Well Logging quantification of the fluids that occupy the pore space (water

brine, hydrocarbons, methane), and the fluid viscosities.

There are two basic modes of well logging: wireline logging
and logging-while-drilling. In wireline logging operations, a
Martin D. Hürlimann* string of sensors are connected to a long armored cable that
Schlumberger-Doll Research, Cambridge, MA, USA
acts simultaneously as a mechanical support, a conduit of
power, and a link to communicate with the sensors and to
retrieve the information from the sensors. The assembly is first
lowered to the bottom of the borehole. As it is then brought
back up to the surface at a controlled rate that is typically in
1 Introduction 1
the range of 2 to 30 cm s−1 , the sensors perform continuous
2 Sensors for NMR Logging 2
measurements. This results in ”well logs” of the formation.
3 Measurement Technique 5
Under special circumstances, it is also possible to perform
4 Applications and Interpretation 6
”station stops” and to make measurements while the sensors
5 Conclusion 9
are stationary.
6 Related Articles 9
In logging-while-drilling operations, the sensors are directly
7 References 9
integrated into the drill string and the measurements are
performed while the borehole is being drilled. This mode
enables the measurements to guide the drilling operation
1 INTRODUCTION and to greatly improve the accuracy of the placement of
the borehole at the desired location in the earth formation.
NMR well logging deals with the characterization of fluids This logging mode is particularly challenging for NMR
filling the pore space of earth formations.1 This problem lies measurements due to tight space constraints, very limited
at the heart of a number of important applications in the communication capabilities, and the presence of often severe
geosciences. Hydrology is concerned with the distribution vibrations induced by the drilling operation.
and transport of water in the subsoil. The exploration and Given the harsh environment in a borehole, it is at first
production of oil and gas reservoirs rely on the ability to sight not obvious that NMR is at all a suitable technique
identify and quantify the hydrocarbons, to determine their for well-logging applications. The measurements have to be
composition, and to assess the flow properties of the fluids performed on a sample that is located outside the sensor. As a
through the reservoir. For carbon sequestration applications, it consequence, it is not possible to apply a strong, uniform mag-
is essential to understand the connectivity of the pore space netic field to the sample. Measurements have to be performed
and to monitor the injected fluid inside the earth formations. while the sensor is moving relative to the sample. The whole
For shallow aqueducts and reservoirs, it is possible to sensor and the associated electronics have to fit into the bore-
obtain detailed information from surface measurements such hole that has a typical diameter of 20 cm and have to be able to
as ground-penetrating radar or NMR measurements in the withstand extreme conditions of temperature and pressure that
earth’s magnetic field. Similarly, seismic measurements can can reach 175 ◦ C and 140 MPa. Furthermore, the measurement
be used to determine the general geological structure of has to be sufficiently robust so that it can be performed in a
the underlying earth formation. However, the acquisition highly automated manner without the assistance of an NMR
of high-resolution information on earth formations deeper specialist in a wide range of environments (on-land, off-shore,
underground or off-shore requires measurements from sensors tropical to arctic regions) and in boreholes filled with fluids
in boreholes. The technique of measuring physical properties of varying electrical conductivities ranging from salt-saturated
of the subsurface using specialized sensors lowered into brines to nonconducting hydrocarbons. The only mitigating
boreholes is referred to as well logging. factor is that the samples encountered in NMR logging are
The evaluation of hydrocarbon reservoirs is the most highly larger than those typically encountered in laboratory NMR
developed application of well logging. A multitude of sensors experiments. Nevertheless, given all these challenges, it is
has been developed that probe different properties of the truly remarkable that NMR well logging has become a com-
earth formation, including the mechanical, electromagnetic, mercially successful technique that now routinely provides
acoustic, nuclear, and NMR properties.1 These sensors have critical quantitative information on fluid and flow properties
to be able to withstand the extreme conditions of oil and gas in earth formations that are many kilometers underground.
reservoirs that can be found at depths exceeding 10 km and at This success has been made possible by crucial develop-
temperatures up to 175 ◦ C. ments in a number of different areas over the last few decades.
NMR measurements are particularly useful for the charac- They span the areas of hardware development, measurement
terization of the reservoir properties because they probe the techniques, and the underlying science that relates the mea-
fluids directly. NMR logging has the capability to determine sured NMR quantities to the properties of the earth formation.
a wide range of important physical properties of earth for- Advances in hardware included the development of specialized
mations. They include the porosity, i.e., the fraction of the pulsed NMR sensors with novel configurations of permanent
formation occupied by fluid (typical ranges are between magnets and radiofrequency (RF) coils suitable for borehole
3% and 30%), the distribution of pore size, an estimate of operation. The progress in measurement techniques was made
the flow properties of the formation, the identification and possible by the detailed understanding of spin dynamics in
grossly inhomogeneous fields and the development of new
∗
Update based on original article by Robert L. Kleinberg, Encyclopedia of Magnetic techniques in data inversion. This has resulted in new schemes
Resonance, © 1996, John Wiley & Sons Ltd. to measure novel multidimensional distribution functions of

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
2 WELL LOGGING

relaxation and diffusion. Furthermore, much progress has been approach is closely related to standard NMR measurements.
made in the fields of physics of porous media and the physics The generic instrument consists of an array of magnets and
of complex fluids. This has been crucial to enable the direct an RF coil that is connected to an RF transmitter and a
extraction of the physical parameters of interest in the earth sensitive receiver. The magnets are designed to generate a
formation from the measured NMR relaxation and diffusion strong magnetic field in the formation. The RF coil is used
quantities. An extensive historical review of the development to transmit strong RF pulses into the formation that generate
of NMR logging up to 2001 can be found in Ref. 2. Related net precessing magnetization, which is then detected with
developments in the field of single-sided NMR measurements the same RF coil. The pulsed technique can overcome the
have been described in Ref. 3 and in a book.4 The progress of main limitations of measurements based on the earth’s field
physics in porous media as applied to magnetic resonance is mentioned above.
covered by the biannual conference series of Magnetic Reso-
nance in Porous Media (MRPM).5 • Using externally applied magnetic fields, it is possible to
achieve a strong spatial localization of the NMR signal
with a well-defined volume of investigation. Signal is only
generated in regions where the local Larmor frequency
2 SENSORS FOR NMR LOGGING (proportional to the local field strength) is close to the
RF frequency of the RF pulses used. This eliminates the
2.1 Earth’s Field NMR Technique need to dope the drilling fluid with paramagnetic salts and
it avoids the uncertainties associated with the borehole size
The first practical implementation of NMR logging was and invasion profiles of the paramagnetic ions encountered
based on signal detection in the earth’s magnetic field.6 The in the earth’s field technique.
central element of the logging apparatus was a large coil • The detection frequency is up to three orders of magnitude
consisting of 1000 turns. Using a 2 kW power supply, the higher than in the earth’s field measurements. In current
coil was first energized with a large current to produce a logging instruments based on modern permanent magnets,
strong static magnetic field that increased the polarization of the operating frequencies range from about 150 kHz to
the nuclear spins in the surrounding formation. The current 2 MHz. As a consequence, the electronic recovery times
in the coil was then abruptly turned off. This nonadiabatic are much shorter compared to earth’s field operation and
removal of the applied field generated a free induction decay it becomes feasible to detect rapidly relaxing components
in the earth’s magnetic field that was detected with the same and to determine the total porosity.
coil. The signal frequency was the Larmor frequency in the • The pulsed NMR technique allows flexibility in the type
local earth’s magnetic field and was in the range of 1.0 to of measurements to be performed. The pulse sequence and
2.8 kHz depending on the geographical location. pulse parameters are now routinely tailored to the specific
With this earth’s field approach, it is possible to generate application of interest. As discussed below, measurement
transverse magnetization from a relatively large volume techniques are now available to determine not only porosity
resulting in an adequate signal-to-noise ratio despite the low but also relaxation and diffusion properties quantitatively
detection frequency. However, the earth’s field method has in grossly inhomogeneous fields. In addition, it is possible
some intrinsic shortcomings that limit the information that can to vary the spatial location of the volume of investigation
be quantitatively extracted from the measurements. First, the with the choice of the carrier frequency of the RF
deadtime of the electronics after turning off the large current pulses.
is typically about 25 ms. In many formations, there are signal
components with relaxation times shorter than this deadtime.
In such formations, only a fraction of the proton NMR signal is
2.2.2 General Considerations for Tool Design
therefore detectable with this method. Second, there is only a
weak capability for the spatial localization of the NMR signal
It is critical to find configurations of magnets and RF
and it is necessary to eliminate the NMR signal of the fluid in
coils that create sufficiently large resonant volumes that
the borehole. This was achieved by doping the borehole fluid
in turn result in sufficiently strong NMR signals with
with sufficiently high concentrations of paramagnetic salts.
adequate signal-to-noise ratios. Before discussing specific
However, the shape of the borehole and the exact invasion
sensors currently used in NMR logging, we first review the
profile of the paramagnetic ions into the formation are variable
general considerations that control these designs.
and not known to high accuracy. This results in an uncertainty
Since the sample is outside the NMR apparatus, it is not
in the relationship between the amplitude of the NMR signal
possible to apply uniform B0 and B1 fields to the sample. This
and the formation porosity, even when there are no rapidly
leads to a large distribution of the local Larmor frequency
relaxing components.
ωL (r ) = γ B0 (r ) and nutation frequency ω1 (r ) = γ B1,⊥ (r )/2.
In inhomogeneous fields, the in-phase voltage V induced in
the RF coil by the normalized magnetization m⊥ (r ) is given
2.2 Pulsed NMR Technique in an Applied Magnetic Field by Hoult and Richards7

2χ ω1 (r )
2.2.1 Advantages of Pulsed NMR Technique V =φ dr B02 (r ) F (ω0 (r )) m⊥ (r ) (1)
μ0 I
Modern NMR logging devices use an alternative approach Here φ is the porosity, χ is the susceptibility of hydrogen
that is based on pulsed NMR techniques in an externally nuclei, I is the current in the antenna, and F (ω0 ) is the
applied magnetic field. Conceptually, this pulsed NMR frequency response of the antenna and includes the response

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
 WELL LOGGING 3

of any additional filters. We use the notation ω0 (r ) ≡ ωL (r ) − 2.2.3 Configurations of NMR Logging Tools
ωRF to indicate the offset of the local Larmor frequency from
the applied RF, ωRF , in the rotating frame. The factor ω1 /I In 1980, Jackson and collaborators at Los Alamos National
is the detection efficiency of the antenna according to the Laboratory9 – 11 described the first practical solution of this
reciprocity theorem. optimization problem and demonstrated that it is feasible to
To produce a large signal, it is beneficial to use magnet use the pulsed NMR technique for NMR logging. Since then,
assemblies that generate the highest possible B0 static field in a number of additional configurations have been proposed
the formation. In equation (1), one factor of B0 is due to the and developed into commercial logging tools. The different
Boltzmann factor that describes the polarization effect, while configurations of the magnet and RF coils can be classified
the other factor of B0 comes from the detection efficiency according to the symmetry of the resonant volume (azimuthal
based on Faraday’s law. versus one-sided) and the profile of the magnetic field across
Furthermore, the resonant volume has to be maximized. This the sensitive region (saddle point versus gradient).
is the volume where the Larmor condition is fulfilled and Logging tools with azimuthally symmetric resonant regions
where significant coherent transverse magnetization m⊥ can probe the formation all around the borehole and are operated
be generated. The resonant volume can be approximated by while being centralized in the borehole. They are often
the region in space where the magnitude of the offset Larmor designed in a few sizes that are optimized for different sizes
frequency ω0 (r ) is less than the nutation frequency ω1 . This of the borehole. Logging tools with one-sided sensitivity
condition illustrates a general fact in NMR logging: an increase patterns are pushed against the borehole wall during the
in available RF power increases the NMR signal that can be logging operation. In this case, the depth of investigation is
generated. An increase of the instantaneous RF power results independent of the diameter of the borehole and the same tool
in larger values of ω1 and therefore increases the resonant can be used in a range of borehole sizes.
volume. In current commercial logging tools, typical levels of There are also trade-offs between tools with saddle points
RF power are of the order of kilowatts. versus those with gradient profiles. A saddle point profile
The resonant volume depends on the profile of the magnetic generates a relatively wide resonant region that makes the
field in the formation. It would be ideal to use a magnet measurements less sensitive to lateral motion than in a tool
configuration that produces a profile with a broad maximum based on a gradient design. Measurements in a gradient tool
in the formation. However, Earnshaw’s theorem states that are generally more affected by diffusion effects. Depending on
there are no such solutions.8 Nevertheless, it is possible to the application, this can be considered either an advantage or a
generate fields that have a relatively uniform magnitude over disadvantage. In a gradient tool, it is relatively straightforward
to change the depth of investigation by changing the RF
an extended region. In particular, this occurs with field profiles
frequency. In a saddle point design, the RF frequency has
characterized by a saddle point where the local gradient
to be set to the Larmor frequency at the saddle point. It is
vanishes. Such profiles can be generated by a superposition
important to track any drifts in this Larmor frequency due to
of the fields from different magnet components that generate
possible changes in the magnet temperature or accumulation
field gradients in opposite directions. While this compensation
of magnetic debris on the magnets.
approach increases the resonant volume, it will unfortunately
also reduce the absolute field strength B0 . The designer of
logging tools therefore has to find a compromise between 2.2.4 Centralized Logging Tool with Saddle Point Magnetic
optimizing the absolute field strength and optimizing the Field Profile
volume of the resonant region.
There is a long list of additional requirements for a The magnet configuration introduced in the pioneering work
successful design of an NMR logging tool. It is essential by Jackson et al.9 – 11 consists of two cylindrical magnets that
that the resonant region is completely confined to the earth are arranged coaxially with a spatial separation and that are
formation surrounding the borehole and that it does not polarized along the axis of symmetry in opposite direction.
intersect the borehole. It is desirable that the resonant region In the plane of symmetry perpendicular to the axes of the
is located as far away from the borehole wall as possible to magnets, they produce a purely radial field with azimuthal
minimize the impact of possible formation damage induced by symmetry. The field strength starts at zero in the center, rises to
the drilling operation on the measurement. A large depth of a maximum, and then falls off at larger distance. By adjusting
investigation also reduces the fraction of the formation fluid the magnet spacings, it is possible to achieve a relatively
within the sensitive region that is being replaced by invasion uniform field profile in a toroidal region. An efficient way
of borehole fluid. The design criterion for a large depth of to generate an RF field perpendicular to the static field in
investigation has to be balanced with the goal of maximizing the resonant region is to place a tuned solenoid coaxially
the signal-to-noise ratio of the measurement. With current between the magnets. The resulting resonant region is in the
commercial logging tools, the largest depth of investigation form of a toroid and has azimuthal symmetry, but not a large
is about 10 cm from the borehole. extent along the sensor axis. Therefore, this configuration is
The sensor design also has to take into account that in well suited for logging-while-drilling applications where the
logging mode, the apparatus is moving during the measure- dominant sensor motion is circular, but less suited for wireline
ments and the resonant region has to extend along the direction logging where the sensor is moved along the borehole. Several
of the motion. For wireline applications, the resonant region commercial logging-while-drilling devices12,13 are based on
has to extend along the axis of the apparatus, whereas for modified versions of this basic tool configuration. A schematic
logging-while-drilling applications, the resonant region is re- drawing of the device described in Ref. 12 is shown in
quired to have azimuthal symmetry. Figure 1.

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
4 WELL LOGGING

Sensitive region Sensitive Sensitive

(14" diameter × 6" height) volume No. 1 volume No. 2
RF coil
Mud flow

Borehole Formation

Annular
magnets Optional stabilizer
B0 B0

Figure 1 Schematic view of the NMR logging-while-drilling device

with an azimuthally symmetric sensitive region described in Ref. 12. Frequency Frequency
The two permanent magnets, shown in green, are magnetized along the band No.1 band No.2
6"
tool axis in opposite directions and generate a magnetic field that is
radial in the sensitive region (Courtesy of Schlumberger)
16"

2.2.5 Centralized Logging Tool with Gradient Magnetic Field

Profile
Figure 2 Schematic cross section of a wireline NMR logging tool15
Shtrikman and Taicher14 developed an alternative tool based on the design by Shtrikman and Taicher. The permanent magnet,
configuration that is based on two-dimensional dipole profiles shown in gray, is magnetized in the direction of the arrow. The tool
for both the static and RF magnetic fields. The static is centered in the borehole (shown in red), and the sensitive regions
field is created by a long cylindrical magnet, magnetized are thin cylindrical shells. In this design, B0 is perpendicular to B1
perpendicular to its axis. When end effects are neglected, everywhere in space. (Courtesy of NUMAR Corp.)
the resulting magnetic field is in the transverse plane with
an amplitude that depends only on the distance from the both magnetized transverse to the sonde axis. The resulting
tool. The direction of the field varies along the azimuth. magnetic field is predominantly radial into the formation
This static field is matched by an RF field generated by and its profile has a saddle point about 25 mm inside the
a coil that is wound around the magnet and oriented in formation. The field strength in the resonant region is 55 mT,
such a way that it produces a two-dimensional dipole RF so the Larmor frequency is 2.3 MHz. The RF coil occupies a
field that is everywhere perpendicular to the static field. The semicircular cylindrical cavity on the face of the sonde. It can
magnets are built from nonconducting ferrite so that they are be thought of as a half coaxial cable 15 cm long that irradiates
transparent to the RF fields. This elegant design results in the formation with a field transverse to the static field. It is
an azimuthally symmetric resonant region that consists of a loaded with ferrite to improve the reception of weak signals
thin cylindrical shell extending along the tool. The diameter from the nuclear spin system. This device, shown in Figure 3,
of the shell is controlled by the operating RF frequency. In has been commercialized for wireline applications.
practical implementations, the magnets are extended beyond
the RF coils to attenuate the impact of the end effects in
the field profiles and to better polarize the spins before 2.2.7 Single-sided Logging Tool with Gradient Magnetic Field
they enter the measurement zone. The symmetry of this Profile
design makes it suitable for both wireline (see Figure 2) and
logging-while-drilling applications. Commercial devices have When a single-sided logging tool with a saddle point is
been developed for both applications.15,16 In these devices, the operated at an RF that does not correspond to the saddle
operating frequencies are in the range of 500 kHz to 750 kHz, point, the sensor becomes a one-sided logging tool with a field
and the corresponding diameters of the resonant regions are in profile that is characterized by a gradient. More generally, any
the range of 13.5 to 17 in. Within the resonant region, the field sensor that has an RF coil located on one side operates as a
profile is characterized by a well-defined gradient that falls in single-sided gradient tool unless special care is taken to design
the range of 140 mT m−1 to 210 mT m−1 . This leads to thin it as a saddle point tool. Such a configuration has the advantage
shells with a typical thickness of about a millimeter. that it can be easily configured to have a sensitive region
at different depths away from the borehole simply by using
different RFs. Examples of such sensors used in commercial
2.2.6 Single-sided Logging Tool with Saddle Point Magnetic wireline operations are described in Refs 18 and 19. As an
Field Profile illustration, Figure 4 shows the wireline NMR logging tool of
Ref. 18 that can be operated at different RFs between 1000 kHz
Kleinberg et al.17 introduced a single-sided design with a and 500 kHz, corresponding to depths of investigations ranging
magnetic field profile characterized by a region of a saddle from 1 inch to 4 inches. In each case, the field profile is well
point that is extended along the tool direction. This field characterized by a single gradient that varies between 380 mT
profile is achieved with the use of two SmCo magnets that are m−1 and 120 mT m−1 .

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
 WELL LOGGING 5

For this reason, NMR logging focuses on the measurements

of the relaxation and diffusion properties of the earth
Permanent magnet formation. Since the pore geometry in these samples is often
Bowspring Borehole
strongly heterogeneous and the composition of the fluids is
eccentralizer wall variable, it is generally not possible to describe the relaxation
property by a single relaxation time T1 or T2 , and the
Electronic Antenna
diffusion by a single diffusion coefficient D. Instead, it is
cartridge
necessary to use distribution functions such as f (T2 ), f (T1 ),
and f (D) to characterize the behavior of the samples.21,22
14 ft

In addition, these measurements have been extended to the

determination of multidimensional distribution functions, such
CMR skid as f (D, T2 ) and f (T1 , T2 ).23,24 Such relaxation–diffusion and
T1 − T2 measurements have proved to be particularly useful to
Wear
plate
distinguish different fluids in porous media and to characterize
6 in.
their spatial configuration and chemical composition.
In grossly inhomogeneous fields, the decay time of the
Permanent magnet free induction decay is very short: T2∗ is only of the or-
Sensitive zone
der of the pulse duration. This makes the signal generated
by a single pulse difficult to detect. For practical applica-
tions, it is essential to use multipulse sequences that gen-
Figure 3 Schematic view of the NMR logging tool designed by
Kleinberg et al.17 The permanent magnets are magnetized in the
erate echoes. In particular, the Carr–Purcell–Meiboom–Gill
direction of the arrows. The sonde is pressed against the borehole wall. (CPMG) sequence25,26 and related sequences form the basis
The magnitude of the magnetic field forms a saddle point about 25 mm of most NMR well-logging measurements. The basic CPMG
inside the formation. The resulting sensitive region is shown as an insert. sequence consists of an initial 90◦ excitation pulse followed
The intensity of the color is proportional to the local signal sensitivity by a long train of identical refocusing pulses that are phase
(Courtesy of Schlumberger) shifted by 90◦ relative to the excitation pulse. Echoes form
between the refocusing pulses and they have a width of about
the duration of a refocusing pulse. In order to extract quanti-
tative information from these multipulse measurements, it is
essential to consider the unavoidable off-resonance effects and
to fully account for the spin dynamics across the entire excited
volume.27,28
Auxiliary RF coil Sensitive regions:
Multiple depths of investigation
Permanent magnet (controlled by RF frequency)
Auxiliary RF coil 3.1 T2,eff Relaxation Measurements
Permanent
Main RF coil RF coil The measurements of the overall amplitude and transverse
magnet
relaxation times are based on the CPMG sequence. In
grossly inhomogeneous fields, it is unavoidable that the RF
pulses excite multiple coherence pathways simultaneously.
Borehole
The inference of the different contributions (e.g., direct echoes
and various stimulated echoes) results in a nonmonotonic
Figure 4 Schematic view of the NMR wireline logging device and oscillatory dependence of the echo amplitudes. Even
described in Ref. 18. The magnet (shown in blue) generates a gradient with a sample characterized by a single relaxation time, the
field in the formation. The main RF coil can transmit and receive decay of the CPMG echo amplitudes is not exponential as
at different radiofrequencies, which correspond to different sensitive expected for on-resonance conditions. In general, the form
regions as shown by the differently colored surfaces. In addition, there of the decay depends on the details of the inhomogeneities.
are two auxiliary RF coils that are operated at a higher radiofrequency
This complicates the extraction of relaxation information from
and are associated with a shorter sensitive region. The sensor is pushed
again the wall of the borehole by a spring (Courtesy of Schlumberger)
CPMG data acquired in inhomogeneous fields. Fortunately,
in grossly inhomogeneous fields encountered in logging tools
when ΔB0  B1 , the situation simplifies.27,28 Only the first
3 MEASUREMENT TECHNIQUE two or three echoes show a pronounced transient behavior.
For higher echoes, the response follows an asymptotic regime
The static B0 and RF B1 fields that NMR logging where the echoes exhibit a stable shape and the amplitudes
tools generate across the earth formation are necessarily decay with an effective relaxation rate 1/T2,eff that is a
nonuniform. In such grossly inhomogeneous fields where weighted sum of the 1/T1 and 1/T2 relaxation rates:
ΔB0  B1 , all RF pulses act as slice-selective pulses. Thus, 1 1 1
NMR logging can be considered a particular application of = (1 − α) +α (2)
the stray field technique20 (see Stray-Field (STRAFI) NMR: T2,eff T2 T1
Imaging in Large Field Gradients). The standard modalities The parameter α is a tool-specific parameter and is typically
of pulsed NMR measurements, namely spectroscopy and in the range of 0.09 to 0.12.28 Relative to the asymptotic
magnetic resonance imaging (MRI), are difficult to implement. behavior, the amplitude of the first echo is systematically

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
6 WELL LOGGING

low by around 20% and the second echo is high by a few Diffusion can be measured in the intrinsic inhomogeneities
percent.27,28 The exact size of the initial transient effect of the static field of the logging tool and it does not require
is controlled by the B0 and B1 inhomogeneities and is the added complexity of pulsed-field gradients. The first mea-
independent of the properties of the sample. It can therefore surements were based on the dependence of the measured
be easily calibrated and accounted for. T2,eff relaxation time on the echo spacing tE . For on-resonance
In many cases, earth formations show a wide distribution of CPMG measurements, it is well known that
relaxation times that can span from submillisecond to several 1 1 1
seconds. In order to capture the full relaxation decay, it is = + γ 2 g 2 DtE2 (4)
necessary to use an echo spacing, tE , that is at least as short as T2,meas T2 12
the shortest relaxation time. At the same time, it is necessary where D is the diffusion coefficient and g is the local gradient.
to acquire a large number of echoes in order to characterize Over the diffusion length, the local inhomogeneities of the
the longest relaxation times. The use of short echo spacings static field can be well approximated by a gradient g. In
has the added benefit to increase the effective signal-to-noise grossly inhomogeneous fields, this expression is only a rough
ratio and to minimize the diffusion effect on the measured approximation and underestimates the diffusion-induced decay
relaxation time. In current logging operations, it is common rate as additional coherence pathways contribute to the signal
to acquire several thousands of echoes with echo spacings as decay.33
low as 200 μs. A more quantitative technique to measure diffusion in
After correcting for the transient of the first two echo grossly inhomogeneous fields is based on a two-dimensional
amplitudes, the measured echo amplitudes Mn of the CPMG measurement technique where only the first few echo spac-
sequence can be expressed in terms of the distribution of ings in a CPMG sequence, tE,1 , are increased systematically.23
relaxation times, f (T2,eff ), by This approach selects the desired coherence pathways during

the diffusion encoding part with well-defined diffusion sen-
Mn = dT2,eff f (T2,eff ) e−ntE /T2,eff (3) sitivities. This allows the extraction of diffusion–relaxation
distribution functions f (D, T2,eff ). In standard logging modes,
The distribution function f (T2,eff ) is one of the main it is advantageous to encode diffusion with a two-echo se-
quantities of interestas it is related to the pore size distribution. quence so that motion effects are eliminated to first order. In
The integral M0 ≡ dT2,eff f (T2,eff ) corresponds to the initial that case, after correcting the first two echo amplitudes for
amplitude and is directly related to the overall spin density transient effects, the measured echo amplitudes M(tE,1 , ntE )
and porosity. are related to f (D, T2,eff ) by
The extraction of the distribution f (T2,eff ) from the  
experimental data Mn is an ill-conditioned problem and M(tE,1 , ntE ) = dD dT2,eff f (D, T2,eff )kD (tE,1 )kT2 (ntE )
requires care. Even with very high signal-to-noise ratios,
there are many different solutions of f (T2,eff ) that fit the (5)
experimental data within the experimental uncertainties. A  
common solution is to include a regularization term in the where the relaxation kernel kT2 (nt E ) = exp −nt E /T2,eff and
cost function of the inversion.29 This approach extracts a the diffusion kernel kD has contributions  from the direct

and the stimulated echo: kD (tE,1 ) = α exp −γ 2 g 2 DtE,1
2
/6 +
representative solution that has the fewest features among all  2 2 2 
solutions that fit the problem. The reader is urged to consult β exp −γ g DtE,1 /3 . Here α and β are tool-specific co-
the literature for further details.30 – 32 efficients that depend on the field inhomogeneities and de-
scribe the fractions of signal that are excited into the two
coherence pathways of interest during the diffusion encod-
3.2 T1 Relaxation Measurements ing time. The simple form of equation (5) with the separable
kernel for diffusion and relaxation makes it possible to in-
T1 measurements can be performed by adopting vert the two-dimensional data M(tE,1 , ntE ) and extract the
saturation—recovery or inversion—recovery experiments. diffusion–relaxation distribution function f (D, T2,eff )23 using
A typical implementation consists of CPMG sequences an efficient inversion algorithm.30
that are run consecutively with a variable delay time. In In current state-of-the-art logging operations, the
inhomogeneous fields, it is important to take into account diffusion–relaxation measurements are further combined
that the saturation or inversion of the magnetization is with T1 -encoding sequences, which enables the extraction
nonuniform.33 It is beneficial to use a full CPMG train for of multidimensional measurements of f (D, T1 , T2,eff ). In
detection. This allows an enhancement of the signal-to-noise addition, measurements acquired at different RF frequencies
ratio and enables the extraction of the T1 − T2,eff distribution are multiplexed.35 This technique is analogous to multislice
function.23,24 T1 measurements are not affected by diffusion measurements in MRI and enables spatial profiling of the
and are therefore particularly useful for the characterization NMR properties away from the borehole.
of components with long relaxation times.

4 APPLICATIONS AND INTERPRETATION

3.3 Diffusion and Multidimensional Diffusion–Relaxation
Measurements In basic wireline NMR logging operations, the sensor
continuously performs CPMG measurements while being
Akkurt et al.34 first demonstrated that diffusion measure- moved along the borehole. At each depth, the echo amplitudes
ments can be used to clearly identify gas in the formation. are processed and the relaxation time distribution function

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
 WELL LOGGING 7

f (T2,eff ) is extracted. The basic “NMR log” consists of a Depending on the fluid composition, the average relaxation
display of the resulting distribution functions as a function time can vary widely between a few milliseconds and several
of depth and it describes the NMR relaxation properties of seconds. Furthermore, hydrocarbons can contain molecules
the earth formation. In order to extract the full value from with a wide distribution of molecular size. This leads to a
these NMR logs, an additional step is required. The NMR corresponding distribution of relaxation times f (T2,bulk ). It has
information has to be translated and used to infer valuable been shown that it is possible to use the relaxation data to infer
rock and fluid properties. For the customer of NMR logs, the the distribution of molecular sizes in the fluid.38 For brine, the
determination of these properties is the main motivation for bulk relaxation time is a few seconds. This is typically much
NMR logging. longer than the surface relaxation time.
For protons in molecules of the wetting phase, the frequent
collisions of the molecules with the grain surfaces result in the
dominant relaxation mechanism. In the limit of fast diffusion,
4.1 Amplitude: Porosity the rate of this surface relaxation is proportional to the local
ratio of surface area to pore volume, VSp . This ratio is inversely
The initial amplitude of the CPMG decay, or equiva- proportional to the pore size. The proportionality constant ρ is
lently
 the area of relaxation time distribution, M(t → 0) = known as the surface relaxivity and it characterizes the strength
dT2,eff f (T2,eff ), is directly proportional to the number of mo- of relaxation of the rock surfaces. It has been found empirically
bile hydrogen nuclei in the formation. Since the density of that ρ shows a surprisingly small variability among typical
hydrogen nuclei in water and alkanes is almost identical, this reservoir rocks, especially within a given well. It tends to be
amplitude can be used to infer the porosity of the formation, φ. centered in the range of 1–10 μm/s for carbonate rocks and
The porosity is defined as the fraction of the rock volume 4–40 μm/s for sandstones.
that can be occupied by fluid and it is one of the most When relaxation is dominated by surface relaxation, the
important parameters characterizing a hydrocarbon reservoir. relaxation time of a given component is an indicator of the
There are several other logging techniques for measuring size of the pore it resides in. In this case, the distribution of
porosity. They include techniques that measure the attenuation relaxation times f (T2 ) can be interpreted as the distribution
of γ -rays in the formation or the interaction of neutrons with of the relative pore sizes.21 The pore size distribution is a
the formation.1 However, these techniques require information key property of any porous media. The NMR relaxation log
about the mineralogy of the earth formation. In contrast, can quickly reveal changes in the average pore size of the
NMR has the unique capability of detecting directly the fluid earth formation. As an example, the relaxation times in a shale
molecules. interval are much shorter than in a sand interval.
In order to capture the complete porosity by NMR logging, In sufficiently small pores, the capillary pressure required
the CPMG data are generally acquired with the shortest echo to mobilize water by immiscible displacement becomes larger
spacing tE possible. Protons in solid matrix materials such than a few atmospheres. For practical purposes, the water in
as gypsum have a solid-like T2 of the order of 10 μs and these pores can be considered effectively immovable and is
do not contribute to the signal detected in NMR logging. referred to as bound fluid. Water occupying larger pores is
Similarly, protons in bitumen decay faster than the dead time referred to as free fluid. The corresponding fractions φBF and
and are not included in the NMR porosity. It is useful to φFF can be determined from a simple partitioning of the T2
compare the porosity derived from NMR logs with porosity distribution functions
determinations based on nuclear logs. Differences in these  T2,cutoff
values can indicate that an incorrect lithology was assumed φBF = f (T2 )dT2 (7)
in the analysis of the nuclear measurements. Alternatively, it T2,min
can indicate the presence of methane or bitumen with a lower
density of NMR-detectable protons. φFF = φ − φBF (8)
The default value for the cutoff relaxation time T2,cutoff has
been found empirically to be 33 ms for sandstones and 90 ms
4.2 Relaxation: Pore Size Distribution, Viscosity for carbonates.
Another common estimator derived from the relaxation time
The relaxation rate of a fluid molecule inside porous media distribution is the hydraulic permeability of the rock. This
such as a rock can be generally written as a sum of two important quantity is controlled by the geometry of the pore
contributions that are due to bulk relaxation and surface space in the rock, in particular, the size of the pores and their
relaxation: connectivity. For a given porosity and geometry of the pore
1 1 S network, the hydraulic permeability scales with the square
= +ρ (6) of the pore size. It is therefore reasonable to expect that
T2 T2,bulk Vp
the hydraulic permeability is correlated with the square of
Here T2,bulk is the bulk relaxation time, ρ is the surface the average relaxation time. Two common estimators of the
relaxivity, S is the pore surface area, and Vp is the pore hydraulic permeability are of the form21,39,40 :
volume. For the hydrocarbon and nonwetting phase, the first  2
term generally dominates, whereas for brine and the wetting kSDR = a T2,eff φ 4 (9)
phase, the second term typically dominates.
The bulk relaxation time T2,bulk is solely a property of the  2
φFF
fluid filling the pore space. To first order, the average relaxation kTC = bφ 2
(10)
time is inversely proportional to the viscosity of the fluid.36,37 φBF

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
8 WELL LOGGING

Bound fluid
10−8 Brine saturated
Core permeability/ T2,eff distribution /
NMR porosity
md ms

NMR permeability NMR free fluid 10−9

D / m2/s
10 102 103 30% 15% 0 1 10 1001000

10−10
xx450 ft

10−11
10−3 10−2 10−1 100 101
(a) T2,eff / s
xx500 ft

10−8 Drainage

xx550 ft
10−9

xx600 ft D / m2/s 10−10

(a) (b) (c)

Figure 5 Example of an NMR log. (c) Distribution f (T2,eff ) as a 10−11

function of depth. The red line indicates the value of T2,cutoff . (b) NMR 10−3 10−2 10−1 100 101
porosity is shown as a black dashed line, the free fluid porosity φFF (b) T2,eff / s
is shown as a magenta line, and the bound fluid φBF is shown shaded
between the two lines. (a) Permeability estimator kTC is shown as blue
lines, and red dots show results of permeability measurements performed 10−8 Imbibition
in the laboratory on cores. The two most significant digits of the depth
scale have been replaced by xx
10−9
D / m2/s

Here a and b are coefficients that depend on the surface

relaxivity.
Figure 5 shows an example of a basic NMR log based on 10−10
relaxation measurements.

10−11
4.3 Diffusion–Relaxation: Fluid Typing, Pore 10−3 10−2 10−1 100 101
Connectivity (c) T2,eff / s

Figure 6 Measurements of diffusion–relaxation distribution function

From an NMR log based on relaxation measurements alone, f (D, T2,eff ) on a sandstone in three different saturation states: fully brine
it is in general not possible to distinguish the NMR sig- saturated (a), oil saturated after drainage with a crude oil (b), and mixed
nals from the hydrocarbon and brine phase unambiguously. saturation after forced imbibition with brine (c). The horizontal dashed
However, this can be achieved by adding diffusion infor- line indicates the molecular diffusion coefficient of water, whereas the
mation and in particular by measuring the two-dimensional diagonal dashed line shows the hydrocarbon line. Adopted from Ref. 41
diffusion–relaxation distribution function. To demonstrate the
wealth of additional information obtainable from such mea-
surements, Figure 6 shows laboratory measurements of the In fine grained rocks, the effects of restricted diffusion re-
diffusion–relaxation distribution function f (D, T2,eff ) on a duce the measured diffusion coefficients from the molecular
rock sample saturated with different mixtures of brine and diffusion coefficient of the fluid.43 This effect gives informa-
crude oil.41 tion on the configuration of the fluids within the pore space
There is a clear separation of the signals from the two and can be used to give an independent value of pore size. The
phases. The brine signal is located close to the horizontal shape of the diffusion–relaxation distribution function of the
line that indicates the molecular diffusion coefficient of water, hydrocarbon phase can be used to infer detailed information
whereas the hydrocarbon signal lies on a diagonal line, the about the fluid composition including the presence of asphal-
so-called hydrocarbon line.23 The relative intensity of the two tene or dissolved methane, and the detection of emulsions.42,44
contributions can be used to infer the oil saturation in the Diffusion–relaxation time measurements are now being
sample. Furthermore, the relaxation times associated with the performed routinely in commercial logging operations.45 A
brine phase indicate the sizes of the pores occupied by the particularly powerful application is the mapping of the fluids
water phase, whereas for the oil phase, it is directly related to not only as a function of depth along the borehole but also
the composition of the hydrocarbon phase.38,42 as a function of the depth of investigation away from the

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
 WELL LOGGING 9

borehole.35,46 This enables the in situ imaging and monitoring 8. S. Earnshaw, Trans. Camb. Philos. Soc., 1842, 7, 97–112.
of the fluids in the near borehole environment and allows the 9. R. K. Cooper and J. A. Jackson, J. Magn. Reson., 1980, 41,
quantification of invasion of the borehole fluid into the earth 400–405.
formation and its interaction with the formation of fluids. 10. L. J. Burnett and J. A. Jackson, J. Magn. Reson., 1980, 41,
406–410.
11. J. A. Jackson, L. J. Burnett, and J. F. Harmon, J. Magn. Reson.,
1980, 41, 411–421.
5 CONCLUSION
12. J. Horkowitz, S. Crary, K. Ganesan, R. Heidler, B. Luong,
J. Morley, M. Petricola, C. Prusiecki, M. Poitzsch, J. R. Scheibal,
The progress in the development of NMR well logging and M. Hashem, In Transactions SPWLA 43th Annual Logging
over the last 20 years has been remarkable. Key contributions Symposium, Paper EEE, Oiso, Japan 2002.
include the design of several distinct sensors for pulsed NMR 13. M. Borghi, F. Porrera, A. Lyne, Th. Kruspe, V. Krueger, H. Thern,
applications that exhibit different field profiles and that contain and R. Chemali, In Transactions SPWLA 46th Annual Logging
ever more sophisticated electronics. In parallel to the advances Symposium, Paper HHH, New Orleans, USA 2005.
in hardware, new measurement concepts have been introduced 14. S. Shtrikman and Z. Taicher, U.S. Pat. 4,710,713, (1987).
that allow for the first time the systematic separation of the 15. M. N. Miller, Z. Paltiel, M. E. Gillen, J. Granot, and J. C. Bouton,
NMR signal into the contributions from the different fluid In SPE Annual Technical Conference and Exhibition, Paper SPE
phases. These measurements give in situ information on the 20561, New Orleans, USA 1990.
fluid properties and they also serve as a probe of the geometric 16. M. G. Prammer, G. D. Goodman, S. K. Menger, M. Morys, and
arrangements of the different fluids within the pore space. S. Zannoni, In SPWLA 41th Annual Logging Symposium, Paper
As these techniques are further developed and the underlying EEE, Dallas, USA 2000.
physics of the relaxation and diffusion properties is understood 17. R. L. Kleinberg, A. Sezginer, D. D. Griffin, and M. Fukuhara,
in more details, it will be possible to address completely new J. Magn. Reson., 1992, 97, 466–485.
and more detailed questions about the fluids and their flow 18. L. DePavia, N. Heaton, D. Ayers, R. Freedman, R. Harris,
properties within the porous media. B. Jorion, J. Kovats, B. Luong, N. Rajan, R. Taherian, K. Walter,
Finally, the techniques and technologies that have been and D. Willis, In SPE Annual Technical Conference and Exhibition,
developed for well logging are rather general and can be Paper SPE 84482, Denver, USA 2003.
adapted to the study of other samples that cannot be easily 19. S. Chen, E. Beard, M. Gillen, S. Fang, and G. Zhang, In SPWLA
brought to the laboratory for examination. The successful 44th Annual Logging Symposium, Paper ZZ, Galveston, USA
2003.
implementation of inside-out NMR measurements many
kilometers underground at elevated temperatures and pressures 20. P. J. McDonald, Prog. Nucl. Magn. Reson. Spectrosc., 1997, 30,
69–99.
demonstrates that it is completely realistic to apply similar
techniques to applications in a range of other fields such as 21. W. E. Kenyon, P. I. Day, C. Straley, and J. F. Willemsen, SPE
Form. Eval., 1988, 3, 622–636, Paper SPE 15643-PA, Erratum:
process control, material science, food science, biology, and Soc. Petrol. Eng. Form. Eval., 1989, 4, 8.
medicine.4
22. K. P. Whittall and A. L. MacKay, J. Magn. Reson., 1989, 84,
134–152.
23. M. D. Hürlimann and L. Venkataramanan, J. Magn. Reson., 2002,
6 RELATED ARTICLES 157, 31–42.
24. Y.-Q. Song, L. Venkatarmanan, M. D. Hürlimann, M. Flaum,
Diffusion and Flow in Fluids; Stray-Field (STRAFI) P. Frulla, and C. Straley, J. Magn. Reson., 2002, 154, 261–268.
NMR: Imaging in Large Field Gradients; Brownian Motion 25. H. Y. Carr and E. M. Purcell, Phys. Rev., 1954, 94, 630–638.
and Correlation Times; Oil Reservoir Rocks Examined by 26. S. Meiboom and D. Gill, Rev. Sci. Instrum., 1958, 29, 688–691.
MRI; Porous Media Studied by MRI; Relaxometry: Two- 27. G. Goelman and M. G. Prammer, J. Magn. Reson., Ser. A, 1995,
Dimensional Methods; Terrestrial Magnetic Field NMR. 113, 11–18.
28. M. D. Hürlimann and D. D. Griffin, J. Magn. Reson., 2000, 143,
120–135.
7 REFERENCES 29. E. J. Fordham, A. Sezginer, and L. D. Hall, J. Magn. Reson., Ser.
A, 1995, 113, 139–150.
1. D. V. Ellis and J. M. Singer, ‘Well Logging for Earth Scientists’, 30. L. Venkataramanan, Y.-Q. Song, and M. D. Hürlimann, IEEE
Springer, Dordrecht, The Netherlands 2007. Trans. Signal Process., 2002, 50, 1017–1026.
2. R. J. S. Brown, R. Chandler, J. A. Jackson, R. L. Kleinberg, M. N. 31. C. L. Epstein and J. Schotland, SIAM Rev., 2008, 50, 504–520.
Miller, Z. Paltiel, and M. Prammer, Concepts Magn. Reson., 2001, 32. M. Prange and Y. Q. Song, J. Magn. Reson., 2010, 204, 118–123.
13(6), 335–411. 33. M. D. Hürlimann, J. Magn. Reson., 2001, 148, 367–378.
3. B. Blümich, J. Perlo, and F. Casanova, Prog. Nucl. Magn. Reson. 34. R. Akkurt, H. J. Vinegar, P. N. Tutunjian, and A. J. Guillory,
Spectrosc., 2008, 52, 197–269. In Transactions SPWLA 36th Annual Logging Symposium, Paris,
4. F. Casanova, J. Perlo, and B. Blümich, eds, ‘Single-Sided NMR’, Paper N, 1995.
Springer, 2010. 35. N. Heaton, H. N. Bachman, C. C. Minh, E. Decoster, J. LaVigne,
5. P. Fantazzini, V. Bortolotti, J. Kärger, and P. Galvosas, eds, In AIP J. White, and R. Carmona, Petrophysics, 2008, 2, 172–186.
Conference Proceedings 1330, Leipzig, Germany 2011. 36. C. E. Morriss, R. Freedman, C. Straley, M. Johnston, H. J. Vinegar,
6. R. J. S. Brown and B. W. Gamson, Trans. SPE AIME , 1960, 219, and P. N. Tutunjian, Log Anal., 1997, 38/2, 44–59.
201–209. 37. R. L. Kleinberg and H. J. Vinegar, Log Anal., 1996, 37/6, 20–32.
7. D. I. Hoult and R. E. Richards, J. Magn. Reson., 1976, 24, 71–85. 38. D. E. Freed, J. Chem. Phys., 2007, 126, 174502-1–174502-10.

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2
10 WELL LOGGING

39. A. Timur, J. Petrol. Technol., 1969, 21, 775–786. 43. P. N. Sen, Concepts Magn. Reson., 2004, 23A, 1–21.
40. C. R. Coates, D. Marschall, and D. Mardon, Log Anal., 1998, 39/1, 44. M. D. Hürlimann, D. E. Freed, L. J. Zielinski, Y.-Q. Song, G.
51–63. Leu, C. Straley, C. C. Minh, and A. Boyd, Petrophysics, 2009, 50,
41. M. D. Hürlimann, M. Flaum, L. Venkataramanan, C. Flaum, 116–129.
R. Freedman, and G. J. Hirasaki, Magn. Reson. Imaging, 2003, 45. R. Freedman and N. Heaton, Petrophysics, 2004, 45, 241–250.
21, 305–310. 46. J. White and M. Samir, In Transactions SPWLA 49th Annual
42. A. R. Mutina and M. D. Hürlimann, J. Phys. Chem., 2008, A 112, Logging Symposium, Paper X, Edinburgh, Scotland 2008.
3291–3301.

Encyclopedia of Magnetic Resonance, Online © 2007–2012 John Wiley & Sons, Ltd.
This article is © 2012 John Wiley & Sons, Ltd.
This article is published in the Encyclopedia of Magnetic Resonance in 2012 by John Wiley & Sons, Ltd.
DOI: 10.1002/9780470034590.emrstm0593.pub2