nuclear magnetic resonance comes out of its shell

4 Oilfield Review

Nuclear Magnetic Resonance Comes Out of Its Shell

Ridvan AkkurtSaudi AramcoDhahran, Saudi Arabia

H. Nate BachmanChanh Cao MinhCharles FlaumJack LaVigneRob LeveridgeSugar Land, Texas, USA

Romulo CarmonaPetróleos de Venezuela S.A.Caracas, Venezuela

Steve CraryAl-Khobar, Saudi Arabia

Eric DecosterBarcelona, Venezuela

Nick HeatonClamart, France

Martin D. HürlimannCambridge, Massachusetts, USA

Wim J. LooyestijnShell International Exploration and Production B.V.Rijswijk, The Netherlands

Duncan MardonExxonMobil Upstream Research Co.Houston, Texas

Jim WhiteAberdeen, Scotland

Oilfield Review Winter 2008/2009: 20, no. 4. Copyright © 2009 Schlumberger.AIT, CMR, MDT, MR Scanner, OBMI and Rt Scanner are marks of Schlumberger.MRIL (Magnetic Resonance Imager Log) is a mark ofHalliburton.

Advances in measurement technology, along with improved processing techniques,

have created new applications for nuclear magnetic resonance (NMR) logging. A new

NMR tool delivers conventional NMR-based information as well as fluid-property

characterization. These NMR data identify fluid types, transition zones and production

potential in complex environments. Placing this information into multidimensional

visualization maps provides log analysts with new insight into in situ fluid properties.

Petrophysical evaluation involves a lot of scienceand a bit of art. The scientific basis of a newmeasurement technique often develops from stepchanges in technologies. The art of applicationsometimes plays catch-up while interpretationtools are developed to fully exploit new measure -ments. Attempts to integrate new forms of datainto existing workflows may be met withresistance by those skeptical of the added value ofthe new information. In addition, the learningcurve inherent in adopting new concepts is oftensteep, which can be at odds with the timedemands of busy geologists and petrophysicists.

Nuclear magnetic resonance logging is anexample of the physics of measurement—thescience—being understood and developed beforepetrophysical analysis—the art—integrated themeasurements into standard workflows. AlthoughNMR was initially introduced in the 1960s, it took30 years to develop an NMR acquisition tool thatcould deliver the information that physicistsknew was available. The first successfullydeployed pulsed-NMR tool was introduced in theearly 1990s by the NUMAR Corporation, now asubsidiary of Halliburton. Equipped with

permanent, prepolar izing magnets, these loggingtools use radio frequency (RF) pulses tomanipulate the magnetic properties of hydrogennuclei in the reservoir fluids. Schlumbergerfollowed soon after with the CMR combinablemagnetic resonance tool.

In general, NMR measurements were notaccepted enthusiastically because the data didnot always assimilate well with existing inter -pretation schemes. However, early adoptersfound applications for the new measurement,and, as tools evolved, petrophysicists establishedthe value of NMR logging to the interpretationcommunity—creating an expanding niche in theoil and gas industry. Today, most servicecompanies offer some form of NMR logging tool,and LWD NMR tools have been developed toprovide reservoir-quality information in real timeor almost real time.

Magnetic resonance tools measure lithology-independent porosity and require no radioactivesources. They also provide permeability estimatesand basic fluid properties. Initially, the fluidproperties were limited to free-fluid volume andimmovable clay- and capillary-bound fluid

23810schD4R1.qxp:23810schD4R1 2/25/09 1:53 AM Page 4

Winter 2008/2009 5

volumes. Although physicists were aware thatmuch more information about the fluids could becoaxed from NMR data, downhole tools capableof providing more advanced acquisition andprocessing techniques were needed to extractfluid properties in a continuous log.

This article discusses developments in mea -surement techniques that provide NMR-based insitu formation-fluid properties. A recently intro -duced downhole tool capable of making thesemeasurements in both continuous and stationarymodes is described, along with NMR logging theoryapplicable to these new measurements.1 Withthese new NMR capabilities, fluid-characterizationmeasurements resolve interpretation ambiguitiesas demonstrated by case studies from SouthAmerica, the North Sea, the Middle East and west Africa.

A Bit of ScienceAll NMR tools share some common features. Theyhave strong perma nent magnets that are used topolarize the spins of hydrogen nuclei found inreservoir fluids. The tools generate radiofrequency pulses to mani pulate themagnetization of hydrogen nuclei and then usethe same antennas that generate those pulses toreceive the extremely small RF echoesoriginating from the resonant hydrogen nuclei.

Because of their magnetic moments,hydrogen nuclei behave like microscopic barmagnets. Upon exposure to the static magneticfield, B0, of the NMR tool’s permanent magnets,the hydrogen’s magnetic moments tend to alignin the direction of B0. Exposure time is referredto as the wait time, WT, and the time required forpolarization to occur is influenced by various

formation and fluid properties. The buildup inthe resulting magnetization is represented by amulticomponent exponential curve, eachcomponent of which is characterized by a T1

relaxation time.

Gas

WaterOil

1. For more on NMR theory and logging: Kenyon B, Kleinberg R, Straley C, Gubelin G andMorriss C: “Nuclear Magnetic Resonance Imaging—Technology for the 21st Century,” Oilfield Review 7, no. 3(Autumn 1995): 19–33.Allen D, Crary S, Freedman B, Andreani M, Klopf W,Badry R, Flaum C, Kenyon B, Kleinberg R, Gossenberg P,Horkowitz J, Logan D, Singer J and White J: “How to Use Borehole Nuclear Magnetic Resonance,” OilfieldReview 9, no. 2 (Summer 1997): 34–57.Allen D, Flaum C, Ramakrishnan TS, Bedford J,Castelijns K, Fairhurst D, Gubelin G, Heaton N, Minh CC,Norville MA, Seim MR, Pritchard T and Ramamoorthy R:“Trends in NMR Logging,” Oilfield Review 12, no. 3(Autumn 2000): 2–19.


Clay-boundwater

Capillary-bound water

T2

Ampl

itude

Free water

Time

Total porosity

Small poresLarge pores

Ampl

itude

Ampl

itude

Time

Inversion

North

South Antenna

NS

NS

NS

NS

NS

NS

NS

NS N

S

NS N

S

NS

NS

T1 buildup

T2 decay

Echo TE

CPMG sequenceWT

A B C D

E

F

G

> Basic NMR theory. Hydrogen nuclei behave like tiny bar magnets and tend to align with themagnetic field of permanent magnets, such as those in an NMR logging tool (A). During a set waittime (WT), the nuclei polarize at an exponential buildup rate, T1, comprising multiple components (C).Next, a train of RF pulses manipulates the spins of the hydrogen nuclei causing them to tip 90° andthen precess about the permanent magnetic field. The formation fluids generate RF echoes betweensuccessive pulses, which are received and measured by the antenna of the NMR tool (B). The timebetween pulses is the echo spacing (TE) (D). The amplitudes of the echoes decay at a superpositionof exponential relaxation times, T2, which are a function of the pore-size distribution, fluid properties,formation mineralogy and molecular diffusion (E). An inversion technique converts the decay curveinto a distribution of T2 measurements (F). In general, for brine-filled rocks, the distribution is relatedto the pore sizes in the rocks (G).

After a given WT, a train of electromagneticRF pulses manipulates the magnetic moments ofthe hydrogen nuclei and tips their direction awayfrom that of the B0 field. The process of sendinglong trains of RF pulses is referred to as a CPMGsequence.2 A key feature of this sequence isalternating the polarity of the received signal toeliminate electronics-related artifacts. Duringthe CPMG measurement cycle, the hydrogennuclei in the formation generate detectable RFechoes at the same frequency used to manipulatethem.3 The echoes occur between RF bursts. Thetime between bursts is the echo spacing, TE.

The amplitude of the echoes is proportionalto the net magnetization in the plane transverseto the static field created by the permanentmagnets. The amplitude of the initial echo isdirectly related to the formation porosity. Thestrength of the subsequent echoes decreasesexponentially during the measurement cycle.The exponential decay rate, represented by therelaxation rate, T2, is primarily a function of poresize, but also depends on the properties of thefluid in the reservoir, the presence ofparamagnetic minerals in the rock and thediffusion effects of the fluids. In typical cases,the decay of the echo amplitudes is governed bya distribution of T2 times, similar to the T1 timesfound in the buildup curve. An inversiontechnique fits the decay curve with discreteexponential solutions. These solutions areconverted to a continuous distribution of

relaxation times representative of the fluid-filledpores in the reservoir rock (above).4

When the fluid in the sensed region is brine,the T2 distribution is generally bimodal,particularly in sandstones. Small pores andbound fluid have short T2 times, and free fluidsin larger pores have longer relaxation times. Thedividing line between bound and free fluid isreferred to as the T2 cutoff. Oil and gas in thepore spaces introduce a few complications intothe model.

The three primary mechanisms thatinfluence T2 relaxation times are grain surfacerelaxation, relaxation by bulk-fluid processes andrelaxation from molecular diffusion.5 Grainsurface relaxation is a function of pore-sizedistribution. Relaxation effects from moleculardiffusion and bulk-fluid properties are directlyrelated to the type of fluid in the pores.

Tar has an extremely short relaxation timeand may not be measurable with downhole NMRtools. Heavy oils have short relaxation times,similar to those of clay- and capillary-boundfluids, but may also be too short for NMRacquisition (next page). Lighter oils have longerT2 times, similar to those associated with freefluids. Gas has an even longer relaxation timethan oil. During the measurement process, oiland gas signals are detected along with signalsfrom movable and irreducible water. While the T2

times from the oil and gas signals may have norelationship to the producibility of the

6 Oilfield Review


Winter 2008/2009 7

hydrocarbons, they do help characterize the fluidtype. Techniques have been developed to exploitthe fluid response and identify the presence andtype of hydrocarbons.

In the past, there were two primary tech -niques by which NMR data were used to identifyfluids: differential spectrum and enhanceddiffusion.6 The differential spectrum techniquecombines measurements with two different waittimes. Short WTs underpolarize formation fluids,such as gas and light oil, which have long buildupand decay times. Measurements from fluids withshort relaxation times are not affected by achange in WT. Differences between sequentiallogging passes identify the presence of lighthydrocarbon, making the differential spectrumtechnique most effective in gas or condensateenvironments. Logging sequences have also beendeveloped that acquire the data in a single pass.

Enhanced diffusion exploits changes in fluidresponse that occur when different echo spacings,or TEs, are used.7 Water and oil generally havesimilar relaxation times when measurements are

acquired using short TEs, but water often relaxesfaster than oil when longer TEs are used. Toisolate the oil signal, a measurement with a shortTE is compared with an echo train with a longerTE, chosen to enhance the diffusion differences ofthe fluids in the formation. The water signaldecreases with longer TEs, leaving primarily theoil signal. This diffusion sensitivity provides aqualitative indication of the presence of oil,although the measurement may sometimes bequantitative as well.8

Both differential spectrum and enhanceddiffusion rely on traditional T2 relaxationmeasurements to identify hydrocarbons. Thislimits the results to a one-dimensional aspect ofthe fluids, and fluid type can only be inferred, notdirectly quantified. Also, prior knowledge of theexpected fluids is necessary to choose the correctacquisition parameters. The primary limitation ofthe relaxation dimension is the difficulty indistinguishing water from oil (see “Dimensions inNMR Logging,” next page). But, the fact that oiland gas signals are included with the water signal

in the total distribution introduces an exploitabledimension to the relaxation distributions. Removethe water contribution and only the hydro carbonsignal remains.

Molecular diffusion is the key to unlockingfluid properties from the NMR data. Gas andwater have characteristic diffusion rates that canbe calculated for given downhole conditions. Oilhas a range of diffusion values based on itsmolecular structure. This range can also bepredicted from empirical data derived fromdead-oil samples.

The T2 measurement provides the totalvolume of fluid—bound and free. The addition ofdiffusion discriminates the type of fluid present.A graphical presentation—the diffusion-T2, or D-T2, map—displays these data in a 2D spaceformed by the diffusion dimension and therelaxation dimension. The water signal can beseparated from that of the hydrocarbons. Theintensity of the components in the D-T2 mapprovides fluid saturations. Maps can also begenerated using T1 relaxation data.

This quantification of diffusion is madepossible by a new acquisition technique,diffusion editing (DE), which alleviates thelimitations of previous methods, such asenhanced diffusion and differential spectrum.Distinguishing water and hydrocarbon by theirdiffusivity differences not only permits the

2. CPMG refers to the physicists who successfully deployedthe RF pulse sequence used in NMR devices—Herman Carr, Edward Purcell, Saul Meiboom and David Gill.

3. During the CPMG sequence, hydrogen atoms are manipulated by short RF bursts from an oscillatingelectromagnetic field. The frequency of the RF pulses isthe Larmor frequency.

4. Freedman R and Heaton N: “Fluid Characterization UsingNuclear Magnetic Resonance Logging,” Petrophysics 45,no. 3 (May–June 2004): 241–250.

5. Kleinberg RL, Kenyon WE and Mitra PP: “On the Mechanismof NMR Relaxation of Fluids in Rocks,” Journal of

Magnetic Resonance 108A, no. 2 (1994): 206–214.6. Akkurt R, Vinegar HJ, Tutunjian PN and Guillory AJ:

“NMR Logging of Natural Gas Reservoirs,” The LogAnalyst 37, no. 6 (November–December 1996): 33–42.

7. Akkurt R, Mardon D, Gardner JS, Marschall DM andSolanet F: “Enhanced Diffusion: Expanding the Range ofNMR Direct Hydrocarbon-Typing Applications,”Transactions of the SPWLA 39th Annual LoggingSymposium, Houston, May 26–29, 1998, paper GG.

8. Looyestijn W: “Determination of Oil Saturation fromDiffusion NMR Logs,” Transactions of the SPWLA 37thAnnual Logging Symposium, New Orleans, June 16–19,1996, paper SS.

> The effects of oil on T2 distributions. For brine-filled pores, the T2distribution generally reflects the pore-size distribution of the rock. Thisdistribution is often bimodal, representing small and large pores (left ). Thesmall pores contain clay- and capillary-bound fluids and have shortrelaxation times. The large pores contain movable free water and havelonger relaxation times. The dividing line between bound and free fluids isthe T2 cutoff. When oil fills the reservoir pore spaces, the measured T2

distribution is determined by the viscosity and composition of the oil(middle). Because of their molecular structure, tar and viscous heavy oilshave fast decay rates, or short T2 times. Lighter oils and condensate have aspectrum of T2 times, overlapping with those of larger brine-filled pores.Mixed oil and water in the reservoir result in a combination of T2 timesbased on both pore size and fluid properties (right ).

Brine T2 Distributions

T2 cutoff

Clay-boundwater

Tar +clay-boundwater

Lightoil +freewater

Heavy oil +capillary-bound water

Tar Lightoil

Heavy oil Intermediate oil Intermediate oils +free water

Capillary-bound water

Free water

Pore size Viscosity and composition

Oil T2 Distributions Total Distribution

(continued on page 13)


8 Oilfield Review

Humans generally visualize in three dimensions,and geometric relationships are understoodas adding levels of complexity with eachdimension. For instance, a 1D image may havelength, 2D adds width, 3D adds depth, and 4Dadds the element of time.1 Analogous tospatial relation ships, NMR measurements canbe described using dimensionality, with eachdimension adding a degree of complexity.

The 1D NMR distribution refers to T2

transverse relaxation time measurements. T2

distributions are obtained by inverting rawNMR echo-decay signals. The distributionscontain information about both fluidproperties and pore geometry. However,signals from different fluids often overlap,and it is not always possible to distinguishwater from oil, or water from gas purely onthe basis of the T2 distribution.

The T1 relaxation measurement, from thebuildup of polarization, also provides a 1Ddistribution. A single echo (or a small numberof echoes) is acquired for a series of differentwait times, WTs.2 The observed increase inecho amplitude with increasing WT is thepolariza tion buildup, which is governed by thedistribution of T1 relaxation times (above).With a mathematical inversion similar to theone employed to derive T2 distributions fromecho-decay signals, the T1 distribution isextracted from the polarization buildup.

During the T1 acquisition, a full T2 echo-decay signal can be acquired for each WT,rather than a single echo or a short series ofechoes, and thus a 2D dataset can begenerated with T2 and T1 data. The individualecho-decay signals are inverted to obtain aseparate T2 distribution for every WT. Each T2

component follows its own characteristicbuildup with increasing WT, governed by theT1 distribution (buildup) associ ated with that

T2 component. In practice, a 2D inversiondirectly transforms the original echo datasetto a 2D T1-T2 distribution, some times referredto as a T1-T2 map (next page, top left).3

For many fluids, T1 and T2 distributions arevery similar since they are governed by thesame physical properties. Under typicalmeasurement conditions the T1/T2 ratio forwater and oil ranges between 1 and 3.However, an important difference betweenthe two relaxation times is that T2 times areaffected by molecular diffusion whereas T1

times are free of diffusion effects. In NMRmeasurements, diffusion causes a reduc tionin echo amplitude and therefore shortens T2

times. The magnitude of the diffusion effectis a function of the molecular-diffusionconstant of the fluid, D, and the echospacing, TE. TE is an adjustablemeasurement parameter defining the timebetween consecutive radio frequency (RF)pulses in the measurement sequence.

Diffusivity is an intrinsic fluid property,depending only on the fluid composition,temperature and pressure. Once quantified, itidentifies the fluid type.4 For water, D isprima rily related to temperature, and fornatural gas, it is determined by both tempera -ture and pressure. Crude oils exhibit adistribution of diffusion rates governed by themolecular composition, temperature andpressure. Diffusion, therefore, is the key toidentifying fluid type with NMR. For example,T2 relaxation times for gas are much shorterthan T1 times because of diffusion. Byidentifying the differ ence between T1 and T2

measurements in a gas reservoir, thehydrocarbon type can be inferred.

Diffusion distributions are determined bymeasuring echo-amplitude decays for echotrains acquired with different echo spacings,TEs. However, increasing the TE to allowdiffusion to take place comes at a price. Theincreased time between echoes means there

Dimensions in NMR Logging

> T1 logging. Traditional NMR CPMG sequences measure T2 distributions and begin after asufficient WT for polarization of hydrogen nuclei. For T1 acquisition, a succession of short CPMGcycles with WTs selected over a range of values is used. In a departure from earlier T1 acquisitionmethods, the MR Scanner tool acquires full T2 echo trains for each chosen WT value, and thusthe resulting data can be subjected to a multidimensional inversion and provide both T1 and T2distributions. T1 logging is especially useful in low signal-to-noise environments and for fluidswith long polarization times, such as are found with light hydrocarbons and in large pores. Inaddition, T1 distributions, unlike T2 distributions, are free of diffusion effects and provide more-accurate results in highly diffusive fluids.

Full CPMG (1)

WT (n)

Pola

rizat

ion WT (3)

WT (2)

WT (1)Full CPMG (2)

Full CPMG (3) Full CPMG (n)

Time


Winter 2008/2009 9

are fewer echoes over an equivalent timespan, reducing the data density. This alsoresults in more rapid signal decay—T2 timesare shorter—because of the diffusion effects.The end result is a reduction in the amount ofusable data, and the inversion becomes morechallenging because of the lower signal-to-noise ratio.

The diffusion-editing (DE) techniqueovercomes these limitations by combining twolong initial TEs—during which diffusion iseffective in reducing the NMR signal—followed by an extended train of short TEs,during which diffusion effects are minimized(above right). A large number of echoes can

be acquired, and the effective signal-to-noiseratio is maximized.

Analogous to the T1-T2 measurementdescribed earlier, a 2D experiment can bedesigned to extract diffusion information.

Rather than acquiring echo trains for anumber of sequential WTs, an echo train isacquired with different initial long TEs. Thedata are subjected to the inversion processingand can then be used to generate D-T2 maps,

1. Beyond classical physics, there are other applicationsthat describe four dimensions and beyond. Stringtheory, for example, predicts 10 dimensions, includinga zero dimension.

2. The wait time is the time allotted for the alignment of protons within the static magnetic field of thepermanent magnet of an NMR logging tool during themeasurement cycle.

3. Song YQ, Venkataramanan L, Hürlimann MD, Flaum M,Frulla P and Straley C: "T1–T2 Correlation SpectraObtained Using a Fast Two-Dimensional LaplaceInversion,” Journal of Magnetic Resonance 154, no. 2(February 2002): 261–268.

4. Freedman and Heaton, reference 4, main text.

> Two-dimensional NMR data. The 2D nature of T1-T2 maps is highlighted by overlappingsignals from the two sets of distributions. Thecrossplotted signals are at maxima, indicatedby the color variation from blue to dark red, inthe center and right of this plot. The dataconverge along the center line in the middle—their agreement indicating similar fluidmeasurements from both T1 and T2. But,divergence of the longer time components ofthe two sets of data, resulting from moleculardiffusion, moves the plot away from the centerline at the right corner. If there were nodiffusion effects, the crossplot would becentered along the dividing line.

T1 T2

> Diffusion editing. With traditional CPMG sequences and short echo spacing (TE), oil (green) and water (blue) signals relax, or decay, at similar rates (top). Lengthening the TE value (middle)enhances the diffusion effect preferentially for the fast-diffusing water compared with slower-diffusing oil. However, long TEs correspond to fewer echoes and a lower signal-to-noise ratio.Diffusion editing (bottom) is a variant of the multi-TE CPMG method, where only the first twoechoes are lengthened to enhance the diffusion effect, while maintaining the advantage of theshort TE for better signal-to-noise ratio.

TE

WaterOil

WaterOil

TE

t TE2 x TE

Diffusion (D) Transverse relaxation (T2)


10

which are a graphical means to identify fluidtype and quantify saturations (above).5

Although the two-dimensional D-T2

measure ment is effective in separating the oilsignal from that of the water, it is less robustfor distinguishing between highly diffusive

fluids, such as gas, condensate or water athigh temperatures. The problem arisesbecause diffusion can dominate the T2

relaxation mechanism for these fluids, even atthe shortest echo spacing available from thelogging tools. The underlying “diffusion-free”

T2, which contains important complementaryinformation about composition, such asgas/oil ratio (GOR) and pore-size informationfor water, cannot be measured. This limitationis overcome by invoking a third dimension tocombine with diffusion, T1 relaxation times.6

The 3D NMR measurements acquire echo-train data at multiple WTs (for T1) andmultiple TEs (for diffusion). Sufficientinformation is then available to create 3Dmaps of D-T1-T2, in which the T2 axis refers toa transverse relaxa tion time with diffusioneffects removed (next page, left). The map istherefore a 3D correlation of intrinsic fluidproperties for T1, T2 and D. In practice, NMRfluid maps are typically presented in a 2Dformat, plotting D with either T1 or T2, or onoccasion plotting T1 with T2.

Overlain on the maps are default fluid-response lines for the D of gas and watercomputed from their diffusion coefficient atformation temperature and pressure. The oilline is derived from the estimated dead-oilresponse at downhole conditions.

The fourth dimension in NMR logging, theradial distance from the borehole wall, resultsfrom acquisition at multiple depths ofinvestigation (DOIs). Data from two or threeDOIs are simultaneously inverted. Resultsfrom the shallow DOI are used to correct datafrom deeper DOIs, improving the outputsaffected by missing information and poorersignal-to-noise ratios.

NMR logging tools acquire data from aregion often affected by filtrate invasion,which alters the original fluid distribution.The 4D NMR processing is based on theassumptions that bound-fluid volume andimmovable hydrocarbon volume are invariantto DOI. The shallow-measurement data areused to constrain the inversion for the deepermeasurements by fixing the bound-fluidcomponents. (For examples of 4D NMRprocessing, see pages 16 and 17.) Diffusionand T1 (or T2) data from 4D NMR are used toproduce fluid maps at multiple DOIs. Fluidchanges that take place as filtrate invades thereservoir rock are graphically displayed and

Oilfield Review

5. Hürlimann MD, Venkataramanan L, Flaum C, Speier P,Karmonik C, Freedman R and Heaton N: “Diffusion-Editing: New NMR Measurement of Saturation andPore Geometry,” Transactions of the SPWLA 43rdAnnual Logging Symposium, Oiso, Japan, June 2–6,2002, paper FFF.

6. Freedman and Heaton, reference 4, main text.7. Cao Minh C, Heaton N, Ramamoorthy R, Decoster E,

White J, Junk E, Eyvazzadeh R, Al-Yousef O, Fiorini R

and McLendon D: “Planning and Interpreting NMRFluid-Characterization Logs,” paper SPE 84478,presented at the SPE Annual Technical Conferenceand Exhibition, Denver, October 5–8, 2003.

8. For more on wettability: Abdallah W, Buckley JS,Carnegie A, Edwards J, Herold B, Fordham E, Graue A,Habashy T, Seleznev N, Signer C, Hussain H, Montaron Band Ziauddin M: “Fundamentals of Wettability,”Oilfield Review 19, no. 2 (Summer 2007): 44–61.

> D-T maps. Diffusion plotted with T2 (or T1) provides 2D reservoir-fluid maps that can resolve oil,gas and water. In this example, the diffusion dimension (right ) is the key to identifying the fluids,which otherwise overlap in the T2 dimension (top left ). The amplitudes of the signals along onedirection of the two-dimensional map result in 1D distributions, which then can be converted tofluid saturations. As an aid to interpreting the 2D maps, fluid-diffusion coefficients aresuperimposed on the map (bottom left ). The gas line (red) is computed using downhole pressureand temperature inputs. The water line (blue) is calculated using the downhole formationtemperature. The oil line (green) shows the position of oil at different viscosities, with the lowerleft being heavy oil, trending to light oil and condensate at the upper right. The interpretation ofthis map is that the reservoir contains gas, oil and water.

Gas diffusioncoefficient

Gas

Water

Water

Gas

Diffusion distribution

T2 (or T1) distributions

Oil

Oil

Water diffusioncoefficient

Oil diffu

sion coefficient

viscosity correlation

T2

D


Winter 2008/2009 11

allow petro physicists to detect oil mobility,wettability effects and fluid interactions(above right).

Although the interpretation of maps createdfrom 3D or 4D NMR data might seem simple,complications do exist. The results rely on aforward-model approach that assumes thefluid and the reservoir meet certain criteria.When nonideal fluid properties or atypicalreservoir conditions are encountered, theresponse deviates from the model, andconflict ing or erroneous results may ensue.7

In some cases, nonideal effects can bedetected and even quantified by inspection of the D-T maps. Relevant parameters in theforward model can be adjusted once theseeffects are identified.

In another problem, when diffusion of fluidmolecules in small pores is restricted, themeasured values of diffusion are reducedfrom those of the ideal model (next page).While the signal from fluids in large pores

appears as expected in the D-T maps, fluids in the small, poorly connected pores may plotat lower diffusivity values. The problem ismost common for water diffusion in fine-grained carbonate rocks. If the effect is notidentified, the calculated oil saturation maybe overly optimistic. However, once therestricted-diffusion effect is detected, modelparameters can be adjusted according to theobserved 2D map results and fluid-saturationestimations corrected.

Another anomalous effect results frominternal magnetic-field gradients caused byparamagnetic and ferromagnetic materials inthe rocks, either in the matrix or coating thegrains. These are often associated with highchlorite content and create significantlocalized field gradients, resulting in fasterrelaxation times. Because the inversion modelis based on the tool’s fixed magnetic-fieldgradient, the D-T map responses of the fluidsin these rocks are shifted to higher diffusion

rates, the opposite of the restricted-diffusioneffect. For example, water signals may appearabove the water line. Fortunately, it is usuallypossible to identify these effects by inspectionof the maps, and model parameters can thenbe adjusted to provide correct interpretations.

The wettability state also affects D-T maps.Under water-wet conditions, the oil viscositydetermines the position of the oil signal alongthe oil line of the map. The trend is fromheavy oil at the bottom left to lighter oils andconden sate at the top right of the line. Oil-wet rocks and those with mixed wettabilitytend to have shorter relaxation times becauseof the addi tional surface relaxation of thehydrocarbon in direct contact with the poresurface. Although this can compromise theaccuracy of oil viscosity estimated from theNMR data, it can also be a usefulmeasurement for petrophysicists inunderstanding the nature of the reservoir.8

> Three dimensions of NMR. Diffusion, T1distributions and T2 distributions presented in a 3D format provide intrinsic fluid properties.The cube is used to identify diffusion effectsand may aid the interpreter in deciding whichmodel best describes the fluid properties.

T2

D

D-T2 map

T1-T2 map

D-T1 map

T1

> NMR in four dimensions. The fourth dimension of NMR logging is depth. Bound-fluid volumes,associated with both clay- and capillary-bound fluids (yellow), do not generally change whenfiltrate from the drilling fluid invades the reservoir. Tool or measurement limitations, however, canresult in changes in computed fluid properties that do not represent the true fluid distributions.Constraining the volumes of bound fluid measured by deeper-reading shells to be equivalent tothat of the more-precise shallower shells and reapportioning the total porosity across the fluidspectrum provide more-accurate fluid analysis. Use of 4D NMR processing is especiallybeneficial in interpreting data from heavy-oil reservoirs.

T2

DOI

Shallow

Ampl

itude

Deep


12 Oilfield Review

Fluids with a high GOR tend to plot aboveand to the left of the oil line. This can be seenin native fluids and gas-bearing zones invadedby oil-base mud filtrate (OBMF). OBMFshould plot as a moderate to light hydrocarbon.The response of OBMF mixed with native gasfrom the reservoir plots between the oil andwater lines.

The D-T maps are powerful tools for inter -preting fluid types in the reservoir. In manycases the interpretation is straightforward, but

consideration must be given to external factorsthat lead to nonideal behavior and mislead anovice interpreter. For this reason, it isimportant to rely on experts adequately trainedin processing and interpreting NMR data.

A surgeon relies on a trained radiologist tointerpret MRI images. A clean break of a boneis easy to spot, even by a novice user, butexperience helps a radiologist differentiatebetween bone fragments and calcification.The differences may be indistinguishable to

the untrained eye. In a similar manner, thelog analyst may easily determine the presenceof water or gas in a D-T map, but there areoccasions when a skilled NMR expert shouldbe called in to help analyze the results. Withthe aid of fluid maps and an understanding ofthe measurement physics, the petrophysicistcan diagnose conditions that are hidden fromthe inexperienced observer.

> Interpreting the maps. After inversion, two-dimensional crossplots ofthe data identify the presence of oil, water and gas. When the plot ofthe response from formation fluids (shown as color contours) conformsto the interpretation model, the response will fall on or near theexpected gas, oil or water lines. This affords a straightforwardinterpretation. Water falls on the water line as shown (middle).However, signals frequently fall off the lines as a result of competingpetrophysical effects, which include internal gradients, restricteddiffusion, wettability and high gas/oil ratios (GOR). Because internalgradients shorten the relaxation times, the plots tend to shift upward(top left ). Restricted diffusion causes the measured diffusion rate toincrease, and, as a consequence, the plots will trend down and away

from the expected fluid line (bottom left ). Plots from oil-wet reservoirstend to shift to the right of the oil line, as do reservoirs with mixedwettability (bottom right ). The maps from oil-wet and mixed-wetreservoirs tend to have a spectrum of responses, which produce abroader image. Because the model is built with dead-oil responses,maps from high-GOR oil reservoirs may not respond as expected. Theseplots are shifted away from the oil line toward the gas line (top right ). Ina reservoir that contains only gas, OBM filtrate may mix with the nativegas and produce a response similar to that of high-GOR oil. Deepermeasurements often help experts refine their interpretation of thesemaps because filtrate generally diminishes away from the wellbore andgas response increases.

Gas

Water

Internal gradient

Restricted diffusion

Internal gradient High GOR

High GOR

Boundwater

OBMFwith gas Native

oil

Mixed wettability

Oil

Restricteddiffusion

Unrestricteddiffusion

Mixed wettability


Winter 2008/2009 13

computation of fluid saturations, but also helpsinfer fluid viscosity from the T2 contribution ofthe fluid (right).

Diffusion-editing sequences from the new MR Scanner service supply a water-saturationoutput that is independent of that traditionallyderived from resistivity and porosity measure -ments. In contrast to a saturation derived fromArchie’s equation, NMR-based saturationmeasurement techniques are useful in freshwater or formation waters of unknown salinity.Wettability can also be inferred from NMR data.One drawback to using NMR measurements forfluid characterization is that the measurementcomes from a near-well region referred to as the flushed zone, where mud-filtrate effects are strongest.

The MR Scanner Tool Although NMR measurements may come fromonly a few inches into the formation, they can stillprovide formation-fluid properties. To measurecontinuous in situ fluid characteristics—including fluid type, volume and oil viscosity—much more information is needed than wasprovided by previous-generation NMR tools.9 Forthis reason, fluid characterization was a key driverin the development of the MR Scanner service.

In the past, there were two basic designs forNMR tools: pad-contact tools and centralizedconcentric-shell tools. The pad device, representedby the CMR tool, measures NMR properties of acigar-sized volume of the reservoir fluids at afixed depth of investigation (DOI) of approxi -mately 1.1 in. [2.8 cm]. The NUMAR MRILMagnetic Resonance Imager Log tool measuresconcentric cylindrical resonant shells of varyingthickness and at fixed distances from the tool,with the DOI determined by hole size and toolposition in the wellbore.

The MR Scanner design offers the fixed DOIof a pad device with the flexibility of multipleDOIs of resonant shells.10 It consists of a mainantenna optimized for fluid analysis and twoshorter high-resolution antennas best suited foracquiring basic NMR properties (right). Themain antenna operates at multiple frequenciescorresponding to independent measurementvolumes (shells) at evenly spaced DOIs.

9. Heaton NJ, Freedman R, Karmonik C, Taherian R,Walter K and DePavia L: “Applications of a New-Generation NMR Wireline Logging Tool,” paper SPE 77400, presented at the SPE Annual TechnicalConference and Exhibition, San Antonio, Texas,September 29–October 2, 2002.

10. DePavia L, Heaton N, Ayers D, Freedman R, Harris R,Jorion B, Kovats J, Luong B, Rajan N, Taherian R,Walter K, Willis D, Scheibal J and Garcia S: “A Next-Generation Wireline NMR Logging Tool,” paper SPE 84482, presented at the SPE Annual TechnicalConference and Exhibition, Denver, October 5–8, 2003.

> Viscosity transform. The T2 (or T1) relaxation time for crude oil is a functionof viscosity. The relaxation time can be converted to viscosity using anempirically derived transform. Because of diffusion effects, the viscositymeasurement for heavy oils below 3 cP [0.003 Pa.s] is influenced by the echospacing (TE) of the measurement. Thus, T2 times may be tool dependent forheavy oils if the tool is not capable of shorter TEs. As a consequence ofdiffusion, T1 and T2 values in light oils diverge above 100 cP [0.1 Pa.s].

T 1 or T

2, s

10

10.1 100Viscosity, cP

10 100,00010,0001,000

1

0.1

0.01

0.001

0.0001

T2 (TE = 0.2 ms)

T1

T1

T2

T2 (TE = 0.32 ms)

T2 (TE = 1 ms)

T2 (TE = 2 ms)

Permanent magnet

Antenna

Sensed region

Main antenna

High-resolution antennas

> MR Scanner service. The MR Scanner tool has three antennas. The main antenna operates atmultiple frequencies and is optimized for fluid-property acquisition. The sensed region consists ofvery thin shells that form arcs of approximately 100° in front of the 18-in. [46-cm] length of the antenna.The thickness of the individual shells is 2 to 3 mm. The two high-resolution antennas are 7.5 in. [19 cm]long and provide measurements with a DOI of 1.25 in. [3.17 cm]. The MR Scanner tool is run eccenteredwith the antenna section pressed against the borehole.


Although multiple frequencies are availablefrom the main antenna, the three mostcommonly used are Shells No. 1, No. 4 and No. 8,which correspond to 1.5-in., 2.7-in. and 4.0-in.

[3.8-cm, 6.8-cm and 10.2-cm] DOIs, respectively.A recently introduced three-shell simultaneousacquisition mode eliminates the need for multiplepasses to acquire data from all three DOIs.

Profiling Fluid PropertiesThe three primary frequencies commonly used bythe MR Scanner tool correspond to threeindependent DOIs, providing measurements atdiscrete radial steps into the formation. Thefrequency of the RF pulse, along with the fieldstrength of the magnet, determines the DOI ofthe shell (above). A key advantage of the MRScanner shells is that the measurement comesfrom a thin cylindrical portion of the formation—an isolated slice—and is not generally affected bythe fluids between the tool and the measurementvolume. This allows interpretation of near-wellbore fluid properties in a manner that isunique in formation evaluation.

Multiple DOIs introduce new concepts forNMR logging—radial profiling and saturationprofiling (left). Profiling incorporatesmeasurements from successive DOIs to quantifychanges in fluid properties occurring in the firstfew inches of formation away from the wellbore.

Borehole rugosity and thick mudcake can invali -date shallow NMR measurements but rarely affectthe readings from the deeper shells. NMR porosityfrom a deep shell has been used as a substitute forformation density porosity when that measure -ment was compromised by borehole rugosity.

Fluid-property changes resulting from mud-filtrate invasion may also be observed andquanti fied using radial profiling. Often more thanjust filtrate from the drilling mud invades theformation. Whole mud and mud solids canreplace existing fluids in the near-wellboreregion. NMR-derived porosity and permeabilitymay be reduced by the presence of these solids,but the effects diminish deeper into theformation. Radial profiling identifies andovercomes these effects.

Saturation profiling delivers advanced fluid-characterization measurements such as oil, gasand water saturations, fluid type and oilviscosity—all at discrete multiple DOIs. Oneapplication of saturation profiling is theevaluation of heavy-oil reservoirs.

Viscosity and RugosityOf the world’s known reserves, 6 to 9 trillionbarrels [0.9 to 1.4 trillion m3], are found in theform of heavy- or extraheavy-oil accumulations.11

This is triple the amount of world reserves ofconventional oil and gas combined. Heavy-oilreser voirs pose serious operational concerns forproper evaluation of fluids and productionpotential. Sampling with wireline-conveyed toolsor drillstem tests may not be possible because of the difficulties in getting the oil to flow. NMR measurements can provide essentialinformation on in situ fluid properties to evaluateheavy-oil reservoirs.

Located south of the Eastern Venezuelanbasin, the Orinoco heavy-oil belt holds anestimated 1.2 trillion barrels [190 billion m3] ofheavy oil. NMR logs have been an integral part ofevaluation programs of the wells drilled in thisregion, but with recognized limitations.12 Therelaxation times of high-viscosity oils are veryshort and may not be fully measurable usingNMR tools. Also, borehole conditions in theOrinoco basin are often poor, with rugosity thataffects pad-contact tools.

Although the early introduction of NMR fluidcharacterization with the CMR tool showedpromise, the measurements were limited to asingle, shallow DOI. Techniques were developed to use NMR data to estimate the oil viscositythroughout a sand interval based on thelogarithmic mean of T2 distributions. Encouraging

14 Oilfield Review

> Radial profiling. The MR Scanner tool sensesfluid from multiple, thin DOI shells. The spacing isoptimized to avoid shell interaction. Fluidproperties vary in the first few inches into theformation as a result of flushing by mud filtrate.Deeper shells are less affected by filtrate, mudinvasion and borehole rugosity than are shallow-reading shells.

Shell No. 8

Shell No. 1

Shell No. 4

1.5 in.

0 in.

DOI

2.7 in. 4.0 in.

> Gradient tool and DOI. The MR Scanner tool is referred to as a gradienttool because the magnetic-field strength (B0 , blue) from the permanentmagnet, although uniform over the sample region, decreases monotonicallyaway from the magnet. The tool’s magnet extends along the length of thesonde section. A constant, well-defined gradient simplifies fluid-propertymeasurements. The DOI is determined by the magnetic-field strength and thefrequency of RF operation, f0 . Although multiple frequencies are availablefrom the tool, standard operating procedure is to acquire data from the 1.5-in,2.7-in. and 4.0-in. DOI shells, referred to as Shell No. 1, Shell No. 4 and Shell No. 8, respectively. Three shells are shown, with their respective DOI-related operating frequencies. The frequency associated with Shell No. 1 isshown in green.

DOIDecreasing B0 strength

ƒ0

Distance from magnet

Borehole

Mainantenna

Mag

net

1.5 in.

2.7 in.

4.0 in.

ShellNo. 1

ShellNo. 8

ShellNo. 4


Winter 2008/2009 15

results were obtained but fell short of true in situviscosity. There was no calibrated transform tolink the logarithmic mean of the T2 distributionsto the viscosity at downhole conditions that wouldalso account for the apparent hydrogen index (HI)of the oil.13 The importance of using HI and anempirical transform was demonstrated by recentlaboratory work.14

The MR Scanner tool was included in a morerecent logging program, in part, to overcome someof the limitations of the CMR measure ments.Radial profiling was found to be beneficial in zones where rugosity affected the 1.5-in. DOI

measurement from Shell No. 1, which iscomparable to the CMR tool’s DOI. The 2.7-in.measurement from Shell No. 4 was only slightly

affected by rugosity. The 4.0-in. Shell No. 8 datawere not affected because they were acquired froma region beyond the rugosity (above).

11. Alboudwarej H, Felix J, Taylor S, Badry R, Bremner C,Brough B, Skeates C, Baker A, Palmer D, Pattison K,Beshry M, Krawchuk P, Brown G, Calvo R, Cañas Triana JA,Hathcock R, Koerner K, Hughes T, Kundu D, López deCárdenas J and West C: “Highlighting Heavy Oil,”Oilfield Review 18, no. 2 (Summer 2006): 34–53.

12. Decoster E and Carmona R: “Application of Recent NMRDevelopments to the Characterization of Orinoco BeltHeavy Oil Reservoirs,” Transactions of the SPWLA 49thAnnual Logging Symposium, Edinburgh, Scotland, May 25–28, 2008, paper VVV.

13. Carmona R and Decoster E: “Assessing ProductionPotential of Heavy Oil Reservoirs from the Orinoco Beltwith NMR Logs,” Transactions of the SPWLA 42ndAnnual Logging Symposium, Houston, June 17–20, 2001,paper ZZ.

14. Burcaw L, Kleinberg R, Bryan J, Kantzas A, Cheng Y,Kharrat A and Badry R: “Improved Methods forEstimating the Viscosity of Heavy Oils from MagneticResonance Data,” Transactions of the SPWLA 49thAnnual Logging Symposium, Edinburgh, Scotland, May 25–28, 2008, paper W.

> Radial profiling with filtrate and whole-mud invasion. The interval fromX,170 to X,255 ft (red shading) is a clean water sand beneath a heavy-oilreservoir in the Orinoco heavy-oil basin. The fluid properties from the 1.5-in. DOI, Shell No. 1 (Track 5) have spurious volumes of bound water(light brown). Even at 2.7 in., Shell No. 4 indicates more bound fluid thanexpected for a clean sand (Track 6, light brown). The differences areattributed to whole-mud invasion. The Shell No. 8 data come from a region

beyond the whole-mud invasion and provide more-representativeinformation (Track 7). The total porosity measurements from the three shellsappear to be unaffected by the presence of whole mud. Permeabilitiescalculated from the shallower shells (Track 8, blue, green) are lower thanthat of the deepest shell (Track 8, red) because the measurements areaffected by the solids that are filling the pore spaces.

Free Water Free Water Free Water

Heavy Oil Heavy Oil Heavy Oil

Oil Oil Oil

RXO

ohm.m

X,100

X,150

X,200

X,250

0.2 2,000

90-in. Array

ohm.m

Depth

ft

0.2 2,000

T1 Cutoff

ms1 9,000

T1 Distribution,

Shell No. 1

T1 Cutoff

ms1 9,000

T1 Distribution,

Shell No. 4

T1 Cutoff

ms1 9,000

Porosity, Shell No. 1

%40 0


%40 0


%40 0

Shell No. 8

mD100,000 1

T1 Distribution,

Shell No. 8 Bound Water Bound Water Bound Water

60-in. Array

ohm.m0.2 2,000

30-in. Array

ohm.m0.2 2,000

20-in. Array

ohm.m0.2 2,000

10-in. Array

Resistivity

ohm.m0.2 2,000

Caliper

in.6 16

Shell No. 4

mD100,000 1

Shell No. 1

mD100,000 1

Permeability

X,200

X,250

23810schD4R1.qxp:23810schD4R1 4/11/09 2:32 PM Page 15

The independent measurements at variousDOIs from the MR Scanner tool read deeper intothe formation than the CMR tool. Not only has theMR Scanner tool overcome rugosity problems, italso has verified a condition—previouslytheorized from the CMR data—of partial- orwhole-mud invasion effects on the bound-fluidvolumes and permeability. These effects wereespecially noticeable in water zones. The mudsolids did not appreciably alter NMR porosity, butthe bound-fluid measurement was too high. As aninput to the NMR permeability calculation,incorrect bound-fluid volume resulted inpermeability outputs that were too low.

Deeper-reading NMR measurements over comerugosity but have some trade-offs. Because the

formation signals from the deeper shells areweaker, noise has the potential to corrupt theprocessed data. Vertical resolution is degradedbecause data must be averaged or stacked over alonger interval to overcome the noise effects. Themeasurement from deeper shells is also acquiredat longer echo spacings because of tool powerlimitations. The CMR tool uses 0.2-ms echospacing so that in 10 ms it generates 50 pulses.This provides sufficient data to resolve some heavyoils such as those found in the Orinoco wells.However, the 1.0-ms echo spacing available fromthe MR Scanner tool’s 4.0-in. shell provides only10 pulses and echoes in an equivalent time frame.The result is a decrease in signal-to-noise ratiobecause there are fewer echoes to work with.

One solution to this dilemma comes in theform of four-dimensional (4D) NMR processingin which DOI is the fourth dimension.15 Thisprocessing simultaneously inverts the NMR datawithin the portion of the relaxation-timedistribution that should be common for all DOIs.For the Orinoco wells, the time interval to allowthe oil signal to decay is always below 10 ms. Thebad-hole and whole-mud effects begin after20 ms. Imposing a common solution on each shellfor the first 10 ms forces the deeper-reading 4.0-in. shell measurement to be equivalent to thehigher-resolution 1.5-in. shell reading in thiscommon-data area. The result is improvedcoherence between the shells and more-accuratereadings from the deeper shell (above). This is

16 Oilfield Review

> 4D NMR processing. Standard processing results in a lack of coherence between bound-fluid volumes measuredby Shells No. 1, No. 4 and No. 8 (top, Track 1). The same is true for the free-fluid volumes (Track 3). Using 4D NMRprocessing, the bound-fluid volumes, which should remain constant across each DOI, are constrained and theporosity contributions are reassigned. The result is improved coherence for both the bound-fluid (Track 2) and free-fluid (Track 4) volumes. The fluid properties are affected by hole conditions from X,120 to X,135 ft (red shading) asevidenced by the increased porosity measured from the shallower shells (Tracks 5, 6, 8 and 9). Shell No. 8 (Tracks 7and 10) senses from beyond the washout and provides more-accurate data. The D-T1 maps used for saturationcomputation for each shell demonstrate the effectiveness of 4D processing. The standard 2D NMR processing(bottom left panel ) results in similar fluid volumes in Shells No. 1 and No. 4. Shell No. 8 has less bound fluid, but allthree shells should have equivalent volumes because bound fluid should not change with DOI. The 4D NMRprocessing (bottom right panel ) constrains the fluid volume to be the same below 30 ms. Reapportioning theporosity to account for the bound-fluid volume delivers a more-accurate measurement from the deepest shell. As aresult, the 4.0-in. measurement furnishes fluid properties from a region less influenced by mud-filtrate invasion.

Shell No. 8

%

Depth

Washout Bound Water Bound Water Bound Water Bound Water Bound Water Bound Water

Free WaterFree WaterFree WaterFree WaterFree WaterFree Water

Oil

Heavy Oil Heavy Oil Heavy Oil Heavy Oil Heavy Oil

Standard 2D NMR Processing 4D NMR Processing

Heavy Oil

Oil Oil Oil Oil Oil

ft 50 0

Shell No. 4

%50 0

Shell No. 1

Standard Bound Fluid

%50 0Caliper

in.6 16

Shell No. 8

%50 0

Shell No. 4

%50 0

Shell No. 1

4D Bound Fluid

%50 0

Shell No. 8

%25 0

Shell No. 4

%25 0

Shell No. 1

Standard Free Fluid

%25 0

Shell No. 8

%25 0


%50 0


%50 0


%50 0


%50 0


%50 0


%50 0

Shell No. 4

%25 0

Shell No. 1

4D Free Fluid

%25 0

X,100

X,150

Morecoherence

Morecoherence

Shell No. 1 Shell No. 4

Diffu

sion

Ampl

itude

Shell No. 8

T1, ms T1, ms T1, ms


Diffu

sion

Ampl

itude

Shell No. 8


Standard 2D NMR Processing 4D NMR Processing

OilWater

OilWater


Winter 2008/2009 17

valid for borehole-rugosity and thick-mudcakeeffects, but heavy oil impacts the NMRmeasurements even when the borehole is in good condition.

Because heavy oils have short relaxationtimes and rapidly decaying signals, NMR toolsinvariably fail to measure all the heavy oil. This istrue even with the shortest echo spacingscurrently available from downhole tools.Sequences with longer echo spacings miss evenmore of the heavy oil. The MR Scanner tool’sdeeper shell measurements have longer echospacings than those of the shallower shells.Consequently, the volume of heavy oil that ismeasured by the tool decreases with DOI. The oilvolumes will always be underestimated in theseheavy-oil environments.

Despite this shortcoming, the effects of heavyoil on the measurement can still be used tounderstand the formation fluids. The measuredNMR porosity decreases with DOI as a result ofthe missing heavy-oil signal. The measuredvolume of immovable bound water will notchange with DOI. Invading filtrate will displaceonly movable water or movable hydrocarbon.Thus, the 4D inversion can be used in a mannersimilar to that used with hole rugosity, but the interpretation focus will be on the changes in free fluid and total porosity rather thanborehole effects.

The 4D processing provides a markedimprovement over that of conventional 3Dinversion. The first 30 ms of the inversion are

constrained to be common across all three shellsbecause it is assumed that bound fluid and theheavy-oil signals are stable in this time range ateach DOI. This is in contrast to the processingused when borehole rugosity or whole-mudinvasion is a problem; here, only the first 10 msare constrained.

Well data show the free-water signal above100 ms decreasing progressively from shallow todeeper shells (above). This leads to aninterpretation that the source of the free-water

15. Heaton N, Bachman HN, Cao Minh C, Decoster E,LaVigne J, White J and Carmona R: “4D NMR—Applications of the Radial Dimension in MagneticResonance Logging,” Transactions of the SPWLA 48thAnnual Logging Symposium, Austin, Texas, June 3–5,2007, paper P.

> The big picture in heavy oil. The D-T1 maps from X,155 ft show bound-fluid and heavy-oil signals in the Shell No. 1plot (bottom right ). The free-water signal above 100 ms decreases progressively from shallow to deeper DOIs. Thefluid analysis (top, Tracks 5 through 7) shows a steady decrease in free water from Shell No. 1 to Shell No. 8. Theinterpretation is that the source of the water signal is the mud filtrate, which displaced heavy but movable oil in thereservoir; the water signal would remain constant if filtrate were displacing formation water. For the zone from X,020to X,050 ft, the interpretation is more difficult. The resistivity is lower (Track 1), and there is a water signal at each DOI.D-T1 maps from X,040 ft (bottom left ) provide fluid information. Because the water signal from filtrate invasion ispresent in Shells No. 1 and No. 4 but disappears in Shell No. 8, the interpretation is that filtrate displaced heavy oil thatcannot be measured by the NMR tool. The strong water signal present in all three shells is from irreducible water,so the zone should produce water-free oil.


Coal

Clay

Clay Water

Shell No. 1 Shell No. 4Depth X,155

Diffu

sion

Ampl

itude

Shell No. 8


OilWater

Depth

ftX,000

X,200

X,150

X,100

X,050

Caliper

6 16in.

T1 Cutoff

1 9,000

T1 Cutoff

1 9,000

T1 Cutoff

ms ms ms1 9,000


Bound Water Bound Water Bound Water

40 % % %0


40 0


40 0 psig

Pressure

700 900

Shell No. 1 Shell No. 4 Shell No. 8

T1 DistributionsFree Water Free Water Free Water

Oil Oil Oil

Heavy Oil Heavy Oil

Magnetic Resonance 4D Fluid Analysis

Heavy Oil

Water

Moved Oil

Oil

Quartz

Shell No. 1 Shell No. 4Depth X,040

Diffu

sion

Ampl

itude

Shell No. 8

T1, ms T1, ms

100 ms 100 ms

T1, ms

OilWater

RXO

ohm.m0.2 2,000

90-in. Array

ohm.m0.2 2,000

60-in. Arrayohm.m0.2 2,000



10-in. Array

Resistivity

ohm.m0.2 2,000

signal is the mud filtrate and that it hasdisplaced heavy oil in the reservoir, although theheavy oil is invisible to the MR Scanner tool. Ifthe filtrate were displacing movable formationwater, the water signal would be constant atdeeper DOIs.

In a lower interval, the resistivity is high,exceeding 100 ohm.m, which leads interpretersto conclude that filtrate displaced oil. However,for the upper zones with lower resistivity values,the answer is less obvious. Lower resistivityvalues would suggest the presence of waterrather than oil. NMR data provide the missingfluid information. Hydrocarbon in the form ofheavy oil was displaced by the filtrate. Becausethe water signal from the filtrate is present in the1.5-in. Shell No. 1 measurement but disappears

in the 4.0-in. Shell No. 8 measurement, thesezones should produce water-free oil. A strongwater signal remains in the D-T1 maps at eachDOI, but its source is irreducible bound water.16

Based on the answers provided by the 4Dprocessing, the operator can confidently producefrom the upper and lower sections with anexpectation of little or no water production.Minimizing water production decreases upfrontcosts for surface equipment and, because waterremoval and disposal are not required, reducescosts over the life of the well.

Fluid CharacterizationFluid type directly affects the economic value ofa field, and surface facility decisions depend onan accurate understanding of reservoir fluids.

However, many reservoirs contain more than onefluid type: Fluid composition may varycontinuously or discontinuously across areservoir interval. Fluid gradation is not alwaysapparent with conventional well logs, andsurprises can occur in both the early and laterstages of production.

A North Sea exploration well was drilled toevaluate a reservoir that, based on an offset well,was believed to contain gas condensate.17

Adjacent gas-handling infrastructure made theprospect an inviting target. Resistivity anddensity-neutron logs clearly indicated that thisexploration well had a significant hydrocarbondeposit with approximately 48 feet [15 m] of netgas pay.

MR Scanner data were then acquired in asaturation-profiling mode. Diffusion and T1 distri -butions, extracted from multiple wait time,variable echo-spacing sequences, were derivedfrom the data. T2 distributions were computed, butT1 distributions proved better for analyzing thelong relaxation times of the fluids in this reservoir.

Water and hydrocarbon saturations at 1.5-,2.7- and 4.0-in. DOIs were computed from thedata acquired in two separate logging passes.The D-T1 maps at sequential depths were plottedand, although there is a clear gas signal in theupper part of the reservoir, the NMR interpre -tation concluded that a significant portion of thereservoir contained light oil—not condensate as originally anticipated (left).

Closer analysis of the data and the D-T1 mapsidentified the gas/oil contact in the reservoir, aswell as a stratigraphic feature assumed to be avertical-permeability barrier. A 4D inversionenhanced the deeper shell measurement. The1.5-in. shell data indicate a significant volume ofoil-base mud filtrate (OBMF), but the deep shelloutput is less affected by the OBMF (next page).

The porosity and permeability derived fromthe MR Scanner data were immediately availableto the client in the field. The acquired data weresent to a Schlumberger computing center foradvanced processing. NMR fluid-property datawere returned in time to assist in picking depthsfor pressure and sample points. Pressure plotsindicated three different fluid gradients. Fluidsamples confirmed the presence of gas in theupper interval and oil in the lower interval.Unfortunately, filtrate contamination preventedan accurate PVT analysis.

A two-stage drillstem test (DST), conductedafter the well was completed, confirmed thepresence of oil in the lower interval. The NMRdata provided improved understanding of thecomplex nature of the reservoir fluids. The

18 Oilfield Review

> Determining fluid type. The resistivity and porosity data indicate a hydrocarbon interval from 822 to872 ft. Mud logging during drilling suggested gas or condensate throughout the interval. D-T1 mapsfrom data acquired from the MR Scanner tool provide a different fluid analysis. The lowest interval(bottom right ) contains connate water (white circle) and oil-base mud filtrate (OBMF). Successivelyhigher points indicate a transition from light oil to gas (black circles). Based on interpretation of theNMR D-T1 maps, this reservoir contains oil below 840 ft, rather than the expected condensate and gas.

Tony—Figure 10/11_1

in. 2010

Bit Size

ft

Depth

gAPI 1000

Gamma Ray

ohm.m 100.5

Shallow Resistivity

ohm.m 100.5

Deep Resistivity

% 030

Neutron

g/cm3 2.72.2

Density

Crossover

µs/ft 40240

Compressional ΔT

in. 2010

Caliper

800

900

850

Depth, 832 ft

Diffu

sion

T1 time

Depth, 847 ft

Diffu

sion

T1 time

Depth, 874 ft

Diffu

sion

T1 time

Depth, 886 ft

Diffu

sion

T1 time

Depth, 823 ft

Diffu

sion

T1 time


Winter 2008/2009 19

calculated volume of gas and condensateavailable for export from the reservoir wasgreatly reduced. The initial objective of the well,to develop and produce a gas reservoir, wasmodified along with the development plans forthe field.

Finding the Oil/Water ContactLaminated sand-shale sequences, referred to aslow-resistivity, low-contrast (LRLC) pay, arefamiliar to log analysts. They are often over -looked or improperly evaluated because the payis not obvious using conventional logging tools.There are, however, high-resistivity, low-contrast(HRLC) reservoirs where everything looks likepay, and these provide an entirely different set ofchallenges. In HRLC reservoirs, resistivitychanges caused by water-salinity variations andpoor resistivity contrast between oil and waterzones make determination of an oil/water contact(OWC) extremely difficult. Correct deter mination

of the OWC directly affects reserves calculations,completion designs and production decisions.Mistakes are costly, especially when high watercut reduces oil production while adding addi -tional water-disposal costs.

Traditional evaluation techniques are basedon resistivity contrasts between oil and salineformation water. Reservoirs containing fresh orbrackish water may exhibit little or no resistivitycontrast between the fluids. NMR fluid-saturation measurements are based on thevolume of each fluid and are not depen dent onwater salinity. Thus, oil and water saturationsderived from NMR data offer an ideal solution forevaluating HRLC reservoirs and determiningfluid contacts.

In a Middle East HRLC reservoir drilled withoil-base mud (OBM), determination of hydro -carbon saturation and the OWC was not possibleusing resistivity and porosity logs.18 The methodfor finding the OWC was to use pressure

measurements from wireline-conveyed pressure-sampling tools to determine fluid gradients. Thecondition of the borehole often deteriorated duringdrilling, and pressures were difficult to obtainbecause of seal failures, yielding inconclusiveresults. Costly and time-consuming DSTs werethen performed to pinpoint the contact.

Saudi Aramco discovered the subject field inthe 1960s, but there had been no recent

> Gas, oil and water. The permeability (Track 2) is consistent throughout the zone from 820 to 880 ft except for two areas withlower permeability at 845 and 860 ft. The viscosity (Track 3) indicates light oil, but its source is the OBM filtrate. The continuousfluid-saturation logs from Shell No. 1 (Track 4) and Shell No. 8 (Track 5) show fluid changes occur with deeper DOI. The well wasdrilled with oil-base mud, and OBMF (dark green shading) displaces native fluids. In contrast, the Shell No. 8 data show morenative oil (light green) and gas (red). There is very little free water in the wet interval below 880 ft, perhaps because the freewater has been flushed by OBMF. The well has more oil than originally anticipated, and consequently, less gas.

Gas

Oil

Gas


Oil

Oil Filtrate Oil Filtrate

gAPIft

Gamma RayDepth

1000

810

820

830

840

850

860

870

880

mD

Permeability

1,0000.1 ms

T1 Cutoff

9,0001ms

T1 Cutoff

9,0001cP

Oil Viscosity

100.1 %

Porosity, 1.5-in. Shell

040 %

Porosity, 4-in. Shell

Bound WaterBound Water

040

ms

T1, Log Mean

9,0001ms

T1, Log Mean

9,0001

T1 Distribution

630

T1, Distribution

Shell No. 8Shell No. 1

630

16. Irreducible bound water in the reservoir remains inplace during production, and hydrocarbons alone are produced.

17. White J and Samir M: “Continuous Characterization of Multiple Fluids in a North Sea Gas CondensateReservoir by Integrating Downhole NMR with Downhole Sampling,” Transactions of the SPWLA 49thAnnual Logging Symposium, Edinburgh, Scotland, May 25–28, 2008, paper X.

18. Akkurt R, Ahmad NA, Behair AM, Rabaa AS, Crary SFand Thum S: “NMR Radial Saturation Profiling forDelineating Oil-Water Contact in a High-Resistivity Low-Contrast Formation Drilled with Oil-Based Mud,”Transactions of the SPWLA 49th Annual LoggingSymposium, Edinburgh, Scotland, May 25–28, 2008,paper Y.


production. The company reactivated the fieldand then drilled evaluation wells to properlycharacterize the reservoir and determinelocations for horizontal multilateral producingwells. In the first well logged with the MRScanner tool, the tool was part of a logging suite that consisted of resistivity, densityporosity, neutron porosity, acoustic andspectroscopy tools along with a formationpressure and sampling program.

In the past, conventional logging tools hadbeen unable to identify the OWC. Fluid gradientsfrom pressure data were not conclusive. NMRsaturations were acquired using data from the

MR Scanner tool. However, the saturationsindicated that the entire zone was pay, which wasknown to be incorrect based on production fromoffset wells.

The inability of the NMR tool to identify theOWC was attributed to less-than-optimal acquisi -tion parameters for the challenging case of thisHRLC pay. The reservoir contains low-viscositylight oil, and NMR experts concluded that the lackof success in locating the OWC resulted from usinga wait time that was insufficient to fully polarizethe native oil. A new acquisition sequence wascreated to address the underpolarization.

In the next well, the MR Scanner tool usedthis modified sequence, acquiring data from the

1.5-in. and 2.7-in. shells (above). The resultsagain indicated pay throughout the interval. OBMfiltrate had flushed out native water and oilthroughout the interval. Upon closer inspection ofthe results, a subtle increase in the computedwater volume was observed in the water leg fromthe 2.7-in. shell data compared with thatcomputed from the 1.5-in. shell.

The presence of OBM filtrate explained theoil in the water leg. The 2 to 3% of free water seenin the known oil leg was not so easily explained.It was assumed that the signal-to-noise ratio wasinsufficient for accurate volumetric calculationsand the increase in the computed water volumewas due to noise.

20 Oilfield Review

> OWC not found. Early logging attempts with the MR Scanner tool produced inconclusive results. In the well shown, the fluid saturations were computed using data from Shell No. 1 (Track 4) and ShellNo. 4 (Track 5). Oil (green) is indicated from top to bottom of the interval, but it is OBMF, not native oil.Free water (blue) is also seen throughout the interval in data from both shells. A DST located the OWCat X,204 ft. From the NMR data, its location is not obvious. The presence of free water above the OWCwas attributed to noise in the data and influenced the decision to take stationary measurements forfuture wells.

Free Water

%50 0

%50 0

%50 0

Free Water

%50 0

Oil Oil

%50 0

Bound Fluid Bound Fluid


%50 0

Gamma Ray

gAPI

Depth

ft0 200

Formation Density

g/cm31.95 2.95

%45 0

Shallow Resistivity

ohm.m0.1 1,000

Deep Resistivity

ohm.m0.1 1,000

X,100

OWCX,200

X,300

X,400

X,500

Neutron Porosity

Bad

Hole

Fla

g

Density Correction

g/cm3–0.2 0.2


Winter 2008/2009 21

In a third well, the petrophysicists acquireddata from the 4.0-in. shell, and from the twoshallower shells, using stationary measurements.Station data can be stacked to obtain a bettersignal-to-noise ratio. Continuous data wereacquired from the 1.5- and 2.7-in. shells. Asbefore, the data from the two shallower shellslooked similar, with a strong OBM filtrate

response. However, the stationary data from the4.0-in. shell clearly indicated free water in thewater leg. Surprisingly, looking an additional 1.3 in. [3.3 cm] into the formation made a bigdifference in identifying the OWC.

Given the success of using the stationarymeasurements from the deep shell, whichclarified the fluid distributions and explainedvertical trends in the reservoir fluids, three-shell

measurements were added to the standardlogging program. A first-of-its-kind triple-shellactivation sequence to simultaneously measureall three DOIs in stationary and continuouslogging modes was also introduced. This replacedmultiple passes and multiple stations previouslyrequired to obtain three DOIs (above). Today, MRScanner data play a critical role in the evaluation

> Pinpointing OWC with station logs. After the inconclusive results from early attempts to use MR Scanner datato locate the OWC, stationary measurements were used. Station logs permit stacking of data to achieve highersignal-to-noise ratios. Continuous fluid analysis from Shell No. 1 (Track 2) shows oil throughout the interval, butthe source is the OBM filtrate. A three-DOI sequence allows Shell No. 8 data to be acquired simultaneously withShells No. 1 and 4. The deeper shell detects native oil when invasion is not too deep. The two top stations, atX,103 and X,138 ft, are in the oil leg and have a clear oil signature, with little or no water signal from all threeshells. The three bottom stations, at X,142, X,165 and X,185 ft, are in the water leg, as evidenced by a strong free-water response. The absence of movable water in the upper two maps pinpoints the OWC at X,140 ft. Formationsamples confirmed the interpretation.

Oil

%50 0

Bound Fluid

Shell No. 1

%50 0GammaRay

gAPI0 200

Shallow Resistivity Depth

ohm.m ft0.1 1,000

Deep Resistivity

ohm.m0.1 1,000

Shell No. 1DOI = 1.5 in.



X,100

X,200

X,300

T1 time T1 time

Diffu

sion

Diffu

sion

T1 time

Diffu

sion

T1 time T1 time

Diffu

sion

Diffu

sion

T1 time

Diffu

sion

T1 time T1 time

Diffu

sion

Diffu

sion

T1 time

Diffu

sion

T1 time T1 time

Diffu

sion

Diffu

sion

T1 time

Diffu

sion

T1 time T1 time

Diffu

sion

Diffu

sion

T1 time

Diffu

sion


program of the field, and use of three-shell datahas been adopted as a best practice.

However, one limitation of the MR Scannertool’s ability to identify fluid type came to lightduring the analysis of the data from this field. NMRfluid properties, acquired from just a few inchesinto the formation, can be of little help inidentifying fluid contacts when invasion exceedsthe 4.0-in. DOI of Shell No. 8. In these cases, MDTmodular formation dynamics tester fluid gradientsand DSTs are required to provide the neededinformation.

Resolution SolutionThe trend in log interpretation and tool design hasbeen toward resolving and measuring increasinglythinner beds. This is critical in the interpretation

of anisotropic, laminated sand-shale sequences.However, it is not possible to resolve extremely thinlaminations with NMR tools because of the generalrequirement to stack successive measure ments toachieve an adequate signal-to-noise ratio. With aCMR tool, the smallest aperture window is a 6-in.[15.2-cm] station measurement. The MR Scannertool’s main-antenna measure ment window is 18 in.Recent developments demonstrate that even atthis lower resolution, NMR data are still useful inanalyzing and interpreting laminated sequences.NMR data provide complementary petrophysicalmeasure ments of the reservoir fluid propertiesthat resistivity and porosity measurementscannot.19

There are three general methods used toanalyze thin-bed reservoirs using well logs.20

Traditionally, imaging logs are used to characterizethe laminations, separating sand from shale.Other, lower-resolution data are then deconvolvedusing the higher-resolution image data. Theseoutputs are used in Archie’s equation to computewater saturation. One drawback to this method isthat imaging tools make very shallow readings andthus rely on good borehole conditions to acquirequality data. Furthermore, full quantitativeanalysis, using deconvolution techniques, is ofteninconclusive, and fluid properties and type arerarely quantifiable.

In a second method, NMR data quantify thetype and volume of fluids in a reservoir section. Butsince it is not possible to resolve thin beds withNMR tools, this method lumps the fluids togetherand differentiates bound fluid from free fluid.

22 Oilfield Review

> Integration of data in an anisotropic reservoir. A laminated reservoir isinferred from the OBMI image data (between Tracks 1 and 2). Processingbegan by calculating sand volumes (Track 1) from density-neutron andNMR data. Horizontal, Rh , and vertical, Rv , resistivities (Track 3), derivedfrom the Rt Scanner tool, were used to compute the electrical anisotropy(Track 4, green). The shales and the laminated sand-shale intervals exhibitanisotropy. The T2 distributions from the NMR data (Track 5) indicate bimodalfluid distributions in the intervals where sand is present, but not in theshales. Free fluid is to the right of the T2 cutoff, and clay-bound fluidassociated with the shale is to the left. Sand fractions were computed withinputs from both the NMR and the Rt Scanner data (Track 6). Sandresistivity (Track 7) was calculated from Rt Scanner data using intervals

with free fluid as defined by the MR Scanner tool. The NMR fluidsaturations indicate oil, water and OBM filtrate (Track 8). Hydrocarbon (HC)volumes are displayed for comparison (Track 9). They are computed fromNMR data (green), the Rt Scanner data (red) and Archie’s water saturationequation using traditional AIT array induction imager tool outputs derivedfrom the Rt Scanner data (black). Traditional Archie saturation underestimatedthe HC volume throughout the interval, significantly reducing the calculatednet pay and hydrocarbon in place. Finally, the petrophysicist identified thelaminated pay intervals using a modified Klein plot (inset) that incorporatesRt Scanner data, NMR data and high-resolution porosity measurements.Productive zones are highlighted on the log (Track 3, magenta).

PhisandPhisand NMR Rv , Rh Anisotropy

OBMI

GR

T2

DistributionFsand

Fsand NMRRt Scanner Rsand

NMR Rsand NMR Fluids HC Volume

Pay

Zone

s

900

1,000

Dept

h, m

800

700

0.5 10 0.4 0.2 0 0 0 0 0 0 0 0.2 0.4 0 0.2 0.410 1000.5 110 100 1,0005 10 1510 100

20 ft

Shale TotalPorosity

Cutoff

Sand

OBM

Water

Neutron Density

Rt ScannerData

AIT DataNMR Data

101

Rh, ohm.m

R v, o

hm.m

10–1

10–1

100

101

102

103

100 102 103

Fshale

Rsand

Rshale-v = 3.4Rshale-h = 0.58

Shale

Oil

19. Cao Minh C and Sundararaman P: “NMR Petrophysics in Thin Sand/Shale Laminations,” paper SPE 102435,presented at the SPE Annual Technical Conference andExhibition, San Antonio, Texas, September 24–27, 2006.Cao Minh C, Joao I, Clavaud J-B and Sundararaman P:“Formation Evaluation in Thin Sand/Shale Laminations,”paper SPE 109848, presented at the SPE AnnualTechnical Conference and Exhibition, Anaheim,California, USA, November 11–14, 2007.

20. Claverie M, Azam H, Leech R and Van Dort G: “AComparison of Laminated Sand Analysis Methods—Resistivity Anisotropy and Enhanced Log Resolution fromBorehole Image,” presented at the Petroleum GeologyConference and Exhibition (PGCE), Kuala Lumpur,November 27–28, 2006.

21. Anderson B, Barber T, Leveridge R, Bastia R, Saxena KR,Tyagi AK, Clavaud J-B, Coffin B, Das M, Hayden R,Klimentos T, Cao Minh C and Williams S: “TriaxialInduction—A New Angle for an Old Measurement,”Oilfield Review 20, no. 2 (Summer 2008): 64–84.

22. For more on the use of modified Klein plots: Cao Minh C,Clavaud J-B, Sundararaman P, Froment S, Caroli E,Billon O, Davis G and Fairbairn R: “Graphical Analysis of Laminated Sand-Shale Formations in the Presence of Anisotropic Shales,” Petrophysics 49, no. 5 (October 2008): 395–405.

23. Hürlimann MD, Freed DE, Zielinski LJ, Song YQ, Leu G,Straley C, Cao Minh C and Boyd A: “HydrocarbonComposition from NMR Diffusion and Relaxation Data,”Transactions of the SPWLA 49th Annual LoggingSymposium, Edinburgh, Scotland, May 25–28, 2008,paper U.


Winter 2008/2009 23

Within laminated sands, bound fluid is associatedwith shale laminations, and aggregate free-fluidvolume is associated with sand laminations.Diffusion data can provide fluid properties whensufficient quantities are available in the reservoirrocks. Although the depth of investigation isdeeper than that of imaging tools, NMRmeasurements are still quite shallow.

A recently introduced third method of evalu -ating laminated sand-shale sequences incorporateshigh-resolution porosity informa tion and inductiontool data, such as those from the Rt Scannertriaxial induction service.21 This tool measureshorizontal, Rh, and vertical, Rv, resistivities. Inlaminated sands, the hydrocarbon-bearing sandlaminations exhibit electrical anisotropy, indi -cated by a high Rv /Rh ratio, whereas water-bearing sand-shale sequences have low ratios.Reservoir evaluation based on the electricalanisotropy alone is not sufficient to prove thepresence of hydrocarbon. Anisotropic shalesexhibit high Rv /Rh ratios even in the absence ofhydrocarbon-bearing sand layers because offormation compaction.

In an even newer technique, MR Scannerfluid measurements are combined with the RtScanner Rv /Rh method to provide criticalinformation for properly analyzing complex sand-shale reservoirs. This method yields sandfraction (net-to-gross), sand porosity, sandresistivity and hydrocarbon saturation. The keyto maximizing the value of the information isintegration of the data from the various sources.

This complementary integra tion techniquewas recently demonstrated in a west Africareservoir. Characterized as having a thin sand-shale sequence, the well in this case study wasevaluated with MR Scanner data, Rt Scannerinformation and high-resolution porosity datafrom formation density and neutron logs.

Once thin beds with hydrocarbon potentialhad been identified from image data, thepetrophysicist followed an established workflowto interpret them:• Compute the sand fraction, Fsand, from the

porosity data.• Derive sand resistivity, Rsand, from Rv and

Rh data.• Compute an NMR Fsand from the T2 distributions.• Derive a new Rsand. • Compute the porosity of the sand layers, Phisand,

from both the density-neutron data and theNMR T2 distributions.

• Compare water saturations computed withinputs from the NMR data with those derivedfrom the Rt Scanner and high-resolutionporosity data.

With this workflow, the data analysis for thiswell began with identifying laminations from theimage log of the OBMI oil-base microimager(previous page). Rv values are greater than Rh

values, indicating electrical anisotropy. But, theRv /Rh ratio is high in clean shales as well as inlaminated sand-shale sections.

Shales have only bound fluids and thus aunimodal T2 distribution. Sand-shale sequencesexhibit a bimodal distribution, indicative ofmovable fluids in sand laminations. Thus, thesand fraction, Fsand, can be derived from the free-fluid portion of the NMR T2 data. This value isthen compared with the Fsand value derived fromthe density-neutron data. Rsand values arecomputed from the triaxial induction tool datafor both Fsand inputs. The NMR fluid analysis ofdata from the 2.7-in. shell indicates native oil andOBM filtrate.

Three analysis methods for water saturationare used to calculate the hydrocarbon volume:Archie’s water saturation equation, the Rv and Rh

method using triaxial induction tool data, andNMR fluid saturations. Rv and Rh crossplots arepresented using a Schlumberger modified Kleinplot technique.22 The selected points from thecrossplot are transposed on the log, identifyingquality reservoir intervals. Anisotropic shales plotin the nonpay region and may be ignored. Withthis technique, the log analyst quickly assessesthe reservoir and identifies potential pay zones.

This integrated approach resulted in an 80%increase in calculated net-to-gross values ascompared with classical resistivity-porositymethods. An increase of 18 net hydrocarbon feet[5.5 m] is derived from the NMR-based satura -tions, and there is an increase of 15 net feet [4.6 m]using the triaxial induction technique withoutNMR data. The conclusion from the log evalua -tion is that NMR data enhance the calculation forhydro carbons in place, while corroborating theresults of triaxial induction-based laminated-sand analysis. The technique offers thepetro physicist the ability to identify qualityreservoir intervals and eliminate nonproductiveanisotropic shale intervals from evaluation.

Mapping the Future of Magnetic ResonanceMagnetic resonance logging has transcended itsniche market status and attained a high level ofacceptability within the petrophysical com -munity. It will never supplant resistivity andnuclear porosity measurements; nor should it.NMR data offer the oil and gas industry analternative source for certain measurements,including porosity and fluid saturations, yetthere are limitations inherent in the physics—asthere are in all petrophysical measurements.

New enhancements to the software formaking 2D maps of reservoir fluids create staticsnapshots that can then be added to 2D logs. Inaddition, the software has the capability topresent saturations and fluid properties in avideo format, allowing visualization of thechanges that take place laterally along thewellbore and horizontally into the formation.

Laboratory-based NMR fluid measurementshave been and will continue to be transferred tothe downhole environment. Obtaining the proper -ties of sampled fluids while tools are still downholeoffers the closest approximation to in situmeasurements available. An NMR oil classificationsystem exists based on molecular properties, andapplying that classi fication to downhole fluids willaid in proper reservoir development.23

But one of the unique aspects of NMRmeasurements is that they offer the onlytechnique that can see and distinguish differentfluids in situ, without flowing them. Evendownhole samples may not provide true fluidproperties because of changes to the fluidsduring flow. Sampled fluids do not reflect thetrue distri bution of fluids in the reservoir, onlythose that are mobile. As NMR techniques areapplied, and fluid variations within reservoirs areidentified, the complicated nature of oil and gasproduction is better understood. With reservoirunder standing come informed productionpractices, greater efficiencies and higherrecovery rates.

Viable options exist for future development.Researchers continue to develop NMR-basedcarbonate answers. Deeper measurements are agoal, but tools to acquire them are years away.Although there may never be an NMRmeasurement from the virgin reservoir, fluidproperties from LWD tools offer a glimpse into thereservoir fluids unaffected by mud filtrate. Such asolution would overcome the problem identified inthe Saudi Aramco HRLC wells. Other challengesawait further research and development.

It took 30 years to develop a workablemagnetic resonance tool for downhole environ -ments. NMR measurements have continued toevolve along with the tools used to acquire thedata. The most recent developments put acolorful visualization technique in the hands ofthe petrophysicist. The art and science of NMRapplications have been combined to provide analternative to static 2D logs of the past. Thesenew dimensions in NMR logging have ushered ina powerful tool for reservoir analysis. But thebest may be yet to come. —TS


nuclear magnetic resonance comes out of its shell

Documents