the true value. As illustrated in Fig. 11 the layer pressure is usually well constrained. The leftmost points on the SIP plot are those operating rates and pressures while the well is shut in at surface, and crossflowing downhole. The size of the reservoir compartments are very large in comparison to the crossflow rate, so the rate of crossflow will stabilize and remain relatively constant over very long periods of time (months). The “fill up effect” in the more depleted Bashkirian due to crossflow is minor. Because we shut the wells in for several days prior to commencing a PLT survey, by the time the shut in PLT passes are made, the rates and pressures are quite stable.

Fig. 11: Uncertainty in layer pressures is usually low using SIP technique because the shut-in passes are at stable pressures. The uncertainty in drawdown pressure is mostly due to uncertainty in stable flowing bottom hole pressure.

The crossflow we observe during shut-in conditions helps

the SIP analysis by providing a definitive operating point on the left side of the SIP plot.

SIP Pressure Confidence. The stabilization time required for the flowing passes is sometimes an issue. We now open the well to flow with PLT tools above all perforations, and monitor rates and pressure for stability before commencing flowing passes. A real-time log-log pressure derivative plot is used in the assessment of stability. When analyzing historical data where this was not done, we may still assess pressure stability by comparing pass to pass pressure repeatability.

Correction for Transient Flow Instability. On some wells with lower permeability, the pressure does not stabilize with a practical amount of time. In practice the flowing passes begin within 8-12 hours of opening the well to flow even if the flow has not become perfectly stable by that time. If a well has not stabilized by that point in time, an additional 12 hours is unlikely to make a significant difference. After the flowing passes have been completed, the well is shut-in for pressure buildup. A log-log derivative pressure diagnostic plot is used to ensure the Infinite Acting Radial Flow period has been reached before terminating the buildup. The rate and pressure data may then be analyzed with pressure transient analysis software. In doing so, we have constructed an interpretation model that can be used to extrapolate to what the flowing bottom hole pressure would have stabilized at had we continued to flow (Fig. 12). It is this estimate of flowing bottom hole pressure that we use in the calculation of apparent permeability. Although this technique is not exact, we believe

SPE 102894 7

this is a reasonable method of compensating for instability, and the residual error is reduced to an acceptable level.

Fig. 12 – Well not stable during time allotted for PLT. Modeled stable pressure is used for calculation of APERM.

It is unlikely to have more than a 20% error in drawdown

pressure, after using the above correction method if necessary. Because permeability is proportional to drawdown, this will only produce an error in APERM of 20%. This only affects the total KH, not the vertical distribution of it, which is the more important issue.

It is also important to note that the error in drawdown, expressed as a percentage, remains more constant over a range of drawdowns. Wells with better permeability stabilize quicker, have a smaller error bar on pressure, but also have lower drawdowns. Lower permeability wells have larger error bars on pressure, but usually have very high drawdowns. The error bar as a percentage of drawdown may be similar between the two cases.

4) Drainage Radius & Wellbore Radius. Because these factors are inside a logarithmic term, the calculated Apparent Permeability is very insensitive to errors in radius. For typical Tengiz assumptions, a 20% difference in either radius makes only a 1.5% difference in calculated permeability.

5) Flow Rate. When possible, we try to flow the well through a test separator during a PLT but this is not always possible due to operational constraints. The PLTs that are conducted with stable flow measured with accurate surface meters typically match the surface rates with a difference of less than 10%. Because rate is proportional to calculated perm, this again has a negligible effect on APERM.

6) Skin Factor: The assumptions made about skin factor typically receive the greatest amount of attention, for this can have the impact of twofold difference in permeability, or more (Fig 13).

Productivity Ratio vs Skin

0

0.5

1

1.5

2

2.5

-5 0 5 10 15 20

Skin

Prod

uctiv

ity R

atio

Fig. 13 – Impact of skin factor, expressed as a proportion of the productivity the well would have with a skin of zero.

A single skin factor equal to that value calculated from the

pressure buildup test is used as a starting point for the calculation of APERM. Under ideal circumstances, this can be a valid assumption. The ideal case is one where the PLT was run after the well was stimulated with several reciprocations of a coil tubing acid wash resulting in an even stimulation and a calculated skin factor of -4 or less (highly stimulated) based on the post stimulation pressure build up analysis. Approximately half of our APERM interpretations fall into this category. The rest are complicated by a variety of factors. Any time a well has had a workover on it, where the well was killed with drilling mud, and the well may now have, for example, a +2 skin factor, there is reason to question the assumption of the same skin factor for all intervals. A well with +2 skin may have a 0 skin over a higher pressure zone with shallow invasion and a +5 skin over lower pressure, more deeply invaded zone.

Our certainty factor on computed APERM is influenced by how the APERM value compares to that predicted by permeability transform. If a highly positive boost must be applied to the transform permeability to match the PLT perm there is more confidence that the high perm streak is real and not the product of over-stimulation, especially if we assumed a negative skin in the calculation of APERM. With matrix acidizing, the skin factor cannot be much more negative than about -5. If a negative skin is assumed in the APERM calculation and a boost factor of 10 is required to match the PLT_perm, the only possible explanation can be that the permeability is indeed as high as it appears to be.

Low Permeability vs. Damage. If the productivity of a zone is much less than predicted, then we must assess whether the permeability really is low, or whether this layer is damaged or unstimulated. To make this assessment, we must take into account factors such as; drilling history (whether fractures were encountered, amount of mud losses into those fractures indicating high permeability), workover history, prior flow tests (especially pre-workover tests), stimulation history (type and volumes), as well as vintage and quality of open hole logs.

For example, when a well that has been drilled and logged with modern logs of high quality on the reservoir platform (where the porosity is primarily intercrystaline and the transform permeability is usually fairly accurate), shows much

8 SPE 102894

less inflow than was predicted by the transform perm, and the measured skin factor was -2, we would be inclined to conclude that the reason for the low productivity was a high skin factor over this interval, and leave the transform permeability un-adjusted.

Other wells may have poor open hole log quality, limited logging suites, or logs made with by tools that have been poorly characterized by western standards. In these cases, we often have much less confidence in the resulting predicted permeability, and so would be less hesitant to adjust the permeability downward. Fig. 14 is an example of this. Track 1 shows porosity from open hole logs (red line), porosity from PNC ratio normalized to PHIT (black line), with difference shaded in grey. Acid effect (explained later in this text), is shaded in magenta. Track 2 is a wellbore diagram, showing perforations and two strings of casing. Track 3 is transform permeability from open hole logs (green), apparent permeability from PLT (black bars), and the resulting adjusted APERM (red). Track 4 shows the multiplicative permeability adjustment factor used to calibrate the transform perm to PLT. Positive boosts are shaded in magenta; negative boosts are shaded in grey. Also posted in this track is the difference between the open hole and PNC porosity (linear scale). Track 5 is the PLT results.

Fig. 14 – Example of reduction of transform permeability after calibration to PLT perm.

In this case we felt confident in adjusting the permeability

downward because we believed the original open-hole porosity (from Soviet era uncalibrated neutron log with poor excentralization) was too optimistic over the lower section. The porosity values were higher than offsetting wells, whereas the PNC porosity was consistent with offsets. The difference between the two tracks well with the required downward adjustment factor. A reciprocating coil tubing acid stimulation was conducted over the entire interval, and the PNC acid effect shows the entire interval was stimulated, so we do not

suspect damage is responsible for the lower than predicted flow.

Flank Wells – fractured reservoir. In the rim and flank areas of the Tengiz reservoir, fractures are frequently the primary source of productivity. Matrix porosity is very low, in the range of 0-5% versus 10-15% for the reef platform.

KH is often under predicted by 2 or 3 orders of magnitude by the permeability transform. The permeability transform is designed to predict matrix permeability only. The “excess permeability” due to fractures is difficult or impossible to predict. At best, we can recognize the presence of fractures with FMI image logs which are surprisingly good in the oil-based mud. However, both healed and open fractures are shown as high resistivity and it is difficult to separate them. To classify open fractures, we are using other fracture indicators such as: Pef (barite invasion in the fracture), Stoneley wave reflection and mud losses.

Matrix Permeability. Permeability transforms were based on the statistical relationships between log porosity and core permeability. Klinkenberg core permeability was corrected for overburden, assuming an average net overburden pressure of 3000 psi.

A total of 8056 plug data from 16 wells (5 platform and 11 Flank) constituted an initial database. The plugs with indication of fractures (mostly drilling induced) were rejected to obtain an accurate representation of matrix permeability.

Two transform types were created: porosity-based (linear or non-linear) and multivariate.

Porosity based transforms were used to populate permeability in the geologic model. Multivariate transforms (with porosity generally as a main predictor) were used for the definition of APERM.

A total of 22 transforms were defined for combinations of stratigraphic horizons and reservoir gross facies (inner platform, outer platform, upper slope, middle slope and toe of slope).

Fig. 15 shows an example of matrix permeability transform for zone 2 in the platform. Application of multivariate transform improves permeability prediction error from 0.9 log cycles (porosity based) to 0. 5 log cycles (multivariate) for matrix permeability on wells with modern logs.

SPE 102894 9

Matrix Permeability Transform for Serp-Unit 1

0.5 10.5Petrophysical Rock Types

0.0

00

0.0

00

0.0

30

0.0

30

0.0

60

0.0

60

0.0

90

0.0

90

0.1

20

0.1

20

0.1

50

0.1

50

0.1

80

0.1

80

0.2

10

0.2

10

0.2

40

0.2

40

0.2

70

0.2

70

0.3

00

0.3

00

0.001 0.001

0.01 0.01

0.1 0.1

1 1

10 10

100 100

Perm

eab

ilit

y (

mD

)

Porosity (frac)

909876

2

31

0 0

)3252.8692738.2297596.5034802.3(104sm_K φ∗+φ∗−φ∗+−=

( )SGR0692.0)Cxolog(239.0CMFF412.17503.210Perm ∗−∗−∗+−=

Fig. 15 – Example of permeability transform for zone 2. Porosity transform is defined by 3rd order polynomial and multivariate transform by NMR Free Fluid Volume from NMR log, spectral GR, and logarithm of conductivity.

Excess Permeability

Permeability derived from PLT is in agreement with matrix permeability in most of the reservoir zones. However, several inflow zones yield a boost factor >1 which results in APERM higher than log derived matrix permeability. The excess permeability named EPERM is related to secondary porosity which corresponds most likely to the vuggy/moldic porosity in the platform and fractures in the flank (Fig. 16).

Porosity

Perm

eabi

lity

Matrix transform from Core

APERM values from PLT

Eperm events Matrix Perm Prediction Area

Porosity

Perm

eabi

lity

Porosity

Perm

eabi

lity

Matrix transform from Core

APERM values from PLT

Eperm events Matrix Perm Prediction Area

Fig. 16 – Excess permeability is defined as a PLT permeability which falls above matrix permeability prediction range.

Confidence Factor

A confidence factor is assigned to each interpretation to communicate the level of certainty in the final APERM value. In the example illustrated in Fig. 4 the permeability of the upper zone has a positive boost. A key point with this method is that when a much higher permeability than predicted by transform is encountered there is higher confidence that this is real and significant.

Open-Hole Log Quality, and confidence in the Petrophysics.

Before the TengizChevroil joint-venture was formed in 1994, a total of 67 wells were drilled and logged using Soviet technology. Half of the wells were logged with full logging suites including density, neutron and sonic logs. However, calibration of the Soviet logs was not up to the western standard which triggered a need for the normalization effort. Since the joint venture was formed, a total of 41 wells were drilled and logged using a modern suite of logs. These logs were used to normalize Soviet log data, first in 1999 then in 2005.

Uncertainty analysis indicates that maximum error in porosity determination for modern logs in 1.5 PU while the wells with Soviet era logs can have porosity uncertainty up to 4 PU. A major source of uncertainty was inability to predict bitumen content from Soviet logs. In modern wells bitumen is estimated using Nuclear Magnetic Resonance logging and advanced multimineral modeling. Fig. 17 shows comparisons of log versus core porosity for 2 key platform wells.

0.0

0.2

0.4

0.6

0.8

1.0

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.0

00

0.0

30

0.0

60

0.0

90

0.1

20

0.1

50

0.1

80

0.2

10

0.2

40

0.2

70

0.3

00

Wells:Statistics

Phi_av= 8.63 %

Core Porosity

0.0

0.2

0.4

0.6

0.8

1.0

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.00

0

0.03

0

0.06

0

0.09

0

0.12

0

0.15

0

0.18

0

0.21

0

0.24

0

0.27

0

0.30

0

W llSt ti ti

Phi_av= 8.97 %

Log Porosity0.0

0.2

0.4

0.6

0.8

1.0

0.00

0.01

0.02

0.03

0.04

0.05

0.06

-0.1

00

-0.0

80

-0.0

60

-0.0

40

-0.0

20

0.0

00

0.0

20

0.0

40

0.0

60

0.0

80

0.1

00

W llSt ti ti

Delta Phi_av = - 0.0017

Phi_cor- phi_log

Fig. 17 – Comparison of Core porosity with Log-derived porosity for 2 key wells. Multimin porosity yields unbiased values with low uncertainty.

Using Fig. 9, a porosity uncertainty of 1.5 PU translates to a matrix permeability uncertainty of 0.3 logarithmic cycles. The 4 PU porosity uncertainty of Soviet logs would translate to almost one order of magnitude uncertainty in matrix permeability.

EPERM Prediction from OH Logs

Matrix permeability was predicted using core permeability vs porosity transforms (Fig 15, 16). An attempt was made to predict EPERM (secondary permeability) from logs in order to predict this dynamic property in wells without PLT. Since, secondary perm is usually driven by properties other than porosity (amount of connected vugs, fractures, rock types etc), predictive variables were selected to consider all logs which might carry this information.

Use of Stepwise Multiple Regression allows to evaluate the predictive significance of input variables. The dependent variable used was EPERM defined as :

EPERM = APERM - Kmatrix (3)

Where: K_matrix – matrix permeability predicted from logs using Core PHI-k transforms

10 SPE 102894

Input independent logs included:

– Porosity corresponding to different T2 cutoffs: o FFI(1-8)for:8;16;32;64;128;256;512;1024

(>cutoff) and BVI – below the same cutoffs. o K_sdr,K_tim – permeabilities from CMR

logs. – Ct,Cxo and Delta_con (short and deep

conductivities and their difference ) – PHIE, PHIE-vug (porosity from Multimin and vug

porosity derived from sonic using Lucia method) – Raw logs such as: RHOB,T2LM, CGR,SGR, U,

Pef, CMFF,CMRP For variables exhibiting lognormal distribution a

logarithmic transform was applied. Multiple regression was run separately for Bashkirian and Serpukhovian formation and resulted in predictive regression equations. The most significant message from this statistical exercise was:

1. EPERM can be predicted from logs in statistically reliable way ( R= 0.8-0.85 standard prediction error approx 1 log cycle)

2. Top predictive variables include consistently variables such as FFI7,FFI8, (porosity corresponding to high T2 cutoffs), Phi_vug – which indirectly indicate that the EPERM in platform is caused mostly by connected vugs.

Those are encouraging results showing possible link between “static” log data and dynamic PLT–derived permeability. Regression analysis is based on data from three wells, and we plan to obtain more PLT data in wells with modern logs to validate the prediction and create more robust transforms.

Predicted EPERM was than used to obtain predicted APERM using:

APERMpred = EPERMpred + kmatrix (4)

Acid Effect with PNC Sigma

A key factor in assessing whether low productivity is due to low permeability or high skin factor is the effectiveness of the acid stimulation. Most of the wells in Tengiz receive acid stimulations, either reciprocating coil tubing acid washes or bullhead diverted treatments. We have run several Pulsed Neutron Capture (PNC) logs in the central area of the Tengiz field as baseline data in preparation for a future hydrocarbon miscible flood. Although not the primary reason for running the PNC logs, we have found the sigma curve useful in assessing placement of acid stimulations.

The PNC log records the thermal neutron capture cross section of the formation by measuring the thermal neutrons’ rate of decay. Chlorine is one of the strongest neutron absorbers of the common earth elements so the response of the PNC log is determined primarily by the chlorine present in the formation water (usually as NaCl). The sigma reading also has a pronounced effect due to chlorine in the acid reaction products in the form of calcium chloride11. We have found a significant increase in sigma due to the HCL acid stimulations that are conducted on most wells in Tengiz (some wells are prolific enough that stimulation is not needed). Nearly all wells in Tengiz produce water free. The irreducible connate

water saturation is approximately 10% of pore volume. The salinity of the connate water is low, only about 60 kppm. When an interval is acidized, the acid reaction products are essentially saturated with calcium chloride. The elevated chlorine concentration imbibes into the irreducible water increasing its chlorine concentration in the near wellbore area. The spent acid water flows back during clean up but the elevated chlorine levels in the irreducible water remain, and may be detected with a PNC sigma log. We believe this effect is permanent, or nearly so, because we have observed elevated sigma readings due to acid several years after the stimulation. With only clean oil production, there is no mechanism to dilute the chlorine concentration with time in the near wellbore area.

Most of our wells only have sigma data acquired after the acid stimulation has been completed so a comparison of measured sigma before and after stimulation was not possible. We have been able to overcome this by comparing the measured sigma to a synthetic baseline sigma generated with a predictive transform. A few wells did have pre-stimulation sigma logs, and a multivariate transform was developed to predict sigma from commonly available open-hole wireline data (GR, NPHI, DT, and Rt). The predictive algorithm works well, with a standard deviation of error of 0.5 cu. Fig. 18 illustrates the good match between predicted and measured sigma on one well logged with a PNC tool prior to stimulation.

Fig. 18 – Comparison of Measured vs. Predicted Sigma on Unstimulated Well.

SPE 102894 11

The post stimulation sigma is compared with synthetic baseline sigma to determine acid effect. The difference between the two curves is computed and used as an indicator of acid stimulation. Fig. 19 shows the sigma increase (magenta) rescaled to overlay on porosity (green).

Fig. 19 – Example of PNC acid effect showing poor stimulation effectiveness, confirming prior interpretation of high skin with analysis of time lapse PLT’s

The post-stimulation sigma logs are a great help in validating our assumptions about distribution of acid penetration. Sometimes, these sigma comparisons show us the reservoir has been unevenly stimulated, causing us to use a different skin factor by zone in our computation of APERM. Fig. 19 shows the utility of the sigma acid effect in assessing stimulation effectiveness. Track 4 on the right shows the results of a PLT conducted in 2001. A pressure transient test at the time yielded a skin factor of +1 (blue curve). After a scale problem was identified, a bullhead acid treatment with no diverter was conducted the following year to restore production. In 2006, the PLT illustrated in track 3 was conducted. A noticeable reduction in flow contribution was observed below 4287m, and a variable skin of up to +20 (blue curve) had to be used to reconcile the 2006 results with the flow observed in the 2001 PLT. A PNC log was then run to assess the effectiveness of the prior acid treatment. The measured sigma was compared to a synthetic baseline sigma, and a difference curve was calculated. This difference curve (sigma increase) is over posted on porosity in track 1. The PNC acid effect in track 1 clearly shows that little or no acid went below 4287m, as one might expect from an undiverted bullhead acid treatment. This lack of visible acid effect confirmed our prior interpretation of damage in this zone. Although the reduced flow clearly indicates the region below 4287m has become damaged, the mechanism for this damage is not certain. We speculate it may be due to wellbore scale deposition.

This example, and other similar ones, have increased our confidence in this diagnostic technique, which helps in computing apparent permeability and the frequent decision whether lower than anticipated flow rate is due to lack of stimulation or permeability that is lower than predicted.

The very distinct acid signature still visible 4 years after the treatment supports our belief that the acid effect is permanent in a reservoir with low salinity connate water at irreducible saturation.

Fig. 20 is an example of a well with uneven acid penetration that was caused by crossflow from the higher pressure lower zone into the lower pressure upper zone during the coil tubing acid stimulation. The crossflow out of the lower zone during the stimulation did not allow matrix stimulation over that interval, and all the acid was swept up into the upper interval.

In this example, for the calculation of APERM, rather than assuming the skin of -2.8 from the pressure transient test was evenly distributed, we assumed a skin of -4 over the upper interval, and a skin of 0 over the lower interval. As shown in Fig. 13, this will increase the apparent permeability of the lower zone by a factor of about 2.

Fig. 20 – Example of uneven acid penetration of

reservoir due to crossflow during stimulation.

Constraining APERM with MLT Analysis Another technique used for determining individual layer

permeability and skin is with Multi-layer transient pressure testing12. These procedures are operationally more complex, but further constrain the APERM solution. We have either

12 SPE 102894

two or three layers which each form separate pressure compartments. Multi-layer transient tests are integrated into a multi-rate Production Logging procedure13.

1. Shut the well in two days before PLT operations begin to allow stabilized crossflow to develop. Record up and down passes to measure the stabilized shut-in flow profile with crossflow.

2 Place the tool above the top layer and record the drawdown with the low flow rate until IARF is observed on the log-log plot (typically 4-6 hours). At the end of this drawdown, record up and down flowing passes with the PLT to measure the flow profile and determine the zonal contributions.

3. Station the PLT above the middle layer. Increase the production rate to the medium flow rate. Record the drawdown until IARF is observed on the log-log plot (4-6 hours). At the end of the drawdown, record up and down passes with the PLT to measure the flow profile and determine zonal contributions.

4. Place the PLT above the bottom layer. Increase the production rate to a high flow rate. Record the drawdown until IARF is observed on the log-log plot (4-6 hours). Record up and down passes with the PLT at the end of the transient.

5. Place the PLT between at the point of maximum crossflow that was observed in step 1 (this is usually the same location drawdown number 2. Close the well for shut in and record the pressure buildup. At the end of the transient record one shut in up and down pass with the PLT across the reservoir. Remove PLT from well and rig down. It is assumed that the transients produced by changing the surface flow rates from low to medium will induce a sufficient transient to be measured by the PLT pressure gauges and flowmeter and that they are within their respective resolution and tolerance. If the transients are not of sufficient magnitude, then the drawdowns should be recorded with two different flow rates: low and high. In this scenario the drawdowns should be recorded with the first drawdown at the maximum rate and the second one with the low rate and the third one with the maximum rate. Due to operating time restrictions the drawdown transient between zone 1 and 2 is omitted in Tengiz. This coincides with the point of maximum crossflow so the build-up transient is used to constrain layer 2 properties.

Tool Position Drawdown 1

Zone

1

Tool Position Drawdown 2and final build up

Zone

2

Tool Position Drawdown 3

Zone

3

Fig. 21 – Depiction of PLT tool position during Multi-layer transient testing.

Analysis of the transient data proceeds from the bottom

zone to the top zone. Analysis of the transient data from the drawdown 3 is a single layer analysis of the lowest layer. The reservoir properties of the lowest layer is thereby constrained, and used in a multilayer analysis of the second drawdown, which is a combination of layers 2 and 3. The properties of layer 2 can then be determined. The properties of layers 2 & 3 are input as constraints into the analysis of the first drawdown, which is the combined response of layers 1, 2 & 3, to solve for the properties of the top layer. Layer pressures that are required for the multilayer analysis are taken from the SIP analysis that was previously described. When the second drawdown is not recorded we use the pressure and crossflow recorded during the final build-up to estimate layer 2 properties. In this fashion, the permeability and skin of each layer may be determined, and used to constrain the APERM analysis.

Fig. 22 is an example of a well where the results of a multi-layer transient test match well with the APERM interpretation. Track 1 shows the porosity trace (green shading) with the acid effect from calculated sigma increase over-posted in magenta. The porosity on this well was also computed from a Soviet era uncalibrated neutron log, increasing uncertainty on the resulting predicted permeability. Track 2 is APERM (red), transform perm (green), and apparent perm from PLT in black. Track 3 is the multiplicative factor used to adjust the transform perm to the PLT perm. Note the adjustment factors are all highly positive. This is likely because the open hole porosity is reading too low, resulting in a low predicted perm. Track 4 is the PLT results. Also posted in track 4 is a log of the calculated skin factor from the multi-layer transient test, which was a value of -3 in the upper zone and +1 in the lower zone. This is consistent with the acid effect observed in the upper zone but not in the lower.

SPE 102894 13

Fig. 22 – Example of good match between APERM, Acid Effect, and Multi-Layer Pressure Transient Interpretations.

Construction of the Static Model using APERM

The successful use of reservoir simulation is critical to the proper management of Tengiz reservoir. The efforts associated with delineating, developing, and producing Tengiz reservoir are substantial. Optimizing these efforts with a robust reservoir simulation can have large positive effects on improved oil recovery, and optimizing expenditures. For the reservoir simulation model to be successful, it must be populated with accurate geologic model data, rock property descriptions, fluid properties, and well information. Additionally, the detail captured in the simulation model must match the scope of the studies for which the modeling effort will be applied.

Several generations of reservoir models have been constructed for Tengiz Field. These models have accurately predicted the primary depletion of the field, and have successfully been used to guide infill drilling. Tengiz reservoir models have conventionally been built using porosity and permeability transform, which is based on core plugs. However, to model secondary recovery processes (such as sour gas injection) even more accuracy is required.

The acquisition of modern wireline logs and PLT logs in the Tengiz platform has provided a rich source of data to properly calibrate the reservoir model. APERM traces are input into the reservoir simulation model to improve the reliability of the permeability field, and improve the accuracy of secondary recovery estimates.

Geostatistical methodogies have been used to populate APERM data into Tengiz reservoir models. In the area of investigation at Tengiz, most of the wells have reasonable quality suite of wireline logs. Construction of the model follows several discrete steps in a sequential workflow. Wireline porosity was first distributed in a 3D earth model using Sequential Gaussian Simulation (SGS). Porosity is a fair predictor of permeability in Tengiz, and provides a guide

to the APERM geostatistical population process. Both SGS with collocated-cokriging and SGS with cloud transform methods have been used to populate APERM. An example of the “cloud” of APERM data for a shallow reservoir interval at Tengiz is shown in Figure 22.

Fig. 23. – Cross plot of porosity and APERM for a shallow interval in the Tengiz platform.

These well established geostatistical techniques honor the

spatial variability of the APERM data, while following the trend of the background porosity as "soft" data. These processes provide a way to condition the model at the well location with the input APERM traces while honoring the spatial variability of the APERM data in each stratigrahic interval of the reservoir. They provide a set of equiprobable images of the permeability between the measured APERM traces. The resulting permeability images are calibrated with all the known geologic and engineering data, and provide a reasonable (albeit non-unique) realization for predicting reservoir performance.

Restoration of Natural Heterogeneity:

As illustrated in figure 9, carbonate reservoirs are naturally heterogeneous, with a permeability variiance often exceeding 1 order of magnitude for a given porosity value. The very nature of regression based transforms reduces this heterogeneity to a single value for a given facies/porosity pair, leading to models that are inherently too homogeneous. This is compounded by the tendency to either under-recover core in fractured high permeability intervals, or undersample in non-permeable intervals. The excess homogeneity results in predicted flood responses that are too “piston-like”, and actual breakthrough times earlier than predicted. As illustrated in figure 23 above, the APERM method restores that heterogeneity. This should lead to more realistic predictions, and improved reservoir management planning.

Results of Incorporating APERM into the Static Model

Introduction of the APERM data into the reservoir model has substantially improved the reservoir model. Compared to model built with a traditional matrix permeability transform, the new APERM based model has measurably higher permeability. Fig. 24 compares two reservoir models, one constructed with a conventional transform, and another with APERM traces. The hotter colors in the right figure indicate the presence of higher K values after the introduction of APERM.

14 SPE 102894

Fig. 24 – Conventional Reservoir model for Tengiz (left) and model constructed using APERM (right)

The introduction of APERM data has also helped calibrate

the permeability levels in different stratigrahic intervals of the reservoir. Some specific stratigrahic intervals are observed to have more statistical difference from the conventional matrix permeability transform. A cross section through the central platform area of the Tengiz Field (Fig. 25) shows the tendency of zone 1 to have a positive permeability correction (blue color) relative to the conventional matrix transform.

The introduction of APERM data has also substantially increased the heterogeneity of the reservoir model relative to a conventional permeability transform in the platform. Dykstra-Parsons coefficient calculations in some areas of the Tengiz platform have mean values as low as 0.50 for a conventional permeability transform. This level of heterogeneity implies a fairly homogeneous permeability distribution. However, after the introduction of APERM data, the Dykstra-Parson coefficients are much larger, with some calculations having a mean value as high as 0.95, suggesting a very significant level of heterogeneity.

Fig. 25 – Difference between APERM and conventional permeability transform for one area of Tengiz field.

Match to well test KH

One benefit from the integration of APERM data into the reservoir models is quite clear when comparing cumulative permeability (KH) from the reservoir model to well test (PTT) derived cumulative permeability. After the introduction of APERM, the Kh of the reservoir model in the platform area now has a much better match to the Kh observed in well tests. Accumulated Kh from the reservoir model has a poor match to well-tests, when the permeability is not properly calibrated.

This is clearly shown by the scatter in Fig. 26 A. Note that before the use of APERM, there is considerably more scatter, and there is tendency for many wells to have inadequate KH in the model compared to well test KH. However, after the permeability field was constructed using the APERM traces there is a much better match between the well test Kh and model derived Kh, as shown in Figure 25 B.

Fig 26 – Crossplot of KH from welltest in the Tengiz platform versus KH from a full field reservoir model. A shows the comparison from a model without APERM calibration, and B shows the same wells after APERM calibration Validation of APERM Model Improvement with Pulse Tests

The Tengiz injection wells have permanent downhole pressure gauges and have been shut-in while preparing for gas injection (Fig. 27). Surrounding producers are pulsed and the pressure response is measured at the injector. The results have shown both higher levels of connectivity than previously modeled and different levels of connectivity between various wells pairs (colored arrows).

5442220

5442220

*

*

*

*

* Tag Fill above Serp

MOA

SGI

4431’, 1351m

5442220

5442220

*

*

*

*

* Tag Fill above Serp

MOA

SGI

4431’, 1351m

5442220

5442220

*

*

*

*

* Tag Fill above Serp

MOA

SGI

4431’, 1351m

< 50 50 to 100 > 100

Diffusivity (kh/phih)

Fig. 27 – Sour Gas Injection pilot area with results of pulse testing

permeability (md) .001 .01 .1 1 10 100

Conventional Matrix Perm APERM

A

B

Kh well

Kh Model With

APERM

Kh Model without

APERM

SPE 102894 15

showing relative connectivity between wells. Injectors are in the center of the patterns (red arrows).

The pressure response to the pulse test flow sequences were simulated using the flow-calibrated APERM based reservoir model and compared to the simulated results from the prior model, where the permeability was based solely on transforms from wireline logs.

A significant improvement in the predictive accuracy was observed with the new model after the introduction of APERM. This important history match calibration confirms that the simulation model now has approximately the correct amount of connected pore space and permeability. Fig. 28 shows a comparison of actual pulse test results (black lines) and simulated pressure response with both the APERM based earth model (red lines) and the previous transform perm based model (blue). The bottom panel shows the flow sequence. The top panel is the pressure response, and the middle panel is the derivative of pressure. This analysis shows that the model after the introduction of APERM has a much better match to the pulse tests than the conventional reservoir model. In general, the model with the introduction of APERM provided a significantly improved match of the inter-well connectivity as indicated by the pulse tests.

Fig. 28 – Simulated vs. Actual Pulse Test Responses with APERM Based Model and Prior Transform Perm Based Model. Ongoing Tengiz Modeling Efforts

A full field reservoir modeling effort is underway at Tengiz, and the APERM data are providing an important data source to calibrate the permeability. For this effort, and for all future modeling efforts, it is anticipated that the APERM method will continue to be used and the methodology further refined.

An aggressive program of acid stimulation and PLT is underway which will contribute to the current database of APERM wells. This database of high quality permeability measurements will further improve the confidence of the reservoir model, and be used to help optimize reservoir management activities at Tengiz.

Acknowledgements

We wish to thank Tengizchevroil (TCO) and it’s Joint-Venture Partners for permission to publish this paper. We also

wish to thank Peter Hillock of Exxon Mobil for his thoughtful collaboration and insight. Dennis Fischer with TCO provided critical geologic support and assisted constructed the earth model. Dennis has been an enthusiastic supporter of the APERM process. Rob Kimmel provided a synthesis of Tengiz geological setting. Steve Johnson with Chevron conducted the reservoir simulation and pulse test matching for gas injection monitoring and provided helpful feedback. Travis Billiter with TCO demonstrated the ability of the APERM data to reach a rapid history match for the Tengiz full field model. We also wish to thank our colleagues Ed Neubauer and John Clarke for modeling and analysis of the multilayer transient data and many fruitful discussions that led to improved estimates of layer properties.

References

1. Skalinski, M., and Sullivan, M.: “Application of Improved Method for Permeability Estimation in Complex Lithology Reservoirs”. Paper C, presented at the 2001 SPWLA Annual Logging Symposium, 17-20 June 2001, Houston, Texas.

2. Lucia, F. J., 1995b, Rock-fabric/Petrophysical Classification of carbonate pore space for reservoir characterization; American Association of Petroleum Geologists Bulletin, V. 79, P. 1275-1300

3. Al-Henshiri, M., Arisaka, K., Al-Hassani, H., Al-Yasi, T., and Fujita, N.: SPE 93475 “Integration of Dynamic & Geostatic Data Improves Reservoir Characterization”.

4. Ahmed, U., Crary, S., and Coates, G.; SPE 19604 “Permeability Estimation: The Various Sources and Their Interrelationships”.

5. Bahar, A., Ates, H., Al-Deeb, M. Salem, S., Badaam, H., Linthorst, S., and Kelkar, M.: SPE 84876 “An Innovative Approach To Integrate Fracture, Well-Test, and Production Data Into Reservoir Models”.

6. Sibley, M., Bent, J., and Davis, D.: “Reservoir Modeling and Simulation of a Middle Eastern Carbonate Reservoir”. SPE Reservoir Engineering, May 1997.

7. Mezghani, M., Van Lingen, P., Cosentino, L., and Sengul, M.: SPE 65122 “Conditioning Geostatistical Models to Flow Meter Logs”

8. Chambers, K.T., Hallager, W.S., Kabir, C.S., and Garber, R.A.; SPE 72598 “Characterization of a Carbonate Reservoir with Pressure Transient Tests and Production Logs: Tengiz Field, Kazakhstan”.

9. Stewart, G., Wittmam, M.J., and Lefevre, D.: “Well Performance Analysis: A Synergetic Approach to Dynamic Reservoir Description,” paper SPE 10209 presented at the 1981 SPE Annual Technical Conference and Exhibition, San Antonio, Oct. 5-7.

10. Harrison, C. Personal correspondence: Technique for production log analysis using simplified assumptions.

11. Al-Saif, A.S., Cochrane, J.E., Edmundson, H.N., and Youngblood, W.E.; SPE 5443: “Analysis of Pulsed Neutron Decay Time Logs in Acidized Carbonate Formations”

12. Ehlig-Economides, C.A., and Joseph, J.A.: "A New Test for Determination of Individual Layer Properties in a Multilayered Reservoir," SPEFE, Sept., 1987

13. R. R. Jackson, and R. Banerjee, “Advances in Multilayer Reservoir Testing and Analysis using Numerical Well Testing and Reservoir Simulation”. SPE Annual Technical Conference and Exhibition, Dallas, Texas, 1–4 October 2000.

16 SPE 102894

14. Kuchuk, F.J., Karakas, M., and Ayestaran, L.: “Well Testing and Analysis Techniques for Layered Reservoirs”, SPEFE, Aug. 1986

ABOUT THE AUTHORS Michael J. Sullivan is a Petroleum Engineer working on

the Tengiz reservoir surveillance in Kazakhstan. E-mail: msul@chevron.com. Prior to moving to Tengiz, Mike worked as a Petrophysicist at ChevronTexaco Exploration and Production Technology Company in Houston, Texas. He had previous assignments as Open Hole Formation Evaluation Specialist and Cased Hole Formation Evaluation Specialist for Chevron in Cabinda Angola, and Cased Hole Formation Evaluation Specialist for Chevron Canada Resources in Calgary. He has also held a variety of positions in Petroleum and Production Engineering starting in 1979. His interests include integration of reservoir surveillance into reservoir characterization studies. Mike has a BS in Petroleum Engineering from Montana College of Mineral Science and Technology. He is a member of the SPWLA, SPE, Petroleum Society of CIM, and CWLS

David L. Belanger is a Petroleum Engineer working on Tengiz reservoir surveillance in Kazakhstan. E-mail: dbelanger@chevron.com. His main interests include reservoir surveillance and the integration of dynamic data in reservoir characterization. David was previously a Senior Staff Formation Evaluation Specialist at ChevronTexaco Exploration and Production Technology Company in Houston, Texas, where he provided formation evaluation support to ChevronTexaco operations worldwide. David is a member of the SPWLA and SPE and he is a former Review Chairman for the SPE Formation Evaluation Journal. He holds a BS degree in Mechanical Engineering from the University of California at Berkeley

Mark Skalinski is currently a Formation Evaluation Supervisor with Tengizchevroil in Atyrau, Kazakhstan. Email: mtsk@tengizchevroil.com. He has M.Sc. (1971) and Ph.D. (1979) degree in Geophysics, from Mining University in Cracow. His previous assignments include: Chevron Canada Resources in Calgary, CABGOC in Cabinda, Angola, Husky Oil in Calgary, ONAREP in Morocco (as a Petrophysicist) and University of Mining & Metallurgy in Cracow Poland (as assistant professor). Mark’s interest includes: petrophysical multimineral modeling, application of statistical methods for facies, permeability and rock type prediction and integrated petrophysical field studies. Mark is a member of SPE and SPWLA

Steven D. Jenkins is an Earth Scientist working on the exploration and reservoir geology of Tengiz Field, Atyrau, and Republic of Kazakhstan. Prior to moving to Atyrau, Steve worked as a geologist with Chevron Nigeria Limited in Lagos Nigeria, with Caltex Pacific Indonesia in Sumatra, Indonesia and with Chevron USA in Denver Colorado. Steve’s interest includes: reservoir characterization using seismic methods, the application of statistical methods for facies, permeability and rock type prediction and integrated petrophysical field studies. He has held a variety of positions in Exploration and Development Geology, and in Reservoir Geophysics starting in 1981. Steve has a BA in Geology from the University of

Tennessee and MA in Geophysics from Indiana University. He is a member of the SEG and the AAPG.

Peter C. Dunn is currently the Sour Gas Injection Team Leader for Tengizchevroil located in Atyrau, Kazakhstan. E-mail: pcdu@tengizchevroil.com. Prior to moving to Kazakhstan, Peter worked for ChevronTexaco Exploration and Production Technology Company in San Ramon assessing and evaluating acquisition opportunities for Chevron’s worldwide operations and in Western Canada as a petroleum and reservoir engineer for Chevron Canada Resouces. His interests include field operations support, reservoir surveillance and the art of reservoir simulation. He holds a BASc degree in Chemical Engineering from the University of Waterloo.