See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/240780111

Advances in NMR logging

Article  in  Journal of Petroleum Technology · January 2006

DOI: 10.2118/89177-MS

CITATIONS READS
59 3,056

1 author:

R. Freedman
Freedman Formation Evaluation Consultants, LLC
98 PUBLICATIONS   1,086 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NMR Logging View project

All content following this page was uploaded by R. Freedman on 10 February 2015.

The user has requested enhancement of the downloaded file.

 DISTINGUISHED AUTHOR SERIES

Advances in NMR Logging

Robert Freedman, Schlumberger Oilfield Services

Introduction
This summary of the state of the art in nuclear-magnetic-resonance
(NMR) well-logging technology is aimed at nonspecialists who
would like to gain some knowledge of the formation-evaluation
capabilities of NMR logging tools. The objective is to explain the
basic measurement principles and interpretations needed to under-
stand NMR formation-evaluation techniques and to discuss a few
examples of these methods.
Introduction of pulsed-NMR logging tools in the 1990s provided
the industry with unique, even revolutionary, new methods for ana-
lyzing reservoir fluids, rocks, and fluid/rock interactions. The intro-
duction of this technology came at an opportune time. It coincided
with rapidly declining production after the 1970s drilling boom and
the need for new tools to evaluate the more complex reservoirs
being explored and developed. Pulsed-NMR logging tools brought a Fig. 1—The Schlumberger MRX Magnetic Resonance eXpert
wealth of new and unique formation-evaluation applications, and wireline tool has a multifrequency main antenna for fluid-
this technology has grown rapidly since its inception. Today, major characterization applications and two high-resolution
service companies (e.g., Baker Hughes, Halliburton, and antennas for providing rock-quality and producibility infor-
mation (after Ahr et al.10).
Schlumberger) offer NMR logging services.
measurements like those used in modern laboratory-NMR instru-
Historical Perspective1 ments. The value of these measurements is that they are extremely
The potential value of NMR logging was first recognized in the flexible and can be tailored to fit many different formation-evalua-
1950s, leading to development of nuclear-magnetic-logging (NML) tion applications.
tools in the early 1960s. NML tools had many limitations and even- The Los Alamos tool demonstrated feasibility but did not meet
tually were retired from service in the late 1980s. In spite of these the requirements for a commercial tool because the signal-to-noise
limitations, laboratory research conducted to support NML logging (S/N) ratio was too low and the magnet and radio-frequency (RF)
anticipated many formation-evaluation applications in use today. coil design produced a large borehole signal. Soon after this demon-
These applications include estimation of permeability, pore-size dis- stration of feasibility, Numar Corp., a company founded in 1983,
tribution, free-fluid volume, oil viscosity, and wettability. and Schlumberger began independent research efforts to design
The modern phase of NMR logging can be traced to the initiation NMR magnets and RF antennas that would be suitable for commer-
of an NMR borehole-logging research project at Los Alamos Natl. cial NMR logging measurements.
Laboratory in 1978. The goal of this project was, in part, to build These efforts came to fruition in the early 1990s when both com-
and test a borehole NMR logging tool that would overcome the lim- panies began field tests of prototype wireline tools. These tools were
itations of the NML tools. The Los Alamos experimental tool used vastly superior to the NML tools and quickly had an effect on for-
strong permanent magnets and performed pulsed-NMR spin-echo mation evaluation. Since introduction of the first commercial tools,
both companies have introduced advanced NMR wireline tools as
well as logging-while-drilling (LWD) NMR tools. Numar was sold to
Robert Freedman, SPE, is a scientific adviser and project man-
Halliburton in 1997 and operates today as a wholly owned sub-
ager in the Resistivity/NMR Dept. at the Schlumberger Sugar
sidiary. In 2001, Halliburton introduced an NMR fluid analyzer that
Land Product Center. He holds a PhD degree in condensed mat-
is part of its wireline fluid-sampling tool. Halliburton and
ter physics from the U. of California at San Diego. During his
Schlumberger introduced LWD tools in 2000 and 2002, respective-
30-year career in the petroleum industry, he has contributed to
ly. Baker Hughes introduced a wireline NMR tool in 2004 and an
the physics, mathematical modeling, and interpretation of
LWD NMR tool in 2005.
NMR, resistivity, and dielectric tools; modeling of deep-reading
resistivity tools for relief-well drilling; and development of
Modern NMR Logging
mathematical inverse methods. Freedman has served as associ-
Pulsed-NMR Logging Tools. The sensor (i.e., magnet and antenna)
ate editor of the SPE Formation Evaluation Journal, as SPE
is the heart of a pulsed-NMR logging tool. It has a significant effect
Distinguished Lecturer, and is a member of the SPE Petroleum
Professional Certification Subcommittee. He is a recipient of the Copyright 2006 Society of Petroleum Engineers
This is paper SPE 89177. Distinguished Author Series articles are general, descriptive rep-
2004 SPE Cedric K. Ferguson Award. Freedman has authored resentations that summarize the state of the art in an area of technology by describing recent
43 industry publications and has been awarded 21 patents on developments for readers who are not specialists in the topics discussed. Written by individuals
recognized as experts in the area, these articles provide key references to more definitive work
well-logging technology. and present specific details only to illustrate the technology. Purpose: to inform the general
readership of recent advances in various areas of petroleum engineering.

60 JPT • JANUARY 2006

 φ
M
M0o 180° 180° 180° 180° T2 Decay
T1 buildup 90° • • • FID
• • •
Echo1 Echo2 Echo3
Tp 0 TE Time 0 Time

(a) Polarization Period (b) RF-Pulse Sequence (c) Spin Echoes

Fig. 2—The NMR-measurement process. The first step (a) is the polarization period during which the reservoir fluids are
magnetized. The second step (b) is the application of RF pulses to the formation, which creates the spin-echo signals (c).

on important tool characteristics including S/N ratio, minimum zation is the longitudinal relaxation time, which is referred to as T1.
echo spacing, depth of investigation (DOI), logging speed, and ver- The T1 buildup of the magnetization during the polarization time is
tical resolution. Available tools all have somewhat different sensor shown in Fig. 2a.
designs. Further differences are electronics, firmware, pulse In reservoir rocks, a distribution of T1 values is needed to describe
sequences, data processing, and interpretation algorithms. Detailed the magnetization process. The T1 distribution reflects the complex
logging specifications for NMR tools can be found on service com- compositions of hydrocarbons and the distribution of pore sizes in
pany websites. sedimentary rocks. A polarization time equal to at least three times
Fig. 1 is a schematic of Schlumberger’s NMR wireline logging the longest T1 is used to ensure that adequate magnetization is
tool. This tool has three antennas and a fully programmable pulse achieved. If a polarization time is too short, NMR-derived porosities
sequencer and can perform a large variety of different measure- underestimate true formation porosities.
ments.2 Two 6-in. antennas are used for making high-resolution Immediately following the polarization time, a train of RF
measurements of NMR-derived total, bound-fluid, and free-fluid pulses is applied to the formation. The first RF pulse is called a
porosities. The high-resolution antennas are also used to detect gas 90° pulse because it rotates the magnetization vector, which ini-
and light hydrocarbons and to provide estimates of permeability and tially is parallel to Bo, into the transverse plane perpendicular to
pore-size distributions. The main antenna is 18 in. long. It provides Bo. Once the magnetization is in the transverse plane, it rotates
a variety of NMR measurements made at multiple frequencies for around Bo, producing a time-varying signal in the same antenna
different formation-evaluation applications. Each frequency corre- used to create the pulses. An NMR free-induction-decay (FID)
sponds to a different DOI in the range from 1.5 to 4 in., measured signal first occurs immediately after the 90° pulse but decays too
from the borehole wall. The formation-evaluation applications pro- fast to be detected. The 90° pulse is followed by a series of even-
vided by the main antenna include all of those provided by the two ly spaced 180° pulses that are used to refocus the magnetic
high-resolution antennas, and it is used for radial profiling of fluid moments of the hydrogen nuclei to form coherent spin-echo sig-
types, fluid volumes, and oil viscosities. nals. The spin echoes are recorded between each pair of 180°
Some features are common to all commercial NMR tools. pulses. The RF pulses and spin-echo signals are shown schemat-
For instance, the tools all use powerful samarium cobalt perma- ically in Figs. 2b and 2c, respectively. The signals are called
nent magnets that are relatively insensitive to changes in tem- echoes because they reach maximum amplitude at the midpoint
perature. The magnets are used to polarize (i.e., magnetize) the between each pair of 180° pulses and then decay rapidly to zero
hydrogen nuclei (protons) in hydrocarbon and water mole- before the following pulse, which refocuses the magnetic
cules. Another common feature is that they all perform pulsed- moments to produce the next echo.
NMR measurements. The RF pulses and associated spin echoes in Figs. 2b and 2c are
known as the Carr-Purcell-Meiboom-Gill (CPMG) sequence. It is
Measurement Principles. The NMR measurement comprises two the most widely used NMR logging sequence. The envelope of the
steps. The first step is to create a net magnetization of the reservoir spin-echo signal decays exponentially with a characteristic time
fluids.3 As the logging tool moves through the borehole, the mag- constant, T2, known as the transverse or spin-spin relaxation (i.e.,
netic-field vector, Bo , of the magnet polarizes the hydrogen nuclei decay) time. The amplitude of the spin-echo decay curve extrapo-
in the reservoir fluids creating a net magnetization. The magnetiza- lated back to zero time (immediately following the 90° pulse) is
tion is along the direction of Bo , which is called the longitudinal equal to the NMR-derived total porosity, assuming that the hydro-
direction. The magnitude of Bo typically is a few hundred gauss in gen index of the fluid is equal to 1.
the near-wellbore region (within a few inches of the borehole wall). An important specification for an NMR logging tool is its mini-
The magnitude of Bo decreases with the radial distance from the mum echo spacing. The minimum echo spacing plays an important
magnet, which causes a magnetic-field gradient or distribution of role, together with the S/N ratio, in determining the T2 sensitivity
gradients over the measurement volume. As discussed below, the limit—the shortest T2 value that can be measured by the tool. Short
magnetic-field gradient is used to identify and characterize the flu- minimum echo spacing is essential for accurate and repeatable mea-
ids in the reservoir. surements of NMR total porosity in formations containing clay-
Before exposure to Bo , the magnetic moments of the hydrogen bound and small-pore waters (i.e., for measuring T2 values shorter
nuclei are randomly oriented so that the fluids have zero net mag- than approximately 3 milliseconds).4 Minimum echo spacings for
netization. During the polarization time, Tp, the magnetization available tools range from about 0.2 to 1.2 milliseconds.
grows exponentially toward its equilibrium value, Mo. The time The number of echoes and the echo spacing, TE, in a CPMG
constant that characterizes the exponential buildup of the magneti- sequence are programmable acquisition parameters. Both are select-

JPT • JANUARY 2006 61

 Pore Size A good practice is to perform enough averaging to achieve a preci-
sion of at least ±1.0 porosity unit on NMR total porosities.
Depending on the drilling-fluid and formation conductivities and
Clay- Capillary - Free
bound bound water tool specifications, averaging three to nine depth levels usually is
water water required. The number of levels averaged, the antenna length, and
the sampling interval determine the vertical resolution of the mea-
surement. As part of prejob planning, service companies can pro-
A(T2) vide the vertical resolution of the tool for different logging modes
and environments.

NMR-Log Interpretation
T2 Distributions. T2 distributions provide much useful informa-
tion about reservoir-rock and -fluid properties and constitute the
basic outputs displayed on an NMR log. Most other NMR-log out-
T2 puts can be computed from these distributions. T2 distributions
computed from NMR echo data are used to compute NMR total,
Fig. 3—Partitioning of the T2 distribution of a typical water- bound-fluid, and free-fluid porosities, and also are used for perme-
saturated sandstone into bound (irreducible) and free
water by use of empirically determined cutoffs.
ability and reservoir-quality estimation. T2 distributions are com-
puted by fitting spin-echo signals to a sum of approximately 30 sin-
gle-exponential functions, each with amplitude, A(T2), and associ-
ed on the basis of logging objectives and expected formation and ated decay time, T2. The fitting procedure is achieved by a mathe-
fluid properties. In a typical NMR measurement, thousands of echoes matical technique known as inversion. The outputs of the inver-
are acquired over a period of approximately 1 second. The number sion are the amplitudes, A(T2), in porosity units corresponding to
of echoes depends on the expected formation T2 relaxation times. In each T2 value. A semilog graph of A(T2) vs. T2 is known as a T2 dis-
formations with long T2 relaxation times (e.g., formations with light tribution. The area under the T2 distribution equals NMR total
oils or rocks with large pores and/or vugs), more echoes are needed porosity. Fig. 3 shows a T2 distribution for a typical water-saturat-
to measure the long T2 values accurately in the T2 distribution. In ed shaly sandstone.
practice, the diffusion of molecules in the tool’s magnetic-field gradi- In water-saturated rocks, T2 distributions are qualitatively related
ent causes an additional T2 diffusion-decay mechanism that places to pore-size distributions. The T2 values typically span several
an upper limit on the longest T2 that can be measured. Longitudinal decades from less than 1 millisecond to several seconds. The wide
relaxation times, T1, are not affected by diffusion. range of T2 values observed in sedimentary rocks is caused by the
broad distributions of pore sizes. As a first-order approximation,
Importance of Prejob Planning. An important part of running a each T2 in the T2 distribution is proportional to a pore-size diame-
successful NMR logging job is prejob planning. Prejob planning ter. Thus, the short T2 values in a T2 distribution are associated with
involves close communication between the service company and signals from water in small pores, whereas the longer T2 values cor-
the client. The service companies have developed job-planning soft- respond to signals from water in larger pores. T2 distributions are
ware that is tool specific and can be used to select optimal NMR used to predict total porosity, bound-fluid porosity, free-fluid poros-
acquisition modes, measurement parameters, and logging speeds on ity, permeability, and pore-size distribution. Fig. 4 shows scanning-
the basis of the client’s objectives. The logging speed of wireline electron-microscope (SEM) images of two sandstones that have
NMR logging tools depends on the logging modes that are being nearly identical porosities but measured brine permeabilities that
run, which depend on the logging objectives. One of the most differ by a factor of approximately 37. The T2 distributions clearly
important factors determining logging speed is how much polariza- reveal the sandstone of better quality. The lower-permeability sand-
tion time is needed, which depends on T1. Formations that contain stone has shorter T2 values and more pore-filling clay as indicated
gas or low-viscosity oils with T1 values of several seconds need long by higher bound-water volume (e.g., the gray shaded part) than the
polarization times, which results in slower logging speeds (i.e., typ- more permeable rock.7
ically speeds in the range from 250 to 900 ft/hr). In many forma- Comparisons frequently are made between the pore-size infor-
tions (e.g., shaly sands having oils with viscosities greater than 10 mation contained in T2 distributions and mercury-injection capil-
cp) logging speeds of 1,800 ft/hr or faster are possible. lary pressure curves. It is important to note that capillary pressure
The S/N ratio of an NMR measurement determines the repeata- curves provide information on pore-throat sizes, whereas T2 distri-
bility of the measurement. As noted above, the S/N ratio of an butions are related to pore-body sizes. T2 distributions have been
NMR-logging-tool measurement depends on the sensor design found to provide information that complements capillary pressure
(i.e., on the magnitude of Bo , magnitude of the RF magnetic field, curves in many sandstones that have pore-body and pore-throat
and the volume of the formation measured by the tool). Special sizes that are well correlated.
pulse sequences are used to enhance the accuracy and repeatabili-
ty of NMR measurements.5,6 These sequences can improve mea- Lithology-Independent NMR Total Porosity. One of the most sig-
surement S/N ratios by a factor of approximately two compared nificant recent advances in formation evaluation is lithology-inde-
with a standard CPMG sequence.6 Highly conductive drilling flu- pendent NMR total porosity. The measurement of lithology-inde-
ids, low-porosity formations, and high temperatures can reduce the pendent total porosity is unique to NMR tools because porosities
S/N ratio significantly. Service companies can provide information derived from density, neutron, and sonic measurements depend on
about the effects of conductive boreholes on measurements made knowing rock-matrix properties. In heterogeneous formations hav-
by specific tools. NMR logging measurements are averaged to ing mixed or unknown lithology, the use of NMR logs for accurate
enhance the S/N ratio and improve the repeatability of the results. porosity prediction is highly recommended. Fig. 5 shows the advan-

62 JPT • JANUARY 2006

 Fig. 5—Porosity logs in mixed-lithology carbonates show the
advantage of NMR lithology-independent total porosity.
NMR total porosity is not sensitive to lithology and reads cor-
rectly over the entire interval. The density tool reads correct-
ly only in intervals in which the assumed matrix density
(lithology) used to compute the density-log porosity matches
Fig. 4—SEM images for two sandstones that have nearly iden- the actual matrix density of the formation.
tical porosities but very different permeabilities. The lower
permeability sandstone has more pore-filling clays and small-
er pores than the higher permeability sandstone. The differ- ble to all sandstones. Laboratory NMR measurements made on
ences in the rocks are revealed by the T2 distributions—the water-saturated core samples, before and after centrifuging, often are
lower permeability rock has shorter T2 values and more bound
water (i.e., the gray part of the distribution). used to obtain more-accurate T2 cutoffs for a specific rock formation.
In water-saturated carbonates, T2 cutoffs are significantly greater
(e.g., hundreds of milliseconds) than the cutoffs for sandstones.
tage of lithology-independent NMR porosity in mixed-lithology car- The computation of bound and free water from T2 distributions
bonates. Track 1 contains logs of rock mineralogy and fluid vol- in carbonates is more complex than in sandstones because in many
umes. The upper portion of the interval is mostly dolomite with carbonates, there are signals with long T2 decay times that come
traces of clay and quartz. The lower part of the interval is predomi- from bound water in isolated vugs. Another complication is that dif-
nantly limestone with varying amounts of dolomite. Track 2 con- ferent pore-size systems (e.g., dual-porosity pore systems with
tains two density-log porosity curves—computed either by assum- micro- and macroporosities) often are found in carbonates.
ing 100% limestone or 100% dolomite. The NMR total porosity also Molecules can diffuse from the micropore system to the macropore
is shown in Track 2. Note that the density-log porosities computed system during the NMR measurement time, obscuring the true
assuming a dolomite matrix agree with NMR total porosities in the pore-size distribution.8
upper dolomite section but read incorrectly in the lower section that The partitioning of T2 distributions into bound water and free
contains limestone mixed with dolomite. Similarly, the density-log water discussed above for water-saturated rocks assumes that short
porosities computed assuming limestone read incorrectly in the T2 relaxation times are associated with clay-bound, capillary-bound,
upper section and approach the NMR porosities only at the very and small-pore waters. This assumption is not valid in hydrocarbon-
bottom of the section. The NMR-log porosities are insensitive to the bearing rocks, which contain, for example, viscous but movable oil
lithology changes and read correctly over the entire interval. with T2 values less than the cutoff.9 Obviously, this movable oil
NMR total porosity equals actual formation porosity in most would be incorrectly tagged as bound water if one applied a cutoff
hydrocarbon-bearing and wet rocks, including shales. However, to the total or composite T2 distribution. One approach for solving
exceptions can occur in heavy-oil reservoirs with oil viscosities on this problem is to separate the two T2 distributions by use of diffu-
the order of 10,000 cp or greater. Such oils can have T2 distributions sion-based fluid-characterization methods.3,9
with significant amplitudes below the T2 sensitivity limits of NMR
tools. NMR-derived porosities in these heavy-oil reservoirs underes- Estimating NMR-Derived Permeability in Sandstones. Two
timate true formation porosities. The deficit of NMR-derived porosi- empirical permeability equations are used widely in the industry—
ties compared with density-log porosities can be used to infer the the Schlumberger-Doll Research (SDR) equation and the Timur-
presence of heavy oil and, with some assumptions, to place bounds Coates equation. The NMR permeability equations provide esti-
on the oil viscosity and oil saturation. mates of brine permeabilities in water-saturated sandstones. A good
correlation between pore-body and -throat sizes exists for many
Estimating Bound and Free Water. T2 distributions in water-satu- sandstones, which is the underlying basis for NMR-derived perme-
rated rocks can be partitioned into bound (i.e., irreducible) and free ability. Both permeability estimators contain model parameters (e.g.,
water by use of empirically determined T2 cutoffs. The partitioning exponents and proportionality constants). The default parameter
of the T2 distribution divides the total porosity into bound- and free- values used by service companies usually provide NMR permeabil-
water-filled porosities. The bound water for shaly sandstones con- ity logs that are qualitatively accurate (i.e., they can be used to pre-
sists of clay- and capillary-bound water as shown in Fig. 3. For sand- dict that one zone is more permeable than another zone). Although
stones, a default value of 33 milliseconds is commonly used for the this information is valuable, the estimated permeabilities can devi-
T2 cutoff that separates the bound water from the free water. This ate significantly from those of the formation. To estimate quantita-
works well in many cases; however, the default value is not applica- tively accurate log permeabilities for a specific type of rock, it is rec-

JPT • JANUARY 2006 63

 ed by shale or rock mineralogy, and, therefore, the DMR method is
a more reliable gas indicator.
Fig. 6 shows an example of the DMR method. The logged inter-
val consists of a shale overlying a shaly gas-bearing sandstone. Note
that the neutron/density crossover effect in the gas sand is sup-
pressed. The NMR-tool response in the gas zone is not affected by
shale and exhibits an NMR-/density-log crossover effect. The vol-
ume of gas shown in Track 1 and the gas-corrected total porosity
shown in Track 2 were computed with the DMR gas equations.11
The DMR method provides a simple and robust technique for
evaluating gas-bearing reservoirs; however, other methods are need-
ed to differentiate water from viscous oils. These methods are dis-
cussed in the next two sections.

Diffusion-Based NMR Fluid-Characterization Methods. Akkurt

et al.12 first recognized that diffusion in a magnetic-field gradient
could be used as a mechanism for fluid characterization and pro-
posed a standalone NMR diffusion-based method for differentiating
gas from water. The early diffusion-based methods were limited in
their range of applicability and were superceded by more-accurate
and general model-based inversion methods.9 Today, diffusion-
based methods provide a means for detecting and evaluating zones
that are difficult or impossible to accurately interpret by use of con-
Fig. 6—A log showing the DMR method in a shaly gas-bear- ventional log-analysis methods. Applications include low-resistivity
ing sandstone. The density- and magnetic-resonance- pay zones, which can be missed using conventional resistivity-tool
porosity logs exhibit a crossover effect in the gas-bearing interpretation; formations with low-salinity connate water or water
sandstone. Thermal neutron absorbers in the shaly sand- with unknown or variable salinity; and formations with complex
stone cause the neutron porosities to read too high and
lithology for which Archie parameters are variable or unknown.
suppress the neutron/density crossover. The gas volume
(Vgas) in Track 1 and gas-corrected porosity in Track 2 are Modern fluid-characterization methods use data suites consisting
computed by use of the DMR equations.11 of diffusion-encoded pulse sequences that are simultaneously invert-
ed to provide continuous logs of fluid (water, oil, and gas) satura-
ommended that measurements on core samples be analyzed to tions and oil viscosities.3 These data suites are sensitive to both the
determine optimal permeability-equation parameters. Estimates of diffusion coefficients, D, and the relaxation times of the fluids.
SDR and Timur-Coates permeability both are affected by hydrocar- Molecular diffusion is the random thermal motion of molecules in a
bons, and more research is needed to develop equations to predict fluid. Diffusion-encoded sequences are modifications of CPMG
relative permeabilities to hydrocarbons and water. sequences in which one or two echoes have an increased spacing.13
The SDR and Timur-Coates permeability equations are based on Molecular diffusion in a magnetic-field gradient shortens the T2
assumptions that are not valid for carbonates because pore-body relaxation times, and this effect is amplified during the increased
sizes measured by NMR and pore-throat sizes are not well correlat- echo spacings of the diffusion-encoded sequences. By acquiring sev-
ed in carbonates. Carbonates also can have NMR signals from fluids eral diffusion-encoded measurements with different long echo spac-
in isolated vugs, which contribute to NMR porosity but not to per- ings, it is possible to measure molecular-diffusion rates directly.
meability. Permeability estimation in carbonates is one of the most
active and challenging areas of NMR-logging research.10 Multidimensional NMR Fluid Characterization. Recent advances
in NMR data acquisition and processing use suites of diffusion-
Methods for Evaluating Hydrocarbon-Bearing Reservoirs encoded data to compute 2D or 3D maps that can be used to visu-
Density/Magnetic-Resonance (DMR) Method for Gas-Bearing ally identify fluids present in a reservoir on the basis of contrasts in
Reservoirs. The DMR method combines density- and NMR-log either the relaxation time and/or diffusion-coefficient distributions
porosities to predict gas-corrected formation total porosities and of the fluids.13–15 The term multidimensional refers to the measure-
flushed-zone water saturations.11 The DMR method exploits the fact ment and display of NMR-signal amplitudes as functions of two or
that density-log porosities read too high and NMR total porosities three of the fluid parameters T1, T2, T1/ T2, and D.
read too low in gas-bearing reservoirs. Density porosities read too Multidimensional maps have proved essential for accurate fluid
high because of the reduced density of gas. NMR total porosities analysis of reservoirs invaded with oil-based drilling-mud filtrate
read too low because of the reduced hydrogen index of gas (i.e., den- (OBMF). The multidimensional inversions used to construct maps
sity of protons of a fluid relative to water). This effect causes densi- do not rely on empirical models, which is an advantage when
ty- and NMR-porosity logs to exhibit a crossover effect in gas zones attempting to identify and characterize several miscible fluids. If
similar to that exhibited on neutron/density logs. OBMF is present in the formation, then it can contribute to the NMR
The DMR method is superior to the traditional thermal neu- signal, together with any residual native hydrocarbons and
tron/density method for detecting and evaluating gas zones. The unmoved connate water. The presence of OBMF complicates the
thermal neutron gas response in shaly sands can be suppressed by interpretation of the NMR data. The severity of the interpretation
the presence of thermal neutron absorbers, which cause neutron problem posed by the invasion of OBMF depends on whether it
porosities to read too high. As a result, neutron/density logs can miss mixes with the native hydrocarbons. The invasion dynamics and
gas-bearing zones in shaly sands. NMR total porosities are not affect- fluid properties that determine if mixing will occur are not well

64 JPT • JANUARY 2006

Fig. 7—The D-T2 map from a clay-free Bentheim Fig. 8—D-T2 maps from a water-saturated sandstone (left) and oil-bear-
sandstone saturated with water and a dead ing sandstone (right) in a deepwater Gulf of Mexico well drilled with
(i.e., zero GOR) North Sea stock-tank oil. The OBM. Note that the peak corresponding to the OBM filtrate falls on the oil
water and oil signals are well separated line in the water-saturated sandstone but moves to higher D and shorter
because water molecules have a larger diffu- T2 in the oil-bearing sandstone because of the effects of dissolved gas
sion coefficient than the bigger and less-mobile (after Freedman and Heaton3).
hydrocarbon molecules in the oil.

understood, but experience with the maps has shown that mixing responds to water that was not flushed by the invading filtrate. The
does occur in some cases. If the OBMF and native hydrocarbons D-T2 map on the right was obtained in an oil-bearing sandstone in
remain separate, and if they have significantly different relaxation- the same well. Note that the strong filtrate peak does not fall on the
time or diffusion-coefficient distributions, then the maps can be oil line but has shifted to a significantly higher diffusion rate and
used to identify the separate fluids and predict the saturation and shorter T2 relative to its position in the water-saturated sandstone.
viscosity of the native oil. If the OBMF and the native hydrocarbons This shift is the result of dissolved gas, which reduces the viscosity
mix, the fluids can lose their identities and coalesce into a single of the filtrate and causes an increase in its diffusion rate. The dis-
fluid with properties not representative of the native reservoir. solved gas also reduces the T2 of the filtrate. The same effect is
Fig. 7 shows a 2D map of signal amplitude vs. D and T2 for a known to occur in live oils with high GORs. The second peak that
clay-free Bentheim sandstone core partially saturated with water appears at the right edge (long T2) of the map above the water dif-
and dead [i.e., zero gas/oil ratio (GOR)] North Sea stock-tank oil. fusion line corresponds to the light hydrocarbon. Note that the
The oil gravity was 33°API, and the oil saturation in the core was OBMF and native oil yield separate identifiable peaks, confirming,
57%. The NMR data were acquired with a laboratory NMR instru- in this case, that the fluids have not mixed. The long T2 and high
ment. The colors of the map indicate the signal amplitude at each diffusion rate of the native-oil peak indicate that the oil is very light.
point in the D-T2 plane. The dark blue background corresponds to When interpreting these maps, one must be aware of deviations
zero signal amplitude. Maximum amplitude is red. The horizontal from ideal-fluid responses. For example, in some rocks, the water sig-
reference line indicates the diffusion coefficient of free water (i.e., nal is observed to lie significantly above the free-water diffusion line.
for unrestricted diffusion). A strong water signal can be seen to lie This effect is caused by the presence of “internal gradients” that are
on the free-water diffusion line. The oblique line is the so-called induced by the magnetic field of the NMR magnet. Iron-rich chlorite
dead-oil line, on which signals from dead crude oils typically lie. clays or other magnetic minerals in the rock matrix cause the internal
For live oils, the oil signal is shifted away from the dead-oil line gradients. Internal gradients tend to be large in small pores and can
toward the northwest. The magnitude of the shift depends on the perturb the magnetic-field gradient of the tool, which leads to uncer-
GOR. A strong signal from the dead North Sea oil can be seen to lie tainties in measured diffusion coefficients. In some rocks, water sig-
on the dead-oil line. nals lie significantly below the free-water diffusion line. This effect is
The two subplots below and to the right of the map are the pro- caused by “restricted diffusion” of water molecules in small pores.
jections of the map amplitudes onto the T2 and D axes, respective- This effect is common in carbonates and other rocks with microp-
ly. These projections are composite T2 and D distributions that rep- orosity and occurs when the distance that water molecules diffuse
resent all of the fluids in the rock (e.g., in this case, oil and water). during the measurement time becomes comparable to the pore size.
Observe that the oil and water signals are not well separated by the The effects of restricted diffusion can complicate interpretations
T2 distribution because the water and oil T2 distributions overlap. because restricted water can be mistaken for oil.15 Other log data (e.g.,
The water and oil show separate peaks in the D distribution because from resistivity and fluid-sampling tools) often are used in conjunc-
the diffusion coefficient of the water is approximately 10 times larg- tion with NMR data to develop an unambiguous interpretation.
er than the diffusion coefficients of the North Sea oil. This is a good For heavy oils, with small diffusion coefficients (e.g., D≤0–7 cm2/s),
example of how contrasts in the diffusivities, D, of water and oil can it is not possible to measure the diffusion accurately with NMR logging
be used for fluid identification, even when the relaxation times are tools. There is a lack of sufficient diffusion sensitivity for oils with T2
similar or identical, which occurs frequently. values below approximately 20 milliseconds.
The next example is a field log from the paper by Freedman and
Heaton.3 The two D-T2 maps in Fig. 8 were obtained in a deepwa- Ongoing Research and Possible Future Applications
ter well drilled with an OBM in the Gulf of Mexico. The left map Recent and ongoing research on methods for inferring rock wetta-
was acquired in a water-saturated sandstone and shows a typical bility from NMR looks promising. It is hoped that this work will
OBMF response on the oil line with a T2 of approximately 1.5 sec- lead to useful downhole techniques for inferring wettability.16,17
onds. The faint peak with higher diffusion rate and shorter T2 cor- Another promising research area is the use of NMR measurements

JPT • JANUARY 2006 65

 to estimate pressure/volume/temperature (PVT) properties of reser- Acknowledgments
voir fluids. It is anticipated that this research will eventually lead to I am indebted to the many scientists, engineers, and technicians
predictions of in-situ PVT properties of reservoir fluids including who have painstakingly contributed to the development of pulsed-
molecular composition of crude oils.18 This method would have a NMR logging tools and interpretation methods. I want to give spe-
large effect on improving conventional formation-evaluation and cial thanks to my Schlumberger colleagues in research, engineering,
well-completion results. and operations who have contributed invaluably to my knowledge
Also, promising research efforts are under way to exploit NMR of NMR logging. I also benefited from long-term collaborative
diffusion measurements to better define rock/pore-space connectiv- research with G.J. Hirasaki and his graduate students at Rice U.
ity and structure.13 Such research can lead to a better understand-
ing of producibility in complex reservoirs. References
Looking further into the future, there is potential to use NMR 11. Brown, R.J.S. et al.: “The History of NMR Well Logging,” Concepts in
methods to image fluids in reservoirs in the same way that magnet- Magnetic Resonance Logging, 13, No. 6, 340.
ic-resonance imaging is used in medicine to image soft tissues. On 12. De Pavia, L. et al.: “A Next-Generation Wireline NMR Logging Tool,”
the distant horizon are possible applications of NMR spectroscopy paper SPE 84482 presented at the 2003 SPE Annual Technical
to well logging, such as in-situ measurements of the aliphatic/aro- Conference and Exhibition, Denver, 5–8 October.
matic ratios in crude oils. 13. Freedman, R. and Heaton, N.: “Fluid Characterization Using Nuclear
Magnetic Resonance Logging,” Petrophysics (May/June 2004) 45, No. 3, 241.
Conclusions 14. Freedman, R. et al.: “Measurement of Total NMR Porosity Adds New
NMR logging was first introduced during the 1960s when NML Value to NMR Logging,” paper OO presented at the 1997 Annual Meeting
tools were used to make crude free-precession measurements in the of the Soc. of Professional Well Log Analysts, Houston, 15–18 June.
Earth’s magnetic field. The NML tools were retired from service in 15. Prammer, M. et al.: “Measurements of Clay-Bound Water and Total
the late 1980s. Modern pulsed-NMR tools were introduced in the Porosity by Magnetic Resonance Logging,” paper SPE 36522 presented
early 1990s, and this technology has had a major effect on forma- at the 1996 SPE Annual Technical Conference and Exhibition, Denver,
tion evaluation. A wide variety of wireline and LWD tools and ser- 6–9 October.
vices is available. 16. McKeon, D. et al.: “An Improved NMR Tool Design for Faster Logging,”
The proper use of NMR technology requires good communica- paper CC presented at the 1999 Annual Meeting of the Soc. of
tion between oil and service companies to ensure that logging objec- Professional Well Log Analysts, Oslo, Norway, 30 May–3 June.
tives and data-quality expectations are met. The latter involves care- 17. Allen, D. et al.: “How to Use Borehole Nuclear Magnetic Resonance,”
ful prejob planning that includes if necessary, plans to mitigate bore- Schlumberger Oilfield Review (Summer 1997) 9, No. 2, 34.
hole environmental effects (e.g., high mud conductivities and high 18. Ramakrishan, T.S. et al.: “Forward Models for Nuclear Magnetic
temperatures) on data quality. Resonance in Carbonates,” The Log Analyst (1999) 40, No. 4, 260.
There are unique and powerful standalone NMR formation-eval- 19. Freedman, R. et al.: “A New NMR Method of Fluid Characterization in
uation methods available for evaluating fluid types, saturations, and Reservoir Rocks: Experimental Confirmation and Simulation Results,”
porosities in complex reservoirs. Additionally, NMR logs provide paper SPE 75325 SPEJ (December 2001) 452.
estimates of bound- and free-water volumes, oil viscosities, and for- 10. Ahr, W.M. et al.: “Confronting the Carbonate Conundrum,”
mation permeabilities subject to the caveats discussed in this article. Schlumberger Oilfield Review (Spring 2005) 17, No. 1, 18.
In complex reservoirs, NMR and other log data should be integrat- 11. Freedman, R. et al.: “Combining NMR and Density Logs for
ed with all other available reservoir information (e.g., especially data Petrophysical Evaluation in Gas-Bearing Formations,” paper II present-
from fluid-sampling and pressure tools and core data) to provide the ed at the 1998 Annual Meeting of the Soc. of Professional Well Log
most accurate picture of the reservoir. Analysts, Keystone, Colorado, 26–29 May.
12. Akkurt, R. et al.: “NMR Logging of Natural Gas Reservoirs,” The Log
Nomenclature Analyst (1996) 37, No. 6, 33.
A(T2) = Amplitude in a T2 distribution corresponding to relaxation 13. Hürlimann, M.D. et al.: “Diffusion-Editing: New NMR Measurements of
time T2, porosity units. Saturation and Pore Geometry,” paper FFF presented at the 2002 Annual
Bo = Magnetic field vector* in the formation produced by the Meeting of the Soc. of Professional Well Log Analysts, Oiso, Japan, 2–5 June.
logging tool as it move through the borehole, gauss. 14. Heaton, N.J. et al.: “Saturation and Viscosity from Multidimensional
D = Molecular diffusion coefficient of a water, gas, or oil mole- Nuclear Magnetic Resonance Logging,” paper SPE 90564 presented at
cule, cm2/s. the 2004 SPE Annual Technical Conference and Exhibition, Houston,
Mo = Equilibrium value approached by the proton magnetiza- 26–29 September.
tion at infinite polarization time, gauss. 15. Guru, U. et al.: “Low-Resistivity Pay Evaluation Using Multidimensional
TE = Echo spacing (also equal to pulse spacing) in a CPMG and High Resolution Magnetic Resonance Profiling,” paper OOO pre-
spin-echo sequence, milliseconds. sented at the 2005 Annual Meeting of the Soc. of Petrophysicists and
T1 = Longitudinal relaxation time used to characterize the Well Log Analysts, New Orleans, June 26–29.
build-up of the proton magnetization during the polariza- 16. Freedman, R. et al.: “Wettability, Saturation, and Viscosity From NMR
tion time, seconds. Measurements,” paper SPE 87340, SPEJ (December 2003) 317.
T2 = Transverse relaxation time used to characterize the decay 17. Flaum, M., Chen, J., and Hirasaki, G.J.: “NMR Diffusion-Editing for D-
of the NMR spin-echo signal, seconds. T2 Maps: Application to Recognition of Wettability Change,”
Tp = Polarization time during which the proton magnetization Petrophysics (April 2005) 46, No. 2, 113.
is allowed to build up prior to application of the 90° pulse, 18. Freed, D.E., Durcaw, L., and Song, Y.Q.: “Scaling Laws for Diffusion
seconds. Coefficients in Mixtures of Alkanes,” Physical Review Letters (2005) 94,
φ = Porosity, fraction 067602, 1. JPT

* Boldface type is used to denote that the magnetic field is a vector.

66 JPT • JANUARY 2006

View publication stats