SPWLA 51st Annual Logging Symposium, June 19-23, 2010

A New Method for Estimating Formation Water Rw Using NMR Logging Data

Songhua Chen and Gabor Hursan

Baker Hughes Inc.
Copyright 2010, held jointly by the Society of Petrophysicists and Well Log analyzed. A continuous depth-wise monitoring of
Analysts (SPWLA) and the submitting authors.
formation salinity is desired for many situations
This paper was prepared for presentation at the SPWLA 51st Annual Logging
Symposium held Perth, Australia, June 19-23, 2010.
described in the previous paragraph.

Water resistivity correlates well with water salinity and

ABSTRACT temperature, but resistivity measurement alone cannot
be a reliable tool for salinity determination, for it also
Saturation determination from a resistivity-based depends on porosity and hydrocarbon saturation, and is
technique requires the information of formation water affected by the formation mineralogy, such as pore-
resistivity, Rw or, equivalently, the reservoir formation lining clay. An independent measurement of porosity
water salinity. If the reservoir salinity varies with depth, and saturation, as well as borehole resistivity, is
it is important that one has the capability of estimating required. In many practical situations, one often tries to
salinity or Rw in-situ. We introduce a new method of measure the formation resistivity from a clear-sand
estimating formation water salinity using NMR logging water leg and applies this value in the hydrocarbon-
data. Specifically, we use the quality factor, Q, a bearing zone, hoping they are identical. This approach
quantity often used as a tool QC indicator, for is valid if formation water resistivity is kept constant
estimating formation water resistivity. We describe the for clean sand, but it cannot handle a situation where
method of identifying the effects on Q due to formation water salinity varies. Moreover, for T
instrument, formation, and the borehole. We also horizontal wells the drilling trajectory often does not T
discuss the application envelope of this technique in pass a known clean-sand water leg. T
terms of borehole environment and formation
conductivity ranges. Examples are provided to NMR logging provides an independent measurement of
demonstrate the efficacy of the technique. porosity and it is widely used for reservoir fluid typing
and quantification. As NMR logging is a shallow depth-
INTRODUCTION of-investigation (DOI) measurement, the saturation
from NMR logging represents the invaded-zone
Formation water salinity is a critical reservoir saturation, which is often not consistent with resistivity-
parameter for resistivity-based water saturation based saturation measured at different DOI.
estimation. The water salinity in many oil and gas
reservoirs varies from location to another, from one This paper describes a new approach for obtaining the
reservoir unit to another, and sometimes even at necessary information on formation water resistivity
varying depths. On the other hand, in many oil- from a byproduct of NMR logging. For quality control
producing or mature fields discerning water-injected purposes, the NMR logging tool provides a quality
zones from intact zones is important for estimating factor, Q. The value of Q is affected by both the tool
remaining reserves and production potentials from hardware and the borehole and formation loading due to
newly-drilled wells. In some part of the world, sea conductive loss. The electronic circuit’s Q can be
water is often used as the injection water and the determined by experimentation and modeling. The
salinity of the native formation water is at or close to environment and formation loading effects on Q, on the
salt saturation. The large salinity difference between other hand, can be used for estimating the formation
injection and native formation water is potentially salinity. Even though the conductive loading utilizes the
useful for discerning residual hydrocarbon zones from same principle on which EM logging tools are based,
virgin zones. NMR logging measures independent hydrocarbon
saturation and porosity values, and these properties
correspond to the same depth-of-investigation (DOI) as
An in-situ fluid analyzer tool, such as IFXSM, is capable
that of the Q-based resistivity analysis. These three
of analyzing formation water salinity from the collected
quantities are sufficient for formation water resistivity
samples after the filtrate contamination is pumped out.
estimation.
Due to the lengthy time required to pump contaminated
fluids out before the measurement, such in-situ water
As NMR logging is a shallow DOI measurement, the
salinity analysis is not used extensively. Even when it is
technique works best for wells drilled with oil-based
used, only a limited number of sampling points are
1
SPWLA 51st Annual Logging Symposium, June 19-23, 2010

mud (OBM). It is also applicable to low-permeability An electrically conductive environment reduces Q

wells drilled with water-based mud where the invasion value significantly. Thus borehole and formation
is negligible, making the mud filtration irrelevant to our loading effects on the overall Q also needs to be
application. The log examples in the paper demonstrate considered. If the mud is salty, borehole loading is
that the NMR Q measurement is capable of providing dominant. The borehole loading is dependent of the
the formation water resistivity when the mud condition mud salinity (or, equivalently, the mud resistivity Rm),
is met. Moreover, by using the multi-frequency NMR and the geometries of the logging tool and the borehole.
tool, in principle, one can combine the saturation and Q For the MReX tool the effective borehole Qbh can be
variations corresponding to different NMR DOI, expressed empirically by:
thereby discerning filtrate invasion vs. formation water
should the two types of water have different salinities. β
Qbh = α ⋅ ( Rm f ) ⋅ ( dbh − dOD )
λ
(2)

METHODOLOGY Where dbh and dOD are the diameter of borehole and
Chen et al, (2006) described a method of using Q as a MReX tool, respectively, and dOD = 5 1 8 " .
hydrocarbon indicator and as a filtrate invasion α , β , and λ are coefficients that also can be determined
indicator. U.S. patent 7,425,827 described the details of by calibration and simulation.
an NMR logging tool antenna Qa, and the borehole and
formation-loading effects on the overall Q The measured Q value can be treated as the collective
measurement. In this section, we first summarize these effects of antenna Qa , borehole Qbh , and formation
effects, then outline the method leading to the
estimation of flushed zone resistivity. We then discuss Q fm by:
the required conditions to enable this technique to
work. At the end of the section we describe one data 1 1 1 1 T
interpretation method to obtain the formation Rw. = + + (3)
Q Qa Q fm Qbh T
T
Effect of Antenna, Borehole, and Formation on Q
The quality factor, Q, of NMR logging is often used as Thus, the effective formation Q fm can be computed by:
a QC tool for monitoring the tool performance in Q −fm1 = Q −1 − Qa−1 − Qbh
−1
. (4)
various mud salinity environments. If the tool is placed
in a non-conductive environment, the Q of an NMR
logging tool, such as the MR ExplorerSM (MReXSM) Consequently, the formation Rxo can be expressed as:
(Chen et al, 2003), is a measure of the resonance circuit
(antenna) performance and is resonance frequency Rxo,MReX = fn( f , dbh ⋅ Q fm ) . (5)
dependent. As temperature can affect the performance
of the circuit elements (inductor, capacitor, and The empirical equation can be determined by
resistor), it is also dependent on temperature. The performing a series of experiments in a calibration tank
MReX logging tool operates in a relatively narrow with varying known formation salinity.
frequency range of approximately 500-975 kHz, so we
can approximate the antenna Qa as being linearly It is important to note that the idea described above
dependent of frequency f. The temperature effect can applies to all types of NMR logging tools conceptually.
also be account for in accordance with the following The actual equations, though, and in particular the
empirical expression: coefficients, vary according to the specific tool design.
Not only are the antenna and operating frequencies
b
 ( f − flo )   300  different for different NMR logging instruments, the
 ( )
Qa , f ≈  Qa , f hi + Qa, f hi − Qa, flo ⋅
c
 ⋅   (1)
geometry and DOI of the sensitive volumes are also
   TK 
unique to each tool. For a given hardware
configuration, such as MReX, the antenna Q variation
where f hi and f lo are the highest and lowest frequencies from one tool to another is usually small.
of MReX operation and Qa , f hi and Qa, flo are the
corresponding antenna Qa values, respectively. TK is the Sensitivity of Q
absolute temperature, and b and c are coefficients that Fig. 1 is a simulated Q contour plot as a function of
can be determined by calibration. borehole conductivity and formation conductivity,
computed for a borehole size of 6.375 in. For a non-
conductive borehole mud, such as the yellow, dashed
line in Fig. 1, the Q technique is sensitive to formation
2
 SPWLA 51st Annual Logging Symposium, June 19-23, 2010

conductivity for a wide range of Q values, Quality of Q Measurement

corresponding to the formation conductivity of Although it is common knowledge that NMR
approximately 0.1−100 S/m. On the other hand, for a measurements are often subject to low signal-to-noise
very conductive borehole, such as the blue dashed line, ratio (SNR) problem because of the intrinsically weak
the measured Q value is much less sensitive to the NMR signal, the Q measurement does not suffer from
formation conductivity. the same problem. The reason is that the Q
measurement does not directly detect the NMR proton
Two conditions must be met for this technique to be signal. Figure 2 shows typical Q curves for three
effective. First, it is important that the borehole effect frequencies over a depth interval of more than 100 ft. In
must be minimal or, at least non-dominant. That is, only each track, multiple Q curves are plotted that
fresh mud or oil-based mud environment, or air-drilled correspond to the same frequency, but are detected at
wells qualify the use of the Q-based methods for any different time when the corresponding echo trains are
formation property estimation. An OBM environment is acquired at different TW, TE, or echo length. It is
the ideal case for the present application. If OBMF obvious from the figure that they superimpose each
invasion occurs in the NMR-sensitive volumes, it is other very well. Note that these curves are plotted
immiscible with the formation water. As long as we can without performing vertical stacking (i.e., the running
use the NMR fluid-typing technique to determine the average of 1), yet the repeatability is remarkably good.
oil-and OBMF-occupying pore volume vs. water- The quality of Q curve is very important as the
occupying pore volume, the formation water Rw can be variation of Q in response to the formation water
estimated. On the other hand, if water-based mud is salinity change depends on it.
used, the invaded fresh water-based mud filtrates
(WBMF) is miscible with native formation water,
resulting in dilution of the reservoir formation water
salinity. Consequently, the method works only if the
T
amount of filtrates is insignificant in the NMR-sensitive
T
volume or if the volume of invaded filtrates can be
T
quantified. Those cases do occur if the permeability is
low and most formation water is irreducible.

The second condition for the effective use of this

technique is that the formation water conductivity must
be greater than the borehole conductivity and the
formation water-bearing porosity must not be very
small so that it will produce detectable loading effect on
the overall Q. The quantitative limitations of these
parameters are the subject of a subsequent paper.

Fig. 2. Example of Q curves acquired with different

frequencies. Plotted on each track are Q curves
Fig. 1. Simulated contour plot for MReX tool Q as a corresponding to the times when different echo trains are
function of borehole and formation conductivities. Shown acquired at the same frequency to illustrate the excellent
here is an example for frequency at 750 kHz, and a 6.375– repeatability of the Q curve.
in. borehole size.
For each frequency the Q value at the same depth
should be identical, therefore, in our analysis we
3
SPWLA 51st Annual Logging Symposium, June 19-23, 2010

averaged all Q curves of the same frequency in a data- where ζ and η are determined by a linear regression of
acquisition sequence. To further improve the quality of the crossplot (Figure 3). If the linear regression holds
Q curve, we can also average all Q curves acquired at well for Q, we can approximate Qfm with Q. Otherwise,
different frequencies. we must first compute Q −fm1 = Q −1 − Qa−1 then replace Q
Both Q and the standard deviation of the noise in NMR with Qfm in Eq. (7). We can then assume that the same
measurement are affected by borehole and formation correlation applies other intervals. Plug Q or Qfm value
loading. Consequently, one can expect that, corresponding to other depths, one can determine
alternatively, it is also possible to use the standard Rxo,MReX. In other words, Eq. (7) is one form of Eq. (5).
deviation of the noise for predicting formation water
salinity or Rw. Another potentially useful curve that is Note that the above shallow-resistivity calibration
familiar to NMR log analysts is “CHI,” the measure of procedure usually is carried out in a low-permeable
goodness of fit which is affected by both the noise and zone to minimize the discrepancy between the DOI of
the data-fit model. In general, the CHI and noise shallow resistivity and NMR logs.
measures are not as direct and straightforward as the Q
measurements in terms of correlating to formation and Next, we use one of the NMR-based fluid-typing
borehole conductivities, because the Q approach is not techniques described in a previous section to determine
complicated by the addition effects that are irrelevant to the part of the porosity that corresponds to
conductivity, such as the data-fitting quality. hydrocarbon-occupying porosity, and using the
difference between the NMR determined total porosity
Method for Taking into Account the Saturation Effect and hydrocarbon-occupying porosity as the water-
Equations (4) and (5) provide the means of deriving the occupying porosity:
flushed zone Rxo,MReX from NMR Q measurement.
Deriving formation Rw from Rxo,MReX requires φw = φT − φhc . (8)
quantification of porosity and saturation, both can be T
T
obtained by NMR logging. Hydrocarbon saturation, We further assume the hydrocarbon-occupying porosity
T
which includes both native oil and OBMF, can be has infinite resistivity so that we can correlate the
derived using NMR-based fluid-typing techniques. For water-bearing φw with Rxo derived with the NMR-based
high magnetic-field-gradient NMR tools such as Q measurement. The examples we used in this paper
MReX, 2D NMR techniques, such as D-T2int (Sun et al, range from clean sands to moderately shaly sands.
2006, DiRosa et al, 2006, Chen et al, 2009) or R-T2app Archie equations are considered approximately valid
(Hursan et al, 2005), can be used. For low-gradient for Rw estimation.
NMR tools such as MagTrakSM (Chemali, et al, 2005),
the light hydrocarbon signal sometimes can be
separated from the water signal on T2 spectra, in which R 2= 0. 85
5
case the hydrocarbon volume fraction can be
determined simply using a defined cutoff method.
Shallo w Re sistivity

Determination of Water Resistivity using Q

The first step determines the near-wellbore resistivity, 2
based on Q. For OBM-drilled wells, Qbh is much larger
than Q, thus it can be removed from Eq. (4).
Furthermore, if the formation water salinity is high and
porosity is also high, Q fm becomes much smaller 1
than Qa , in which case Qa can also be removed from
Eq. (4). When these conditions are met, the overall
Q ≈ Q fm . The Qfm vs. Rx0,MReX correlation (Eq. (5)) can 0.5
42 44 46 48 50
be determined by calibration. Alternatively, a simplified Q
approach of correlating shallow-resistivity log data to
Fig. 3. Shallow resistivity correlates well with Q in Well A.
correlated Qfm at a water-bearing interval can also be
Note that both axes are in logarithmic scale. However, due
used for calibration. Specifically, we can plot Q or Qfm to a narrow Q variation, the logarithmic scale for Q is not
and Rx0,shallow in log-log plots: obvious.

log Rx0,shallow = ζ ⋅ log Q(or Q fm ) +η , (7) Finally, we determine formation water resistivity Rw by
combining the NMR-based Rx0,MReX and water-filled
4
 SPWLA 51st Annual Logging Symposium, June 19-23, 2010

porosities in a reservoir section with using a Pickett- curve. The calibration is illustrated in Fig. 8, using
plot as shown in Figures 4 - 6. which we derived the fitting coefficients of Eq. (7).

LOG EXAMPLES

The first two examples are from an offshore well (Well

A) drilled with oil-based mud. Figure 3 demonstrates a
strong correlation between shallow resistivity and Q.

Figures 4 and 5 correspond to two sections of a same

well but the two reservoir sections are not connected.
One section (Figure 4) is hydrocarbon-bearing, clean
sandstone formation and the other one (Figure 5) is
water-bearing and moderately shaly. OBMF invasion is
observed on NMR 2D maps from both sections. Plotted
on these two figures are water porosity φw vs. NMR Fig. 4. Pickett-plot of a clean sandstone zone from Well A.
log-derived Rxo,MReX. Assuming Archie parameters a=1,
m=2, and n=2, the formation water resistivity is
estimated as approximately 0.018 ohm-m (Figure 4) in
a clean zone. The low scatter in resistivity and porosity
values demonstrates the reliability of the approach. The
method also works for moderately shaly formations
with dispersed clays, as shown in Figure 5, where Rw is
T
estimated to be 0.02 ohm-m. Even though these two
T
sections are not connected, the depositional T
environment is likely to be the same and the native
formation water salinity is expected to be the same. Our
analysis shows indeed they are consistent.

The third example (Figure 6) shows a similar plot of a

clean-sands zone from another offshore well, denoted
as Well B and located in a different geographic region Fig. 5. Pickett-plot of a shaly sand zone from Well A.
than Well A. This well was also drilled with OBM. The
pore water volume varies across the zone;
consequently, the NMR log-derived Rxo,MReX shows a
linear dependence of the water porosity in log-log scale.
This relationship leads to a reasonable estimate of the
formation water resistivity of 0.028 ohm-m. In addition,
it is indicative that the cementation exponent of 2 is a
good approximation. Both the formation water
resistivity and cementation exponent values are typical
in this area.

Figure 7 is a section of log from Well C. We use this

example to illustrate the step-by-step procedure of the
data interpretation method. From left to right, Track 1
shows Gamma Ray, GR, from LWD and GR from WL
NMR passes, respectively. The two GR curves are
plotted together to serve as the depth matching
Fig. 6. Pickett-plot of a clean sand zone from Well B.
reference. The shallow resistivity calibration is done in
the depth intervals marked by the hatched yellow
rectangle symbols in the right side of Track 2. Also
plotted in Track 2 are the Q curve, the shallow
resistivity curve, and the Q-derived, calibrated Rxo,MReX

5
SPWLA 51st Annual Logging Symposium, June 19-23, 2010

method has adequate sensitivity to distinguish

formation water having different salinities. The
conditions with which the method works are also
discussed in the paper.

R = 0. 85
10

Sh allo w Re sistivity
5

1
38 40 42 44
Q

Fig. 8. Well C shallow resistivity and Q correlation.

T
Fig. 7. Well C log example illustrating the section selected
for calibrating coefficient of Eq. (7) and the low-GR T
section where reservoir Rw is estimated. T

We then plot the NMR derived water porosity vs.

Rxo,MReX, shown in Fig. 9. Using Archie’s equation, we
obtained Rw = 0.014 Ohm-m for the reservoir intervals.
Note the clean-sand interval on this figure is a water-
bearing interval. The oil signal shown in the rightmost
volumetric analysis track is the invaded OBMF. The
residual water saturation in the flushed zone is very
small but is sufficient for Q-based Rw estimation.
Note that the use of shallow resistivity-based
calibration is not necessary if the coefficients in Eq. (7)
are obtained via tank calibration using various
salinities. The illustration of the shallow resistivity and Fig. 9. Well C water porosity φw vs. Rxo,MReX for a clean-
Q correlation, however, is direct evidence that Q is an sand interval. The interval is heavily flushed by OBMF
equivalent measure of resistivity. In a subsequent paper, leaving only a few p.u. of water-occupying porosity.
we will present the Rw estimation using NMR log only,
with the application of laboratory calibrated coefficients ACKNOLEDGEMENTS
from simulated-formations.
We thank Baker Hughes for the permission to publish
CONCLUSIONS this paper. We also thank our colleagues, David Beard
and Carl Edwards, for the simulation and lab
We have presented a new method of using an existing,
experiments on MReX Q calibration.
peripheral NMR logging curve, Q, together with NMR-
based porosity and hydrocarbon typing analysis
techniques, for estimating formation water resistivity REFERENCES
Rw. The method is best used for OBM-drilled wells. As Chemali, R., Thern, H., Kruspe, H., Krueger, V.,
the Q curve is always delivered as a quality control Borghi, M., Porrera, F., and Lyne, A, “Magnetic
curve for MReX logging, no additional data acquisition Resonance While Drilling Streamlines Reservoir
is required. We applied this method to analyze several Evaluation,” SPE European Formation Damage
OBM-drilled wells and demonstrated that the new Conference, May 2005.
6
 SPWLA 51st Annual Logging Symposium, June 19-23, 2010

Nanjing Institute of Technology in China and a Ph.D.

Chen, S., Beard, D., Gillen, M., Fang, S., and Zhang, from University of Utah, both in physics.
G., “MR explorer log acquisition methods:
petrophysical objective-oriented approaches,” Paper ZZ Gabor Hursan is a staff scientist in the Integrated
presented at 44th SPWLA Logging Symposium, Interpretation and Petrophysics group at the D&E
Galveston, Texas, June 2003. Houston Technology Center of Baker Hughes. He has
been involved in the development of Baker’s NMR data
Chen, S., Shao, W., Hursan, G., and Sun, B.: processing and interpretation technology. His
“Improvement of NMR Multi-dimensional Inversion background includes geophysical field theory,
Methods for Accurate Petrophysical and Fluid petrophysics and inversion theory. He holds a M. S. in
Quantification Analyses,” Presented at 50th Annual Geophysical Engineering from the University of
APWLA Logging Symposium, Woodlands, Texas, June Miskolc, Hungary and a Ph. D. in Geophysics from the
2009. University of Utah.
Chen, S, Kruspe, T, Blanz, M., Thern, H.F., Edwards,
C.M., Beard, D.R., Method and apparatus for formation
evaluation and borehole size determination, US patent #
7425827, Sept., 2008.
Chen, S., Munkholm, M., Hursan, G., Georgi, D, and
Gillen, M., “Simple, robust NMR-based indicators for
detection of hydrocarbon gas or oil, borehole
contamination, and vugs,” SPE paper 96409 presented
at 2005 SPE ACTE, Dallas, Texas, Oct. 9-12.
T
Di Rosa, D., Gyllensten, A., Chen, S., Li, W., Georgi,
T
D, and Tauk, R.: “Use of the NMR Diffusion Log to
T
Identify and Quantify Oil and Water in Carbonate
Formations,” SPE Reservoir Evaluation & Engineering,
April 2008, pp.238-245.

Sun, B.Q., Olson, M, Baranowski, J., Chen, S., Li, W.,

and Georgi, D., “Direct fluid typing and quantification
of Orinoco belt heavy oil reservoirs using 2D NMR
logs”, paper EE presented in 2006 SPWLA Annual
Symposium and Exhibition, Veracruz, Mexico, June 4-
7, 2006.

Hursan, G., Chen, S., and Murphy, E., “New NMR two-
dimensional inversion of T1/T2app vs. T2app method
for gas well petrophysical interpretation,” Paper GGG
presented at 2005 Annual Symposium of SPWLA, New
Orleans, June 2005.

ABOUT THE AUTHORS

Songhua Chen is the manager of Integrated

Interpretation and Petrophysics at D&E Houston
Technology Center of Baker Hughes. He has been
primarily involved in research and development of
NMR logging technology and integrated petrophysics.
Prior to joining Western Atlas in 1996, he was a
research scientist for five years with Texas Engineering
Experiment Station in College Station, Texas, where he
worked in the area of NMR and MRI applications to
flow in porous media. Songhua holds a B.S. from