NMR petrophysics

This page provides an overview of the mathematical principles behind the

petrophysical aspects of nuclear magnetic resonance (NMR) logging, including
porosity, saturation, hydrocarbon identification, facies prediction, and permeability.

Laboratory studies
Extensive laboratory studies on NMR behavior and on the properties of fluid-
saturated porous media have been conducted since the inception of NMR and
throughout the development of NMR-logging tools. The results from these
investigations have provided the petrophysical foundation for understanding the
logging measurements and for developing interpretation models and applications.
Low-field, bench-top pulse-NMR spectrometers were developed concurrently with
logging tools so that wellbore measurements could be duplicated on core samples
in the laboratory.[1][2] These instruments operate and record data in the same
manner as NMR-logging tools.[3] Because NMR analysis is nondestructive, NMR
and conventional capillary-pressure measurements can be performed on the same
samples, in both the saturated and partially saturated states. Low-field
spectrometers provide the ability to make repeatable measurements of rock- and
fluid-NMR properties. This ability, in turn, permitted correlation and calibration of
laboratory and field measurements and also permitted direct transfer of
interpretation models developed in the laboratory to logging data. Where core is
unavailable for NMR-log calibration, new technology and methods now allow NMR
petrophysical measurements on drill cuttings.[4]
Laboratory NMR studies are routinely conducted for the following purposes:
Verifying formation porosity

Evaluating textural effects, such as microporosity, on NMR-log response

Determining formation-specific models that enhance the accuracy of

determining bulk-volume-irreducible (BVI) water, free-fluid index (FFI), and,
ultimately, permeability
Developing models to identify and quantify hydrocarbons, including residual
oil
Developing models to predict changes in pore size (facies)

Much of this work is summarized in Kenyon,[5] Murphy,[6] Woessner,[7] and Dunn et

al.[8] The most recent laboratory studies have suggested that some established
NMR core-log relationships should be further investigated to better account for
data scatter.[9] A related area of study not dealt with here is NMR imaging of fluid
flow in core.[10]
As in NMR logging, data quality is critical. To achieve the desired objectives,
laboratory NMR studies should include a preplanning phase similar to that used in
logging. (See Job planning for nuclear magnetic resonance (NMR) logging).

Petrophysical properties
The basic petrophysical parameters of porosity, permeability, and producibility can
be determined from either T1 or T2 echo-decay data. Until low-field spectrometers
were developed, T1 was the preferred acquisition method in the laboratory, where
time is not a concern.[3][5] Because T1 measurement requires more time
than T2, T2 became the primary acquisition mode in pulse-NMR logging because it
allowed logging at speeds that were commercially viable. Fortunately, there is a
correlation between T1 and T2[11]; and T1 can be estimated from T2 data by
extrapolating the T2 decay-obtained-polarization pulses of different lengths (i.e.,
using different values of TW; see Fig. 1).

Fig. 1 T1 measurement can be determined from a series of multiple wait-

time T2 measurements. The peak amplitudes from T2 the measurements trace
the outline of the curve (dotted).
Modern logging tools are capable of operating in either T1- or T2-acquisition modes.
The logging mode is dictated by operational factors and job objectives and may, in
fact, switch back and forth, as needed. (See Logging while drilling (LWD)).
NMR properties of fluids
T1 relaxation occurs when the precessing proton system transfers energy to its
surroundings. T2 relaxation occurs through a similar transfer in energy and also
through dephasing. Consequently, transverse relaxation, T2, is always faster than
longitudinal relaxation, T1. The emphasis of the proton NMR techniques used in
formation evaluation is on the NMR fluid response from the pore fluids
where T2 T1. The NMR response in solids (i.e., the non-shale/clay component of
the rock matrix) is very short compared to the pore-fluid signal, and is generally not
measured by laboratory or logging devices.
NMR relaxation of fluids depends on whether the fluid is measured in bulk form as
a wetting-pore fluid within a rock matrix or in a gradient magnetic field. Bulk
relaxation is the intrinsic relaxation property of a fluid that is controlled by viscosity,
chemical composition, temperature, and pressure: T2bulkT1bulk. Fluids contained
within pores have different relaxation characteristics, namely those of surface
relaxation.
Surface relaxation occurs at the fluid-solid interface between a wetting-pore fluid
and the rock-pore walls (Fig. 2) and is different from the relaxation in either the
solid or the fluid, individually. Surface relaxation dramatically decreases
both T2 and T1 and is the dominant component contributing to T2. When a
nonwetting fluid (e.g., oil) is also present in rock pores, the nonwetting fluid may
continue to relax at its bulk-relaxation rate.

Fig. 2 Schematic illustrating surface relaxation.

Surface relaxation is expressed by the following equations (Eqs.1 and 2):

....................(1)
and

....................(2)
where 2 = T2 surface relaxivity (i.e., T2 relaxing strength of the grain
surfaces); 1 = T1 surface relaxivity (i.e., T1 relaxing strength of the grain surfaces);
and (S/V) pore=ratio of pore surface to fluid volume.
Surface relaxivity varies with mineralogy; and, for simple pore shapes (S is the
surface area of a pore and V is the volume of the same pore), S/V is a measure of
pore size. In a brine-wet rock, T2 in smaller pores will be less than T2 in large
pores; consequently, identical pore water in different rocks can have a wide range
of relaxation times because of variations in surface relaxivity. Laboratory studies
have demonstrated that in water-wet rocks, the surface and volume ratio (S/V) is
also a measure of permeability.
Fluids controlled by surface relaxation exhibit T2 values that are not dependent on
temperature and pressure. For this reason, laboratory NMR measurements made
at room conditions are commonly used to calibrate formulas used to estimate
petrophysical parameters such as permeability and bound water.[12][13]
Diffusion-induced relaxation occurs when a significant gradient exists in the static
magnetic field. Molecular diffusion in this gradient causes additional dephasing that
contributes to increased T2 relaxation. In addition to the magnetic-field gradient,
diffusion is also controlled by inter-echo spacing and fluid diffusivity, viscosity,
molecular composition, temperature, and pressure.
Bulk-fluid processes and surface relaxation affect both T1 and T2, while diffusion
only affects T2 relaxation. All three processes are independent and act in parallel
according to the following equations (Eqs.3 and 4):

....................(3)
and

....................(4)
The relative importance of the three diffusion-relaxation mechanisms depends on
(see Table 1):
Fluid type in the pores (e.g., water, oil, or gas)

Sizes of the pores

Strength of the surface relaxation

Wettability of the rock surface



Table 1
T2 decay
Eq.5 states that the T2 decay associated with a single pore size in water-saturated
rocks is proportional to the pore size.[5] In fact, because reservoir rocks typically
comprise a distribution of pore sizes and frequently contain more than one fluid
type, a CMPG T2spin-echo train actually consists of a distribution of T2 decays,
rather than a single T2 decay. In these cases, the exponential decay is described
by Kenyon et al.[14] as follows:

....................(6)
where M(t) = measured magnetization at t; Mi(0) = initial magnetization from the ith
component of relaxation; and T2i = decay constant of the i th component of
transverse relaxation. The summation is over the entire sample (i.e., all pores and
all different types of fluid).
Fig. 3 illustrates the multiexponential decay character of a porous medium
containing pores of different sizes and a single wetting phase. Surface relaxation
dominates when a short inter-echo spacing is used and the formation is only brine
saturated. Under this condition, T2 is directly proportional to pore size. When all
pores are assumed to have similar geometric shape, the largest pores (see Fig. 3,
left column) have the lowest S/V and, thus, the longest T2. Medium-size pores have
smaller S/V, yielding shorter T2values. The smallest pores have the highest S/V
and the shortest T2 values.


Fig. 3 A 100% water-saturated pore (upper left) has a single T2 value (upper
center) that depends on pore size and, thus, its spin-echo train exhibits a
single exponential decay (upper right) that also depends on pore size.
Multiple pores at 100% water saturation (bottom left) have multiple T2 values
(bottom center) that depend on the pore sizes and, thus, their composite
spin-echo train exhibits multiexponential decay (bottom right) that also
depends on the pore sizes.
[15]
Eq.6 can also be expressed as follows:

....................(7)
where (S/V)t is the surface-to-volume ratio of the ith pore. When t = 0, the following
is true:

....................(8)
If the measured magnetization for 100% bulk water, M100%(0), is known, then M(0)
and M0i can be calibrated to porosity by the following equation:

....................(9)
where = calibrated porosity of the formation, i = calibrated porosity associated
with all pores of the ith pore size (also known as the incremental porosity).
Therefore, the T2 distribution (in the form of the amplitudes, M0i, associated with the
time constants, T2i, is calibrated to the porosity distribution (i.e., the individual
pores i with the associated time constants T2i).
If the rock is water-wet and the pores are partially saturated (i.e., the pores contain
oil and/or gas in addition to water), then the total signal comprises contributions
from each component in the following equation:
 ....................(10)
where Moil = magnetization produced by oil protons in the pores, Mgas =
magnetization produced by gas protons in the pores, T2oil = T2 of oil measured with
a CMPG sequence, and T2gas = T2 of gas measured with a CMPG sequence.
In a brine-saturated rock, the T2 decay spectrum represents a pore-size
distribution. However, when nonwetting fluids (e.g., oil or gas) are present,
the T2 spectrum includes a bulk response from the nonwetting fluid, in addition to
the pore-size response. Pores containing the nonwetting fluid either appear in the
spectrum at a decay time that is faster than is normally associated with the pores,
or do not appear at all if the surface layer is too thin. This behavior affects the
appearance of the T2 spectrum and associated T2 distribution but not the total
signal amplitude (i.e., porosity).
Data fit-inversion
The raw data recorded by an NMR device are a series of spin-echo amplitudes
(echo train) as a function of time, usually at fixed time increments (bins). These
NMR measurements are statistical, and stacking is required to improve the
precision of the log outputs. The precision depends on formation-fluid properties,
activation (acquisition) type, and the number of pulses stacked. [16]The data are
mathematically inverted (mapped) by use of a best-fit curve to produce a
distribution of T2 values as a function of relaxation time (Figs. 4 and 5). Initially,
biexponential-fitting algorithms were used[17]; however, T2 decay in fluid-saturated
rocks is multiexponential, because of this, multiexponential models were developed
and are commonly used for inverting the data. [8] The inverse solution
(T2 distribution) is a function of both the measured echo data and the chosen
smoothness (regularization) for the inversion. However, because regularization is
controlled in part by the S/N, the fit that is actually used is not unique; that is, there
can be a number of differently shaped T2 distributions that fit the original echo-
decay curve. In general, the area under the T2-distribution curve (porosity) and the
general location in time of the high-porosity bins are robust; however, caution is
advised when interpreting the fine details of the distribution. [18] Because the spin
echoes are measured over a very short time, an NMR tool travels no more than a
few inches along the wellbore while recording the spin-echo train; thus, the
recorded spin-echo data can be displayed on a log as a function of depth.


Fig. 4 Spin-echo-decay train. The amplitudes of the spin echoes are

recorded over time. A best fit line is applied to the data before inversion. In
this example, the raw spin-echo measurements recorded at TE = 1 ms are
shown together with a data-fitting curve (solid).

Fig. 5 - T2 distribution. Mathematical inversion is used to convert the spin-

echo-decay data to a T2 distribution. This distribution is the most likely
distribution of T2 values that produce the echo train. With proper calibration,
the area under the T2-distribution curve is equal to the porosity. When the
rock is water-saturated, this distribution will correlate with pore-size
distribution. The presence of hydrocarbons will affect the T2 distribution
depending on the hydrocarbon type, viscosity, and saturation.
T2 distribution
The mathematical statement (Eq.9 and Eq.10) that T2 distribution observed in
water-saturated rock represents the pore-size distribution and porosity of the rock
has been confirmed using mercury-injection capillary pressure (MICP) methods
(see Figs. 6 and 7).[19][20]
NMR responds to pore-body size, and MICP responds to pore-throat size. [19][21][22]
[23]
In clastic rocks in which there is a good correlation between pore-body and
pore-throat size, there is often good qualitative agreement between NMR and
MICP data, as Figs. 6 and 7 illustrate. The NMR porosity, reflected in
the T2 distribution, is a spectrum comprising rock matrix and fluid components
(see Fig. 8).

Fig. 6 The correlation between MICP and T2 distribution in water-saturated

sandstones.

Fig. 7 The correlation between MICP and T2 distribution in water-saturated

carbonates.

Fig. 8 NMR-porosity model. These figures illustrate the distribution of pore
size and pore fluids used in NMR logging. The suggested NMR mnemonics
are Halliburtons. Because NMR tools respond to the invaded zone, a mud-
filtrate component is added between movable water and oil. NMR-logging
tools are sensitive to fluids but not to matrix materials and dry clay. NMR
porosity is affected by HI, TW, and TE.
Although matrix minerals and dry clay may contain hydrogen atoms in the form of
hydroxyl groups, the T1 relaxation times of these nuclei are too long to be polarized
by a moving NMR-logging tool, and their T2 relaxation times are too short to be
recorded.[24]The fluid component is subdivided into bound and free
subcomponents. The hydrogen nuclei of clay-bound water are adsorbed on the
surfaces of clay grains. These hydrogen protons can be polarized by NMR-logging
tools and recorded when a sufficiently short TE is used.[25] Similarly, hydrogen
protons in capillary-bound water and movable fluids (e.g., free water, mud filtrates,
oil, and gas), are polarized and recorded by NMR-logging tools with appropriate
values for TE and TW.
The porosity and pore-size information from NMR measurements can be used to
estimate producible porosity (i.e., the movable fluids) and permeability and for
hydrocarbon identification.
Nomenclature

M = magnetization, gauss/cm3

M0 = macroscopic magnetization, gauss/cm3

M0i = magnitude of the initial magnetization from the ith component,

gauss/cm3

M100% = magnitude of the magnetization for 100% bulk water, gauss/cm 3

Mi(0) = magnitude of the initial magnetization from the ith component of

relaxation gauss/cm3

M(0) = magnitude of the initial magnetization, gauss/cm 3

M(t) = measured magnetization at time t, gauss/cm 3

(S/V)i = ratio of pore surface (S) to fluid volume (V), of the ith pore, 1/cm

(S/V)pore = ratio of pore surface (S) to fluid volume (V), 1/cm

t = time, seconds
T1 = longitudinal relaxation time, seconds

T1bulk = pore-fluid bulk-T1 relaxation time, seconds

T1surface = pore-surface T1 relaxation time, seconds

T2 = transverse relaxation time, seconds

T2bulk = pore-fluid bulk-T2 relaxation time, seconds

T2diffusion = pore-fluid T2 relaxation time in a magnetic field gradient, seconds

T2i = pore-fluid surface T2 relaxation time of the ith component,

seconds

T2surface = pore-fluid surface T2 relaxation time, seconds

TE = CMPG interecho spacing, seconds

TW = polarization (wait) time, seconds

1 = T1 surface relaxivity, cm/sec

2 = T2 surface relaxivity, cm/sec

i = calibrated porosity associated with all pores of the i th pore size

Moil = magnetization produced by oil protons in the pores

Mgas = magnetization produced by gas protons in the pores

T2oil = T2 of oil measured with a CMPG sequence

T2gas = T2 of gas measured with a CMPG sequence