2 Oilfield Review

Trends in NMR Logging

For help in preparation of this article, thanks to Rob Badry,Calgary, Alberta, Canada; Kamel Bennaceur, Dylan Davies,Robert Freedman and Bruce Kaiser, Sugar Land, Texas,USA; Dale Logan, Midland, Texas; Robert Kleinberg,Ridgefield, Connecticut, USA; Don McKeon, Montrouge,France; LSD Onuigbo, Lagos, Nigeria; and Lee Ramsey andFrank Shray, Houston, Texas. CMR, CMR-200, CMR-Plus (Combinable MagneticResonance), DMR (Density-Magnetic ResonanceInterpretation Method), FracCADE, FMI (Fullbore FormationMicroImager), MDT (Modular Formation Dynamics Tester),OFA (Optical Fluid Analyzer), PowerSTIM and TLC (ToughLogging Conditions) are marks of Schlumberger. MRIL andMRIL-Prime are marks of NUMAR Corporation.

During the past decade, the remarkable technology of nuclear magnetic resonance

(NMR) logging has improved continually. Oil companies are using NMR measurements

for an ever-growing number and range of applications, such as characterizing formation

fluids during reservoir evaluation and assessing formation producibility. Now the

measurements provided by these tools are dramatically changing well-completion

designs and reservoir development.

In the last decade, petrophysicists applauded thenewly arrived pulsed nuclear magnetic reso-nance (NMR) logging tools for their ability tosolve difficult formation-evaluation problems.Service companies continue to invest significantresearch effort into refining NMR measurements.The result has been a steady stream of tooladvancements and new applications. In the mid-1990s, the introduction of higher rate pulsing

techniques extended the fluid-mobility character-ization capabilities of these tools. Recently, therehave been impressive improvements in data-acquisition capabilities, which have led to signif-icant increases in logging speeds.

A fundamental advantage of the latest gener-ation of NMR tools is the ability to provide awider scope of information about reservoirs than ever before. NMR data can answer manykey questions for nearly everyone involved in

David AllenCharles FlaumT. S. RamakrishnanRidgefield, Connecticut, USA

Jonathan BedfordLondon, England

Kees CastelijnsNew Orleans, Louisiana, USA

David FairhurstSan Antonio, Texas, USA

Greg GubelinNick Heaton Chanh Cao MinhSugar Land, Texas

Mark A. NorvilleMilton R. SeimKerns Oil and Gas, Inc.San Antonio, Texas

Tim PritchardBG Group plcReading, England

Raghu RamamoorthyKuala Lumpur, Malaysia

Autumn 2000 3

exploration and production, including reservoirengineers, completion engineers, geologists andpetrophysicists. For example, completion engi-neers now use NMR measurements to designhydraulic-fracture treatments for reservoir stimu-lation. Reservoir engineers assess rock qualitywith high-resolution NMR data to locate verticalpermeability barriers and improve productionmanagement. Geologists and petrophysicists

have an improved understanding of pore geometryfor depositional analysis from decay-time distri-butions. Hydrocarbon characterization is alsoimproved by interpreting NMR logs in combinationwith other logging measurements. The result is amore accurate evaluation of well producibility.

This article first reviews recent advances inNMR tool technology and discusses how some ofthese developments, such as enhanced precision,

increased logging speed and high-resolution mea-surements, are leading to new NMR applications.Field examples show how this information is usedto design well completions, and how NMR anddata from wireline formation testers are provid-ing highly efficient, low-risk ways to evaluatewell producibility. Finally, we discuss the latestdevelopments in evaluating carbonate formationswith NMR.

New Tool AdvancesThe CMR Combinable Magnetic Resonance tool,introduced by Schlumberger in 1995, is runpressed against the borehole by a bowspring. Ashort directional antenna sandwiched between apair of optimized magnets focuses the CMR mea-surement on a 6-in. [15-cm] vertical zone located1.1 in. [2.8 cm] inside the formation. These fea-tures and the tool’s enhanced signal-to-noisedata-acquisition electronics provide accurate,high-precision measurements in the formationwith high vertical resolution.1

With the increased price of crude oil and sub-stantial rig rates offshore, it’s more importantthan ever to be able to make decisions rapidly.The latest addition to the CMR family of tools,the CMR-Plus tool, addresses this need (left). Thenew tool includes several improvements over itspredecessor, the CMR-200 tool, including a newmagnet design with a longer prepolarizing field thatincreases logging speeds to 3600 ft/hr [1097 m/hr]in fast-relaxation environments. This compact,lightweight, rugged tool is 15.6 ft [4.8 m] longand weighs 450 lbm [204 kg]. A low-profile skiddesign allows logging in holes as small as 5 7⁄8 in.[15 cm]. A new pulse-acquisition sequence, calledthe enhanced-precision mode (EPM), is coupledwith an upgraded electronics package toincrease the signal-to-noise ratio and improvehigh-precision measurements for evaluatingreservoirs (next page, top) .2

The ability to make fast, high-precision NMRmeasurements is changing the way that engineersperceive the producibility of their wells. For exam-ple, zones that might have been considered unpro-ductive because of high water saturation and thepotential for producing excessive water might, infact, be worth examining to determine if the wateris immovable. For example, in a development wellin South America, a CMR-Plus log revealed that azone of apparent high water saturation was irre-ducible—the formation would produce water-freehydrocarbons. Previously, the operator hadbypassed this zone during field development.Based on the new information, the operator perforated the zone and produced dry gas, adding20 Bcf [566 million m3] of gas reserves.

The higher logging speeds made possible bythe CMR-Plus tool enables operators to econom-ically acquire data over longer intervals thatinclude zones that initially were of little interest.

NUMAR Corporation, a subsidiary of Hallibur-ton, developed the MRIL Magnetic ResonanceImaging tool, which incorporates a long perma-nent magnet to create a static lateral field in theformation. This tool is run centralized in the bore-hole, and the measurement volume consists of aconcentric cylindrical resonant shell with a length

4 Oilfield Review

Borehole wall

Antenna

Blind zone

Sensitivezone

Wear plate

Permanent magnet

Permanent magnet

Bowspringeccentralizer

CMR skid

CMR-200 tool

CMR-Plus tool

6 in.

15.6

ft

Sensitive zone

Electroniccartridge

12 in.

30 in.

> CMR tool design. The CMR-Plus tool uses a similar antenna, magnet config-uration and electronics as the CMR-200 tool. The pair of permanent magnetscreates a resonant-field sensitive zone in the formation (top right and lowerleft). However, the CMR-Plus magnets (lower right) are 30 in. [76 cm] long toprovide prepolarization of the hydrogen spins while logging continuously. Thisnew design feature allows the CMR-Plus tool to log faster.

1. 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 toUse Borehole Nuclear Magnetic Resonance,” OilfieldReview 9, no. 2 (Summer 1997): 34-57.

2. Freedman, R: “Dual-Wait Time Processing for MoreAccurate Total and Bound Fluid Porosity,” U.S. PatentApplication 156,417, 1998.McKeon D, Cao Minh C, Freedman R, Harris R, Willis D,Davies D, Gubelin G, Oldigs R and Hurlimann M: “AnImproved NMR Tool for Faster Logging,” Transactions of the SPWLA 40th Annual Logging Symposium, Oslo,Norway, May 30-June 3, 1999, paper CC.

3. Prammer MG, Bouton J, Chandler RN and Drack ED:“Theory and Operation of a New Multi-Volume NMRLogging System,” Transactions of the SPWLA 40thAnnual Logging Symposium, Oslo, Norway, May 30-June 3, 1999, paper DD.

4. Coates G and Denoo S: “The Producibility AnswerProduct,” The Technical Review 29, no. 2 (1981): 54-63.

5. Heaton N, Cao Minh C, Freedman R and Flaum C: “HighResolution Bound-Fluid, Free-Fluid and Total Porositywith Fast NMR Logging,” Transactions of the SPWLA41st Annual Logging Symposium, Dallas, Texas, USA,June 4-7, 2000, paper V.

Autumn 2000 5

of 24 in. [61 cm], and a thickness of approximately0.04 in. [1 mm]. The average diameter of the reso-nant shell is about 15 in. [40 cm] and is set by theoperating frequency of the tool. In a 10-in. [25.4-cm]borehole, this leads to a depth of investigation(DOI) of 2.5 in. [7.6 cm]. A large DOI helps reducethe sensitivity to rugosity in many boreholes.

The latest version of the NUMAR tool, theMRIL-Prime, incorporates improvements toincrease logging speed and efficiency.3 It has 3-ft[1-m] prepolarizing magnets both above and belowthe antenna for logging upward and downward,and a nine-frequency multishell measurementcapability. Each measurement shell can be pro-grammed with a different pulse sequence, and the

measurement can be switched between differentshells by changing the frequency. The total varia-tion in the depth of investigation of the nine shellsis about 1 in. [2.5 cm]. The multifrequency opera-tion provides a total porosity measurement, andmultiparameter data acquisition with differentpulsing sequences in each shell.

The tool is available in two sizes. The standardtool diameter is 6 in.; the length is 53 ft [16 m]; andthe weight is 1500 lbm [680 kg]. A slim tool has a47⁄8-in. [12-cm] diameter, is 50 ft [15 m] long, andweighs 1300 lbm [590 kg]. These tools enable log-ging in hole sizes from 5 7⁄8 in. to 12 1⁄4 in. [31 cm].Logging speed in short-polarization-time envi-ronments is 1440 ft/hr [440 m/hr], and 700 ft/hr[213 m/hr] for the slim version of the tool.

High-Resolution NMRThe identification and quantification of rockgeometry and fluid mobility on the basis of thefluid’s nuclear spin relaxation characteristics areamong the most important contributions of NMRlogging. The separation of porosity into bound-fluid and free-fluid components is essential forevaluating reservoir producibility.4 In thin, lami-nated formations, producibility depends not onlyon the net ratio of bound-fluid to free-fluid vol-umes, but also on the relative location of the twofluid volumes within the different laminatedbeds. Measurements are useful in this capacityonly if they are sensitive to spatial variations ona length scale comparable to the laminationthickness. The production engineer can use high-resolution NMR data to evaluate producibility ofthinly laminated sections, obtain accurate hydro-carbon pore volume and identify vertical perme-ability barriers that can help avoid producingundesired water from nearby aquifers. The com-pletions engineer can use high-resolution data toaccurately position perforating, fracturing andformation stimulation designs.

High-resolution producibility—The verticalresolution of an NMR measurement is deter-mined by the antenna length, the acquisitionsequence signal-to-noise ratio and the loggingspeed. For example, CMR-200 tool measure-ments combine overlapping phase-alternatedpairs (PAP) of Carr-Purcell-Meiboom-Gill (CPMG)pulse-echo sequences and a short antenna toresolve beds as thin as 6 in. (see “Basics ofNMR Logging,” page 12 ). The long prepolariza-tion magnet in the CMR-Plus tool allowsnonoverlapping PAP measurement acquisition atlogging speeds up to 3600 ft/hr with only a minorreduction in vertical resolution. In practice, thevertical resolution for most NMR measurementsis degraded by the depth-stacking techniqueused to improve the signal-to-noise ratio neces-sary for T2-inversion processing.

Recent developments in high spatial-resolu-tion CMR logging stem from a new processingmethod optimized for high-resolution answersand an EPM data-acquisition scheme.5 In high-resolution processing, T2 inversion is performedwithout applying any vertical averaging of theecho data. The high-resolution inversion proce-dure differs from standard inversion in severalaspects. Standard inversion typically usesbetween 30 and 50 T2 components, which encom-pass the entire range of possible relaxation timesexhibited by formation and drilling fluids. High-resolution processing uses only two to five T2

components (above left). Furthermore, these

> Enhanced-precision mode (EPM). The EPM measurement is a new multiwait Carr-Purcell-Meiboom-Gill(CPMG) pulse-echo acquisition sequence designed to improve the short T2, or bound-fluid, measurementprecision. The EPM measurement consists of one long wait-time pulse sequence that measures allthe T2 components, followed by a series of short wait-time sequences optimized for the early T2 com-ponents from small pores. The short wait-time sequences are stacked to decrease the noise of themeasurement, resulting in greater precision in the early echo data. This improves the precision in thebound-fluid volume and total CMR porosity measurements.

> High-resolution, multilevel data-windows processing. Vertical stacking (left) of echo trains andinversion are used to obtain a T2 distribution (middle) for the averaged data. An equivalent T2 distri-bution is formed with a reduced number of T2 bins of approximately equal amplitude (right). The totalporosity, φ, and logarithmic mean T2 value of the reduced distribution are identical to those of theoriginal distribution.

One long wait-time subsequence(to measure the long T2 components)

Ten short wait-time subsequences(to measure the short T2 components)

Long wait-time

Short wait-time Long CPMG

Long CPMG

Short CPMG

Vertical stackingStandard T2 inversion with

stacked dataHigh-resolution T2 inversion with

single CPMG sequence or depth data

φ φ

components are selected by analysis of the stan-dard T2 distribution obtained from the stackeddata at each depth.

For example, if the stacking interval at a par-ticular depth consists of a clean sand giving a sin-gle peak in the T2 distribution centered between50 and 300 msec, then the T2 components used inthe high-resolution inversion will be derivedentirely from this range. By using a small numberof “optimal” T2 values, the high-resolution inver-sion provides total porosity with high precision. Toensure accuracy and consistency with standardprocessing, the high-resolution porosity is scaledsuch that the mean value matches that of the

standard porosity. This strategy is similar to thealpha-processing method used to derive high-resolution density measurements.6

Separation of the high-resolution porosityinto bound- and free-fluid volumes is straight-forward. Because the free-fluid signal decaysslowly, it contributes to a large number ofechoes. Consequently, the free-fluid volume canbe computed with high precision using standardinversion without stacking the data. This pro-vides a high-resolution free-fluid volume esti-mate. A high-resolution bound-fluid volume isobtained as the difference between the high-resolution total porosity and the high-resolutionfree-fluid volume.

High-resolution NMR porosity and perme-ability data are essential for evaluating thinlylaminated reservoirs. For example, data in anAustralian well drilled in a thinly bedded sand-shale sequence show little correlation betweencore porosity and the openhole neutron anddensity porosity logs (above). The core porositydisplays large fluctuations in the lower zone—as much as 20 p.u. over a 1-ft [0.3-m] interval, asmight be expected in a laminated formation. Insuch zones, traditional openhole logs and tradi-tional NMR logging with three- or five-level aver-aging often provide inadequate vertical resolutionand limited ability to pick zones for perforating.

6 Oilfield Review

> Australian well with thinly bedded sand-shale sequence. Traditional density (blue) and neutron(red) logs are shown in Track 1. A five-level averaged CMR total porosity (black) is compared withcore porosity in Track 2. The high-resolution CMR total porosity (black) is compared with core porosityin Track 3. Note how well the high-resolution total porosity captures the sharp porosity variations seenthroughout the core data. The five-level averaged CMR permeability (blue) is compared with core datain Track 4, and the high-resolution permeability (blue) in Track 5. Again, the high-resolution CMR logagrees well with the permeability variations seen in the core data. A gamma ray log is shown in thedepth track, and the CMR T2 distributions are shown in Track 6.

XX80

XX90

X100

X110

X120

Depth,ft

Gammaray

Density porosity

Neutron porosity Core porosity Core permeability Core permeability T2 distributionCore porosity

30 0p.u. 30 0p.u. 30 0p.u.

TCMR, 5-LEV

0.1 1000mD

KTIM, 5-LEV

0.1 1000mD 0.3 3000msec

KTIM-HRTCMR-HR

Autumn 2000 7

The high-resolution total CMR porosity, plotted inTrack 3, captures the sharp porosity fluctuationsand agrees well with the core data. The high-res-olution Timur-Coates permeability estimatesderived from the CMR measurements also com-pare favorably with the core permeabilitiesshown in Track 5. Timur-Coates permeability esti-mates are discussed in detail below.

In Canada, the CMR-Plus tool was run at1200 ft/hr [366 m/hr] in a shaly-sand formationwith fine laminations (above). The high-resolutionfree-fluid and bound-fluid volume logs correlatewell with the fine laminations shown by the FMIFullbore Formation MicroImager display in thelower zone. In this well, the high-resolution free-

and bound-fluid logs anticorrelate and compen-sate each other so much that the high-resolutiontotal porosity log provides little indication of thelaminations. Single CPMG estimates furtherenhance resolution, most notably over the sectionfrom X100 ft to X120 ft. The high-resolution CMRlogs in the upper zone from XX90 ft to X100 ft over-lie the traditionally processed logs and show littleevidence of laminated beds.

This example demonstrates that robust high-resolution NMR logs can be obtained with theCMR-Plus tool at fast logging speeds using thisinnovative processing technique. Laminations of4 to 6 in. [10 to 15 cm] are resolvable, and pre-cisely locate the thin, high-porosity pay zonesbetween the shale beds.

High-precision permeability indicator—Another important contribution of NMR loggingis its capability to provide a continuous per-meability measurement. In thinly laminated formations, permeability can vary by orders ofmagnitude within a few inches. Under these conditions, it is important to obtain a continuouspermeability estimate with the highest possibledepth resolution. The two widely applied perme-ability transforms used today, based on NMR

> High-resolution NMR log correlation with FMI images. In Tracks 1 to 3, the traditional bound-fluid,free-fluid and total porosity logs processed with five-level depth averaging (black) are compared withhigh-resolution curves (green), and corresponding single CPMG-derived estimates (red). Neutron(blue) and density (red) porosity logs are shown in Track 4 and the deep (red) and shallow (green)resistivity logs are shown in Track 5. Over the top zone, from XX90 to X100 ft, the high-resolution logsare relatively featureless and overlie the averaged logs. However, in the lower zone from X100 toX120 ft, the high-resolution logs show increased activity due to the fine laminations that are apparentin the FMI image shown in Track 6. Note that the bound-fluid and free-fluid logs anticorrelate, andcompensate for one another, leading to a total porosity log that provides little indication of the lami-nations. The substantial fluctuations in free fluid and bound fluid in the lower section correlate wellwith the laminations apparent in the FMI image.

6. Galford JE, Flaum C, Gilchrist WA and Duckett S:“Enhanced Resolution Processing of CompensatedNeutron Logs,” paper SPE 15541, presented at the SPEAnnual Technical Conference and Exhibition, NewOrleans, Louisiana, USA, October 5-8, 1986.

XX90

Depth,ft

Gammaray

X100

X110

X120

Bound-fluidvolume

Single CPMG

5-level

HR

Free-fluidvolume

Single CPMG

5-level

HR

Totalporosity

Single CPMG

5-level

HR

p.u.40 0 p.u.40 0 p.u.40 0

Density porosity

Resistivity

Neutron porosity

p.u.40 0

Rxo

ohm-m0.2 2000

Bound-fluid HR

p.u. msec40 0.30 3000

Free-fluid HR

FMI

90-in. T2 distribution

measurements, are the Timur-Coates equationand the Schlumberger-Doll Research (SDR) equa-tion (above). The Timur-Coates equation esti-mates permeability using total porosity and theratio of the free-fluid volume (FFV) to the bound-fluid volume (BFV). The SDR transform is basedon the logarithmic mean of T2 and total porosity.

Although the high-resolution porosity process-ing discussed earlier can be used to derive high-resolution SDR and Timur-Coates permeability,there is an alternative approach that can yield bet-ter results in high-noise environments. It has beenobserved that the sum of all spin-echo amplitudesis proportional to the product of porosity and theaverage T2. This, in turn, correlates well with per-meability (above right).7 In addition, the sum ofechoes has a high signal-to-noise ratio, so it canbe interpreted without stacking. This leads to ameasurement with higher vertical resolution.

The new high-resolution NMR permeabilityindicator is derived from the sum of echo ampli-tudes and is directly proportional to the areaunder the spin-echo decay envelope (right). Thevertical resolution achievable with this noveltechnique is equal to the tool antenna apertureplus the distance traveled during one CPMG

8 Oilfield Review

Permeability increases with increasing porosity...

Echo

am

plitu

de

Time

...and increasing T2 or FFV/BFV

> Permeability and the NMR T2 signal. The top row shows a series of hypotheticalNMR signals—echo-decay envelopes—for increasing porosity and permeabilityin which the T2 decay time remains constant. The bottom row shows a series ofsignals in which the porosity remains constant, but the T2 decay time and com-puted permeability increase from left to right. The area under the echo-decayenvelope and the computed permeability increase with porosity and decay time.

NMR high-resolution permeability indicator.The new indicator is simply the sum of the echoamplitudes (top) and is directly proportional tothe area of the echo-decay envelope. The high-resolution sum-of-echoes permeability indicatoris compared with laboratory-measured perme-abilities on 30 core samples from four wells fromdifferent parts of the world (middle). The linearcorrelation (R2 = 0.95) is good for over six ordersof magnitude, and compares favorably with thatprovided by the standard NMR-based permeabilitytransforms. The high-resolution curve shown inthe graph was calibrated for a sum of 600echoes. The prefactor and exponent used in thecomputation are adjusted according to localconditions. The signal-to-noise gain (bottom),and therefore the optimal number of echoesused in the computation, depends on the T2 signal-decay rate of the formation.

Timur-Coates equationkTIM=aφm (FFV/BFV)n

m~4, n~2

SDR equationkSDR=bφm (T2LM)n

m~4, n~2

> NMR permeability transforms. The Timur-Coatesequation contains total porosity, φ, and the ratioof the free-fluid volume (FFV) to the bound-fluidvolume (BFV). The Schlumberger-Doll Research(SDR) equation also contains total porosity, butuses a logarithmic mean T2 (T2LM) in place of theFFV-to-BFV ratio. The exponents are typically 4and 2, but can vary with local conditions.

10,000

1000

100

10

1

0.1

0.011000

R2 = 0.95

10,000 100,000

Mea

sure

d pe

rmea

bilit

y, m

D

Sum of echoes

35

30

25

20

15

10

5

0100 1000 10,000

SNR

(Ech

oes

sum

)/SN

R (F

irst e

cho)

Number of echoes

T2 = 0.01 sec

T2 = 0.05 sec

T2 = 0.1 sec

T2 = 0.2 sec

T2 = 0.4 sec

T2 = 0.6 sec

Permeabilityindicator

NΣecho(n)n = 1

>

Autumn 2000 9

sequence plus the polarization time, typically 7 to9 in. [18 to 23 cm] for the CMR-Plus tool.

An example from the Ekofisk chalk formationin the North Sea demonstrates the excellentvertical resolution of the new sum of echo-amplitudes permeability indicator (right). While itis widely believed that chalk formations arehomogeneous, the FMI images in this well revealsubtle laminations in the chalk. Gamma ray logs,density-neutron logs and the standard five-levelstacking SDR transform permeability log with 30-in. [76-cm] resolution offer no indication ofthese laminations. However, the high-resolutionpermeability indicator log shows clear perme-ability variations consistent with the laminationsseen in the FMI images.

Improving Well StimulationsWells completed in low-to-moderate perme-ability reservoirs frequently require hydraulic-fracture stimulation to produce at economicrates. Because the fracture treatment often is themain cost driver associated with the completion, operators are looking for ways to ensure cost-effective stimulation designs.

The effectiveness of a stimulation design canbe affected dramatically by the permeability of areservoir. Traditionally, permeability data fromwhole core, rotary sidewall core and percussioncore are used with pressure transient analysisand production history-matching to estimatereserves and design well completions. Docu-menting production histories requires a time-consuming delay, and is useful only for remedialcompletions. Acquiring core presents mechanicalrisks and is often prohibitively expensive.Frequently, laboratory core analysis does not fullyrepresent permeability conditions downhole,and, at best, provides only a small sample thatmight not be representative of the zone of inter-est. The preferred method of pressure transientanalysis involves a flow period followed by apressure buildup test, which involves significantpersonnel and equipment costs. In addition, thereare potential workover expenses for running tub-ing, and the testing process delays post-stimula-tion production.

Since permeability data usually are sparse,the stimulation engineer may derive a compositepermeability. This is typically a bulk-averagedpermeability, sometimes from nonrepresentativesamples from sweet spots or high-permeabilityzones. In addition, when continuous detailed per-meability data are not available, often the strategyis to design a fracture with only one stage basedon the averaged rock strength and permeabilityof that zone. Frequently, this results in an unreal-istic and unsuitable fracture design.

For example, a nonoptimal stimulation design,based on composite permeability, might producea fracture of insufficient length and with a large,impractical vertical extent. An optimal designresults in creating a narrow and deep conduit intothe formation. To improve hydraulic-stimulationmodels and overcome the traditional limitationsinherent in obtaining permeability information,stimulation engineers and operators have beeninvestigating methods of reliably estimatingpermeability profiles with wireline logs.8 Under-standing high-resolution permeability distributionsacross the pay zone optimizes the stimulationtreatment because higher permeability streakscan be mapped and correctly included in the frac-ture design.

NMR-derived permeability can provide accu-rate, continuous input into a multilayer stimula-tion design program, such as FracCADEstimulation software. Permeability governs thetransport fluid that disappears—called leakoff—into the fresh sandface during a fracture stimula-tion operation. The leakoff parameter is critical

to fracture design. If the leakoff parameter is settoo high, then too much fluid is pumped. Thiswastes proppant and fluid and results in unnec-essary costs. If the leakoff parameter is set toolow, then a screenout may occur—resulting ininsufficient fracture length and diminished pro-duction along with mechanical risks to comple-tions integrity and wasted proppant. Inputtingcorrect permeability data is important. Perme-ability can easily change by several orders ofmagnitude within the same sand bed, even whileporosity remains constant.

Depth,ft

Gamma rayAPI

FMI High-resolutionpermeability

Standardpermeability

Density porosity T2

Neutron porosity

High-resolutionpermeability

mapHigh-resolutionpermeability

indicator0 100 0.01 3000100 50 0.30mD p.u. msec

X410

X420

X430

X440

X450

> Ekofisk chalk formation in the North Sea. The FMI image in Track 2 showsthe presence of many thinly laminated chalk beds between X410 and X450 ft.In the FMI image, light yellow indicates resistive low-porosity chalk, and darkbrown indicates more conductive, higher porosity chalk. The high-resolutionpermeability indicator log (blue) is shown in Track 4. The permeability imagein Track 3 is derived from the high-resolution permeability indicator log, andit correlates well with the FMI image. In the permeability image, light yellowindicates low-permeability chalk and dark brown indicates higher permeability.The T2 distributions are shown in Track 6. The gamma ray (black) in Track 1,density (red) and neutron (blue) porosity logs in Track 5, and the standardfive-level stacked NMR permeability (dashed black) in Track 4 show littleevidence of these laminations.

7. Sezginer A, Cao Minh C, Heaton N, Herron M, FreedmanR and Van Dort G: “An NMR High-Resolution PermeabilityIndicator,” Transactions of the SPWLA 40th AnnualLogging Symposium, Oslo, Norway, May 30-June 3, 1999,paper NNN.

8. Fairhurst DL, Marfice JP, Seim MR and Norville MA:“Completion and Fracture Modeling of Low-PermeabilityGas Sands in South Texas Enhanced by MagneticResonance and Sound Wave Technology,” paper SPE59770, presented at the SPE CERI Gas TechnologySymposium, Calgary, Alberta, Canada, April 3-5, 2000.

Kerns Oil and Gas, Inc. has been using totalCMR porosity and permeability measurements toprovide critical input parameters to their stimu-lation designs for tight gas-sand wells in southTexas, USA. For example, two FracCADE frac-ture-stimulation designs were prepared for onewell (below). Both designs used the same rockstrengths and amounts of fluid and proppant.The first design incorporated traditional perme-ability data from sidewall core estimates and

published data. The resulting high, short fracturedesign is unsuitable for effective production. Thelateral extent of the fracture is 600 ft [183 m]into the reservoir.

A second fracture design, based on the con-tinuous logged NMR permeability data, leads toa longer fracture—penetrating 800 ft [244 m]deep into the formation—with effective prop-pant placed at least twice as far into the forma-tion as the first design.

If the continuous high-quality NMR perme-ability data were not available, the stimulationengineer would have been misled by simulationresults such as those from the first model.Achieving the desired fracture length wouldrequire increasing the pumping schedule—higher pumping rates and increased fluid andproppant volumes. The result is a more expensiveand less efficient stimulation job. The possibilityof screenout is much higher with the overde-signed job. A detailed profile of permeability withdepth provided by NMR logs is helping Kerns Oiland Gas continue its remarkable track record of92% success in achieving or exceeding their pro-duction objectives with stimulation treatments(next page, top).

In some cases, the cost of gathering all thedata for optimizing fracture geometry prohibitsits full implementation, leading to a suboptimalstimulation. Here, continuous NMR permeabilitydata allow the stimulation engineer to considermultistage stimulation designs. For example, thehighest permeability zones can be stimulatedeffectively with a smaller, shorter, lower costfracture. The payout from immediate productionis realized early. Then, after this zone has beenunder production for some time, the reservoirpressure in the high-permeability zone drops,which results in an increased stress contrastbetween the gas zone and the shale boundary

10 Oilfield Review

> Stimulation designs. The FracCADE program was used to compare two fracture stimulationdesigns in a tight gas-sand formation. The first design (top) used permeability information fromcore samples and local knowledge. It overestimated the permeability by a factor of 10, resulting in an excessively high and short fracture unsuitable for optimal production. The second design(bottom) was based on detailed continuous NMR permeability logging data. It resulted in a longerfracture, 800 ft [244 m], of limited height more suitable for enhanced gas production. The prop-pant in the second design was placed twice as deep as that predicted by the first design.

9. Kerchner S, Kaiser B, Donovan M and Villareal R:“Development of a Continuous Permeability MeasurementUtilizing Wireline Logging Methods and the ResultingImpact on Completion Design and Post CompletionAnalysis,” paper SPE 63259, presented at the SPE AnnualTechnical Conference and Exhibition, Dallas, Texas, USA,October 1-4, 2000.

10. The September 21, 2000 gas price was about $5.28/Mscf.

< 0.0 lbm/ft2

0.0-0.1 lbm/ft2

0.1-0.2 lbm/ft2

0.2-0.3 lbm/ft2

0.3-0.4 lbm/ft2

0.4-0.5 lbm/ft2

0.5-0.6 lbm/ft2

0.6-0.7 lbm/ft2

0.7-0.8 lbm/ft2

> 0.8 lbm/ft2

X4800

X4900

X5000

X5100

X5200

X4800

X5000

Dept

h, ft

Dept

h, ft

X5100

X5200

X4900

3600 4400Stress, psi Fracture width at wellbore, in. Fracture half-length, ft

800-ft half-length

Greatest concentrationof proppant

Greatest concentrationof proppant

0.1 0 0.1 0 400 800 1200

3600 4400Stress, psi Fracture width at wellbore, in. Fracture half-length, ft

0.1 0 0.1 0 400 800 1200

< 0.0 lbm/ft2

0.0-0.1 lbm/ft2

0.1-0.2 lbm/ft2

0.2-0.3 lbm/ft2

0.3-0.4 lbm/ft2

0.4-0.5 lbm/ft2

0.5-0.6 lbm/ft2

0.6-0.7 lbm/ft2

0.7-0.8 lbm/ft2

> 0.8 lbm/ft2

600-ft half-length

Proppantconcentration

Proppantconcentration

Autumn 2000 11

layers. This increased stress contrast allows asecond fracture stimulation to penetrate deeperinto the higher permeability zone without risk ofincreasing fracture height.

Such an approach was developed by Conocoduring a project to understand the economicimpact of using continuous NMR permeabilitydata on hydraulic-stimulation projects.9 An eco-nomic model based on net present value (NPV)was used to test the sensitivity of different variables in the fracture optimization (see“Taking a Calculated Risk,” page 20 ). In onewell, a traditional single-stage design required288,192 gal [1090 m3] of fluid and 935,216 lbm[424,214 kg] of proppant to achieve a 795-ft [242-m] optimum fracture length at a cost of$320,000. The estimated recovery in three yearswould total 2.2 Bcf [62 million m3] of gas.

By first targeting the high-permeability zonewith a small stimulation treatment and followinglater with a second treatment in the lower per-meability zone, the two hypothetical treatmentsused a total of 186,383 gal [705 m3] of fluid and438,079 lbm [198,713 kg] of proppant. The firstzone fracture half-length was 388 ft [118 m] andthe second was 1281 ft [390 m] at a total cost of$254,000. The estimated reserves recovery inthree years was 2.5 Bcf [70 million m3] of gas

(above). The two-stage stimulation design usingcontinuous NMR permeability data results in a$66,000 cost-savings and increased productionof 292 MMcf of gas, which is worth approxi-mately $1.5 million at current gas prices.10

Combining continuous NMR permeability log-ging data with hydraulic-fracture stimulationdesigns is one of the objectives of thePowerSTIM initiative, which will be discussed inan upcoming Oilfield Review article.

> Well production results using NMR logging data. Before stimulation, several wells producedno flow. Production estimated using permeabilities derived from NMR logging correlates wellwith observed post-stimulation production.

2500

3000

2000

1500

1000

500

00 200 400 600 800 1000

Time, days

Cum

ulat

ive

prod

uctio

n, M

Msc

f

Composite permeability Continuous measured permeability

> Cumulative production of traditional and NMR-based stimulation designs. The cumulativeproduction was computed using ProCADE software based on the height of the net pay,bottomhole static pressure and permeability. The red curve shows the production basedon the traditional composite permeability estimates and a single-stage fracture-stimulationdesign. The green curve is based on the continuous NMR permeability data and a two-stage stimulation design.

Well Formation Prestimulation Estimated fromFracCADE prediction

Post-stimulation

1

2

3

4

5

6

7

8

San Miguel

San Miguel

San Miguel

Olmos

Olmos

Olmos

San Miguel

Olmos

100 Mcf/D

800 Mcf/Dand 25 BOPD

1000 Mcf/Dand 20 BOPD

No flow

No flow

No flow

No flow

No flow

400 Mcf/D

1200 Mcf/D

1600 Mcf/D

200 Mcf/D

350 Mcf/D

500 Mcf/D

300 Mcf/D

300 Mcf/D

500 Mcf/D

1550 Mcf/Dand 200 BOPD

2000 Mcf/Dand 45 BOPD

410 Mcf/D

370 Mcf/D

330 Mcf/D

320 Mcf/Dand 18 BOPD

340 Mcf/D

(continued on page 14)

12

Modern NMR logging tools use large permanentmagnets to create a strong static magneticpolarizing field inside the formation. The hydro-gen nuclei of water and hydrocarbons are elec-trically charged spinning protons that createweak magnetic fields—like tiny bar magnets.When a strong external magnetic field from thelogging tool passes through a formation contain-ing fluids, these spinning protons align them-selves like compass needles along the magneticfield. This process, called polarization, increasesexponentially with a time constant, T1, as longas the external magnetic field is applied.1 Amagnetic pulse from the antenna rotates, ortips, the aligned protons into a plane perpendic-ular, or transverse, to the polarization field.These tipped protons immediately start to wob-ble or precess around the direction of the stronglogging-tool magnetic field, just as a child’s spin-ning top precesses in a gravitational field.

The precession frequency, called the Larmorfrequency, is proportional to the strength of theexternal magnetic field. The precessing protonscreate oscillating magnetic fields, which gener-ate weak radio signals at this frequency. Thetotal signal amplitude from all the precessinghydrogen nuclei—typically a few microvolts—is a measure of the total hydrogen content, orporosity, of the formation.

The rate at which the proton precessiondecays is called the transverse relaxation time,T2, and is the second key NMR measurementbecause it reacts to the environment of thefluid—the pore-size distribution. T2 measures

the rate at which the spinning protons lose theiralignment within the transverse plane. It dependson three things: the intrinsic bulk-relaxationrate in the fluid; the surface-relaxation rate,which is an environmental effect; and relaxationfrom diffusion in a polarization-field gradient,which is a combination of environmental andtool effects.

In addition, the spinning protons will quicklylose their relative phase alignment within thetransverse plane because of variations in the staticmagnetic field.2 This process is called the free-induction decay, and the Carr-Purcell-Meiboom-Gill (CPMG) pulse-echo sequence is used tocompensate for the rapid free-induction decaycaused by reversible transverse dephasing effects.3

The three components of the transverse-relaxation decay time play a significant role inthe use of the T2 distribution for well-loggingapplications. For example, the intrinsic bulk-relaxation decay time is caused principally bythe magnetic interactions between neighboringspinning protons in the fluid molecules. Theseare often called spin-spin interactions.

Molecular motion in water and light oil israpid, so the relaxation is inefficient with corre-spondingly long decay-time constants. However,as liquids become more viscous, the molecularmotions are slower. Then, the magnetic fields,fluctuating due to their relative motion approachthe Larmor precession frequency, and the spin-spin magnetic relaxation interactions becomemuch more efficient. Thus, tar and viscous oilscan be identified because they relax relatively

efficiently with shorter T2-decay times thanlight oil or water.

Fluids near, or in contact with, grain surfacesrelax at a much higher rate than the bulk-fluidrelaxation rate. Because of complex atomic-levelelectromagnetic field interactions at the grainsurface, there is a high probability—character-ized by the surface-relaxivity parameter—thatthe spinning proton in the fluid will relax whenit encounters a grain surface. For the surface-relaxation process to dominate the decay time,the spinning protons in the fluid must makemultiple encounters with the surface—causedby Brownian motion—across small pores in the formation. They repeatedly collide with the surface until a relaxation event occurs.

The resulting T2 distribution leads to a natural measure of the pore-size distribution(next page). Traditionally, the total porosityseen in formations is subdivided into threemajor components: free-fluid porosity with longT2 components, capillary-bound water with T2

greater than 3 msec and less than the T2 cutofffor the free-fluid component, and finally, thefast decaying clay-bound water below 3 msec. AsNMR tool technology has improved over the lastdecade with shorter echo spacing, more compo-nents of porosity now can be measured, includ-ing the fastest clay-bound water signal below3 msec. Today, for example, the CMR-200 andCMR-Plus tools can measure T2 down to the0.3-msec range while logging continuously, andto 0.1 msec during stationary measurements.

Oilfield Review

Basics of NMR Logging

Autumn 2000 13

Relaxation from diffusion in the polarization-field gradient is a technique frequently used todifferentiate oil from gas.4 Because the spinningprotons move randomly in the fluid, any mag-netic-field gradients will lead to incompletecompensation with the CPMG pulse-echosequence. For example, between spin-flippingpulses, some protons will drift—due to theirBrownian motion—from one region to anotherof different field strength, which changes theirprecession rate. As a result, they will notreceive the correct phase adjustment for theirprevious polarization environment. This leads toan increase in the observed transverse dephas-ing relaxation rate. Gas has relatively highmobility compared with oil and water, andtherefore, the spinning protons in gas have amuch larger diffusion-in-gradient effect. It isimportant to know that a uniform magnetic-fieldgradient isn’t required to exploit the diffusion-in-gradient effect. All that is required is a well-defined and mapped gradient volume todifferentiate gas from oil.5

> Using NMR T2 distributions (bottom) to identify fluid components (top) insandstone reservoirs. In water-wet sandstone rock, the T2 time distributionreflects the pore-size distribution of the formation. Producible fluids include freewater (light blue), or pockets of oil (green) in the larger pores. Free water andlight oils contribute to the longer T2 time components. Capillary-bound water(dark blue) is held against sand grains by surface tension and cannot be produced.Clay-bound water (black) is also unproducible. Shorter T2 time components arefrom irreducible water that is closely bound to grain surfaces.

1. The time constant for the polarizing process, T1, is tradi-tionally known as the spin-lattice decay time. The namecomes from solid state NMR, in which the crystal latticegives up energy to the spin-aligned system.

2. The observed fast decay, called the free-induction decay,is due to the combined components of irreversible trans-verse-relaxation decay interactions and the reversibledephasing effect caused by variations in the static magnetic field.

3. The pulse-echo technique used in today’s tools is called the CPMG sequence, named after Carr, Purcell,Meiboom and Gill, who refined the pulse-echo scheme.See Carr HY and Purcell EM: “Effects of Diffusion on FreePrecession in Nuclear Magnetic Resonance Experiments,”Physical Review 94, no. 3 (1954): 630-638, and Meiboom Sand Gill D: “Modified Spin-Echo Method for MeasuringNuclear Relaxation Times,” The Review of ScientificInstruments 29, no. 8 (1958): 688-691.

4. 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.

5. Flaum C, Guru U and Bannerjee S: “Saturation Estimationfrom Magnetic Resonance Measurements inCarbonates,” Transactions of the SPWLA 41st AnnualLogging Symposium, Dallas, Texas, USA, June 4-7, 2000paper HHH.

T2 distribution

0.3

T2, msec

Capillary-boundwater

Clay-boundwater

Producible fluids

Total CMR porosity

3-msec CMR porosity

CMR free-fluid porosity

3.0 33 3000

Sandstone

Predicting Reservoir FlowQuantitative petrophysical data are increasinglybeing applied to numerical simulators for reservoirmanagement.11 For example, CMR measurementswere used to help predict flow characteristics in awell in the central North Sea by BG Internationaland Phillips Petroleum. First, laboratory NMR andconventional core analysis were compared tooptimize a new NMR-based permeability algo-rithm. Next, the CMR-derived total porosity andirreducible water were used in the new algorithmto determine permeability continuously across theentire reservoir (left). From this, the flow perfor-mance of the reservoir was determined by com-paring a plot of the flow capacity, defined by theproduct of the permeability and thickness, withthe storage capacity, defined as the product of thehydrocarbon pore volume and thickness. Thisgraphical tool, called a Lorenz plot, offers a guideto how many flow units are necessary to honorthe geologic framework (below left).

The results show that the lowest zone in thiswell, Zone A, had 17% of the total storagecapacity and 74% of the flow capacity. Zone Bcontained 60% of the storage capacity with 24%of the flow capacity. The upper zone, Zone C, con-tained 23% of the storage capacity but only 2%of the flow capacity. These results indicate thatwhen the well is put on production, the initialflow rate will decline sharply as Zone A isdepleted. The long-term performance of the wellwill be dominated by production from Zones Band C. Furthermore, a detailed analysis of theplot indicates that a minimum of ten flow unitsshould be employed in the development of thereservoir model. CMR-derived permeability andporosity helped define the flow and storagecapacities of each of these ten reservoir units.

14 Oilfield Review

> CMR-derived permeability in the North Sea. Track 1 contains the CMR-derivedporosity subdivided into oil (red), capillary-bound water (green), clay-boundwater (white) and producible water (blue). Track 2 contains the saturationanalysis based on integrating CMR data with resistivity measurements showingirreducible water saturation (green) and total water saturation (blue curve). The CMR-derived permeability is shown in Track 3. Laboratory-measured per-meability (yellow) is shown for the core samples (red circles). Track 4 shows thenormalized flow capacity over the three zones in the reservoir.

Lorenz plot—flow performance based onCMR-derived permeability. The shape of theplot is indicative of the flow performance of thewell and reservoir. Segments with steep slopeshave a greater percentage of the reservoir flowcapacity relative to the storage capacity, and by definition have a higher reservoir productionrate. Segments with flat slopes have a higherstorage capacity, but little flow capacity, andtherefore may form reservoir baffles if laterallyextensive. Similarly, segments with neither flowor storage capacities are reservoir seals if later-ally extensive. Individual flow units (right column)can be identified from the location of inflectionpoints. These plots can help define the minimumnumber of flow units to use in the developmentof reservoir models.

XX560

XX660

XX760

Dept

h, ft

XX860

XX960

XY060

Porosity, p.u.0 10 20 30 110-1 100 10+4

Permeability, mD0 50 100

Saturation, %0 50 100

Normalized flow profile

Zone C

Zone B

Zone A

1.0

0.8

0.6

0.4

0.2

00 20 40 60 80 100

Low reservoir-flow capacityand low storage-capacity unit.Reservoir seal.

Low reservoir-flow capacityand high storage-capacity unit.May form reservoir baffles iflaterally extensive.

High reservoir-flow capacityand low storage-capacity unit.

Zone C

Zone B

Zone A

Flow units

Flow

cap

acity

, % to

tal k

hei

ght

Storage capacity, % total hydrocarbon pore volume height

>

MDT points picked from CMR login zones of high free fluid

SDR permeability

Timur-Coates permeabilitymD0.1 10,000

Deep resistivity, LLD

ohm-m 10000.01

Shallow resistivity

p.u. 050

msec 30000.3

Free fluid T2 cutoff (33 msec)

CMR 3-msec porosity

Total CMR porosity

Density porosity

Neutron porosity

Clay-bound water

Capillary-bound water

T2 distribution

Autumn 2000 15

NMR and Wireline Formation Testers Wireline formation testers are used widely toevaluate formation fluids and producibility.12

Logging tools such as the MDT ModularFormation Dynamics Tester tool provide down-hole estimates of permeability in each pay zonebased on dynamic-pressure and fluid-drawdownmeasurements. Fluid samples, permeability andpressure measurements made with thesedevices often provide the first look at a well’sability to produce. Recent studies using the MDTtool have shown that the OFA Optical FluidAnalyzer module can provide in-situ crude-oiltyping for gas/oil ratio, API gravity and colorationestimates. The MDT tool is also capable of deter-mining the pressure, volume and temperature(PVT) behavior of reserves in place.13 Knowingthese properties forms the foundation for plan-ning efficient field development.

Zone selection is critical in the sampling pro-cess. The objective is to obtain a representativesample of the reservoir fluid. Wireline formationtester surveys require stationary measurementsand long pumping times to flush mud-filtrate inva-sion and ensure that PVT-grade formation fluid

samples are obtained. Conventional resistivity-density-porosity logging suites and core data canhelp identify potential pay zones. However, it iscritical to determine correctly which zones willbe permeable; otherwise, the tester will not beable to draw any fluid from the formation, or thesampling will take excessive time.

Sampling efficiency can be improved by usingNMR permeability measurements to select themost productive zones for sampling (above left).CMR bound- and free-fluid logs also help deter-mine the best points for obtaining MDT pressuremeasurements and samples. High-resolutionCMR data are especially effective in laminatedsequences, and high-resolution CMR permeabilitylogging is the recommended practice for determin-ing MDT sampling programs in many locations.

For example, acquiring pressure data in oneregion offshore China had been problematicbecause of probe plugging in unconsolidatedshaly-sand formations (above right).14 Althoughthe resistivity and density-neutron porosity logsshowed little variation throughout the formation,the CMR free- and bound-fluid logs show sub-stantial variations—making it easy to identify

> Combining NMR logging with MDT downholeformation testing. Choosing formation testing andsample points on the basis of NMR log data pro-vides a reduced margin of error whenever thetwo tools are run together. Running the two toolstogether improves efficiency and reduces the riskof deteriorating borehole conditions. The CMRhigh-resolution permeability indicator helps identifypermeability streaks for sweet-spot positioning ofthe MDT tool.

> Improving pressure-sampling efficiency.The resistivity logs in Track 1 and thedensity (red), total CMR porosity (solidblack) and neutron porosity (dashed black)logs in Track 2 show little variationthroughout this unconsolidated shaly-sand formation. However, there is consid-erable variation in the free-fluid volume(purple) log in Track 2, making it easy to identify zones of high permeability. InTrack 1 the NMR permeability estimatesfrom the Timur-Coates transform (dashedblue) and the SDR transform (solid blue)agree well with fluid mobility determinedby the MDT drawdown measurements(solid blue circles). Track 3 shows the CMRT2 distributions. The MDT samples werechosen at points in the formation that hadthe best NMR log permeability estimates.

11. Gunter GW, Finneran JM, Hartmann DJ and Miller JD:“Early Determination of Reservoir Flow Units Using anIntegrated Petrophysical Method,” paper SPE 38679,presented at the SPE Annual Technical Conference andExhibition, San Antonio, Texas, USA, October 5-8, 1997.

12. Crombie A, Halford F, Hashem M, McNeil R, Thomas EC,Melbourne G and Mullins O: “Innovations in WirelineFluid Sampling,” Oilfield Review 10, no. 3 (Autumn 1998):26-41.

13. Hashem MN, Thomas EC, McNeil RI and Mullins OC:“Determination of Hydrocarbon Type and Oil Quality inWells Drilled with Synthetic Oil-Based Muds,” paperSPE 39093, presented at the SPE Annual TechnicalConference and Exhibition, San Antonio, Texas, USA,October 5-8, 1997.Felling MM and Morris CW: “Characterization of In-SituFluid Responses Using Optical Fluid Analysis,” paperSPE 38649, presented at the SPE Annual TechnicalConference and Exhibition, San Antonio, Texas, USA,October 5-8, 1997.

14. Castelijns C, Badry R, Decoster E and Hyde C:“Combining NMR and Formation Tester Data forOptimum Hydrocarbon Typing, Permeability andProducibility Estimation,” Transactions of the SPWLA40th Annual Logging Symposium, Oslo, Norway, May 30-June 3, 1999, paper GG.

MDT tool

CMR tool

Sweet spot

the most permeable intervals. The MDT samplepoints were selected on the basis of the highestpermeability, which are the zones with low bound-fluid volume. All six pressure tests were made suc-cessfully, and three fluid samples were recoveredin a traditionally difficult sampling environment.

Combining the MDT and CMR tools in onelogging pass reduces operating and rig time,resulting in substantial savings, especially inpipe-conveyed, offshore and remote operations.Limiting operating time also reduces the risk ofstuck pipe as hole conditions deteriorate.

Fluid characterization—Pressure measure-ments and fluid-sample analysis combined withspecific NMR data signatures—such as long T2

times in light-hydrocarbon environments or adeficit in the NMR total porosity in tar zones

compared with density-neutron logs—can providepositive identification of formation fluids.

For example, the CMR-MDT combination wasrun in a well drilled with oil-base mud in the NorthMonagas area of eastern Venezuela to validatehydrocarbon identification (above). The gas-corrected formation porosity was determinedusing the DMR Density-Magnetic ResonanceInterpretation Method.15 Interpretation based ondensity-neutron and CMR data confirmed thatZones A, C and H are gas-bearing. The other zonesin this field have no density-neutron crossover andhave been interpreted as oil-bearing. However,Zones B, D, E, F and I show large CMR-porositydeficits (shaded blue) in Track 4 compared withdensity porosity and are interpreted as tar-bearing.Tar has high viscosity and relaxes the NMR signal

quickly. Unfortunately, the resistivity logs cannotdistinguish between tar and lighter hydrocarbonzones in this oil-base mud environment.

However, independent confirmation of tarzones was provided by the MDT measure-ments—all four attempts to sample fluids inthese zones with the MDT tool resulted in drytests. In contrast, all MDT measurements in thegas or light-oil zones produced pressure andmobility readings that agree with the CMR per-meability log derived from the total porosity andbound-fluid measurements. CMR-MDT data incombination with triple-combo log data provideda conclusive petrophysical analysis of the com-plex gas, oil and tar formation.

In four deepwater Gulf of Mexico wells drilledin the same structure, the CMR tool was com-bined with the MDT tool to confirm locations offluid contacts in the reservoir. The tools were runin deviated wellbores on drillpipe using a TLCTough Logging Conditions system, while tradi-tional openhole logs were obtained using logging-while-drilling (LWD) measurements. In one well,the MDT pressure data and the pressure-gradienttrends show evidence of a gas-oil contact (GOC)at X270 ft, which is consistent with the porositydeficit of the NMR measurement in the gas zone(next page, bottom). Similar results were found ineach of the other wells.

The total CMR porosity overlies the LWDporosity measurement in the water zone below X335 ft, indicating that the protons in theinvading mud filtrate—a synthetic oil-base mudsystem—are fully polarized. In the oil zone fromX270 to X335 ft, the total CMR porosity readslower than the density porosity because ofincomplete polarization of the light reservoir oil.The T2 distribution over this zone indicates an oilsignal with an average T2 in the range of 2 to3 seconds and a downhole oil viscosity of around0.3 centipoise, which agrees well with laboratorymeasurements of MDT samples obtained fromthis zone.16 The density-CMR porosity overlayshows increased separation above the GOC atX270 ft. The use of NMR data confirms the inter-pretation of the MDT pressure data, providesfluid identification and helps establish the loca-tion of each fluid contact.

The MDT permeability derived from draw-down-pressure measurements was used in thefour-well study to determine the correct T2 cutofffor differentiating bound fluid from free fluid. TheT2 distributions in each of the wells display abimodal shape that is commonly seen in wellsdrilled with oil-base mud systems. Providing theformation is water-wet, the higher peak in the T2

distribution comes from invaded oil-base mud filtrate and reservoir hydrocarbons, which decay

16 Oilfield Review

Gammaray, API

Depth,ft

CMR permeability

MDTdry test

MDTpermeability

XX200

XX250

XX300

XX350

XX400

XX450

A

B

C

D

E

F

G

H

I

0

0.1 mD 1000

150 40 0 40 0p.u. p.u.

Tarflag

Crossover

Density porosity

Neutron porosity

Density porosity

CMR porosityGasflag

> Evaluation of gas, oil and tar zones. The density-derived porosity (red) and neutron porosity (black)logs are shown in Track 3. A tar flag (black) in Track 2 is derived by comparing the density-neutronporosity (Track 3) and the CMR porosity deficit log (Track 4). The gas flag (red) in Track 2 is computed by comparing the volume of gas seen by the CMR tool and that seen by the density-neutron crossover.The CMR log confirms that Zones A, C and H are gas-bearing. The CMR porosity deficit log (blue shading)computed from the difference between density porosity and the CMR porosity is shown in Track 4.Track 1 shows the permeability derived from the total CMR porosity and bound-fluid measurementsusing the Timur-Coates equation. The ten MDT permeability measurements (green circles) in the high-permeability zones agree well with the CMR-derived permeability log. The four MDT measurementsin the tar zones (red stars) were dry (unproductive) tests.

Autumn 2000 17

at their bulk-relaxation rates. The lower peak rep-resents the bound fluid. A reasonable estimate forthe T2 cutoff is given by the position of the valleybetween the two peaks. In these T2 distributions,the valley appears to be between 60 and 70 msec,whereas typical T2 cutoff values derived fromNMR measurements on core samples from thesame formations were only 11 msec.

To establish the correct T2 cutoff value for thisformation, two approaches were tried. First, the T2

cutoff value was varied to determine the amountof irreducible and free water from the logs in thehydrocarbon zones. The core-derived T2 value of11 msec produces a considerable amount of unex-pected free water, and the cutoff needs to beincreased to 40 msec to remove all the free waterin the hydrocarbon zone. Similar discrepancieshave been seen in other reservoirs.17

In the second approach, an optimal cutoffvalue was determined by varying the T2 cutoff

and comparing the resulting Timur-Coates perme-ability values with the MDT drawdown perme-ability measurements (above). Varying the T2

cutoff changes the NMR-derived permeabilityprofile in each well, as the bound-water volumewill vary at different depths.18 The NMR-derivedpermeability profile using a 40-msec T2 cutoffbest fits the MDT permeability measurements inthis study.

Wellsite-computed NMR permeability can beused to optimize the design of the formationtester pressure measurements and fluid sampling,and better evaluate the production potential ineach well.

> Finding fluid contacts. The MDT pressure data plot (left) shows the pressure-gradient changesdue to fluid contacts. The gas-oil contact (GOC) is at X270 ft and the oil-water contact (OWC) is atX335 ft. The formation volume analysis is shown in Track 1 (right). The NMR-derived permeability(purple) is compared with MDT drawdown-measured mobility (red circles) in Track 2. In Track 3,the density-derived porosity (red) is compared with the total CMR measured porosity (black). Thegas-corrected formation porosity (green) shown in Track 3 was computed using the DMR Density-Magnetic Resonance Interpretation Method. A complete fluid analysis based on the CMR free-fluid, bound-fluid, DMR porosity and other openhole data is shown in Track 4. The CMR-derived T2distributions are shown in Track 5.

> NMR permeability compared with pressure drawdown permeability. The NMR permeability wasderived using the Timur-Coates equation. The effect of varying the T2 cutoff to discriminate between thefree- and bound-fluid components was investigated. The two plots show the effect of varying the cutofffrom 40 msec (left) to 60 msec (right). The data in the gas zone show a wider spread, possibly due to thefact that gas can alter the viscosity of the invading fluid, causing error in the MDT permeability derivedfrom drawdown measurements. The optimal correlation was found for the 40-msec T2 cutoff comparison.

15. Freedman R, Cao Minh C, Gubelin G, Freeman J,McGinness J, Terry B and Rawlence D: “CombiningNMR and Density Logs for Petrophysical Analysis inGas-Bearing Formations,” Transactions of the SPWLA39th Annual Logging Symposium, Keystone, Colorado,USA, May 26-29, 1998, paper II.

16. Morriss CE, Freedman R, Straley C, Johnston M, VinegarHJ and Tutunjian PN: “Hydrocarbon Saturation andViscosity Estimation from NMR Logging in the BelridgeDiatomite,” Transactions of the SPWLA 35th AnnualLogging Symposium, Tulsa, Oklahoma, USA, June 19-20,1994, paper C.For a discussion of viscosity effects on relaxation time:see Allen et al, reference 1.

17. Allen et al, reference 1.18. Effects of varying and unknown fluid viscosity on MDT

drawdown mobility measurements were minimized bycomparing data only in oil and water zones and avoidinggas zones, where gas dissolved in the invading filtratescould alter fluid viscosity. The effects of differing verticalresolution in comparing the values of CMR-derived per-meability with those from the MDT measurements wereminimized by eliminating data close to bed boundaries.The leading coefficient in the Timur-Coates equationwas adjusted to calibrate the NMR permeability to thedrawdown permeability.

10,000

1000

100

10

1

0.1

10,000

1000

100

10

1

0.1

Oil/water zones Gas zones

K-M

DT

k-M

DT

0.1 1 10 100 1000 10,000 0.1 1 10 100 1000 10,000

k-NMR k-NMR

40-msec T2 cutoff 60-msec T2 cutoff

MDT data

TVD, ft

X222

X150

X200

X250

X300

X350

63406300Pressure, psi

Shale analysis MDT mobility DMR porosity

TCMR porosity

Density porosity Water

CMR permmD0 1

T2 cutoff

T2 distributionmsec0.3 3000

p.u.50 0

mD1 10,000

X249

X276

X303

X330335 ft

270 ft

X357

Quartz

Illite

Clay-boundwater

Irreduciblewater

Oil

Gas

Depth,ft

GOC

OWC

NMR in CarbonatesOne of the benefits of NMR-measured porosity isthat it is independent of the mineralogy of the for-mation rock. The echo amplitudes depend only onthe hydrogen content of the formation fluids, andare not affected by the rock bulk properties like den-sity or neutron cross sections. This helps to make apetrophysical analysis in complex mineralogy—such as evaluating water saturation in limestonewith anhydrite inclusions—much easier.

However, there is some concern within theindustry that NMR doesn’t work as well asexpected in carbonate reservoirs. The problem cen-ters on uncertain mapping between T2 distributionsand pore-size distributions in carbonates. Thisresults in inconsistent T2 cutoff values needed todistinguish bound and free fluids, and leads to unre-liable permeability and water-cut predictions. Someof the suspected causes are diffusion of spinningprotons between micro- and macropores, variationsin surface relaxivity with temperature and mineral-ogy, and variable, complex pore systems.

Traditional interpretations of NMR measure-ments in water-saturated reservoir rocks assumethat the T2 distribution and pore-size distributionsare directly related.19 The diffusion of spinning pro-tons between pores is neglected, and the relax-ation in each pore is presumed to be controlled bythe grain surface-relaxivity coefficient. This inter-pretation model results in observed relaxationtime that depends on two components—the bulk-fluid-relaxation process, which is slow for fluidswith high mobility (water and light oil), and thesurface-relaxation process, which depends on theratio of pore volume to surface area. Water, lightoil and gas trapped in large pores decay moreslowly, whereas fluid confined to pores with asmall pore volume to surface area ratio—such asclay-bound fluid—experiences more rapid decaybecause of frequent surface encounters. This pop-ular interpretation model successfully explainsobserved T2 distributions in sandstone reservoirscontaining mixed pore-size distributions.

Carbonate formations also contain multiplepore-size distributions—often consisting ofmicro- or intragranular porosity and macro- orintergranular porosity, as well as isolated vugs.

Nevertheless, logging data in these formationsfrequently yield unimodal T2 distributions thatlimit the capability of the NMR measurement topredict permeability and movable fluid.

Recent developments in NMR researchexplain why the conventional approach breaksdown in grain-supported carbonates which havedual micro- and macropore systems in close prox-imity (above).20 The breakdown is due to diffusionof spinning protons between the micro- andmacropores.21 This result was verified usingnumerical simulations and analytical models toassess the physical processes underlying theNMR measurement on rocks with the same char-acteristics as those typically found in MiddleEastern carbonate reservoirs (next page, top).Diffusion causes the area under the short T2

peak—the porosity fraction associated withmicropores—to decrease; at the same time, theposition of the higher T2 peak shifts towardsshorter times. Acting together, these two effectstend to merge the two peaks and produce a uni-modal T2 distribution that bears little resem-blance to the bimodal distribution one wouldexpect from a dual-porosity system.

Physically, these effects are caused by diffu-sion of the fluid’s spinning protons between thetwo pore systems. If the surface relaxivity issmall enough, then protons originally in themicropores can diffuse back into the macroporesbefore their nuclear spins relax. The decay ofthese spins then proceeds much more slowly andcontributes to the late-time peak, which explainsthe apparent transfer of porosity from early-timeto late-time peaks.

The role of diffusion in carbonates is furthercomplicated by the observation that the T2 distri-butions measured in some carbonate formationshave a temperature dependence.22 Previous labo-ratory work on the effects of temperature wasbased largely on sandstones and concluded thatmeasured decay does not change with tempera-ture, implying that diffusion does not controlrelaxation in those formations.23 In some carbon-ate reservoirs, comparison between boreholeand laboratory NMR measurements shows that,although the total porosity agreement is quitesatisfactory, the laboratory lifetimes—measuredat room temperature—are significantly shifted toshorter values than those observed in logs.

This result has been repeated in laboratorystudies of the temperature dependence of NMRin mud-supported carbonate rocks taken from anumber of wells in the Middle East (next page,bottom). The diffusion process is dominated by afactor called the fluid-diffusion constant. As tem-perature increases, the diffusion constantchanges and leads to a shift in the observed T2

distributions to longer times.

18 Oilfield Review

Three-dimensional model basedon periodic packing of consoli-dated microporous grains. Anumerical model for carbonategrainstones is based on a simplecubic structure containing largegrains of microporous grains,which were made up of smaller,consolidated grains whose cen-ters lie on a sublattice. The NMRphysics (insert), shows that whenpores are coupled by diffusion,spinning protons originally inmicropores will escape to themacropores where they survivelonger. Spinning protons origi-nally in the macropores penetrateinto the micropores where theyencounter more surface interac-tions, decreasing their lifetimes.

19. Allen et al, reference 1.20. Ramakrishnan TS, Schwartz LM, Fordham EJ, Kenyon

WE and Wilkinson DJ: “Forward Models for NuclearMagnetic Resonance in Carbonate Rocks,” The LogAnalyst 40, no. 4 (July-August 1999): 260-270.

21. Typically, the correction for diffusion is due to magnetic-field gradients causing irreversible dephasing. However,even when the field gradient is zero, diffusion can alterthe relaxation significantly.

22. Ramakrishnan TS, Fordham EJ, Venkitaramanan A,Flaum M and Schwartz LM: “New InterpretationMethodology on Forward Models for MagneticResonance in Carbonates,” Transactions of the SPWLA40th Annual Logging Symposium, Oslo, Norway, May 30-June 3, 1999, paper MMM.

23. Latour LL, Kleinberg RL and Sezginer A: “NuclearMagnetic Resonance Properties of Rocks at ElevatedTemperatures,” Journal of Colloidal and InterfaceScience 50, no. 2 (1992): 535-548.

24. Freedman R, Sezginer A, Flaum M and Matteson A: “A New NMR Method of Fluid Characterization inReservoir Rocks: Experimental Confirmation andSimulation Results,” paper SPE 63214, presented at the SPE Annual Technical Conference and Exhibition,Dallas, Texas, USA, October 1-4, 2000.

>

Macropore Micropore

Spinning protonsSpinning protons

Autumn 2000 19

Fortunately, the role of diffusion—with theresultant T2 distribution distortion—does notaffect all carbonates. Depending on the local con-ditions during deposition, the fabric of youngercarbonate formations can vary from grain-sup-ported, in which diffusion effects are significant,to mud-supported carbonates, in which diffusioneffects are not expected to be significant. Second,as formations age, the diagenetic overprint usu-ally increases—resulting in increasing crystal

size and decreasing relationship to the originaldepositional fabric. The decrease in size contrastbetween pores of close proximity diminishes theeffect of hydrogen spin diffusion.

Recent research at Schlumberger-Doll Researchin Ridgefield, Connecticut, USA, shows that diffu-sion is not a factor in either older sucrosic carbon-ate formations—having a granular texture likesugar—or certain fabric-supported formations.Another finding is that the surface relaxivity inthese formations does not vary significantly, and

that NMR distributions exhibit consistent T2 cut-offs that can be used for estimating permeability.It appears that in these formations, the watermolecules remain within either their originalmicropores or macropores and that interpretationscan be performed based on the T2 distributions,although they differ from the laboratory T2 distri-butions due to temperature.

The Next StepWhat is next for NMR logging? It is probably acombination of improved technology and greateracceptance by the oil and gas industry. Today,operators are realizing that these tools are nolonger high-tech niche players, but instead cangive answers that no other tools can, dramati-cally changing the way wells are completed andreservoirs are developed.

The current generation of NMR tools deliversnot only reliable information about formationporosity and permeability, but also supplies rock-characterization information and informationabout the fluids contained within them. Reliablepore-size distributions are obtained from T2 distri-butions measured in clastic sandstone formations.

Despite the progress, many technical chal-lenges remain, particularly in carbonates. Whatis clear, nonetheless, is that ongoing researchand field experience with NMR alone or in com-bination with other tools will inevitably add moreapplications, such as better reservoir characteri-zation, for this innovative technology.24 —RH

ρT2, µm ρT2, µm ρT2, µm10-2 10-1 100 101 102 10-2 10-1 100 101 102 10-2 10-1 100 101 102

Isolated poresSurface relaxivity Coupled pores Middle East formation with coupled pores

1.5 µm/sec

7.5 µm/sec

37.5 µm/sec

187.5 µm/sec

937.5 µm/sec

17 p.u.

17 p.u.

17 p.u.

17 p.u.

17 p.u. 17 p.u.

17 p.u.

17 p.u.

17 p.u.

17 p.u.

14 p.u. 14 p.u. 20 p.u.5 p.u.

5 p.u.

5 p.u.

5 p.u.

5 p.u.

20 p.u.

20 p.u.

20 p.u.

20 p.u.

14 p.u.

14 p.u.

14 p.u.

14 p.u.

14 p.u.

14 p.u.

14 p.u.

14 p.u.

Intragranular Intergranular

Effect of diffusion on inter- andintragranular pore systems. T2distributions from simulated grain-supported carbonate formations areshown for a range of surface relax-ivities, ρ. In the left column, the T2distributions are computed for a forma-tion with14-p.u. intergranular porosityand 17-p.u. intragranular porosity,that are isolated. The middle columnshows the effects of allowing diffu-sion of protons between the two partsof this dual-porosity system. The rightcolumn shows the effect of diffusionon a formation with 5-p.u. intergranu-lar porosity and 20-p.u. intragranularporosity, similar to that seen in manyMiddle East formations. The twosmallest values of the surface relaxiv-ity are representative of those seenin carbonate reservoirs. The x-axis isρT2 and has units of length. It reflectstypical length scales seen in the rock.

> Temperature dependence of carbonate T2 distribution. Laboratory measurements on a core sample taken from a Middle East carbonate formation show the effect of increasing temperature on the T2 distribution.

10-2 10-1 100 101

T2, sec

307010013030 (repeat)

Temperature, °C

T 2 am

plitu

de

>