SPWLA 42 "d Annual Logging Symposium, June 17-20, 2001

THE D O W N H O L E N M R FLUID A N A L Y Z E R
M . G . P r a m m e r , J. B o u t o n and P. M a s a k
NUMAR, Halliburton

ABSTRACT
A recently developed nuclear magnetic resonance analyzer provides in-situ data of formation fluids at true
reservoir conditions. The compact device is compatible
with the wireline Reservoir Description Tool (RDT TM)I,
which is configurable for a variety of sampling, testing
and monitoring purposes.
The measured relaxation times TI and T2 and the
seW-diffusion coefficient D tie directly into important
fluid characteristics such as gas/oil ratio and viscosity.
Knowledge of these parameters is vital to the understanding of a reservoir, in particular when this information is available in real time during the wireline run.
Multiple test points can be taken for downhole analysis
without the actual collection and uphole transfer of a
sample. The q~mlity of samples destined for lab analysis
can be improved by avoiding contamination with filtrate (OBM or synthetic muds).
The ~
chamber is of the flow-through type. It is
not necessary to divert or pinch off a sample for the basic ~
analysis. Operating at 4 MHz, the system prorides near-laboratory signal-to-noise ratios and operating speeds. At the same time, the resonance frequency
is still close enough to that of wireline (MRIL) 1 and
LWD tools (MRIL-WDrM) 1 to provide exact hydrogen
index values for porosity correction and to differentiate
between filtrate and connate fluids in the measurement
space of these logging tools.
INTRODUCTION

Open-hole wireline sampling is the standard way to

collect representative reservoir fluids and to determine
their PVT properties in the laboratory. Problems encountered are difficulties in obtaining uncontaminated
samples, fluid phase changes during retrieval and transport, difficulties in recreating reservoir conditions, and
in general, the significant time delay associated with the
laboratory analysis.
Ideally, this process would be preceded or even replaced by downhole fluid analysis of as many samples
as desired, with the final results instantaneously available at the well site. Obviously, fluid viscosity would
l RDT. MRIL and MRIL-WD are marks of Halliburton.

play a central role in identifying fluids and determining

hydrocarbon properties.
To understand why NMR is important in this scenario, we recall Stokes' equation:
D QckT/rl,

(k=l.38x10 -23 J/K)

(1)

i.e., the self-diffusion coefficient D is inversely proportional to viscosity 11. Viscosity and diffusivity are both
related to the translational motion of molecules and
therefore must be interrelated. At higher temperatures
T, a molecule contains more energy and can move
faster against a given "friction" 11; therefore D oc T. D is
a property that can be precisely determined by NMR
without disturbing or altering the fluid. The relationship
D oc T/rl has been verified over a wide range of viscosities at different temperatures and pressures by
NMR spin-echo experiments (Abragam, 1961; Jonas,
1991, Appel et al., 2000; Lo et al., 2000).
One has to be more careful with relationships involving the NMR relaxation times T1 and T2. The applicability of expressions of the form,
T1, T2 oc kT/rl,

(not universal)

(2)

is more limited than that of Eq. 1. The main reason is

that gas/liquid mixtures such as live oils relax by more
than one relaxation mechanism: dipole-dipole for the
liquid phase and mainly spin-rotation for the gas phase.
In combination, however, NMR relaxometry (measuring Tl and T2) and NMR diffusometry (determining D)
are powerful tools to characterize live oils.
The study of NMR relaxation times with respect to
oil properties began shortly after NMR was first demonstrated (Bloembergen et al., 1948; Brown, 1961).
The practical aspects of how to relate T1, T2 and D to
petrophysical fluid properties such as viscosity and the
gas/oil ratio have attracted interest much more recently
(Kleinberg et al., 1996; Lo et al., 1998; Zhang et al.,
1998; Appel et al., 2000; Lo et al., 2000). These investigations are significant because NMR relaxometry can
be performed at lower field strengths and with much
lower field homogeneity requirements than NMR spectroscopy (1,000 ppm v. <1 ppm). The latter seems to
be out of reach for downhole applications, but the de-

SPWLA 42 "d Annual Logging Symposium, June 17-20, 2001

velopment of a robust and accurate NMR relaxometry/diffusometry system for downhole use proved to be
feasible.
Another key factor was the development of the Reservoir Description Tool (Proett et al., 1999), which is a
modular wireline sampling and testing system that can
readily accept add-on analysis devices. The inclusion
of an NMR fluid analyzer in the RDT system has the
following benefits:
The level of fluid contamination by oil-based muds
is continuously assessed by observing the line shape of
the fluid's T~ distribution.
Estimates for fluid viscosity and gas/oil ratio
(GOR) can be obtained essentially in real time. Viscosity is derived from either T~ or D, and GOR from a
combination of both parameters.
s The NMR analysis takes place at true reservoir
conditions, removing ambiguities associated with sampiing and transport procedure.
s The shared measurement principles behind wireline
NMR tools (e.g., Prammer et al., 1998), the loggingwhile-drilling version (Prammer et al., 2000) and the
downhole NMR fluid analyzer encourage the development of integrated formation evaluation methods (see
also Freedman et al., 2000).

CONSTRUCTION

Figure 1 shows a schematic diagram of the device.

Pumped fluids flow through the device from top to
bottom while measurements are performed on them.
The sensor section is about 1 m long. This dimension is
dictated by the requirement for sufficient polarization
time at all possible pump-out rates. The fluid enters the
system from the top and is initially subjected to a magnetic field of about 2,000 gauss (0.2T) to achieve rapid
polarization of the hydrogen nuclei. The NMR measurement takes place in the lower section, where the
field strength is lowered to 1,000 gauss. The hydrogen
(proton) resonance frequency is approximately 4.2
MHz at this point. Two separate radio frequency coils
are used for pulse transmission and for reception. This
split scheme allows for a transmitter coil that is longer
than the receiver. By pulsing a larger volume of fluid
and by receiving only from the bottom portion, relaxation times can be determined without regard for the actual flow rate.
The magnetic field in the measurement volume is
not entirely uniform. As shown in Figure 2, the volume
is split into an interior region, where the field gradient
is negligible, and a fringe region, where the field
changes with an approximately uniform gradient. The
fringe region comprises about one-third of the total sen-

sitive volume. During T1 measurements and at short

pulse-to-pulse spacings (0.25 ms), the effect of the gradient is not noticeable. To perform a diffusion measurement, the main fluid flow is diverted and a sample is
stagnated within the NMR chamber. Furthermore, the
pulse-to-pulse spacing (To) is increased to induce diffusion-dependent signal dephasing. The uniform and the
fringe regions are large compared to the largest possible
diffusion length, therefore, these regions are essentially
isolated from each other for the duration of a single
pulse-echo train.
The electronics is similar to that of an NMR spectrometer and is shown in Figure 3. The comparatively
low frequency of 4.2 MHz allows many traditionally
analog functions to be realized as digital signal processing (DSP) algorithms. A frequency source, controlled by a pulse programmer, sends its signal to a
power amplifier, which in turn drives the transmitter
antenna. On the receive side, the signal from the receiver antenna is amplified, synchronously demodulated and filtered. All pertinent calibration factors are
stored in non-volatile memory.
OPERATION
Relaxation time measurements. T1 relaxation times
and hydrogen density can be measured continuously
whether or not the fluid is stagnant or flowing. This
mode is most useful during the pump-out period to assess the amount of mud filtrate contamination. Figure 4
illustrates the pulse sequence employed. This is a standard saturation-recovery sequence, where an initial
saauation pulse is followed by a variable delay. The
delay is programmable and is typically stepped through
the values 1, 2, 4, 8, 16. . . . . 16384 ms in cyclical
fashion. The recovered magnetization at the end of the
delay is determined by a short read-out sequence, consisting of two pulses and one spin echo. The height of
the echo, if plotted as function of delay time, traces out
a recovery curve that can be converted into a T1 distribution by standard inversion methods (Prammer, 1994).
It takes 33 seconds to complete a measurement cycle.
Examples of T~ distributions for some example fluids are shown in Figure 5. These data have been acquired by circulating different fluids through the analyzer. Shown from top to bottom are: water (mild brine)
with a single relaxation peak in the "water window" at
2 seconds; next a simple hydrocarbon (diesel) with a
single relaxation peak in the "oil window" at 0.5-1 second; and a complex hydrocarbon (crude), which shows
a characteristic asymmetric distribution that starts in the
few tens of milliseconds and extends to the "oil window." These samples were under atmospheric condi-

SPWLA 42 "d Annual Logging Symposium, June 17-20, 2001

tions at ambient temperature.

In contrast to T~, T2 cannot be determined on a
flowing sample. Distributions of T2 times are determined by standard Carr-Purcell-Meiboom-Gill (CPMG)
sequences on samples that have been stagnated momentarily. Stagnation is achieved by closing a valve
below the analyzer and diverting the flow stream
around the sample chamber. The time required for a T2
measurement is almost entirely determined by the polarization time ("wait time") of 15 seconds.

spacing (typically 0.25 ms) and a long spacing (To) are

alternated. Assuming that the fluid has a T2 relaxation
time (the argmnent also holds for an arbitrary distribution of T2 times) and diffusivity D, the two echo trains
for the short and the long echo spacing can be described
as follows:

Hydrogen density measurement~ The hydrogen density or the total number of hydrogen atoms within the
measurement volume is a by-product of any T~ or T2
measurement. Hydrogen density is automatically converted to hydrogen index IH. The hydrogen index is hydrogen density relative to that of water at ambient conditions. Under the assumption that the oil contains only
hydrogen and carbon atoms, the mass density Pro, the
hydrogen index In, and the hydrogen-to-carbon ratio R
are related as follows (after Zhang et al., 1998):

where 1/TD = 1/12 ('/Go Te)2 D.

Iri -- Pm 9R/(12+R).

(4)

Since the hydrogen index is measure& either the mass

density or the H:C ratio can be computed from an estimate of the other variable.
The hydrogen density in gases is significantly depressed due to the reduced bulk density. Appel et al.
(2000) reported a reduction of about 20% in hydrogen
density on live, unsaturated oil samples at 180F.
Diffusion measurement& The diffusion measurement
is implemented using steady-gradient spin-echo (SGSE)
experiments (Kimmich et al., 1997). It requires that the
fluid flow be temporarily stopped. The idea of using the
fringes of a uniform-field volume for diffusometry is
derived from so-called SSF-SGSE methods. Its main
advantage over pulsed-field gradient spin-echo
(PFGSE) diffusometry is insmmaental simplicity and
superior stability (Kimmich et al., 1997). The main
drawback is a limit on sensitivity, which, for the downhole implementation, is approximately 10-6 cm2/s.
As shown in Figure 2, the sensitive volume can be
divided into an interior, homogeneous region and an
exterior gradient region. The field in the fringe volume,
which makes up about 1/3 of the total volume, can be
approximated by a single field gradient value Go. At
short echo spacings (0.25 ms), the effect of the field
gradient is too small to be relevant. The pulse sequence
used both for diffusion measurements and for diffusivity calibration is shown in Figure 6. Two Carr-PurcellMeiboom-Gill (CPMG) sequences with a short echo

A1 = In exp(-t/T2) and
A2 = In Ko exp(-t/T2)exp(-t/TD) + In (1-Ko)exp(-t/T2),

(5)

The system parameter Ko is the gradient volume divided by the total volume. The hydrogen gyromagnetic
ratio ~, is equal to 26,754 rad/s/gauss. Both Ko and Go
are temperature-dependent and are determined during
calibration. The diffusivity D is derived from Eqs. 5 by
taking the ratio of corresponding echoes:
A2/AI = Ko exp(-t/TD) + (1- Ko).

(6)

This curve is fit to a uni-exponential model plus an offset. In Figure 7, the top two curves are the A~ and A2
signals for water at room temperature. Below is the ratio curve and the best-fit uni-exponential model. Since
D for water is known as 2.5x10 5 cm2/s, these curves
determine the calibration parameters Goand Ko.
Viscosity is determined as (Lo et al., 1998):

N
r I = 5x10 "s T/D.

(7)

In this expression, the viscosity rl is measured in cp, the

temperature T in Kelvin and the diffusivity D in cm2/s.
The temperature is obtained from the RDT fluid temperature sensor. Lo et al. determined the proportionality
factor by fitting Eq. 1 to data from pure alkanes and
methane-alkane mixtures.
Calibration. The system performs its own calibration,
provided the sample chamber is filled with a known
reference fluid (typically oil), and the system is heated
through its operating temperature range. At specific
temperature points, the built-in processor system records the sensor's resonance frequency and pulse-echo
response as function of echo spacing and stores the results in a permanent calibration table.

UNDERSTANDING NMR RELAXATION TIMES

The spin-lattice relaxation time T~ describes the coupling of nuclear spins to energy-absorbing molecular
motions like rotation, vibration and translation. In most

SPWLA 42 "0 Annual Logging Symposium, June 17-20, 2001

fluids at or above ambient temperatures, the coupling to

these modes is very inefficient, resulting in T~'s in the
millisecond to second range. The mechanisms for spinspin relaxation (Tz) include, in addition to those contributing to T~, the exchange of energy between spins.
These effects are small in bulk fluids, and therefore, T2
basically equals T1. Spin-spin coupling is relevant in
heavy oil components such as asphaltenes, resins, etc.
Closed-form expressions for T~ have only been derived for spherical molecules (Abragam, 1961). Translational and rotational relaxation rates are, respectively,
1/T1 (translational) oc Nrl/kT,

and

1/Tl (rotational) oc (a3/b 6) Xrl/kT,

(8)
(9)

where N is the spin density (proportional to the hydrog~n ~nd~x

a is the molecular radius, and b is the
distance between hydrogen spins on the same molecule.
Both mechanisms have the same basic relationship with
viscosity and temperature, which explains the relative
simplicity of Eq. 2.
Downhole pressures over 10,000 psi can induce significant changes in fluid viscosity (Jones, 1991). As a
rule-of-thumb, high pressures and high temperatures
have opposing effects. Higher temperatures expand
fluids and increase mobility and relaxation times, while
increased pressures reduce mobility and relaxation
times in liquids.
Probably the largest effect of high pressures is an
increase in dissolved gas volume. The relaxation times
in the gas phase exhibit the following behavior:
T1 = T2 oc rl/kT.

(methane vapor)

(10)

The reason for not following the relationship in Eq. 2 is

the fact that protons in the gas phase relax by spinrotation and not by dipole-dipole interaction. Increasing
the pressure on gas increases Tl, which is contrary to
the behavior of liquids. For the gas signal to be detectable downhole, the bulk density and the hydrogen density must be relatively high, i.e. only the high-viscosity
end of Eq. 10 is observable. Under these circumstances,
methane relaxation times range from hundreds of milliseconds to several seconds.
Dissolved gas has a profound effect on the Tl of oil
(Appel et al., 2000). Lo et al. (2000) have developed a
mixing-rule model for methane-alkane mixtures that
links GOR to T1 and diffusivity. In this model, T1 consists of two components: one proportional to kT/rl (Eq.
2) and one proportional to rl/kT (Eq. 10). The observed
T~ is a combination of the two, weighted by the proton

fractions of the alkane and the methane gas.

APPLICATIONS

Connate oil v. m u d f ~ a t e differentiation. Tl distributions can be used for qualitative fluid characterization
without invoking much of the relaxation theory. We set
the product tiT1 to 1 at T=300K and establish a simple
viscosity scale from 1,000 cp to 1 cp, corresponding to
the Tl range 1 ms to 1,000 ms (refer to Figure 5). The
water peak is distinctly offset due to the different molecular structure. The complexity of the distribution in
Figure 5.bottom reflects a distribution of internal mobilities due to a mix of short and long hydrocarbon
chains. Contrast this appemance to the simple structure
of Figure 5.center, and we have a method to differentiate between complex oils (crudes) and simple oils (illtrates). The exploitation of these differences is the subject of a follow-up paper (Bouton et al., 2001). It will
describe a more quantitative technique to assess the
contamination level by oil-based mud filtrate.
N M R log interpretation. Both wireline and LWD tools

take their readings in the invaded zone, which is more

or less flushed by filtrates. Porosity measurements by
NMR are in fact hydrogen density readings in the fluid
phase, calibrated to a water sample and corrected for
temperature. Consequently, a lower-than-anticipated
hydrogen index (dissolved gas, etc.) may cause an undercall in porosity. Techniques are available to correct
this effect for entrained methane (Coates et al., 1999),
but it is more precise to sample the invaded zone and
directly determine the hydrogen index of the produced
fluid.
Fluid typing based on wireline/LWD NMR is complicated by the fact that T1 and T2 distributions are
modulated by variations in pore sizes and by the bulk
relaxation response of crude oils. The interpretation
makes simplifying assumptions about the hydrocarbon
phase (non-wetting, single T1 for oil; single T1 for gas).
This approach can be refined by determining the actual
relaxation profiles from fluid samples and by feeding
this information back into the saturation calculation.
Wireline and LWD NMR tools operate at frequencies between 0.5 MHz and 2 MHz, while the fluid analyzer operates at 4 M H . This difference is irrelevant
for the relaxometry of fluids with Tl --- T2. This condition indicates a uniform and fiat energy spectnan in the
employed frequency range.

Hydrocarbon viscosity. Determination of viscosity under true reservoir conditions has always been a challenge. As we have seen, the NMR-derived diffusivity D

SPWLA 42 "d Annual Logging Symposium, June 17-20, 2001

has a universal correlation with viscosity (Eq. 7). In order to get a continuous viscosity reading, however, it is
desirable to also derive a viscosity estimate from T~ and
Irt alone. For the case of no dissolved hydrocarbon gas
(i.e. In close to 1) and no entrained parama~etic oxygen, currently the best available correlation is:
T1 = T2 ~ 9.6x 10-3 T/r 1 (degassed alkanes),

tions. Fluid properties, such as viscosity and gas/oil ratio, can be derived from diffusivities and relaxation
times as measured by NMR. This development not only
aids in the accuracy of the downhole sampling process
and in the interpretation of M R logs, it also paves the
way for a comprehensive fluid characterization at actual
reservoir conditions.

(11)
ACKNOWLEDGEMENT

where T~ and T2 are in s, T is in K, and rl is in cp (Lo et

al., 2000). To compute viscosity from a T~ distribution,
we use:
r I ~, 9.6x10 -3 T/T1 g'm"

(12)

where T~g'm is the geometric mean taken over the T~

distribution. We expect to improve this correlation by
including the hydrogen index once a sufficient body of
downhole NMR data is available.

Gas/oil ratio (GOR). For dead oils (GOR=0), the relationship between T~ and D is linear as can be seen by
comparing Eqs. 1 and 2. Increasing gas contents introduces a deviation from the linear behavior. Modification ofEq. 2 and substituting from Eq. 1 leads to:
T1, T2 oc D/f(GOR),

(13)

where f(GOR) is a function that has been determined

empirically for methane-n-alkane mixtures (Freedman
et al., 2000). The proportionality factor in Eq. 13 is approximately 2x105 (Lo et al., 2000). The C~R can be
determined graphically by means of a crossplot T1 v. D
with GOR as ~ e t e r
(Lo et al., 2000, Fig. 13).
Emerging applications aim at a comprehensive understanding of the reservoir fluid system. At this level, we
are interested in the complexity of diffusion distributions D and how they relate to the distributions of T1
and T2. Each of these measurements probe a different
dimension of molecular interactions and mobilities: D
describes molecular translational movement, T~ relates
both to translational and in-place mobility, and the
contrast between T~ and T2 is most useful for the nearsolid part of the mobility spectrum. We expect methods
to emerge that incorporate complex fluid models that
can be updated in real time based on downhole NMR
analysis.
CONCLUSIONS

It is now possible to determine low-field NMR characteristics of fluids downhole under true reservoir condi-

The authors would like to thank Prof. George Hirasaki

for his helpful comments during the revision of this
manuscript.
REFERENCES

Abragam, A., 1961, Principles of Nuclear Magnetism,

Oxford University Press, Oxford.
Appel, M., Freeman, J.J., Perkins, R.B., and van Diik,
N.P., 2000, Reservoir Fluid Study by Nuclear Magnetic
Resonance, Paper HH: SPWLA, presented at the 41 st
Annual Logging Symposium, Dallas, TX, June 4-7.
Bloembergen, N., Purcell, E. M., and Pound, R. V.,
1948, Phys. Rev. 73, p. 679.
Bouton, J., and Prammer, M. G., 2001, Assessment of
Sample Contamination by Downhole NMR Fluid
Analysis, Society of Petroleum Engineers, accepted for
presentation at the 76th Annual Technical Conference
and Exhibition, New Orleans, LA, Sep. 30-Oct. 3.
Brown, R. J. S., 1961, Proton Relaxation in Crude Otis,
Nature 189, no. 4762, p. 387.
Coates, G. R., Xiao, L., and Prammer, M. G., 1999,
NMR Logging, Principles and Applications, Halliburton Energy Services, Houston, pp. 113.
Freedman, R., Sezginer, A., Flaum, M., Matteson, A.,
Lo, S., and Hirasaki, G. J., 2000, A New NMR Method
of Fluid Characterization in Reservoir Rocks: Experimental Confirmation and Simulation Results, Paper
SPE-63214: Society of Petroleum Engineers, presented
at the 75 th Annual Technical Conference and Exlfibition, Dallas, TX, Oct. 1-4.
Jonas, J., 1991, High Pressure NMR Studies of the Dynamics in Liquids and Complex Systems, in High Pressure NMR, J. Jonas, ed., Springer-Verlag, New York,
85-128.
Kimmich, R., 1997, NMR Tomography, Diffusometry,

SPWLA 42 "d Annual Logging Symposium, June 17-20, 2001

Relaxometry, Springer-Verlag, Berlin.

Kleinberg, R. L., and Vinegar, H. J., 1996, NMR Properties of Reservoir Fluids, The Log Analyst, Nov.-Dec.,
p. 20.
Lo, S.-W., Hirasaki, Kobayashi, R., and House, W. V.,
1998, Relaxation Time and Diffusion Measurements of
Methane and n-Decane Mixtures, The Log Analyst,
Nov.-Dec., p. 43.
Lo, S.-W., Hirasaki, G., House, W. V., and Kobayashi,
1L, 2000, Correlations of NMR Relaxation Times with
Viscosity, D ~ i v i t y , and Gas/Oil Ratio of Methane/Hydrocarbon Mixtures, Paper SPE-63217: Society
of Petroleum Engineers, presented at the 75th Annual
Technical Conference and Exhibition, Dallas, TX, Oct.
1-4.
Prammer, M. G., 1994, NMR Pore Size Distributions
and Permeability at the Well Site, Paper SPE-28368:
Society of Petroleum Engineers, presented at the
69th Annual Technical Conference and Exhibition, New
Orleans, LA, Sep. 25-28.
Prammer, M. G., Bouton, J., Chandler, R. N., Drack, E.
D., and Miller, M. N., 1998, A New Multiband Generation of NMR Logging Tools, Paper SPE-49011: Society
of Petroleum Engineers, presented at the 73 ~d Annual
Technical Conference and Exhibition, New Orleans,
LA, Sept. 27-30.
Prammer, M. G., Drack, E., Goodman, G., Masak, P.,
Menger, S., Morys, M., Zannoni, S., Suddarth, B., and
Dudley, J., 2000, The Magnetic Resonance WhileDrilling Tool: Theory and Operation, Paper SPE62981: Society of Petroleum Engineers, presented at the
75thAnnual Technical Conference and Exhibition,
Dallas, TX, Oct 1-4.
Proett, M. A., Gilbert, G. N., Chin, W. C., and Monroe,
M. L., 1999, New Wireline Formation Testing Tool
with Advanced Sampling Technology, Paper SPE56711, Society of Petroleum Engineers, presented at the
74thAnnual Technical Conference and Exhibition,
Houston, TX, Oct. 3-6.
Zhang, Q., Lo, S.-W., Huang, C. C., Hirasaki, G. J.,
Kobayashi, IL, and House, W. V., 1998, Some Exceptions to Default NMR Rock and Fluid Properties, Paper
FF: SPWLA, presented at the 39th Annual Logging
Symposium, Keystone, CO, May 26-29.

A B O U T THE AUTHORS
Manfred Prammer is president of NUMAR. He has
been with the company in various positions since 1991
and holds numerous patents for the development and
application of NMR wireline tools, for general NMR
techniques and for signal processing methods. Dr.
Pmmmer received the Diplom-Ingenieur degree in
electrical engineering and a Ph.D. degree in technological sciences from the University of Technology, Vienna, Austria, in 1982. From 1980 he held R&D positions with Siemens A.G., and from 1987 to 1991, he
was assistant professor at the University of Pennsylvania, where he developed clinical MRI technology and
MR spectroscopy applications.
John Bouton is director of research at NUMAR. He

holds a BS degree in electrical engineering from

Marquette University (1959). He has been with the
company since 1986, where his research involves
MRIL tool development and interpretation techniques.
Mr. Bouton's other professional interests include image
acquisition and automated image analysis.
Peter Masak is manager of mechanical engineering at

NUMAR. He holds a BASc degree from the University

of Waterloo (1981) in Canada. He has been with the
company since 1997, and has participated in the development of the MRIL-Prime wireline tool and the
MRIL-WD Logging While Drilling tool. Previously he
worked for Schlumberger Well Services, initially as a
field engineer in South America, and later in various
engineering positions both in the wireline and Measurement-While-Drilling divisions.

SPWLA 42 "d Annual Logging Symposium, June 17-20, 2001

~,

Inflow l
,,.,~..

Polarization
Section

Transmitter
Coil

""

Figure 1: Schematic diagram of the downhole NMR

fluid analyzer. The module is part of the RDT tool
string. Pumped fluids enter the device at the top.
Hydrogen nuclei are polarized in a 2,000 gaussstrong field in preparation for a resonance measurement in the lower section at 1,000 gauss. It is possible to determine hydrogen density, relaxation times
and self-diffusivity in a non-intrusive fashiorL

Receiver

"~
1,

Ni]:'.

>

Resonance
Section

Outflow J

Homogeneous
Interior Field

Figure 2" Horizontal cross section through measurement volume. The magnetic field (1,000 gauss) is
transversally oriented (direction of arrow). The field
is homogeneous within a cylindrical interior volume.
This region is surrounded by a fringe volume with an
............ approximately constant field gradient. The fringe field
volume comprises about 1/3 of the total sensitive volume and is used for diffusion measurements.

Gradient

Fringe
Field

N
s~aal
generator
3.8 - 4.4 MHz

Figure 3: Electronics diagram.

The electronics is similar to that
of a conventional NMR spectrometer, with the exception
that many functions are realized
Transmit as digital signal processor
coil
(DSP) algorithms. All timing
functions, like pulse widths and
acquisition windows, are fully
programmable. The system also
performs its own calibration.

Digitalsignal
processor&
systemcontroller

signal
d~,itizer

[ ~.t
i ~ot.

Receive
coil

SPWLA 42 "a Annual Logging Symposium, June 17-20, 2001

Refocusing
pulse \
Excitation
\
pulse

Received
NMR spin echo

Saturation pulse

Time

h~
Y

Variable saturation-recovery
time delay

Figure 4: Saturation-recovery sequence employed for TI measurements.

This sequence is insensitive to fluid flow and can be used to continuously
monitor the Tl profile of pumped fluids. The saturation-recovery time is
stepped through the values 1, 2, 4, 8, .. 16,384 ms and yields 14 points on
a magnetization recovery curve. The magnetization is determined by a
very short, two-pulse sequence, which generates a single spin echo. An
entire measurement cycle takes 33 seconds.

::I

........

0.4

. . . . . . .

0.41,

:or . . . . . . . . . . .
~

. . . . . . . .

lo'r . . . . .

0.3

. . . . .

:e

...........

. . . . . . . .

.......

/,I j
- - - -

. . . . . . . .

,o~

. . . . . . .

1~

ld----

',

::--

lo ~

. . . . . . . .

t I -----. Crude I

o.i!;
10 ~ " ~ ' ~ ' M ' ~ ' -

__
10

........... . i
10
T1 (ms)

10

10

Figure 5: Example Tl relaxation distributions for brine, diesel oil and a

crude oil as seen in the NMR fluid analyzer. The data were acquired according to the sequence in Figure 4 and were inverted from time domain
to T] domain. 53 points were specified for the inversion result The single,
sharp peak at 2-3s is characteristic of water, the rounded peak in the "oil
window" 0.5-1s indicates oil, and the broad response from the crude oil is
characteristic for complex hydrocarbons.

SPWLA 42 "dAnnual Logging Symposium, June 17-20, 2001

Refocusing

NMR

Received
spin echoes

/_x_

Time

Excitation

Time

Echo-to-echo

time

Te

]Figure 6" Pulse sequence employed for diffusivity measurements. Two CPMG sequences with a short and a long echo spacing are alternated. The long echo spacing is
an integer multiple of the short spacing. Echoes line up in time, i.e. occur at the same
elapsed time since the excitation pulse and the ratio of their amplitudes can be formed
to compute the self-diffusion coefficient.

0.8
-

....

,= -7

AI

,
------==mt
~lmltm

0.6

0.4
0.2

[----

Te=0.5 data ]

1
0

1~

200

3~

1
4~

i
200

I
300

I
400

500

L
6~

7~

t
800

900

1000

I
500

i
600

i
700

I
800

i
900

1000

0.8
0.6
0.4

0.2
0

- - - - - Ratio
Fit
i
0
100

I
time (ms)

Figure 7: Example of a diffusivity measuremenL The top two curves are spin echo amplitudes
at different echo spacings. The accelerated decay for the longer spacing is a manifestation of
diffusion in the gradient region of the magnetic field. The ratio curve (below) is the sum of an
exponential and a constant term, corresponding to the gradient-field region and the uniformfield region, respectively. The best-fit model curve is also plotted and is not distinguishable
from the data.