Reservoir-Fluid Sampling and Characterization.pdf

5/23/2018 Reservoir-Fluid Sampling and Characterization.pdf

1

80 JPT AUGUST 2007

AbstractReservoir-fluid properties play a key role in the design andoptimization of injection/production strategies and surfacefacilities for efficient reservoir management. Inaccurate fluidcharacterization often leads to high uncertainties in in-place-volume estimates and recovery predictions, and hence affectsasset value. Reservoir-fluid pressure/volume/temperature(PVT) characterization begins with acquisition of adequatevolumes of representative fluid samples followed by PVT-data

measurement with strict quality-assurance/quality-control(QA/QC) protocols and phase-behavior modeling throughbest-practice methods. In this paper, key steps involved inaccurate fluid characterization are discussed for a wide spec-trum of fluid types ranging from extraheavy oils to highly vola-tile near-critical fluids and lean gas condensates undergoing awide range of production processes from simple depletion tocomplex tertiary recovery. Selection of appropriate samplingmethods and tools, design of tool strings, and customizingprocedures are demonstrated through these examples. Routineand special laboratory-fluid-analysis strategies for various fluidtypes and for different production strategies are highlighted.Fluid-modeling techniques including optimum-component

selection, accurate C7+characterization, robust Gibbs energyminimization, and gravity/chemical equilibrium calculationsare demonstrated through appropriate field examples.

IntroductionReservoir-fluid PVT properties are critical for efficient reser-voir management throughout the life of the reservoir, fromdiscovery to abandonment (See complementary paper byHonarpour et al. 2006.) Reliable PVT properties of in-situfluids are essential for the determination of in-place volumes

and recovery-factor calculations and are key input to reser-voir simulations for technical evaluation of reservoir-devel-opment/-depletion plans. Fluid characterization and distribu-tion within the reservoir help in defining reservoir continuityand communication among various zones. Interpretation ofwell-test data and the design of surface facilities and process-ing plants require accurate fluid information and its variationwith time. In addition to initial reservoir-fluid samples, peri-odic sampling is necessary for reservoir surveillance.

Reservoir-fluid characterization consists of several keysteps: (1) acquisition of representative samples, (2) iden-tification of reliable service laboratories to perform PVTmeasurements, (3) implementation of QA/QC proceduresto ensure data quality, and (4) development of mathematicalmodels to capture fluid-property changes accurately as func-tions of pressure, temperature, and composition. The fluidtype and production processes dictate the type and the vol-ume of required fluid data. This paper outlines recommendedsampling techniques, PVT-data-acquisition strategies, andmodeling methods and presents field examples covering awide range of fluid types from heavy oils to lean gas conden-sates and production processes such as depletion, pressure

maintenance, and miscible recovery.

SamplingMethods, Tools, and Recommended Practice.The main objective of a successful sampling campaign isto obtain representative fluid samples for determining PVTproperties. In addition to PVT samples, adequate volumesshould be collected for plant and process analysis, geochemi-cal analysis for fluid-source identification and reservoir con-tinuity, and crude assay for refinery processes. The criticalsteps in any successful sampling program are avoiding two-phase flow in the reservoir, minimizing fluid contaminationintroduced by drilling and completion fluids, and preservingsample integrity. A sampling program should focus on thekey issues of selecting an appropriate sampling method andassociated tools, customizing the tool string, and developingsound sampling, sample-transfer, and QC procedures. Inaddition, specific sampling issues should be addressed relatedto fluid type and condition, saturated vs. undersaturated, andfluids with nonhydrocarbon components or fluids containingsolid-forming components such as waxes and asphaltenes.

DISTINGUISHED AUTHOR SERIES

Copyright 2007 Society of Petroleum EngineersThis paper, SPE 103501, is based on paper 101517 presented at the 2006 Abu DhabiInternational Petroleum Exhibition & Conference, Abu Dhabi, 58 November. Distinguished

Author Seriesarticles are general, descriptive representations that summarize the state of theart in an area of technology by describing recent developments for readers who are not special-ists in the topics discussed. Written by individuals recognized as experts in the area, these articlesprovide key references to more definitive work and present specific details only to illustrate thetechnology. Purpose:to inform the general readership of recent advances in various areas ofpetroleum engineering.

N.R. Nagarajan, SPE, is an engineering associate atExxonMobils Upstream Research Company with 22 years

of experience in the oil industry. He holds a PhD degreein physics and has served on program committees and theForum Series for SPE. Mehdi Matt Honarpour, SPE, isa senior engineering adviser with ExxonMobil UpstreamResearch Company in Houston. He holds BS, MS, and PhDdegrees in petroleum engineering from the U. of Missouri.Honarpour served as the Chairperson of SPEREEReviewCommittee and as the Chairperson of the SPE SpecialSeries Committee. Krishnaswamy Sampath, SPE, is theReservoir Division Manager at ExxonMobils UpstreamResearch Company in Houston. He has served on programcommittees for the SPE Annual Technical Conference andExhibition and as a technical editor for SPE journals.

Reservoir-Fluid Sampling and CharacterizationKey to Efficient Reservoir ManagementN.R. Nagarajan, M.M. Honarpour, and K. Sampath, ExxonMobil Upstream Research Company


2

JPT AUGUST 2007

Selecting Sampling Method and Tools.The two commonlyused sampling methods are bottomhole and surface sampling.Bottomhole sampling attempts to capture samples close toreservoir conditions, while surface sampling aims at captur-ing gas and oil samples from the separator under stable flowconditions. Separator fluids then are recombined at a mea-sured producing gas/oil ratio (GOR) to prepare representativereservoir fluid. Both methods have challenges and issues thatmust be overcome to ensure high-quality samples.

In bottomhole-sampling operations, adequate cleaning ofnear-wellbore regions and controlled drawdown are criticalfor obtaining uncontaminated representative samples (Wittand Crombie 1999). Controlled drawdown helps avoid two-phase flow in the reservoir. Downhole fluid analyzers areused to monitor sample contamination and ensure single-phase flow prior to sample capture. Accurate calibration ofthese analyzers is essential for accurate estimate of contami-nation levels. In surface sampling operation, proper well con-ditioning with minimum drawdown is the key to acquiringhigh-quality samples (Witt et al. 1999). Well conditioningrequires that the well be flowed at an optimum rate for anextended period of time with a stable producing GOR. Otherfactors that affect sample quality are separator efficiencyand uncertainties in surface oil- and gas-rate measurements.

While bottomhole sampling has the advantage of capturingfluids at reservoir conditions, surface sampling operation hasa potential for obtaining cleaner samples as a result of largevolumes of fluid production before sampling.

Because a variety of sampling tools and techniques are avail-able, careful thought should be given to tool selection andconfiguration as well as sampling procedures to tailor them tospecific reservoirs and fluids. Fig. 1 provides general samplingguidelines that consist of two parts, sampling-method selec-tion and successful implementation. Selection of a samplingmethod requires a critical review of reservoir conditions, rockand fluid type, and several relevant sampling issues listed inFig. 1. Implementation involves details of tool selection andconfiguration, developing procedures, wellsite execution, andQC. For example, the highlighted items in Fig. 1 are an exam-ple of a near-critical-fluid sampling and demonstrate how therock and fluid conditions and other relevant sampling issueslead to the selection of a bottomhole formation-tester sam-pling method. Fig. 1 also shows the steps involved in the jobplanning and preparation. Often, operational considerations,safety issues, and cost are critical in the final decision.

PVT DataRequirement and QC.The objective of the PVT-data-gathering phase is to obtain reliable high-quality data for

81

Fig. 1Reservoir fluid sampling guidelines. OBM=oil-based mud, DST=drillstem test.

Under-saturatedBlack Oil

Near-Critical

Fluids

Gas-Condensate

Heavy Oil

Fluid Type

Consolidated

Reservoir

Rock

High Permeability

Low Permeability

Unconsolidated

Reservoir

Fluid

Nonhydrocarbons

Compositional

Gradients

Saturated

Water

Sampling

Issues

Surface DST

Sampling

Methods

Bottomhole

Conventional

Surface Isokinetic

Bottom Hole

Formation Tester

Sampling-Method Selection

Job Planning

Tool Design

Tool Sticking

Packer vs. Probe

Pumpout

Near-Wellbore Cleaning

Pumpout Volume

Time

Optimum Drawdown

Calculation

Implementation of Successful Sampling Job

Bottomhole

Formation TesterNear-Critical

FluidsHigh Permeability

Execution

Data Monitoring

Pressure

Pump Rate

Fluid Quality

Contamination

On-site Data

Evaluation

Adjust Operating

Conditions for

Successful Sample

Job Preparation

Tool Assembly

Sample Bottles

Cleaning

Well Site

Equipment

Check

Coordination

Meeting With all

Personnel

Under SaturatedUndersaturated

Compositional

Gradients

Drawdown/2-Phase Flow

Sand-ProductionEmulsions

Loss of Nonhydrocarbons

Depth-Dependent

Composition

Safety/Risk Concerns

Two-Phase Flow

Depth-Dependent

Composition

Contamination by OBM


3

82 JPT AUGUST 2007

reservoir evaluation and development. The PVT-data require-ment depends on the fluid type and the expected developmentand production strategies (Whitson and Brule 2000). Table 1

provides a list of required PVT data and the needed accuracyfor various fluid types and production processes along withgeneral guidelines for designing a laboratory PVT program.For example, extraheavy oils require customized PVT cellsand experimental procedures to accelerate the time needed forattaining equilibrium conditions because of the slow gas lib-eration. On the other hand, more-complex near-critical fluidsand miscible-gas-injection processes need special PVT testsand precise measurement techniques to capture the complexphase behavior exhibited by these fluids. Gas condensates inthe presence of water require PVT cells that can handle three-phase mixtures of gas, water, and condensate.

Data Quality.Ensuring high-quality data requires routinelaboratory visits, evaluation of laboratory procedures andmethods, and spot QC as data become available. The QA/QC methods can range from simple graphical techniquesto sophisticated material-balance calculations (Whitson andBrule 2000).

Fluid ModelingRecommended Methodology. Modelingreservoir-fluid PVT behavior is necessary for reservoir-engineering calculations and simulation studies. Severalapproaches including black-oil correlations, pseudocom-positional formulations, and fully compositional equation-of-state (EOS) methods are used to develop fluid models.Most black-oil correlations are based on regional fluid data,and, therefore, caution should be exercised in choosing cor-

relations for specific oils. Also, it should be recognized thatas these correlations are based on measured data and lacka thermodynamic basis, extrapolation outside the range of

data might contribute to large errors. However, EOS modelsbased on thermodynamic principles are useful for reasonableextrapolation beyond the data range (Whitson and Brule2000). When cost and computational considerations dictateuse of black-oil models for simulation studies, it is highlyrecommended to derive black-oil properties by use of EOSfluid models tuned to laboratory data (Whitson and Brule2000). EOS-based fluid modeling involves several criticalsteps including optimum-component selection by means ofC7+ characterization, incorporation of robust phase-equi-librium calculation (energy minimization) and solutiontechniques to ensure convergence, and a rigorous regressionmethod to tune the model to laboratory data. A brief discus-sion of these techniques is provided, and their applicationsare demonstrated through examples.

C7+Characterization and Component Selection.The C7+fraction of the reservoir fluid contains numerous compoundsof different homologues (paraffinic, naphthenic, and aro-matic) and plays a dominant role in determining the PVTbehavior of the fluid. For example, in a gas-condensate fluid,the dewpoint pressure is a strong function of C7+molecularweight and its relative amount in the fluid. In heavy oils, C7+components dictate the viscosity behavior and control theasphaltenes- and wax-deposition characteristics. Similarly, involatile oils and rich condensates, the oil volumes and otherproperties below the saturation pressure are determinedby the amounts of intermediate and heavy components.


TABLE 1PVT-DATA REQUIREMENT FOR VARIOUS FLUIDS AND PRODUCTION PROCESSES

Fluid Type Reservoir and Required PVT Tests, Data, and Accuracy Supplemental

Gravity GOR C7+ Production Composition Tests/Equipment/Procedures Required Data Data Accuracy Tests

API scf/STB mol% Processes

Heavy Oil 725 10200 > 40 Depletion/ C30+, Wax, and Oil PVTCCE, DFL, ST Pb50 psi

Cold Production Asphaltene, % Direct mixing cell/special proc. Rs, o 5%;Solvent Flood Oil and Oil + Solvent PVT Same + Changes by solvent addition Bo, o 2%

Waterflood Oil PVT and Water PVT Same + Bw, w, w F(P)

Steamflood Oil PVT + High-Temperture PVT Same + Steam properties

Black Oil 2535 200 2040 Depletion C30+, Wax, and Oil PVTCCE, DFL, ST Pb, Rs, Bg, Bo, , F(P) Pb50 psi;

1,500 Waterflood Asphaltene, % Oil PVT and Water PVT Same + Bw, w, w F(P) Rs, o 5%;

Gasflood Bo, o2%

Immiscible Oil PVT and Oil + Gas PVT Same + Changes by gas addition

HC or CO Oil PVT and Gas + Oil PVT Same + P/X data (Pb, o, o, Bo, Same + Liquid vol% 2%2FloodMiscible Swe ll ing and fo rward/ L iquid vol%) + Composi ti onal

backward contact tests changes with injected gas and P

Light Oil 3540 1,000 1320 Depletion C30+, Wax, and Oil PVTCCE, DFL, ST Pb, Rs, Bg, Bo, , F(P) Pb20 psi;

2,000 Waterflood Asphaltene, % Oil PVT + Water PVT Same+w, w, and Bw F(P) Rs, o 5%;

Gas Injection Oil PVT + Swelling test Same + P/X data Bo, o 2%

2,000 813 Depletion C30+, Wax, and Oil PVTHigh Precision Pb, Rs, Bg, Bo, , F(P) Same Flow tests for relative permeability

5 ,000 HC-Gas Inj ec ti on Aspha ltene, % Oil PVT + Swe ll ing and forward / Same + P /X Data (Pb, o, o, Bo and Same + Liquid vol% 2%

Miscible backward contact test Liquid vol%), and CompositionalPrecision mesaurements data with gas in ject ion

Compositional Fluid Same as above + miscibility Same as above Pb20 psi;

Gr adien ts Com pos itions ev aluat ion o f var ying oil Rs, o5%; Bo, o 2%;

With Depth composition in the reservoir Liquid vol% 2%

Gas Condensate >50 >5,000 < 8 Depletion C20+, Wax, and Gas-Condensate PVTCCE, Pd, Z-factors, CGR, and Pd50 psi; Liquid vol% 2%;

Asphaltene, % CVD, ST; solubility of gas in liquid dropout; PVT changes fro m Z-Factor 2%;

Water Analysis water and water vaporization water vaporization CGR 1 STB/MMscf

Gas Injection for Nonhydrocarbon Reservoir-fluid + injection-gas Same as above + P/X data ( Pd, Bg, Same as Above

Pressure (N2, CO2, H2S) and PVT; Special 3-phase PVT-cell and CGR changes with injected gas)

Maintenance Sul fu rAnal ys is w ith zero dead volume

Near-Critical

FluidsHighly

Volatile Oil andRich Condensate

Relative permeability, slim-tube, and

coreflood tests for gas injection

Relative permeability at different

Additional slim-tube and co reflood

tests with injected gas

Fluid Properties

Flow tests for relative permeability

Pb, Rs, Bg, Bo, ,

F(P, T, Rs) Core-depletion tests at different ratesFlow tests for relative permeability

Flow tests for relative permeability

capillary numbers

4050






4

JPT AUGUST 2007 83

Therefore, it is important to characterize them accurately.Several techniques are used to lump these components intopseudocomponents for EOS models. The most widely usedmethod is from Whitson (1983) in which the C7+distribu-tion is represented by a continuous gamma distribution thatis optimally discretized into a few fractions (i.e., pseudocom-ponents). Fluid type and the production processes involvedfurther dictated the component selection. When describingnear-critical fluids and miscible processes, it is important tohave more intermediate and volatile components in the fluiddescription to mimic simple revaporization and more-com-plex condensing and vaporizing drives. Specifics of compo-nent selection will be highlighted through examples.

Energy Minimization and Model Optimization. AlthoughEOS-based fluid models predict multicomponent phasebehavior reliably, they still lack the capability to mimicnear-critical behavior because of mathematical singulari-ties encountered in this region. Therefore, more-rigorousmethods such as Gibbs/Helmholtz energy minimization androbust solution techniques are needed to model near-criticalbehavior (Nagarajan et al. 1991). Application of these tech-niques is discussed under various examples. Another criticalstep in fluid modeling is optimizing model parameters tomatch measured data. Generally, an EOS fluid model uses6 to 12 components, of which four to five are C7+pseudo-components. The result is several tens of model parameters(such as component-critical propertiescritical pressures,temperatures, volumes, and acentric factorand severalbinary interaction coefficients) to be regressed on. Systematicgrouping of these parameters is essential for reliable andfaster regression.

Field ExamplesThe fluid-characterization steps discussed above are illus-trated through field examples ranging from heavy oils to leangas condensates as listed in Table 2along with relevant fluidparameters, fluid-characterization challenges, and solutions.

Heavy OilCerro Negro Field. The Cerro Negro field inthe Orinoco oil belt in northeastern Venezuela contains

extraheavy oil in highly unconsolidated sands. The averagereservoir pressure and temperature range between 800 and1,450 psia and 120 and 145F, respectively. The stock-tank oilgravity is approximately 9API. The live-oil viscosity rangesfrom 600 to 3,000 cp at reservoir conditions. The solutionGOR of the oil is 120 to 130 scf/STB. The high oil viscosityimpedes the separation of solution gas from the oil below itstrue bubblepoint pressure, resulting in microbubbles of gasdispersed in the oil until diffusion forces help the gas bubblesto coalescence slowly into a distinct gas phase (Cengiz et al.2004). This unique behavior poses several challenges in fluidsampling and PVT measurement requiring careful choice oftools and procedures.

Sampling Method, Tools, and Procedures.The objectiveof the heavy-oil sampling program was to obtain adequatevolumes of representative single-phase oil samples for labo-ratory analysis. The following sampling challenges had tobe addressed (Reddie and Robertson 2004): adequate near-wellbore cleaning to minimize sample contamination bydrilling-mud filtrate and optimal drawdown to minimizesand production and avoid two-phase flow while mobiliz-ing the oil from the reservoir into the sample chamber.During surface sampling, measurement uncertainty in theproducing GOR is a concern because of large drawdown andincomplete gas separation from the oil. Another challengewith surface samples is the slow dissolution of gas whilerecombining them to prepare reservoir fluid. Many of thesesampling problems can be eliminated by use of bottomholesampling with a wireline formation tester having appropriatetool selection and procedures as identified in Table 3.

The key components of a wireline-formation-tester toolinclude a variable-rate pump-out module, properly sizedscreens to prevent plugging of flow lines by sands and fines,a resistivity cell, and two optical fluid analyzers to monitorfluid quality and detect two-phase conditions. Single-phasesample bottles should be used for sample collection to avoidflashing the samples. The main advantage of bottomholesampling over surface sampling is that the former offers aviable means to capture single-phase samples and eliminateuncertainties associated with surface samples.

TABLE 2FIELD EXAMPLES OF FLUID SAMPLING AND CHARACTERIZATION

Fluid type/

Reservoir

Pressure, T, Gravity, GOR,

Issues MethodsProcess psia F API scf/STB

Formation-fluid sampling,

Extraheavy oil Cerro Negro 1,400 130 89 120130 special apparatus

and procedures

Black oil/CO2flooding

Salt Creek 2,500 124 34 415Extended PVT tests, detailed

fluid characterization

Customized tests, models

by rigorous techniques

Near-critical-fluid/ Sampling, high-precision Formation-fluid sampling,

Miscible-HC-gas Oso 6,300 232 4045 measurements, modeling special apparatus,

injection compositional gradients gradient modeling

Water and Special PVT tests,

Gas Condensate Arun 7,100 352 55 20,000 reservoir-fluid PVT, customized

corrosion modeling

2,500 to

4,000

Sampling, PVT measurements


5

84 JPT AUGUST 2007


A sampling-simulation program was used to estimate

optimum pumping rates to minimize the probability of sandproduction while mobilizing the oil under single-phase con-ditions. By use of estimated reservoir and fluid parameters,the pump-out volume and time required for adequate clean-ing also were estimated. Specific sampling procedures weredeveloped to reduce the possibility of collecting nonrepre-sentative samples. In addition, procedures for on-site sampletransfer and shipping were developed to preserve sampleintegrity. Qualified personnel were on site to coordinate andmonitor the entire sampling operation. As a result, severalhigh-quality samples were captured for detailed composi-tional and PVT analysis.

PVT Measurements and Modeling. Because of slow gas

liberation and dissolution in this heavy oil, special carewas exercised in selecting equipment and procedures forsample preparation and PVT measurements (Cengiz et al.2004). The sample preparation involved removing both freeand emulsified water from the samples by a nonchemicalmethod. The dewatering process consisted of pressurizingthe sample above the reservoir pressure and subjecting it torepeated heating and cooling cycles from room temperature

to double the reservoir temperature, resulting in water con-

tent of less than 1% in the oil.A direct-mixing PVT cell was custom designed to facilitate

faster equilibrium during PVT measurements, especially formeasuring the bubblepoint. However, even with this PVTcell, nonequilibrium measurements were likely if properexperimental procedures were not followed. In an equilib-rium test, the pressure/volume curve exhibits a sharp changein the slope at the bubblepoint. However, as Fig. 2ashows, itis difficult to detect such sharp slope changes in a nonequilib-rium test (Cengiz et al. 2004). Monitoring pressure responseas a function of time and its rate of change in both nonequi-librium (without stirring) and equilibrium (with vigorousstirring) tests, as shown inFig. 2b, would help identify true

equilibrium conditions. A sharp drop in pressure (Segment1 in Fig. 2b) was observed as soon as the cell volume wasexpanded, indicating the behavior of an incompressible liq-uid. As the gas slowly evolved, a gradual pressure buildup(Segment 2) was observed, but equilibrium condition wasnot reached even after 20 hours without stirring. However,with vigorous mixing, equilibrium condition was reachedrapidly as indicated by Segments 3 and 4 in Fig. 2b. By this

600

700

800

900

1,000

1,100

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0

Time, hours

Pressure,psia

1. Expansion2. Slow Pressure Recovery

(no mixing, nonequilibrium)

3. Faster Pressure Recovery

(vigorous mixing, faster equilibrium)

4. Reaching Equilibrium (constant pressure)5. Start of Next Expansion

1

4

3 5

2

0

200

400

600

800

0 400 800 1200 1600

Pressure, psia

Volume,cm

3

Equilibrium TestNonequilibrium

Test

Fig. 2bPressure response vs. time near bubblepointequilibrium and nonequilibrium behavior.

Fig. 2aEquilibrium and nonequilibrium heavy-oilbubblepoint measurement.

TABLE 3WFT TOOL DESIGN FOR HEAVY-OIL SAMPLING

Problem/Issue Requirement Tool SelectionDesign/

Configuration

Sand production and Optimal drawdown, Dual-packer option, Pump configuredtwo-phase flow Control sand production Flow-control pump, close to sampling point

when mobilizing oil Coarse/fine filters

Mud-filtrate Adequate pumpout Pump with variable High rate: cleaning,

contamination volume before sampling rate (high to low) Low rate: sampling

Fluid quality Fluid-quality monitoring, Optical f luid analyzers, Fluid analyzers positioned

Identify contaminants Resistivity cell above and below pump

Single-phase samples Sample-bottle type, Single-phase sample Close to sampling-point

adequate volumes volume, and number bottles location


6

JPT AUGUST 2007 85

approach of ensuring equilibrium, the true bubblepoint pres-sure was measured repeatedly within approximately 50 psi.Similar procedures were followed in other PVT tests.

A capillary-flow viscometer was used to measure the oilviscosity. Because the oil is saturated at each pressure step inthe differential-liberation experiment, small pressure dropsin the capillary viscometer caused by the flow will liberatethe gas. Therefore, it was necessary to conduct several vis-cosity measurements above the saturated pressure and usean extrapolation technique to determine the viscosity at the

desired differential-liberation pressure.PVT-data interpretation and modeling for heavy oilsrequire reliable pseudoization of C7+components becausea majority of components in heavy oils fall in this range(Romero et al. 2001). Solid-forming compounds, such aswax and asphaltenes, should be characterized properly forflow assurance needs. Rigorous viscosity models that corre-

late viscosity as functions of pressure, temperature, and GORare needed to capture large variations of heavy-oil viscositythroughout the operating conditions.

Black OilSalt Creek Reservoir. The Salt Creek field inwest Texas is a carbonate reservoir in the Permian Basincontaining medium-gravity oil (35API). The initial reser-voir pressure and temperature were 2,199 psia and 124F,respectively. The initial GOR of the oil was 415 scf/STBand the live-oil viscosity was 0.8 cp. The fluid exhibited a

bubblepoint pressure of 1,620 psia at 124F. The reservoirwas initially produced by depletion, pressure maintenance,and gas recycling, followed by waterflood and infill drill-ing (Bishop et al. 2004). High remaining oil saturations inseveral parts of the reservoir after waterflood prompted anevaluation of CO2-miscible-flooding potential to recoversome of the remaining oil. Significant solubility of CO2in the

Fig. 4CO2+oil pressure/composition and ternary diagrams at T120F.

(a)

Mole Fraction CO2

Pressure,psi

Critical Point

L1-V

P1

(b)

CO2

C2C13C13+

L1

-V

P1

L1Oil

L2CO2 Rich Oil

VHydrocarbon Gas + CO2

Liquid Phase

Vapor Phase


7

86 JPT AUGUST 2007


oil enhances the recovery through oil swelling and viscosityreduction (Stalkup. 1984). CO2also vaporizes intermediatecomponents leading to multicontact miscibility and highrecoveries (Stalkup. 1984). A systematic PVT-data-acquisi-tion and modeling study was undertaken for Salt Creek toevaluate recovery efficiency to CO2flooding.

CO2+Oil Phase Behavior.A typical phase behavior exhib-ited by CO2+oil mixtures at both high (>120F) and low(


8/

JPT AUGUST 2007 87

critical nature of the fluid contributed to substantial fluidgradients with depth. Early in the development planning,hydrocarbon-gas injection was identified as a necessaryproduction scheme for pressure maintenance and improvedrecovery through near-miscible processes.

Several key fluid-characterization issues were investi-gated for their effect on recovery factors as part of scoping,planning, and development studies. Fluid characteriztion

addressed the following issues: Acquisition of representative samples at various depthsto quantify initial fluid gradients and for PVT studies.

PVT measurements to capture near-critical behavior,evaluate gas-injection strategies, and design the surface-separator train.

Fluid modeling to predict observed near-critical behaviorand property changes during gas injection.

Development of thermodynamically consistent composi-tional-gradient models for use in reservoir studies.

Sampling. Sampling Oso reservoir fluid posed significantchallenges because the fluid was near-critical. The near-criti-cal nature of the fluid required careful design and executionof sampling because small variations in the fluid pressure andtemperature can cause significant changes in fluid composi-tion, particularly near the saturation pressure. Representativefluid sampling required strict isolation of sampling intervalsbecause the fluid properties varied over depth. During the sur-face-sampling operation, low drawdown, complete phase sep-aration in the surface equipment, and accurate measurementof oil and gas rates were critical for obtaining representativeGOR for laboratory recombination. Both bottomhole wireline-formation-tester- and surface-sampling methods were used tocollect several reservoir-fluid samples from different depths.

PVT Data Requirement.Near-critical PVT measurementsrequired expertise in measuring, analyzing, and interpret-ing the data in addition to state-of-the-art PVT equipment.

Longer equilibration times and high-precision measurementswere necessary to capture steep changes in fluid propertiesthat occurred with small changes in pressure, temperature,or composition. A swelling test was conducted to define theP/X phase-boundary and fluid-property changes resultingfrom gas injection. The measured-P/X diagram of Oso fluidwith injection gas inFig. 6shows that with small additionsof injected gas, the in-situ fluid is in a critical region. Ternarydiagrams at several intermediate pressures (P1> P2> P3) con-structed with measured swelling data and EOS calculationsare shown in Fig. 7.These diagrams were used to evaluatevarious injection strategies.

The reservoir pressure at which gas injection begins, alongwith the injection and production rates, affect the thermody-namic path taken by the reservoir fluid during depletion and,hence, the resulting recovery mechanism. For example, if gasinjection begins at the initial reservoir pressure of 6,300 psiaand production takes the path marked AA in Fig. 6 andthe corresponding two-phase region marked P1in Fig. 7, thehydrocarbon fluid will remain a single-phase throughout theinjection process, leading to a first-contact miscible process.During this process, the fluid changes from a volatile oil toa gas condensate as it crosses the critical region in a single-phase state. However, if the injection begins at a lower pres-sure, P2, the fluid enters the two-phase region to the rightof the critical point close to the cricondenbar, path BBinFig. 6 and the ternary diagram corresponding to P2in Fig. 7.As the fluid enters the two-phase region, a small amount ofoil will drop out. As gas injection continues, this oil will re-evaporate and will be recovered at the surface. Finally, if thegas injection begins at a lower pressure, P3 the fluid entersthe two-phase region along path CCin Fig. 6 to the leftof the critical point and the corresponding ternary diagramat P3in Fig. 7. In this process, multicontact miscibility willdevelop. As gas injection continues, the faster-moving gasphase will be continually enriched by vaporizing intermedi-ate components (vaporizing-gas drive) until the gas phasebecomes first-contact miscible with the oil. This analysis wasused to optimize the development plan.

Fig. 7Ternary diagram for Oso fluid and injection gas.

Fig. 6Oso fluid and injection-gas phase diagram.


9

88 JPT AUGUST 2007

Fluid Modeling. The near-critical nature of Oso fluids,coupled with strong compositional gradients, demandsrobust compositional fluid models (Hier and Whitson2001). An EOS fluid model was developed following similarsteps outlined in the previous example. The C7+fraction wasdivided into an adequate number of volatile components tocapture the near-critical behavior. The EOS calculations wereperformed with energy minimization and robust solutiontechniques (Nagarajan et al. 1991). The EOS model also wasrequired to capture observed compositional gradients, which

was accomplished by modifying the chemical-equilibriumequation to include the gravitational effect through a grav-ity/chemical equilibrium model (Hier and Whitson 2001).The compositional gradients and the resulting fluid-prop-erty variations were predicted by use of the gravity/chemical

equilibrium model in the EOS and the results are shown inFig. 8. The predicted methane and C7+compositional varia-tions agreed well with the data. The inset in Fig. 8 showsthat computed saturation-pressure variations with depthagree with laboratory measurements. The PVT data and theEOS model were used to provide guidelines for designingsurface-separator trains. Because of the near-critical nature ofthe fluid, the number of separator stages and the stage-sepa-rator pressures had to be optimized to maximize recovery ofsurface liquids. A series of multistage-separator simulations

were performed to evaluate sensitivity of the liquid yield tothe number of stages and stage pressures. Fig. 9shows thesensitivity of liquid yield to the number of stages and thestage pressures. Two to four stages of separation are necessaryto maximize recovery of surface liquids. However, increasing


Fig. 9PVT-based separator design to maximize liquid yieldnear-critical-fluid example.

200

225

250

275

300

1 2 3 4 5 6

Liquid Yield vs. Separator Pressure in a

3-Satge Separation Process

275

280

285

290

0 400 800 1200 1600

Separatore Pressure (psia)

CondensateYield(STB/MMSCF)

First Stage

Second Stage

Third Stage

200

225

250

275

300

1 2 3 4 5 6

Number of Stages

CGR,STB/

MMscf

Liquid Yield vs. Separator Pressure in a

Three-Stage Separation Process

275

280

285

290

0 400 800 1,200 1,600

Separator Pressure, psia

CGR,STB/MMscf

First Stage

Second Stage

Third Stage

Fig. 8Compositional-gradient and fluid-property variations as a function of depthOso.

11,500

11,000

10,500

10,000

9,500

9,000

5 15 25 35 45 55 65 75

Composition, mol%

TrueVerticalDepth,ft

subsea

C1C7+Points are measured data

Lines are predictions

Oso Saturation Pressure and Reservoir Pressure With Depth

11,500

11,000

10,500

10,000

9,500

9,000

4,000 4,500 5,000 5,500 6,000 6,500 7,000

Pressure, psia

TrueVerticalDepth,feetsubsea

Points are data/

Lines are predictions

Saturation

Pressure

Reservoir

Pressure


1

JPT AUGUST 2007 89

the number of stages beyond four affected the yield nega-tively. In addition, Fig. 9 shows that if a three-stage-separatortrain is selected, the second- and third-stage pressures mustbe selected precisely to maximize surface-liquid recovery.

Gas CondensateArun Field. The Arun Field in Indonesiais a high-temperature lean-gas-condensate reservoir withsurface-liquid yields of approximately 50 to 55 STB/MMscf(Pathak et al. 2004). The original reservoir pressure was7,100 psig at 352F. Initially, Arun gas contained significant

amounts of vaporized water (4 mol%) and CO2(16 mol%).Accurate characterization of gas-condensate/water phasebehavior through an EOS was essential to predict water-pro-duction levels and corrosion potential.

PVT Data and Model. A PVT program was initiated tostudy gas/water and gas-condensate/water phase behavioras functions of pressure and temperature (Ng and Robinson1986; Ng et al. 1988). A specially designed visual PVT cellwas used to measure the amount of water vapor in the gasphase and the condensate/water volumetric behavior asa function of reservoir, wellbore, and surface conditions.The results showed that water content of the reservoir gasincreased four-fold from approximately 4 mol% at 7,100 psigto approximately 16 mol% at 1,000 psig as shown in Fig. 10(Ng and Robinson 1986). This relation corresponds to waterproduction increasing from 6.5 STB/MMscf at the initial res-ervoir conditions to 75 STB/MMscf as the reservoir pressuredeclined to 1,000 psig (Pathak et al. 2004). This informationis critical for designing optimum water-handling facilities andfor corrosion management. Fig. 10 also shows condensatedropout at different reservoir, wellbore, and surface condi-tions, indicating an initially increasing condensate dropoutwith declining pressure and temperature, peaking approxi-mately 2,500 psia, and dropping off at lower pressures as aresult of condensate revaporization (Ng et al. 1988).

Fig. 11 combines the volumetric behavior of condensateand water in terms of condensate/water ratio (CWR) at the

bottomhole (352F), wellhead (200F), and surface (100F)temperatures during reservoir depletion. Generally, the CWRis controlled by the condensate-dropout characteristics dur-ing the early stages of depletion. But, as the reservoir pressuredeclines below dewpoint pressure, the CWR falls more rap-idly as a result of decreasing condensate volume (caused byliquid dropout in the reservoir) and increasing water volume(caused by increased water vaporization). In Fig. 11 at 352F,the condensate volume is zero until the reservoir pressuredrops below the dewpoint pressure, then it increases with

declining reservoir pressure, going through a maximum,and finally decreasing as a result of revaporization at lowerpressures. The water volume, however, increases slowly atfirst and then rapidly as a result of an increased rate of watervaporization at lower pressures and subsequent condensa-tion in the wellbore. At 200F (wellhead) in Fig. 11, thecondensate volume initially increases as a result of both thepressure and temperature drop in the wellbore. However,increased water vaporization during reservoir depletionand subsequent condensation at the wellhead cause a largerincrease in water volume resulting in a sharp drop in CWR.At 100F (surface), the CWR behaves in a similar manneras at the wellhead. However, the crossover of condensateand water volumes (CWR =1) occurs at the surface (pinkcurve) much earlier (reservoir pressure of 4,000 psia) thanat the wellhead, green curve (reservoir pressure 2,800 psia).The implication is that the potential for corrosion is high inthe surface equipment in the early stages of depletion. Thecorrosion potential may increase toward the wellhead anddownhole with reservoir pressure decline.

In the Arun reservoir, water vaporization and three-phasewater/condensate/gas volumetric behavior were modeledwith a three-parameter Peng-Robinson EOS. Because a cubicequation does not model a polar compound such as wateraccurately, the critical properties of water were modified onthe basis of the coordination number to match laboratorydata. Further, special binary-interaction parameters for water

0

10

20

30

0 2,000 4,000 6,000 8,000

Pressure, psia

WaterContent,mol%

0

2

4

6

8

CondensateDropout,vol%

100F

350F

200F

Points are measured data

Lines are EOS predictions

Fig. 10Water content and condensate-dropout characteristics of Arun gas.


1

90 JPT AUGUST 2007

were used in the tuning process. The EOS-model matches areshown in Figs. 10 and 11 as solid lines.

ConclusionsFluid characterization strongly affects in-place-volume,recovery-factor, injectivity/productivity, and well-deliverabil-ity calculations. Accurate fluid characterization minimizestechnical uncertainties and, thus, provides a reliable repre-sentation of the asset value. Four field examples containingfluids from extraheavy oil to lean gas condensates undergo-ing different production processes were presented to high-light key steps in fluid sampling and characterization.

1. Fluid-sampling programs must be tailored to the fluidtype, reservoir-rock and -fluid conditions, and fluid distribu-tion. Special tools and procedures with strict QC will ensureobtaining high-quality representative samples.

2. The fluid type and production processes dictate PVT-data requirements, measurement methods, and data accu-racy. Laboratory methods and procedures must be tailored tospecific fluids with expert QA/QC.

3. The C7+components must be characterized accuratelyfor EOS-component selection. Rigorous modeling methods,such as energy minimization, and robust solution techniquesare needed to model near-critical fluids and processes.

4. Reliable compositional-gradient models are needed tocapture fluid-property variations in reservoirs with highrelief and/or near-critical fluids.

AcknowledgmentsWe gratefully acknowledge the support and encouragementof ExxonMobil Upstream Research Company, ExxonMobilProduction Company, ExxonMobil Oil Indonesia, MobilProducing Nigeria, and ExxonMobil de Venezuela.

Acronyms CCE= constant-composition expansionCGR= condensate/gas ratioCVD= constant-volume depletionCWR= condensate/water ratio

DFL=differential liberation DST= drillstem test EOS= equation of stateGOR= gas/oil ratio HC= hydrocarbonOBM= oil-based mud PVT= pressure/volume/temperature P/X= pressure/composition

QA= quality assurance QC= quality control ST= separator test

Nomenclature Bo= oil formation volume factor Bg= gas formation volume factor Bw= water formation volume factor F(P)=function of pressure F(T)= function of temperatureF(Rs)= function of solution gas/oil ratio p= pressure pb= bubblepoint pressure pd= dewpoint pressure Rs= solution GOR T= temperature Vl= volume of liquid Z= gas deviation factor o= viscosity of oil w= viscosity of water o= density of oil w= density of water

ReferencesBishop, D.L., Williams, M.E., Gardner, S.E., Smith, D.P., and

Cochrane, T.D. 2004. Vertical Conformance in a Mature Carbonate

CO2 Flood: Salt Creek Field Unit, Texas. Paper SPE 88720-MS presented at the SPE Abu Dhabi International Petroleum

Exhibition and Conference, Abu Dhabi, UAE, 1013 October.

DOI: 10.2118/88720-MS.


0.00

1.00

2.00

3.00

4.00

5.00

0 1,000 2,000 3,000 4,000 5,000 6,000 7,000

Reservoir Pressure, psia

CWR,vol/vol

352F

200F

100F

(Near Bottomhole)

(Surface)

(Wellhead)

Fig. 11CWR of produced gas in the wellbore as reservoir pressure declines.


1

JPT AUGUST 2007 91

Cengiz, S., Robertson, C., Kalpacki, B., and Gupta, D. 2004. A Study

of Heavy Oil Solution Gas Drive for Hamaca Field: Depletion

Studies and Interpretations. Paper SPE 86967-MS presented at theSPE International Thermal Operations and Heavy Oil Symposium

and Western Regional Meeting, Bakersfield, California, 1618

March. DOI: 10.2118/86967-MS.El-Mandouh, M.S., Bette, S., Heinemann, R.F., Ogiamien, E.B.,

and Bhatia, S.K. 1993. Full-Field Compositional Simulation ofReservoirs of Complex Phase Behavior. Paper SPE 25249-MS pre-

sented at the 12th SPE Symposium on Reservoir Simulation, New

Orleans, 28 February3 March. DOI: 10.2118/25249-MS.

Hier, L., and Whitson, C.H. 2001. Compositional Grading

Theory and Practice. SPEREE 4(6): 525. SPE 74714-PA. DOI:10.2118/74714-PA.

Honarpour, M.M., Nagarajan, N.R., and Sampath, K. 2006. Rock/

Fluid Characterization and Their IntegrationImplication on

Reservoir Management. JPT58(9): 120. SPE 103358-MS. DOI:10.2118/103358-MS

Nagarajan, N.R., Cullick, A.S. and Griewank, A. 1991. NewStrategy for Phase Equilibrium and Critical Point Calculations by

Thermodynamic Energy Analysis; Part I. Stability Analysis and

Flash; Part II. Critical Point Calculations. Fluid Phase Equilibria62(3): 211.

Ng, H.-J., and Robinson, D.B. 1986. The Influence of Water and

Carbon Dioxide on the Phase Behavior and Properties of a

Condensate Fluid. Paper SPE 15401 presented at the SPE Annual

Technical Conference and Exhibition, New Orleans, 58 October.DOI: 10.2118/15401-MS.

Ng, H.-J., Robinson, D.B., Nagarajan, N.R., Rastogi, S.C., and

Hasan, N. 1988. Phase Behavior of Retrograde Gas Condensate-

Water System Under High Pressure and Temperature Conditions.

Paper presented at the Indonesian Petroleum Congress, Jakarta.Pathak, P., Fidra, Y., Kahar, Z., Agnew, M., and Hidayat, D. 2004. The

Arun Gas Field in Indonesia: Resource Management of a Mature

Field. Paper SPE 87042-MS presented at the SPE Asia Pacific

Conference on Integrated Modelling for Asset Management,Kuala Lumpur, 2930 March. DOI: 10.2118/87042-MS.

Reddie, D.R. and Robertson, C.R. 2004. Innovative Reservoir Fluid

Sampling Systems. Paper SPE 86951-MS presented at the SPE

International Thermal Operations and Heavy Oil Symposium and

Western Regional Meeting, Bakersfield, California, 1618, March.DOI: 10.2118/86951-MS.

Romero, D.J., Fernandez, B., and Rojas, G. 2001. Thermodynamic

Characterization of a PVT of Foamy Oil. Paper SPE 69724-MS

presented at the SPE International Thermal Operations and

Heavy Oil Symposium, Parlamar, Margarita Island, Venezuela,1214 March. DOI: 10.2118/69724-MS.

Stalkup Jr., F.I. 1984. Miscible Displacement. Henry Doherty

Monograph Series. SPE of AIME: Richardson, Texas. 8.

Whitson, C.H. 1983. Characterizing Hydrocarbon Plus Fractions.

SPEJ23(4): 683. DOI: 10.2118/12233-PA

Whitson, C.H. and Brule, M. 2000. Phase Behavior: Monograph

Series. SPE: Richardson, Texas. 20.

Witt, C.J. and Crombie, A. 1999. A Comparison of Wireline and

Drillstem Test Fluid Samples From a Deep Water Gas-CondensateExploration Well. Paper SPE 56714-MS presented at the SPE

Annual Technical Conference and Exhibition, Houston, 36

October. DOI: 10.2118/56714-MS. JPT

From Halliburton, the cementing and drilling

fluids pioneer, comes another innovative, fit-for-

purpose cementing first: our Tuned Cementing

Solutions approach.

Halliburtons Tuned systems deliver the best solution

for any given set of wellbore conditions. For example,if your challenge is to repair wellbore leaks or restore

pressure integrity, our conventional tuned

SqueezeCem or SqueezeSeal (foam) cement

systems can be customized to address your exact

conditionsdelivering superior performance.

For reliability and ingenuity, the one to call is

Halliburton. Whatever your cementing challenge.

For further information, visit us online at

www.halliburton.com/tcs.

Unleash the energy.

For your wellbore

integrity cementingchallenge, the

Tuned solution.one

HALLIBURTON

Drilling and Evaluation

2007 Halliburton. All rights reserved.

R E L I A B I L I T Y . I N G E N U I T Y .

Reservoir-Fluid Sampling and Characterization.pdf

Documents

Transcript of Reservoir-Fluid Sampling and Characterization.pdf