oilfield review autumn 2001 - quantifying contamination...
TRANSCRIPT
24 Oilfield Review
Quantifying Contamination Using Color ofCrude and Condensate
R. John AndrewsHibernia Management and Development Company Ltd.St. John’s, Newfoundland, Canada
Gary BeckBPHouston, Texas, USA
Kees CastelijnsLondon, England
Andy ChenCalgary, Alberta, Canada
Myrt E. CribbsChevronTexacoBellaire, Texas
Finn H. FadnesJamie Irvine-FortescueStephen WilliamsNorsk Hydro, ASABergen, Norway
Mohamed HashemShellNew Orleans, Louisiana, USA
Establishing the level of oil-base and synthetic mud-filtrate contamination in fluid
samples is critical for obtaining meaningful data on fluid properties. New tools and
techniques now allow real-time, quantitative measurement of contamination in gas-
condensate and oil reservoirs.
In deepwater areas, an oil or gas company mayspend tens of millions of dollars drilling a well toprove the presence of hydrocarbons, and thenplug and abandon the well almost immediately.The operator may take years designing and build-ing facilities before drilling another well in thefield. Exploration wells provide a narrow windowof opportunity for collecting hydrocarbon sam-ples to make development decisions; therefore,obtaining high-quality samples is imperativewhether the prospect is in deep water or on thecontinental shelf, in China, Canada, the Caspian,or elsewhere.
Testing well production is a good way toobtain fluid samples, but that is not always fea-sible for economic or environmental reasons.Downhole samples define fluid properties thatare used throughout field development.Estimates of hydrocarbon volume, bubblepointpressure and gas/oil ratio (GOR), simulation ofreservoir flow and placement of wells all dependon formation-fluid properties. Hydrate, asphal-tene and wax formation must be controlled ortreated. Presence of corrosive gases affects thechoice of materials for flowlines and surfacefacilities. These examples illustrate the wideimpact that hydrocarbon composition and behav-ior have on planning a new field.1
A. (Jamal) JamaluddinHouston, Texas
Andrew KurkjianBill SassSugar Land, Texas
Oliver C. MullinsRidgefield, Connecticut, USA
Erik RylanderBelle Chase, Louisiana
Alexandra Van DusenHarvard UniversityCambridge, Massachusetts, USA
For help in preparation of this article, thanks to VictorBolze, Reinhart Ciglenec, Hani Elshahawi, Troy Fields, GusMelbourne, Julian Pop and Rod Siebert, Sugar Land, Texas;Peter Kelley, ChevronTexaco, Houston, Texas; and ToruTerabayashi, Fuchinobe, Japan.AIT (Array Induction Imager Tool), CHDT (Cased HoleDynamics Tester), CMR (Combinable Magnetic Resonance),FFA (Field Fingerprint Analyser), LFA (Live Fluid Analyzer),MDT (Modular Formation Dynamics Tester), OCM (Oil-BaseContamination Monitor), OFA (Optical Fluid Analyzer),Platform Express and TLC (Tough Logging Conditions) aremarks of Schlumberger. RCI (Reservoir CharacterizationInstrument) is a mark of Baker Atlas. RDT (ReservoirDescription Tool) is a mark of Halliburton.1. Joshi NB, Mullins OC, Jamaluddin A, Creek J and
McFadden J: “Asphaltene Precipitation from Live CrudeOil,” Energy and Fuels 15, no. 4 (2001): 979-986.
50387schD03R1.p24.ps 11/17/01 8:30 PM Page 24
Autumn 2001 25
Openhole-wireline or drillstring-conveyed for-mation testers analyze selected fluid propertiesdownhole and acquire small volumes of reservoirfluid for later testing in a laboratory. However,mud filtrate invades the formation during drilling,so these fluid samples usually are contaminated.
During the past few years, real-time methodshave been developed as part of the openhole-logging suite of services to analyze sample contamination. These methods ensure that repre-sentative fluid samples are collected and minimize tool-sticking risks by introducing effi-ciencies in sample collection. Until recently,these sampling methods were unreliable in holesdrilled with oil-base and synthetic muds or in for-mations with high GOR.
This article reviews the requirements andchallenges in sampling reservoirs and reports onadvances in evaluating sample contamination.Except where explicitly stated to be contamina-tion from water-base mud, this article discussesoil-base or synthetic-base mud-filtrate contami-nation. We describe a technique for determining
the time required to collect an acceptable fluidsample at a given sampling station and showhow proven sample-contamination measure-ments can be extended to high-GOR fluids andcondensates. Quantitative contamination mea-surement is illustrated with case histories fromoffshore Newfoundland, Canada, the Gulf ofMexico and the Norwegian North Sea.
Obtaining Downhole Fluid SamplesFormation fluid samples provide important datato optimize operator investment in both upstreamand downstream facilities. Laboratory measure-ments establish standard fluid properties such aspressure-volume-temperature (PVT) behavior, vis-cosity, composition and GOR. In fields destinedfor subsea development, flow assurance is amajor concern, so tests are performed to evalu-ate gas and solids content. Hydrogen sulfide[H2S] and carbon dioxide [CO2] in oil require spe-cial handling and materials. Temperature andpressure changes in pipelines can lead to asphal-tene and wax precipitation and deposition, andlow seafloor temperatures can induce hydrate
Pumpout module
Sample-chambermodules
Multisamplemodules
LFA Live FluidAnalyzer module
Hydraulic-powermodule
Single-probe module
> The MDT Modular Formation Dynamics Testertool configured for fluid-sample collection.
50387schD03R1 11/29/01 3:28 AM Page 25
formation. Commingling different crude oilsthrough satellite tiebacks can dramatically alterfluid properties (above).
The data-acquisition process must includefluid characterization to get the most out of everyprospect. Taking fluid samples early in the life ofa well ensures that fluid composition and proper-ties are available for timely input to field plan-ning decisions. If fluid properties will affectfacilities or transport, accurate fluid analysisgives an operator the opportunity to mitigate oreliminate problems through changes in produc-tion design, or to manage them through ongoingtreatments such as heating pipelines—a choicebetween upfront capital expenditures and on-going operating expenses.
In some fields, fluid samples can be obtainedduring a drillstem test (DST) or, after a well isflowing, a production test. In some cases, a wellmust be completed before a flow test, which can cost tens of millions of dollars in deep-water Gulf of Mexico wells. In areas such as theGrand Banks, offshore Newfoundland, Canada,operators want to minimize operation times toavoid risks such as harsh seas and iceberg haz-ards. Environmental concerns restricting flaringand removing fluids from the rig also restrict use
of DSTs and production tests. The cost and risk ofDSTs lead operators to use wireline tools forfluid-sample acquisition.
A major problem in downhole fluid-samplecollection is contamination from drilling-mud fil-trate entering a tool with reservoir fluids.Contamination from water-base mud (WBM) canbe discriminated easily from reservoir oil. Inmany of today’s high-risk wells, oil-base muds(OBMs) and synthetic oil-base muds (SBMs) areused to ensure compatibility with shales,improve wellbore stability and increase drillingspeed. OBM and SBM filtrates mix with reservoircrude, making quantification of contaminationmuch more difficult than when using WBM. Fluidproperties are often extrapolated to an uncon-taminated condition by mathematically removingthe contaminant from the distribution of con-stituents. However, extrapolation from high lev-els of contamination is risky—most companiesavoid liquid-phase contamination greater than10% on a volume-to-volume basis.
Several commercially available tools havefluid-sampling capabilities, including theSchlumberger MDT Modular Formation DynamicsTester tool, the Baker Atlas RCI ReservoirCharacterization Instrument tool, and theHalliburton RDT Reservoir Description Tool
sonde. Most wireline formation testers press aprobe against the borehole wall at a specifieddepth, pump down the formation and draw influid for evaluation, and then collect sampleswhen desired fluid characteristics are reached.2
With a probe securely pressed against theborehole wall, a short, rapid pressure dropbreaks the mudcake seal. Normally, the first fluiddrawn into the tool will be highly contaminatedwith mud filtrate (next page, top). As the toolcontinues to withdraw fluid from the formation,the area near the probe cleans up, and reservoirfluid becomes the dominant constituent. The timerequired for cleanup depends on many parame-ters, including formation permeability, fluid vis-cosity, the pressure difference between boreholeand formation, and the duration of the pressuredifference during and after drilling. Increasingpump rate can shorten the cleanup time, but therate must be controlled carefully to preserve thereservoir-fluid condition. Because many factorsaffecting cleanup time have unknown values,determining the contamination level during a log-ging job is crucial to obtaining good samples.
The versatile Schlumberger MDT system per-forms a variety of functions, depending on whichmodules are joined together. The tool’s primarypurposes are to obtain formation-fluid samples,to measure formation pressures at given pointsin the reservoir and to estimate permeability insitu. For a description of use of the tool for per-meability measurement and description of othertool modules, see “Characterizing Permeabilitywith Formation Testers,” page 2.
The OFA Optical Fluid Analyzer system in theMDT tool has provided a qualitative measure ofcontamination since its introduction in 1993.Schlumberger has developed the OCM Oil-BaseContamination Monitor technique to predict thetime needed to achieve an acceptably low levelof contamination at a given sampling station.This reliable new technique monitors samplecontamination quantitatively, adding confidenceto these crucial contamination measurements.
26 Oilfield Review
Subsea wellhead
Buildup of solidsin the wellbore
Asphaltene deposition inthe near-wellbore region
Solids in subsea flowlines
Precipitated solidsin the separator
> Transport hazards from reservoir-fluid constituents while flowing to surface.Asphaltenes, waxes and hydrates can form during fluid transport to surface.Depositing such solids clogs tubulars or blocks pores in the formation. Solidsalso precipitate in separators under certain conditions. In addition, commin-gling fluids at wellheads can generate unstable conditions leading to precipi-tation of solids.
> Components of MDT optical analysis modules.
Color channelsGas refractometerWater channelsOil channel
Contaminationmeasurement
Color channelsMethane channelGas refractometerWater channelsOil channel
Methane flagGas flagWater flagOil flag
Gas flagWater flagOil flag
OFA module
OCM module
LFA module
50387schD03R1.p26.ps 11/17/01 8:30 PM Page 26
Autumn 2001 27
The new LFA Live Fluid Analyzer module addsa methane detector that provides a more defini-tive measure of gas content in the oil phase andallows calculation of GOR. This module can beused to ensure that the fluid remains in singlephase during sampling; dropping pressure belowthe bubblepoint would make the fluid unrepre-sentative. The quantitative OCM contaminationmeasurement can be used with either an LFA orOFA module (previous page, right).
Modular reservoir sample chambers (MRSCs)are available to collect large samples (below).Multiple 6-gallon [22,712-cm3] chambers can berun at the bottom of the tool string to act as dumpchambers. Samples for PVT analysis are morecommonly collected in smaller chambers. A mul-tisample module (MRMS) allows collection of sixeasily removable sample bottles (MPSR) that arecertified for transport by the US Department ofTransportation (DOT) and by Transport Canada.The 450-cm3 [0.12-gal] MPSR bottle is reduced to 418 cm3 [0.11 gal] when an agitator is addedto improve fluid mixing in the laboratory. TheSchlumberger Oilphase single-phase multi-sample chamber (SPMC) can be used in theMRMS when keeping a reservoir fluid sample in
2. For more on use of the MDT tool for downhole fluid sam-ple analysis: Crombie A, Halford F, Hashem M, McNeil R,Thomas EC, Melbourne G and Mullins OC: “Innovations in Wireline Fluid Sampling,” Oilfield Review 10, no. 3(Autumn 1998): 26-41.Badry R, Fincher D, Mullins O, Schroeder B and Smits T:“Downhole Optical Analysis of Formation Fluids,” OilfieldReview 6, no. 1 (January 1994): 21-28.
Oil cone
Filtrate Oil cone
t1
Oil
OD
Timet1 t2 t3
t2
t3
Filtr
ate
Filtr
ate
> Drawing in filtrate. The MDT probe pressed against a borehole wall is the source of a pressuredrawdown, pulling fluids into the tool. Filtrate near the probe enters first, but as the pressure sinkexpands, a higher proportion of fluid is reservoir fluid. The optical density (OD) increases as darkercrude oil replaces the more transparent mud filtrate.
MRSC
H2S
Maximum hydrostaticpressure
Volume
Sample pressure
Transportable
Downhole temperature
Surface heatingallowed
Pressurecompensated
20- and 25-kpsi [138-and 172-MPa] options
1- and 2.75-gal [3785-and 10,410-cm3] options
20 kpsi
No
204°C [400°F]
77°C [170°F]
No
14 kpsi [97 MPa]
1- and 2.75-galoptions
14 kpsi
No
204°C
54°C [130°F]
No
10 kpsi [69 MPa]
6 gal [22,712 cm3]
10 kpsi
No
204°C
Not allowed
No
20- and 25-kpsioptions
450 cm3 [0.12 gal]
20 kpsi
Yes
204°C
100°C [212°F]
No
20- and 25-kpsioptions
250 cm3
[0.07 gal]
20 kpsi
No
204°C
204°C
Yes
Non-H2S MPSR SPMC
MRMS
The 25-kpsi limit is for special high-pressure modules, and the sampling must be done in low-shock mode—the bottle is compensated to hydrostatic pressure behind the piston.Only Schlumberger Oilphase is allowed to heat chambers above 54°C [130°F].Six-gallon bottles must be run on the bottom of the string. Several bottles can be combined in one string.Addition of an agitator reduces this volume to 418 cm3 [0.11 gal].Transportable indicates US Department of Transportation Exemption and Transport Canada Permit for Equivalent Safety.Compressed nitrogen is used to compensate the sample pressure so it does not decrease as much upon cooling when brought to surface.
3
5
6
1
2
1 1
4
1
2 3 4 5 6
> Sample bottles available for the MDT tool.
50387schD03R1.p27.ps 11/17/01 8:30 PM Page 27
single phase from the downhole collection pointto the PVT laboratory is necessary. After the MDTpumpout module fills a SPMC chamber at forma-tion pressure, a preset nitrogen charge isreleased. Acting through a piston floating on asynthetic oil buffer, the nitrogen adds sufficientoverpressure to keep the fluid in single phaseduring retrieval to surface.
Black Oil Isn’t Always BlackOils have color—black, brown, red, tan and evengreen crude oils have been seen. The hue andintensity of light transmitted or reflected fromcrude oil or gas condensate depend on the light’sinteraction with molecules and molecular bonds inthe fluid. Measurements of this interaction can beused to distinguish oils of different compositions.
The unit of light absorption or optical density(OD) is the logarithm of the ratio of incident-lightto transmitted-light intensity. Therefore, darkerfluids have higher OD, and a one-unit increase inOD represents a factor of ten decrease in trans-mittance. An OD of zero indicates all light istransmitted, while an OD of two represents 1%transmission. A fluid’s OD varies with the wave-length of incident light.
Reduction of transmitted-light intensity can becaused by one of two physical processes. Somelight is scattered by particles in the fluid; scatter-ing outside the optical path to the detectordecreases intensity. Light also can be absorbed bymolecules in the fluid. The MDT optics relies ondifferences in absorption in visible and near-infrared portions of the electromagnetic spectrumto discriminate fluids in the flowline.
Pure, light hydrocarbons such as pentane areessentially colorless; they do not absorb light inthe visible spectrum. Condensates may be clearor lightly shaded reddish-yellow to tan, becausethey absorb more from the blue end of the spec-trum than from the red end. Heavier crude oils,which contain more complex molecules, absorblight strongly throughout the visible region, mak-ing them dark brown or black.
Light with a wavelength in the visible or near-infrared spectra, referred to as the color region,interacts with a molecule’s electronic energybands. Compared to less complex molecules,larger and more complex aromatic hydrocarbonmolecules, such as asphaltenes and resins,absorb light having longer wavelengths.3 Becauseheavier oils contain more aromatic compounds,they tend to have darker coloring than less denseoils and condensates (above). Waxes are color-less, but if the molecules are long enough, theywill scatter light and appear white.
Despite the differences in optical absorptionof various reservoir oils caused by composition,there is a common behavior. Electronic absorptiongenerally decreases as wavelength increases.The OD decay in the visible and near-infraredregion can be characterized by a single parame-ter, which can be thought of as the color of the oil.
To understand how OD measurements can beused to quantify contamination, it is important todistinguish between absorption in the colorregion by two kinds of hydrocarbons: complexaromatics and saturated aliphatics. Complex aro-matics contain carbon rings with both single anddouble carbon-carbon bonds, which are excitedby visible and near-infrared light. Aliphatic com-pounds are open chains of carbon atoms. If all thecarbon-carbon connections are single bonds andother bonds are with hydrogen, the aliphaticmolecule is termed saturated. Only high-energyultraviolet light can excite saturated aliphaticmolecules, so they have a low OD in the colorregion of the spectrum.
Black oils contain many complex aromaticcompounds, whereas natural OBMs comprisemostly saturated compounds; SBMs are madeonly from saturated aliphatics. The difference inchemical composition between reservoir-crudeoil and drilling-mud filtrate makes OD a goodmeasure of filtrate contamination in crude oil.
Exciting MoleculesWater can be distinguished from oil easily,because it is highly absorbing at near-infraredwavelengths around 1445 and 1930 nano-meters (nm), where oil is relatively transparent(next page, top). Oil has a strong absorption peakaround 1725 nm, where water does not. Thesepeaks come from the interaction of light withvibrational energy bands in carbon-hydrogenbonds for oil and oxygen-hydrogen bonds forwater. Molecules containing such a bond absorbphotons of the proper wavelength, and the pho-ton energy is converted into molecular vibration.Monitoring absorption at these three wave-lengths differentiates between water and oil.
Hydrocarbon compounds comprise linkedchains, branches or rings of carbon atoms, eachhaving hydrogen atoms attached. Typically, a car-bon atom will bond with two other carbon atomsand two hydrogen atoms. Carbon atoms at theend of a molecule will have three hydrogenatoms attached, while those at a branch, con-necting with three other carbon molecules, willhave only one hydrogen bond. Methane is a single carbon molecule with four hydrogen atoms attached.
The oil peak in Channel 8 of the OFA modulemeasures molecular absorption of light by carbonatoms having two hydrogen atoms attached,which are the primary constituents of reservoir
28 Oilfield Review
Optic
al d
ensi
ty
Wavelength, nm
3.0
2.5
2.0
1.5
1.0
0.5
0500 1000 1500 25002000
Asphalts
Condensates
Black oils
> Optical density of various oils. The OD spectrum of hydrocarbons is relatedto the amount of aromatics, which also relates to API gravity. Gas conden-sates have little or no color absorption beyond about 500 nanometers (nm).The range of oils grades through increasingly dense black oils having highercolor absorption out to asphalts, which absorb strongly even into the near-infrared region. All oils and condensates absorb near 1725 nm. The hydrocarbonpeak from 2300 to 2500 nm is beyond the region covered by the MDT channels.
50387schD03R1.p28.ps 11/17/01 8:31 PM Page 28
Autumn 2001 29
oils. A high-resolution optical spectrometerreveals this oil peak in much greater detail, show-ing several absorption peaks in hydrocarbon fluids(right). Although methane has some absorption atthe oil peak, there is no absorption by hydrocar-bons with more than one carbon atom at themethane peak. This provides an ideal discrimina-tor for methane content in live crude oils—uti-lized by a new MDT tool, the LFA Live FluidAnalyzer module.4 The detection channel tuned tothat wavelength replaces the OFA module’s short-est wavelength color band in Channel 0.
3. Mullins OC: “Optical Interrogation of Aromatic Moietiesin Crude Oils and Asphaltenes,” in Mullins OC and Sheu EY:Structures and Dynamics of Asphaltenes. New York,New York, USA: Plenum Press, 1998.
4. A live crude oil evolves significant quantities of gaswhen its pressure and temperature are lowered. A deadoil does not evolve gas at atmospheric pressure androom temperature. Stock-tank oil, the liquid emergingfrom the final surface separator, contains little gas.
Optic
al d
ensi
ty
00
500 1000Wavelength, nm
1500 2000
Channelnumber
1 2 3 4 5 6 7 0' 8 9
1
2
3
4
More
complex molecules
Color or electronicabsorption region
Vibrationalabsorption region
Water peak
Oil peak
Water peak
Methane peak
C HHH
H
H-C-H vibrational peak
CH
HCH
HCH
H
> Absorption spectrum. The MDT tool monitors light absorption starting in visible wavelengths and extending into the near-infrared. The ten chan-nels of the OFA module, numbered 0 through 9, are shown. In the color region on the left, crude oils have a rapidly decaying absorption, caused byinteraction of light with electrons in the molecules. More complex aromatic molecules (green shapes) absorb at longer wavelengths. Channels 6and 9 are tuned in the middle of molecular vibrational peaks for water; Channel 8 is in the molecular vibration peak for the CH2 bond in hydrocar-bons. Channel 0’, which replaces Channel 0 in the LFA module, is tuned to the methane peak.
OD
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0Wavelength, nm
Methanen-HeptaneMethane-heptane mix
Oil peak
Methane peak
> High-resolution vibrational absorption spectrum of heptane, methane and amix of the two. Heptane (green) does not absorb light at the CH4 methanepeak. Methane absorption (red) at the CH2 oil peak is low. Absorption of a mix-ture of the two (black) is the sum of the individual absorptions, according tothe Beer-Lambert law. The LFA module has a channel set at the methane peak.
50387schD03R1.p29.ps 11/17/01 8:31 PM Page 29
The Key to Quantifying ContaminationThe MDT tool includes an optical module withtwo devices designed to monitor contaminationin OBM systems. A gas refractometer uses lightfrom a diode reflected off a sapphire window toqualitatively identify the fluid phase in a flowline(left). At the selected angle of incidence, thereflection coefficient is much larger when gas isin contact with the window than when oil orwater contacts it.5
The second detector in the OFA module usestransmitted light to evaluate absorption charac-teristics of a fluid. A high-temperature tungstenhalogen lamp provides a broadband source oflight that passes along optical guides andthrough a 2-mm thick optical chamber in theflowline. The distribution of transmitted light isrecorded at 10 wavelengths in the visible andnear-infrared spectra. Two of these channelsdetect the strong water-absorption peaks, indi-cating water content in the fluid when comparedwith the strong hydrocarbon-absorption peak.
Discriminating gas and water from oil is sim-pler than distinguishing between crude oil andOBM or SBM filtrate, because crude, OBM andSBM all absorb strongly at the oil peak near1725 nm. Fortunately, oils have different coloraccording to the quantity of large, complex aro-matic compounds they contain. This affectsabsorption in the MDT spectrometer in theshorter wavelength channels constituting thecolor region. Since SBM and OBM contain simplealiphatic compounds, their absorption in thesechannels is small.
In most cases, when the MDT tool first beginsdrawing fluid from a formation, the OD is highdue to light scattering off mudcake solids in thefluid. After a few seconds, the OD falls to a lowvalue, and then increases slowly as the mud fil-trate drains from the formation near the probeand is replaced by darker crude oil.
Particles of mudcake or other solid materialgenerate noise in the absorption channels.Scattering caused by these particles is wave-length-independent, so the effect can beremoved by subtracting a nearby channel. In thecolor region, absorption decreases quicklyenough that skipping a channel and subtractingfrom the next one down removes noise due toscattering without significantly affecting the sig-nal (left). The result is a smoothly varying con-tamination curve.6
The change in OD as reservoir crude replacesmud filtrate in the flowline follows the Beer-Lambert law, which states that a mixture of two
30 Oilfield Review
Fluid flowFluid flow
Gas refractometer
Optical density detectors
LampLight-emitting diode
Water
OilOilGas
> Optical detectors. Light passes through a sapphire window and reflects off the surface in con-tact with the fluid flowline into the gas refractometer. The reflection angle is set so gas reflectsmuch more strongly than oil or water. Another light path passes through a flowline into a seriesof filters to detect absorption, or optical density, in the visible and near-infrared spectra.
0.40
0.36
0.32
0.28
0.24
200 400 600 800Pumping time, sec
1000 1200 1400
OD
Channel 4
Channel 4 minus Channel 6
> Removal of scattering. To remove scattering from the OD signal, a nearbychannel at longer wavelength, which has less color absorption but the sameamount of wavelength-independent scattering, is subtracted. In this case, the signal from Channel 6 (not shown) is subtracted from Channel 4 (yellow)resulting in a data curve (red) that is fit to the OCM prediction (black).
50387schD03R1.p30.ps 11/17/01 8:31 PM Page 30
Autumn 2001 31
oils has an OD that is a linear, volumetricallyweighted combination of the two individual ODs, evaluated at each wavelength. A change in OD is directly related to a change in com-position (right).
Because most OBMs and SBMs mainly con-tain simple aliphatic compounds, their OD iseffectively zero except in the lowest MDT chan-nels. With one endpoint determined, quantitativeevaluation of contamination through OD requiresa method for finding the other endpoint—the ODof uncontaminated crude. This comes fromunderstanding the way fluids move duringcleanup. Fluid withdrawal through the probe cre-ates an expanding pressure sink around the well-bore.7 The OCM analysis fits the cleanup datawith a curve—having a specific shape based onthe physics of the tool and wellbore—to deter-mine the remaining amount of filtrate contamina-tion. In one well, five samples were captured inthe MDT tool at different times during cleanup.The laboratory results show contaminationresults consistent with the OCM model (above).8
5. Badry et al, reference 2.6. Mullins OC, Schroer J and Beck GF: “Real-time
Quantification of OBM Filtrate Contamination DuringOpenhole Wireline Sampling by Optical Spectroscopy,”Transactions of the SPWLA 41st Annual LoggingSymposium, Dallas, Texas, USA, June 4-7, 2000, paper SS.
Wavelength Pumping time
Optic
al d
ensi
ty
100% OBM
filtrate
100% crude
oil
OD1
OD2
OD3
OD4
OD5
OD at specific
wavelength
% O
BM c
onta
min
atio
n
OD1
00 100
0
∞
OD2
OD3
OD4
OD5
> Beer-Lambert mixing. Light absorption for crude oil (brown) is greater than for OBM filtrate (yellow)(left). The Beer-Lambert law says that the optical density (OD) of mixtures of the two (shades from yellow to brown) is related to the relative proportion of the two fluids. As the fluid cleans up, the ODincreases from the OBM value OD1 asymptotically to the crude-oil value OD5 (right).
OD
3.0
2.5
2.0
1.5
1.0500 2000 25001000 1500
Pumping time, sec Pumping time, sec
Sample Pumping time OFA contamination Laboratory contamination
12345
695 sec (12 min)940 sec (16 min)1264 sec (21 min)1681 sec (28 min)2250 sec (37 min)
17%13%12%9%8%
22%17%13%11%10%
0 1.0
3.0
2.6
2.2
1.8
1.410
20
30
40
50
Cont
amin
atio
n, p
erce
nt
Data and OCM fit
Acceptable contamination level
OD
Contamination
Optical density
> Quantitative prediction of contamination. Fluid samples were taken at five times during cleanup. Color channel data from the OFA module are fit using the OCM model (left) to determine contamination cleanup (right). The OCM prediction of contamination levels agrees well with laboratory contaminationmeasurement (table).
7. Hashem MN, Thomas EC, McNeil RI and Mullins O:“Determination of Producible Hydrocarbon Type and OilQuality in Wells Drilled With Synthetic Oil-Based Muds,”SPE Reservoir Evaluation and Engineering 2, no. 2 (April 1999): 125-133.
8. Mullins OC and Schroer J: “Real-time Determination ofFiltrate Contamination During Openhole WirelineSampling by Optical Spectroscopy,” paper SPE 63071,presented at the SPE Annual Technical Conference andExhibition, Dallas, Texas, USA, October 1-4, 2000.
50387schD03R1 11/29/01 3:31 AM Page 31
Like the other optical-detection bands, themethane channel of the LFA module displays ahigh OD as mud solids pass through a tool’s flow-line after pumping begins. Since drilling muds donot contain methane naturally, the initial highconcentration of filtrate drawn into the MDT toolduring cleanup results in a substantial drop in theOD recorded in the methane channel. As reser-voir fluid replaces filtrate in the line, the signalOD increases in proportion to the oil’s methanecontent, generating the same curve shape ascleanup with the OFA tool (left).
Time for complete cleanup cannot be pre-dicted before the logging run, because there aretoo many unknown reservoir variables. For exam-ple, there is not a direct relationship betweenformation permeability and cleanup time.Although fluid can be pumped quickly from ahigh-permeability formation, which would implya short cleanup time, that high permeability mayhave allowed mud filtrate to penetrate deeplyinto the formation before the wireline run. In thatcase, cleanup time could be long. Collecting flu-ids close to a shale stringer can shorten cleanuptime, since the shale provides a flow barrier,allowing collection of less contaminated reser-voir fluid farther away from the wellbore.
The ability of the OFA and LFA modules toquantify contamination levels while pumpingallows sampling decisions to be made in realtime. The OD for all channels is transmitted tosurface at high rate, and the OCM softwareupdates its analysis every 20 seconds. Once suffi-cient data have been acquired, the softwareselects the color channel that will provide thebest fit to the expected trend and shows thedegree of contamination and the time required toachieve an acceptably low level of contamination.
In a Gulf of Mexico well, the MDT probe wasset within a massive sand, and the tool measureda mobility of 87 millidarcies per centipoise(mD/cp). After pumping for 71 minutes, the OCMsoftware predicted an additional 41⁄2 hours pump-ing time to achieve an acceptable level of 10%contamination (left). Rather than wait or waste asample bottle on highly contaminated fluid, theoperator chose to move to another level withinthe same formation.
The tool was moved 44 ft [13 m] lower in theformation. The mobility was higher, 256 mD/cp.Contamination dropped to 9% within 132 min-utes, and samples taken at this location wereacceptable for PVT analysis (next page, bottom).
32 Oilfield Review
1.00 0.21
0.22
0.23
0.24
0.25
0.26
500 1000
Colo
r OD
Met
hane
OD
1500 2000
Pumping time, sec
3500 3000 3500
1.05
1.10
1.15
1.20
1.25Color channel
Methane channel
> Contamination prediction in a Gulf of Mexico well. After noise is removedfrom an LFA color channel (red) and the methane channel (blue), each dataset is fit to the OCM prediction (smooth curves). For this sample, the colordata predict 4.9% contamination and the methane data predict 6.2%. The 5.5%average agrees with 4.3% contamination from a GC in the laboratory.
End
fit
100
90
80
70
60
50
Cont
amin
atio
n, %
Colo
r OD
Optic
al d
ensi
ty
40
30
20
10
00.00
0.25
0
5
4
3
2
1
0.50
0 500 1000 1500 2000
Pumping time, sec
2500 3000 3500 4000
0.75
1.00
1.25
1.50
1.75
2.00
Star
t fit
8
9
7
65
43
21
4000
3000
2000
1000
0
Pumping
tim
e, sec
Channel num
ber
> Avoiding long cleanup times. After the MDT tool had pumped from the for-mation for about an hour, the OCM software indicated about 18% contamina-tion (blue curve), and an additional 41⁄2 hours to achieve less than 10% con-tamination. The inset shows the OD measurement for Channels 0 through 9(shaded green). Channel 4, which has the greatest change in OD duringcleanup, was used for the fit after subtracting Channel 6 to remove scatteringfrom large particles (red curve). The vertical dashed lines at the left and rightof the plot indicate the range over which the OCM method fit the data.
50387schD03R1.p32.ps 11/17/01 8:31 PM Page 32
Autumn 2001 33
Scattering LightScattering from particles smaller than the incident-light wavelength—several hundred nanometersdiameter—depends on the wavelength of inci-dent light. The intensity of this scatteringincreases with decreasing wavelength. Thiseffect, called Rayleigh scattering, gives the skyits blue color.
Wavelength-independent scattering is removedby channel subtraction, but this leaves some wavelength-dependent Rayleigh scattering. For the OCM-color procedure, a longer-wavelength channel is subtracted, but for the OCM-methane procedure, the sub-tracted channel is at a shorter wavelength. Sinceone procedure slightly overcorrects for wave-length-dependent scattering and the otherslightly undercorrects, averaging OCM-color and OCM-methane contamination values fromthe LFA tool tends to remove some of that scattering effect (right).
Discrepancies between the contaminationdeterminations indicate the need to look moreclosely at other channels to identify the causebefore collecting a fluid sample. Methane detec-tion has been shown to be valid for fluids withGOR as low as 700 scf/bbl [126 m3/m3].9
However, in reservoirs containing oil with lowmethane content, color channels may providebetter information on contamination than themethane channel does. For gas-condensate flu-ids, methane detection using the LFA module isessential, because even in the shortest wave-length color channels, OD remains low and theprogression of cleanup using the OCM-color pro-cedure is difficult to assess. In some cases, adrilling-mud filtrate may be darker than the con-densate, and the OCM-color procedure may notbe able to discriminate contamination from reser-voir fluid. The OCM-methane detection in thenew LFA module works well in such cases.
Comparing Contamination at SurfaceSamples are collected to determine properties ofreservoir fluids such as PVT behavior. Mud filtratemixed in the sample must be accounted for toarrive at reasonable estimates of reservoir-fluidproperties. The OFA and LFA modules measure
9. Mullins O, Beck GF, Cribbs M, Terabayashi T andKegasawa K: “Downhole Determination of GOR onSingle-Phase Fluids by Optical Spectroscopy,”Transactions of the SPWLA 42nd Annual LoggingSymposium, Houston, Texas, USA, June 17-20, 2001,paper M.
Cont
amin
atio
n, %
Colo
r OD Op
tical
den
sity
0
10
20
30
40
50
60
70
80
90
100
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0
1
2
3
4
5
0.7
0.8
0.9
1.0
1.1
1.2
0 1000 2000 3000 4000 5000 6000 7000 8000
Pumping time, sec
7
9
8
65
43
21
7500
2000
1000
0
Pumping
tim
e, sec
Channel num
ber
End
fit
Star
t fit
> Obtaining acceptable samples. After about two hours of pumping, contamina-tion had dropped to about 9% (blue curve). OD for all channels is shown in the inset (shaded green). The OCM model was fit to data of Channel 4 minusChannel 6 (red curve) between the start- and end-fit lines (green dashed lines).The increases in OD past the end-fit line occurred during sample collection.
0.16
0.18
0.20
0.22
0.24
500 1000 1500
Colo
r OD
Met
hane
OD
2000 2500 3000 35000.60
0.65
0.70
0.75
0.80
0.85
0.90
Methane channel
Color channel
Pumping time, sec
>Wavelength-dependent scattering. The optical absorption response in the pumping period between 1000 and 1500 seconds indicates some scatter-ing remains even after subtracting a baseline channel. This wavelength-dependent response is stronger in the color channel (purple) than in themethane channel (blue). The noise in the data after 2500 seconds occurredduring sample collection. The OCM method was still able to fit the data, predicting 7% contamination based on the average of color and methane data of 7.9% and 6.0%, respectively.
50387schD03R1 11/29/01 3:32 AM Page 33
contamination in real time before collecting sam-ples. At the rig floor or in a laboratory, samplecontamination can be analyzed further using agas chromatograph (GC), a gel-permeation chro-matograph (GPC), tracer analysis or, less com-monly and not discussed here, a nuclearmagnetic resonance (NMR) spectrometer.
In a GC, a small quantity of sample fluid isinjected into a carrier gas such as high-purityhelium. Light gaseous components are separatedusing a molecular sieve and heavier componentsare separated using a packed chromatographiccolumn. A molecular sieve relies on particle sizefor separation, with smaller molecules staying inthe sieve longer. In a packed column, the gasflows past particles coated with a fluid, termedthe stationary layer in a GC because the gas does not mobilize it. The relative solubility ofcomponents in the stationary layer separatesthem as the carrier gas moves a sample throughthe column. Chromatographs are calibrated forsample components.
The process is similar for a GPC except theinert carrier is a liquid, and constituents do notseparate as well at the detector. Componentpeaks from a GC are typically distinct, but thosefrom a GPC can be smeared together. TheOilphase FFA Field Fingerprint Analyser rigsitedevice incorporates a GPC.
At the end of the column, the carrier gas or liquid containing the sample enters a detector. For hydrocarbons, this is usually a thermal-conductivity detector or a flame-ionization detector. Some detection methods respond tomass and others to the number of carbon atoms inthe molecule.
The distribution of crude-oil constituents nor-mally declines smoothly with increasing carbon
number beyond eight.10 OBM and SBM filtratecontamination causes this distribution to deviatefrom the expected shape. SBMs use a narrowrange of molecular weights, so contaminationcan be discerned with both a GC and a GPC as asharp increase in the frequency of moleculesbetween carbon numbers of C14 to C18 (above).OBMs with a mineral-oil base include a broaderrange of compounds, perhaps ranging from C8 toC20, and are difficult to distinguish using a GPC.Often, these muds can be separated from thecrude-oil signature when using a GC. Drillingmuds that include produced reservoir oil cannotbe distinguished from formation oil using eitherform of chromatography, unless a tracer is addedto the mud.
An OBM or SBM filtrate response also can beremoved from the GC result by separately measur-ing the response of the filtrate, normalizing thetwo signals and subtracting.11 Drilling-mud compo-sition must be maintained while drilling an open-hole section before sampling because variations inmud composition add error to the analysis.
Sometimes, contamination is measured usingtracers, by tagging drilling mud with an isotopeor a molecule that is not present in high concen-tration in reservoir oils. For isotopic tagging ofhydrocarbons, 13C replaces 12C, or deuteriumreplaces hydrogen. Mass spectroscopy measuresthe concentration of an isotope in a reservoir-fluid sample to determine contamination.Detected isotope concentrations must be higherthan those found naturally for this procedure towork. Chemical tagging may use linear alphaolefins, detected using a GC.
Tagging is an expensive procedure that mustbe planned in advance. The isotope or chemicaltag must be in the mud in a constant concentra-tion before drilling into the zone of interest and
must remain in the mud until samples are taken,since all drilling mud that filters into the forma-tion must be tagged to have a meaningful result.Chemical tagging has an added problem: theselected molecules may not behave like reservoircrude. For example, linear alpha olefins are lessstable at high temperature than the correspond-ing alkanes, and may not travel through porousmedia at the same rate.
Results of several contamination-measure-ment techniques have been performed at Hebronfield offshore Newfoundland, Canada, and in Gulfof Mexico wells.12,13 At Hebron field, the syntheticdrilling mud was tagged with deuterium. Fluidsamples from five different zones were collectedusing the OFA module. The OCM-color procedureevaluated contamination while fluid was pumpedfrom the formation. The Oilphase FFA devicedetermined contamination using a GPC at therigsite. Isotope tag concentration was deter-mined using mass spectroscopy, and a laboratoryGC determined the constituents of the fluid.
The LFA module including the OCM analysiswas compared with laboratory GC analysis onlive oils from several Gulf of Mexico wells. Inboth this study and the Hebron field study, thereal-time LFA or OFA measurements generallyagree with the isotope, GC and FFA results(next page, top).
Some discrepancy between methods isexpected, as all methods have potential errors.The FFA device can overestimate contamination ifthe mud is not synthetic; even with SBM, both theFFA results and GC methods assume a distribu-tion of hydrocarbon constituents to determinecontamination. Tagging is expensive and in princi-ple can be accurate, but in practice, it may notobtain reliable results. It is difficult to ensure thatall the drilling mud has a uniform concentration ofthe chemical or isotopic tag and that the taggedmolecules have the same physical and transportproperties as the rest of the filtrate. The OCM-color method has problems when the mud filtratehas significant color or the reservoir oil is color-less, because the method requires a contrastbetween the two. However, the LFA-OCM-methane method provides a solution for suchcases, since it is based on methane concentration.
Even if contamination-detection methodsalways were correct, many errors can occur incollecting samples. The fluid can go through a phase transition as it is drawn into the tool,leaving components behind in the formation, orphases can separate in the tool. Valves can fail,either not opening properly downhole and cap-turing insufficient fluid or not closing completelyand losing pressure and fluid after sample collec-tion. At surface, every time the fluid is transferred
34 Oilfield Review
Oil 1Oil 2
C2 C4 C6 C8 C10 C12 C14 C16 C18 C20 C22 C24 C26 C28 C30+
Trend for Oil 13.1% contamination
Component
0.1
1.0
10.0
100.0Co
ncen
tratio
n, m
ole
%
> Removing contamination. GC results indicated Oil 1 (blue) and Oil 2 (red) from neighboring wells hadsimilar profiles except for the contamination of C16 and C18 from synthetic drilling mud. Contaminationcan be removed by developing the trend line for Oil 1 and decreasing the concentrations of C16 and C18to the trend level. This analysis confirms that the oils came from the same source rock.
50387schD03R1.p34.ps 11/17/01 8:31 PM Page 34
Autumn 2001 35
or a sample bottle is handled, there is potentialfor damaging the sample. Bottles should beheated and agitated for about five days beforeperforming laboratory analyses, but not all labo-ratories follow this recommended procedure.Collecting the right base oil of the drilling mud—used to compare with spectra of contaminatedreservoir oil—is difficult because mud composi-tion often changes during a job as componentsare added to control various drilling problems.
Collection and analysis of fluid samples areimportant; operators must control sources oferror to obtain the best possible data. The OFAand LFA procedures measure properties down-hole in real time before sample collection, a distinct advantage. The few sample bottles avail-able on the tool are not wasted storing bad samples. Since OCM measurements are madebefore any possible transport and handling problems, they provide a check for the quality oflater measurements.
When sufficient information is available fromthe reservoir, measured values of fluid propertiescan be used as an additional check on samplequality. Norsk Hydro conducted a detailed studyof oil samples taken from several North Seafields.14 In a reservoir with a gas cap, both chem-ical tags and the FFA device indicated a highlevel of sample contamination, ranging from8.9% to 25.8%. The OFA-OCM method and GCanalysis indicated lower contamination levels of2.6% to 6.8%. The difference in these tworanges of contamination measurement led NorskHydro to investigate further.
Reservoir saturation pressure, Psat, at thesampling depth was estimated from reservoirpressure and density gradients starting at the gas-oil contact (right). The reservoir satura-tion pressure of the sample, based on PVT
10. Gozalpour F, Danesh A, Tehrani DH, Todd AC and Tohidi B:“Predicting Reservoir Fluid Phase and VolumetricBehaviour from Samples Contaminated with Oil-BasedMud,” paper SPE 56747, presented at the SPE AnnualTechnical Conference and Exhibition, Houston, Texas,USA, October 3-6, 1999.
11. MacMillan DJ, Ginley GM and Dembicki H: “How toObtain Reservoir Fluid Properties from an Oil SampleContaminated with Synthetic Drilling Mud,” paper SPE38852, presented at the SPE Annual TechnicalConference and Exhibition, San Antonio, Texas, USA,October 5-8, 1997.Gozalpour et al, reference 10.
12. Connon D: “Chevron et al. Hebron M-04 ContaminationPrediction Method Comparison,” Released ProjectReport available at Canada-Newfoundland OffshorePetroleum Board, St. John’s, Newfoundland, Canada,May 1, 2001.
13. Mullins et al, reference 9.14. Fadnes FH, Irvine-Fortescue J, Williams S, Mullins OC
and Van Dusen A: “Optimization of Wireline SampleQuality by Real-Time Analysis of Oil-Based MudContamination—Examples from North Sea Operations,”paper SPE 71736, presented at the SPE Annual TechnicalConference and Exhibition, New Orleans, Louisiana,USA, September 30-October 3, 2001.
5c
5b
5a
4b
4a
3c
3b
3a
2b
2a
1c
1b
1a
Sam
ple
num
ber
0 20 40Contamination, %
60 80 100
Gas chromatographOFA measurementFFA analysisIsotopic tag determination
11
12
13
025
045
150
155
157
160
Sam
ple
num
ber
0 20 40Contamination, %
60 80 100
Gas chromatographLFA-color measurementLFA-methane measurement
> Comparing different methods of evaluating contamination. Contamination measurements of fluidsamples from Hebron field (left) and Gulf of Mexico wells (right) indicate agreement among differentmethods for most samples.
True
ver
tical
dep
th, m
2700
2725
2750
2775
2800270 275 280 285
Pressure, bar absolute
290 295 300 305
Gas-oil contact
Sample gas-loss correction
Saturation-pressure
gradient
Pressure gradient
Contaminatedsample
Removalof 3%
contamination
Removal of 9%contamination
> Using reservoir properties to evaluate contamination measurements. The known gas-oil contact,pressure gradient (blue line) and saturation-pressure, or bubblepoint, gradient (green line) intersect atthe gas-oil contact for a North Sea well. The contaminated sample had a bubblepoint pressure ofabout 272 bar [27.2 MPa or 3950 psi] (dark brown). PVT modeling allowed prediction of bubblepointpressure of uncontaminated oil by mathematically removing measured contamination from the sample.Removing 9% contamination, measured using isotopic tagging and the FFA procedure, produced anunphysical result above the reservoir values (purple). Removing only 3% contamination (dark blue),based on the OFA-OCM result, did not raise the bubblepoint enough. Assuming the sample bottle deadvolume of 2.5% was all lost gas provides another factor to adjust the PVT properties of the contami-nated sample (light brown). Combining the 3% contamination correction with the gas-loss correctionbrings the prediction of bubblepoint (light blue) close to the saturation-pressure gradient.
50387schD03R1.p35.ps 01/10/2002 03:55 PM Page 35
properties determined with contaminants in thefluid, was about 20 bar [2 MPa or 290 psi] belowthe saturation-pressure gradient at the samplingdepth. These PVT properties can be mathemati-cally corrected to remove the effect of contami-nants and then compared with the reservoir-gradient calculation.
When using the FFA contamination value of9%, the resulting calculated Psat was greaterthan the reservoir pressure, which is an unphysi-cal result. When using the OFA-OCM value ofcontamination, Psat was about 10 bar [1 MPa or145 psi] below the expected value. This indicatesthe sample may have lost gas before PVT proper-ties were evaluated. Gas could separate from liq-uid in the formation due to near-well pressuredrawdown, but the downhole conditions werenot known well enough to evaluate this effect.The investigation focused on what happened tothe sample coming out of the hole.
The sample bottle did not allow for downholepressure compensation. The fluid could enter thetwo-phase region due to cooling from the reser-voir temperature of 107ºC [225ºF] during trans-port to surface. The sample had probably cooledbelow 102ºC [217ºF]—the temperature at whichpressure in the enclosed chamber decreasedbelow the bubblepoint—and was in two phasesby the time it reached the surface. The 450-cm3
bottle has a dead volume of 2.5% between theisolating valve on the bottle and the valve on thedownhole flowline, which could have been filledwith gas that was lost when the valves wereopened at the surface. The PVT properties of the contaminated samples can be corrected forthis gas loss, increasing the contaminated-sample bubblepoint pressure by 10 bar. Whenthe gas-loss correction is combined with removalof the contamination, as measured by the OCM-color method, Psat increased to within 4 bar
[0.4 MPa or 58 psi] of the expected reservoirvalue, which is reasonable agreement. This anal-ysis could not have been performed without thedownhole OCM-contamination measurement.
Monitoring Gas DirectlyGas-condensate fields engender additional diffi-culties for fluid sampling when OBM and SBMare used. Although they contain single-phase flu-ids in the reservoir, gas condensates separateinto a gas phase and a liquid phase when condi-tions drop below the dewpoint. The liquidderived from gas condensates is a more valuablecommodity than the gas. Surface-separator con-ditions are tuned to optimize the volume andvalue of liquid obtained from condensates. Theseparator designs often are based on fluid prop-erties from wireline samples, so determining thelevel of contamination and correcting the PVTproperties are essential.
OBM and SBM filtrate may mix only partiallywith condensate in a reservoir, leaving mud fil-trate in a liquid-hydrocarbon phase and a gasphase with some of the more volatile compo-nents of the filtrate. A wireline sampling probedraws both hydrocarbon phases into the device,and samples collected contain both reservoirfluid and filtrate contamination. When the fluidpressure is lowered during laboratory testing, thephases separate. All mud filtrate is concentratedin the liquid phase; presence of contaminationstrongly affects a sample’s dewpoint pressure.
To calculate a correct GOR and other reser-voir-fluid properties, the volume of the oil phasemust be adjusted to remove contamination. Thatliquid-phase contamination must be kept low toavoid excessive correction factors, just as with ablack oil. However, to compensate for the con-centration of SBM and OBM contaminants in theliquid phase, many companies set the acceptablelevel of contamination in a gas condensatebelow that for a black oil. The LFA tool providessignificant new information for gas-condensatereservoirs, improving data quality used fordesigning production facilities.15
A gas-condensate prospect in the NorwegianNorth Sea offered one of the first tests for the LFAtool, used in this case without the OCM module.16
A mobile C36+ GC, capable of measuring individualconstituents up to C36 at the rigsite, indicatedcontamination of 32% to 60% in the low-pressureliquid phase. This was comparable to results fromsubsequent FFA analysis onshore. The LFA time-sequence data were later analyzed using the
36 Oilfield Review
Pumping time, sec
00.00
0.25
0.50
0.75
0.1
0.2
0.3
0.4
Met
hane
OD
Cont
amin
atio
n, %
Colo
r OD
Star
t fit
End
fit
1000 2000 3000 4000 5000 6000 7000 8000 90000
25
50
75
100
10,000
> Cleanup curve for gas condensate. This North Sea gas condensate wastransparent. Even the OD of the shortest wavelength channel (top) showedinsufficient contrast to reliably determine OD buildup using the OCM-colormethod (red). Cleanup was more reliable in the methane channel (bottom)(pink), giving a more quantifiable OCM-methane fit to the OD data (black). Calculated values of contamination are shown in the lower plot, with anOCM-color curve (green), an OCM-methane curve (purple) and the average of the two (blue). In this case, the large discrepancy is caused by the lightcolor of the condensate.
50387schD03R1.p36.ps 11/17/01 8:31 PM Page 36
Autumn 2001 37
OCM method. Mud filtrate and the reservoir fluidwere indistinguishable in the color channels. The OCM-methane analysis provided a quantita-tive contamination measurement, about 8% of the live oil (previous page). The operator had littleexperience with the new tool and sought to under-stand the difference.
A subsequent well test proved a gas-conden-sate find. Surface-separator samples collectedduring the flowing well test and analyzed using aC36+ GC indicated stock-tank oil contamination of23%. A full PVT analysis provided the GOR,allowing correction of contamination to single-phase, downhole conditions. The result indicated6 to 7% contamination, which agreed well withthe OCM-methane measurement on the live fluid.
During determination of fluid properties for agas-condensate reservoir drilled with OBM, thebuildup of methane measured with the LFA moduleis essential to obtain accurate, real-time conden-sate-contamination measurement. The alternativesare to conduct a DST or complete a well withwater-base mud to avoid oil contamination altogether. Moreover, using the LFA device alsoprovides a simultaneous measurement of GOR.
The gas refractometer on both the OFA and LFA tools indicates gas only when it is in con-tact with the detector window. Gas bubbles maynot be detected if they are in the center of the
flowstream, or on the opposite side. The refrac-tometer detects all gases, regardless of composi-tion, so CO2 and H2S are flagged.
The LFA module also provides a complemen-tary gas-detection system using measurement ofOD in the methane channel. Although insensitiveto other gases, this detector monitors allmethane passing through the flowline. If live oilis flowing, the volume percentage of methanewill be low. However, if the pressure drops belowthe bubblepoint, gas evolves and methaneabsorption will be high when a bubble passes thelight beam anywhere within the flowline. Thecombination of the gas refractometer andmethane detector makes a robust LFA gas-detec-tion method (right).
The ratio of the methane peak to the oil peakin the LFA module correlates with GOR both formixtures of pure components and for live crudeoils (above). A multiplying factor applied to themethane-heptane mixtures compensates forother hydrocarbon components in the gas phaseof reservoir oils. The tool does not measure CO2 orH2S, so the LFA-GOR measurement may be incor-rect for fluids from reservoirs containing signifi-cant quantities of these nonhydrocarbon gases.
15. Mullins et al, reference 9.16. Fadnes et al, reference 14.
504
522
540
558
576
594
612
630
648
666
684
702
720
738
756
774
792
810
828
846
864
882
900
918
936
954
972
990
1008
1026
1044
1062
1080
1098
1116
1134
1152
1170
1188
1206
1224
1242
1260
1278
1296
1314
1332
1350
1368
1386
1404
1422
1440
1458
1476
1494
1512
GasFlag
Tim
e, s
ec
OilFlag
Fluid Color OD Channels Methane Channel
> LFA gas-detection combination. After an initialcleanup period, the color Channels 1 through 5 inTrack 4 show little absorption, confirming a gascondensate. Channels 6 and 9 also have low OD,which means no water is present. The oil peak inChannel 8 is transformed into an oil flag in Track 2(green), indicating periods when no oil flows, particularly from 1116 to 1188 seconds and 1422 to 1458 seconds. The gas refractometer in Track 1(red) measures all gases, but only when they contact the sapphire window of the refracto-meter. It misses some periods of gas flow. The LFAmethane response from Channel 0, expanded inTrack 5, is sensitive to all methane in the flowline,but not to other gases. The combination of the twogas detectors is more robust than either alone.
0.2
0.3
0.4
0.5
0.1
00 1000 2000 3000
GOR, scf/bbl
Ratio
of m
etha
ne p
eak
to o
il pe
ak
4000 5000 6000
Live oilsBinary mix*0.85Dead oil from GOR run
> GOR measurement derived from molecular vibration peaks. In laboratorytests, the ratio of absorption at the methane peak to the oil peak fits well withGOR for both methane-heptane mixtures (red squares) and live oils (blue circles). The multiplicative factor applied to the methane-heptane mixturesaccounts for the absence of other gases normally present in live oils. The deadcrude oil (orange triangle) was evaluated after gas was removed in the laboratory.
50387schD03R1.p37.ps 11/17/01 8:32 PM Page 37
Real-Time Fluid TypingThe combination of the MDT system and theCMR Combinable Magnetic Resonance toolrevealed new insights about a reservoir operatedby Shell in the Gulf of Mexico. The Yellow sandunit had been depleted for two years. The newdrilling target was an underlying sandstone for-mation, called the Blue sand, separated from theoverlying reservoir by a thick shale.
A logging-while-drilling (LWD) resistivity logrevealed a 10-ft [3-m] water layer on top of theBlue sand oil, which is not a gravitationally stablesituation. A thin hydrocarbon layer sat atop thewater, just below the thick shale (below). Theoperator wanted to know whether water fromabove had broken through. The LWD gamma raylog and standard CMR processing did not explainhow this water could be above the oil (right).Pressures collected with an MDT tool indicatedthat the water zone was not in pressure commu-nication with either the Yellow sand above or theBlue sand below. Reservoir pressure in the waterzone was about 800 psi [5.5 MPa] higher than theBlue sand, and was slightly less than originalreservoir pressure for the Yellow sand.
The depleted Yellow sand placed a limit onthe mud weight that could be used in the bore-hole. This created concerns about the wellbore;the well was not stable enough to leave the MDTtool in place long enough for formation fluid toclean up. The MDT tool was used instead for fluidtyping with the gargling technique developed byShell Deepwater Services.17 In this technique,reservoir fluids from the formation were pumpedfor a short period of time through the OFA mod-ule and out to the wellbore, without collectingsamples in bottles. An OD spectrum from the OFAmodule allowed analysis of these small quanti-ties of reservoir oil. Since oil color relates to APIgravity and GOR, the color pattern from the 10OFA channels enabled discrimination betweenthe oils. In this case, the Yellow sand was a gascondensate with an API gravity of about 40º and
a GOR of 6000 scf/bbl [1080 m3/m3], while theBlue sand held a 35º API gravity oil with a GOR of2000 scf/bbl [360 m3/m3]. Surprisingly, the colorspectrum of the hydrocarbon sitting on top of thewater had the same signature as the Yellow sandabove the thick shale.
The CMR log data were reprocessed toimprove resolution from 18 in. [46 cm] to about8 in. [20 cm], revealing a thin permeability barrierat the base of water, thought to be about 6-in.[15-cm] thick. This led to a rethinking of the dis-tinction between the top and bottom units. Inother wells, the Yellow sand remained above thelarge shale, but in this well, a splinter member ofthe Yellow sand cut below the shale. The trueboundary between the zones was the thin barrier,which appeared to be sand on sand, undifferenti-ated on conventional logs.
38 Oilfield Review
Depletedcondensatein Yellowsand
Shale
Splinter ofYellow sand
Target oil inBlue sand
> Section of Yellow sand below shale. The Yellowsand above the shale is saturated with a conden-sate. The oil-saturated Blue sand did not extendto the shale, but stopped at a thin barrier (thickblack line). The splinter of the Yellow sand unitbelow the thick shale had a water leg (blue) belowa thin layer of condensate.
Pumpout at X61635° API gravity
Pumpout at X59740° API gravity
Pumpout at X64035° API gravity
Pumpout at X48240° API gravity
LWD CMR Standard-Resolution CMRHigh-Resolution
Gamma Ray Resistivity
MDT
Pressure
MDT-OFA
Fluid TypingPermeabilityPermeability Porosity
Blue
sand
Yello
w sa
ndSh
ale
NMR T2
>Water above oil investigated by MDT-OFA fluid typing. There is a water zone over the oil-saturatedBlue sand, located at the blue arrow pointing to the low resistivity in Track 2. The responses in thegamma ray and standard-resolution CMR logs do not explain how this water zone can sit atop oil. Areprocessed high-resolution CMR permeability log (Track 6) shows a thin permeability barrier, indi-cated by the green arrow. MDT logging shows three pressure compartments: the depleted Yellowsand above the shale, the Blue sand below the barrier and the region between the shale and the thinbarrier at high pressure. The MDT color channels, evaluated at the depths indicated by the blackarrows, were used to type the reservoir fluids. The oil atop the water above the barrier has the samecharacteristics as the oil in the Yellow sand. This caused the operator to reevaluate the boundarybetween the Yellow and Blue sands in this well as being at the thin barrier rather than the thick shale.
50387schD03R1 11/29/01 3:35 AM Page 38
Autumn 2001 39
Had this been an exploration well, facilitiesplanning would have relied on results from fluidsampling. Depending on where the samples werecollected, the GOR could have been too high ortoo low, leading to an inefficient design. If thesample GOR measured were lower than theactual production, the facilities would have anundercapacity for gas production, and insufficientcompression and transmission capabilities,resulting in lost or delayed revenues. Significanterror in GOR in the opposite direction could havethe opposite problem—an expensive overdesignwith too much capacity. MDT fluid typing is avaluable means for detecting such situations.
In a well in the North Sea, Norsk Hydro drilleda pilot hole through three horizons prior to drillinga horizontal section.18 The typical log response inthis field made distinguishing the fluid type ineach formation difficult. Precise definition of fluidcompositions was not required, but rapid differ-entiation of gas, oil and water was imperativebecause the rig was idle while the operatorawaited this fluid identification. The operatorwanted to drill a horizontal wellbore into thedeepest oil-bearing zone. The MDT sonde waschosen to identify the fluids in real time.
Pumping fluid into the tool progressed untilthe OFA-OCM method indicated contaminationhad dropped below 8% in the middle zone and to1% in the upper zone. The MDT tool indicatedthat the lower zone was water-filled. The lowcontamination values in the other zones gave theoperator confidence in the tool response, whichshowed that the reservoir fluid was oil. A 3%-olefin tracer placed in the OBM mud beforedrilling the section allowed rapid confirmation ofthese contamination values using a GC at the rig.The surface contamination measurements—5% in the middle zone and 4% in the upper—provided reasonable agreement with the OFA-OCM measurement.
Although additional fluid samples had beencollected for testing onshore, the real-timeresults using the OFA-OCM analysis coupled witha rigsite GC confirmation provided answers thatwere conclusive enough to cancel the onshoretesting program. The horizontal section wasdrilled into the middle horizon immediately aftercompleting the MDT run, resulting in a success-ful well.
Norsk Hydro no longer uses olefin tracers totag drilling mud. Recent wells have relied suc-cessfully on the combination of the OCM methodand a C36+ GC.
Fluid Compartments in Hibernia FieldThe Hibernia field, discovered in 1979 and oper-ated by Hibernia Management and DevelopmentCompany, Ltd. (HMDC), was the first significantoil discovery in the Jeanne d’Arc basin on theGrand Banks of Newfoundland, Canada. Oil pro-duction commenced on November 17, 1997, froman ice-resistant, gravity-based platform in 80 m[262 ft] of water, 315 km [196 miles] east-south-east of St. John’s, Newfoundland (above).
The structure is a highly faulted, south-plung-ing anticline containing approximately 3 billionbarrels [475 million m3] of oil-in-place, with anestimated 750 million recoverable barrels[120 million m3]. Most of these resources are intwo Lower Cretaceous reservoirs, the Hibernia,and the combined Ben Nevis and Avalon sand-stones. The Hibernia reservoir will be depleted
17. Hashem et al, reference 7.18. Fadnes et al, reference 14.
2000 m
100 km62 miles
200 m
NEWFOUNDLAND
Atlantic Ocean
Water depth
Rift-basinoutlines
St John’sHibernia field
C A N A D A
> Hibernia field, offshore Newfoundland, Canada.
50387schD03R1 11/29/01 3:36 AM Page 39
using both waterflood and gasflood processes(above). Delineation drilling of the Ben Nevis andAvalon formations continues; these reservoirswill be produced under waterflood.
HMDC encountered operational problemswhile drilling the first four wells using WBM.Shifting to OBM resulted in improved boreholeconditions, few seal losses while running thelogs and decreased logging-acquisition time.
Extensive faulting makes reservoir continuityuncertain. Early in field development, HMDC ini-tiated a comprehensive data-acquisition plan todetermine fluid compositional variation betweenfault blocks and within the vertically extensivefluid column. Obtaining high-quality sampleswith the MDT tool is an integral part of the program for determining reservoir-fluid proper-ties. MDT pressure measurements establishpressure gradients and locate gas-oil and water-oil contacts.
Fluid samples were collected in three ways—MDT samples, bottomhole samples and separa-tor samples. The MDT string typically wasconfigured to obtain approximately 30 pressurepoints across selected reservoir intervals andincluded six MPSR sample bottles. Several wellswere sampled using 12 sample cylinders: six
40 Oilfield Review
Bonavistaplatform
Mur
re fa
ult
Nautilus fault
N
0
0 1 2 3 miles
1 2 3 4 5 km
> Hibernia water- and gasfloods. The 3D image indicates some of the oil-production (green), water-injection (blue) andgas-injection (red) wells in the highly faulted reservoir (left). The structure map shows distinct fault blocks in the Hiber-nia formation (right). Part of the field is under waterflood (blue) and part under gasflood (red). The section line (black)indicates the location of the cross section shown on page 42.
3550
3600
3650
3700
3750
3800
3850
3900
3950
4000125 175 225
GOR, m3/m3
Dept
h, m
275 325 375
B-16 5 MDT 2B-16 5 MDT 3
B-16 6 MDT 3
B-16 3 MDT 4B-16 2 BHSB-16 6 MDT 1
B-16 1 BHSB-16 1 BHS
B-16 3 BHS
B-16 3 BHSB-16 3 BHS
B-16 1 BHS
B-16 3 MDT 3
B-16 9 MDT 6C-96 DST 4 BHS
C-96 DST 3 BHS
C-96 DST 1 BHS
B-16 9 MDT 3
B-16 11 MDT 6
B-16 7 MDT 2
B-16 7 MDT 3
> Hibernia GOR. Fluid samples from the MDT tool and from bottomhole samples(BHS) from DSTs indicate the trend of GOR with depth. Separator samples fromHibernia are not associated with a specific depth and are not shown here. (225 m3/m3 = 1249 scf/bbl.)
50387schD03R1 11/29/01 3:37 AM Page 40
Autumn 2001 41
MPSR cylinders and six pressure-compensatedSPMC cylinders. The variation of PVT propertiesin the MDT samples helped define depth andareal trends, which were further refined by geo-chemical fingerprinting of the samples. MDTdetection of OBM contamination was importantfor the program. Use of OCM real-time monitor-ing allowed collection of high-quality gas-condensate samples.
Initially, bottomhole samples from the entireperforated interval were collected during produc-tion testing to obtain representative PVT proper-ties. Single-phase flow conditions weremaintained downhole during sampling. Fluidsamples collected from test separators were lessexpensive, allowing continued monthly samplingto monitor compositional changes. Samples fromthe three sources have shown excellent agree-ment in PVT studies and determination of OBM-contamination levels (previous page, bottom).
The operator uses PVT data from thesesources for well-test analysis, reserves determi-nation, material balances, reservoir simulation,production allocation, production monitoring andfluid-metering factors, process simulation andregulatory reporting.
The initial pressure in the Hibernia reservoirwas approximately 40 MPa [5800 psi]. Becausethe bubblepoint varies across the field, the company avoided sampling below bubblepointpressure. The MDT tool monitored pressure during sampling, allowing minimal drawdownand accurate bubblepoint determination fromrecovered samples.
The OFA module detected sample contamina-tion levels to estimate pumping time to achievecleanup. About halfway through the sample-col-lection program, the OCM option became avail-able, providing a quantitative measure of
contamination in real time. The OFA results fromthe previous logging runs were analyzed laterusing the OCM-color methodology to determinecontamination levels (above).
The MDT sampling tool is an effective meansof collecting representative fluid samples to eval-uate variations through long fluid columns. TheHibernia group has successfully run the tool onwireline, but because of wellbore deviations upto 80°, the tool typically was run as part of a TLCTough Logging Conditions superstring. The TLCtool usually includes the Platform Express inte-grated tool, including the AIT Array InductionImager Tool sonde, a caliper and gamma ray tool,and the MDT modules. Logs collected on a firstpass were transmitted in real time to the com-pany office in St. John’s where engineers pickedpoints for MDT pressure determination and asample-collection pass. With fluid columns inexcess of 400 m [1300 ft] thick in areas of the
5-2
5-3
5-4
5-5
5-6
8-1
8-3
8-5
11-5
11-6
12-1
12-2
12-3
12-4
12-5
12-6
14-2
14-3
Sam
ple
num
ber
0 20 40 80 10060Oil-base mud contamination, %
Sam
ple
num
ber
0 20 40 80 10060Oil-base mud contamination, %
20-6
20-5
20-4
20-3
20-2
20-1
19-1.11
19-1.10
19-1.09
19-1.08
Sam
ple
num
ber
0 20 40 80 10060Oil-base mud contamination, %
OFA-OCM measurementGas chromatograph
Gasflood Wells Waterflood Wells Ben Nevis and Avalon Wells
6-1
6-3
6-4
6-5
6-6
7-2
7-3
9-1
9-3
9-5
16-1
16-2
16-4
16-5
16-6
17-1
17-2
17-3
17-4
17-5
17-6
OFA-OCM measurementGas chromatograph
OFA-OCM measurementGas chromatograph
> Comparison of contamination measurements. The OFA-OCM measurement at the wellsite agrees well with laboratory GC measurements for the gasflood(left) and waterflood (middle) zones of the Hibernia formation and Ben Nevis and Avalon formations (right).
50387schD03R1.p41.ps 11/17/01 8:32 PM Page 41
42 Oilfield Review
NW C block SEB block
4200
Dept
h su
bsea
, m
3400
4000
3800
3600
WOC
GOC
B-08
B-16 10z
B-1614
5:1 vertical exaggeration
B-1615z
B-1611
0
0 10.5 1.5 miles
1 2 km
> Cross section spanning Blocks B and C in the Hibernia field gasflood area. The Hibernia formation dips steeply,plunging into the Murre fault in the northwest. The gas-oil contact (GOC) is shown at the crest. The water-oil contact(WOC) is unknown in the southeast; in the northwest it lies between the two marked depths. This section line is indi-cated on the map on page 40.
Dete
ctor
resp
onse
700
600
500
400
300
200
100
00 10 20 30 40
Time, min
50 60 70 80
n-C 10
n-C 15
n-C 30
n-C 22
n-C 20
n-C 18
Ph
n-C 17
Pr
n-C 16
n-C 12
n-C 14
Pr Ph
n-C 20
Hibernia oil sample
Base oil
n-C 25
n-C 30
> Gas chromatograms of reservoir and drilling-mud base oils. The sharp peaks on the curves are spe-cific carbon compounds, such as normal-alkane C30 [n-C30]. Pristane (Pr) and phytane (Ph) are geo-markers found in reservoir fluids. A scaling factor is applied to the base-oil spectrum before subtract-ing it from the reservoir-oil spectrum. The scaling factor is related to the degree of contamination.
50387schD03R1 11/29/01 3:37 AM Page 42
Autumn 2001 43
field, using MDT pressures and fluid-type deter-mination to establish gas-oil and water-oil contacts was important (previous page, top). A significant benefit of the MDT logging program is real-time decision-making on sample collection points.
MDT fluid-sample composition was deter-mined in a PVT laboratory by the GC method. Thechromatogram of the base oil of the mud wassubtracted from the sample GC spectrum (previous page, bottom). The resulting peak-height spectra from different blocks, coupledwith other PVT data such as the bubblepointpressure, GOR and formation volume factor, pro-vided evidence to correlate oil from differentfault blocks, indicating seven distinct fluidregions across the field (above). With this infor-mation, gasflooding and waterflooding can beimplemented more efficiently. Formation pres-sures from openhole MDT runs also indicatedwhether offset production had drawn down for-mation pressure in the new locations. Other mea-surements made on the reservoir fluids, includingwax content, sulfur content, acid number, pourpoint, cloud point and saturates-aromatics-resins-asphaltenes content, also indicated varia-tions by fault block, impacting the production andcompletion strategies.19
A Downhole Chemistry LaboratoryDistinguishing fluid phases may seem like someof the simplest chemistry that can be performed.Doing it from miles away, in a harsh environment,is the significant new accomplishment of theMDT tool. The channels of absorption informa-tion in the OFA tool have allowed correlationwith many more attributes of the fluid: oil-shrink-age factor, bubblepoint pressure, oil compress-ibility, oil density and average molecularweight.20 Minimizing contamination in collectedsamples and controlling phase separation duringcollection to enhance the value of in-situ fluidproperties measurements is an ongoing chal-lenge. The additional capabilities in the new LFAmodule provide direct measurement of methanecontent, allowing estimation of GOR and a morerobust gas flag to avoid taking the fluid into thetwo-phase region.
In addition, obtaining fluid samples frombehind casing is significantly easier now. TheCHDT Cased Hole Dynamics Tester tool can drillup to six holes through casing in one trip and, incombination with other MDT modules, obtainsamples and monitor contamination in real time.It then seals the hole through the casing with a
corrosion-resistant plug rated to 10,000-psi [69-MPa] differential pressure.
Already, significant decisions are made basedon real-time downhole fluid measurements.Continuing development will improve the rangeand reliability of these measurements. —MAA
Bonavistaplatform
Mur
re fa
ult
Nautilus fault
N
0
0 1 2 3 miles
1 2 3 4 5 km
> Fluid regions in Hibernia field. Seven distinct regions are defined for Hibernia fluids, based on constituents and physical properties determined from DST and MDT fluid samples.
19. The pour point is the lowest temperature at which an oil will begin to flow under standard test conditions. The cloud point is the temperature at which paraffinmolecules first start to crystallize from oil, as observed visually.
20. Van Dusen A, Williams S, Fadnes FH and Irvine-Fortescue J: “Determination of Hydrocarbon Propertiesby Optical Analysis During Wireline Fluid Sampling,”paper SPE 63252, presented at the SPE Annual TechnicalConference and Exhibition, Dallas, Texas, USA, October 1-4, 2000.
50387schD03R1.p43.ps 11/17/01 8:33 PM Page 43