Understanding the Waxman-Smits Equations

(for version 2010.1

subject to change by Techsia)

Errors in the Techlog Help File have now been corrected.

The Waxman-Smits equations for the interpretation of electric logs for water
saturation in oil-bearing shaly sands was first proposed in 1966 and verified in 1972.
Until these equations were proposed the industry was saddled with a myriad of
equations based roughly on some measure of Vshale. This seminal work
demonstrated to the world that the correct parameter that controlled electrical
response in shaly sands was Qv, the cation exchange capacity of the clay minerals
surface per unit pore volume. The scientific community of the oil patch quickly
recognized the usefulness of the approach and it quickly became an industry standard.
However, the laboratory science was ahead of the logging industry and to this date
there is no logging tool that can be used to determine Qv in situ. Core to log
correlations are the best available, while in lieu of core measurements indirect
calculation of Qv from log shale indicators is the normal practice. The most widely
accepted empirical methods to use Qv (besides Waxman-Smits) are those proposed
by Juhasz (which applies the Waxman-Smits approach utilizing logs only) and
secondly, the Dual Water method. An earlier, but overlooked method that applies the
Waxman-Smits approach utilizing logs only was published by Thomas and Haley.1
But to apply this concept, we should understand its genesis.
A MODEL TO EXPLAIN CONDUCTIVITY OF SHALY SANDS
When Waxman and Smits presented their shaly sand conductivity model at the
Annual Fall SPE Meeting in 1966, it was but one in a long string of models that had
been proposed to explain the curious resistivity behavior of shaly sands first brought
to our attention by Gus Archie in the late 1930s. The names of the earlier model
authors were 1)Patnode & Wyllie, 2)de Witte, 3)Winsauer & McCardell, 4)Poupon,
Loy and Tixier, 5)Hill & Millburn, and 6)Simandoux. All of these workers
recognized the problem and attempted a solution in a slightly different way and each
built on previous work and on advancements in the industry and science in general.
All were trying to provide help to the engineer interpreting resistivity logs and thus
all fell into the same trap. They were used to seeing resistivity logs and thus
expressed all the data in terms of resistivity that resulted in some awkward log-log or
exponential-log equations, particularly since they had to be solved on slide rules.
Many of the authors had pity on the engineers and presented graphical solutions. It
took two physical chemists to look at the problem a different way and present the
resistivity data as conductivity. (Historical note: Induction logs actually measure
conductivity, but it is converted into resistivity to be presented so that it could easily

be compared to the older electric logs and guard logs whose data were recorded as
resistivity.) When the resistivity behavior of rocks is looked at in terms of
conductivities, the simple relationship is obvious. To demonstrate this let me show
you the relationship presented by Waxman and Smits.
Monroe Waxman working at the Shell Development research labs in Houston and
Lambert Smits working at the Royal Dutch/Shell research labs in Rijswijk, The
Netherlands painstakingly made conductivity measurements on hundreds of small
core samples in the laboratory. The Smits samples are referred to as Group I samples,
while the Waxman set of cores are referred to as Group II samples. Each core was
carefully cleaned and characterized as to porosity, permeability, grain density and
cation exchange capacity. The Group I samples used membrane potential to
determine cation exchange capacity, while Group II used a wet chemistry method to
determine cation exchange capacity. Each core was vacuum saturated with brine and
its conductivity monitored until stable. This process was repeated with different
salinity brines ranging from very dilute to near saturated. All measurements were
carried out at 77 F. Waxman and Smits were building upon the earlier work of Hill
and Milburn that had conclusively demonstrated that resistivity and SP behavior of
shaly sands could be explained by accounting for the effect of cation exchange
capacity. Unfortunately, Hill and Milburn never quantified the exchange capacity
parameter independently, and just fit the resistivity data with a term called b and fit
b to core data, all empirically; the resulting equations were cumbersome, e.g.,
R0 (100 Rw ) b log(100 Rw ) FR 0.01 Rw

Eq. (1)

and
CEC
0.0055
V p

b 0.135

Eq. (2)

When the conductivity of the 100% brine-filled core, C0 , is plotted vs. the
conductivity of the brine in the pores of the core, Cw , most of the data falls on a linear
trend. This trend is shown below.

Figure 1. Shaly Sand Core Conductivity Behavior, after Waxman & Smits
When the data are viewed in this format, the relationships are evident. Lets discuss
some important points. Firstly, the dotted line, labeled Clean Sand shows the
behavior that one would expect for rock without any extra conductivity paths. It is
the classical case that obeys Archies relationship. All the conductivity of the rock is
due to the conductivity of the hydrated ions in the brine, thus when Cw goes to zero,
C0 also goes to zero. Interestingly, if we had plotted these data as resistivities, then it
would take a log-log plot to make the data appear linear. Thus for the dotted line, we
can write Archies equations:
C0

1
Cw
F

Eq. (3)

where 1/F is the slope of the dotted line, and

Eq. (4)
where is the fractional porosity and m is a fitting parameter.
F m

Secondly, the solid line, labeled Shaly Sand shows higher core conductivity at any
given value of brine conductivity than the Clean Sand. Thus there is some
mechanism causing increased conductivity other than hydrated ions in the brine.
Thirdly, the Shaly Sand line has a linear portion at high values of brine
conductivity, the saltier part, but exhibits non-linear behavior in the low values of
brine conductivity, the fresher part. Also, even when the brine conductivity goes to
zero, the core has measurable conductivity. Interestingly, when we dry out the core, it

does not have any conductivity. Thus the mechanism that explains the observed
increased core conductivity requires water molecules.
Fourthly, the non-linear part of the curve, indicates that the increased conductivity
depends upon the brine salinity in a non-linear fashion, and as the salinity of the brine
increases, the increased conductivity reaches a maximum amount then becomes
independent of the brine salinity.
Thus any shaly sand conductivity model must explain all of these observations to be
credible. Monroe and Lambert also required that all terms in their equations be
measurable in the laboratory and that they would limit their relationship to one
additional adjustable parameter beyond what Archie had used. Waxman and Smits
proposed the following explanation:

C0

1
C w Ce
F*

Eq. (5)

where Ce is the conductivity resulting from the clay exchange ions that are producing
the increased conductivity, and 1/F* is the slope of the linear portion of the C0 vs.
Cw plot. Monroe chose to designate the Archie F with an asterisk to signify that the
observed slope has a contribution of clay exchange ions; the slopes of the two lines,
constructed one measurement at a time, are not parallel, i.e., F F*, except at the
unrealistic limit of infinite brine conductivity. (NOTE: Monroe chose the * since it
was the only unused symbol left on his typewriter. To honor Monroe, I always use
the asterisk in typewriter position rather than make it a superscript.) The fact that
1/F* is applied equally to the brine ions and the clay mineral counterions explicitly
assumes that the current path (cell constant) is the same for the two conductivity
components.
It is obvious from Figure 1 above that Ce is complex in its behavior, starting out at a
low value and exponentially growing to a maximum value. The maximum value can
be computed from Eq. (5), i.e., when C0 = 0 , then Ce = -Cw , and we can see this
graphical construction in Figure 1; we extrapolate the slope of the linear part of the
line back to the X-axis, which is -Cw , and is therefore , Ce(Max). Since each core
has a different amount and type of clay minerals, each core has a different value of
Ce(Max). Now the only problem remaining is how to determine Ce in the dilute
region when Ce < Ce(Max).
From basic electrochemical behavior, we can decompose Ce , a conductivity, into two
parts, an equivalent conductance times a concentration, or
C e BQv

where

Eq. (6)

B = ionic equivalent conductance of the exchange cation, (Mhos/Meter)/

(Eq/L),
Qv = concentration of exchange cations per unit pore volume, Eq/L .
Since we can determine the value of Qv experimentally for each core plug in the
sample set, we have one adjustable parameter, B that we can fit to the data set. Thus
the first parameter that can be computed is Bmax [since we know both Qv and
Ce(Max)]. For the Group I samples, Bmax = 4.6, while for the Group II samples
Bmax = 3.8. For many reasons explained in the original paper, the Group II samples
are believed to be more accurate.
When the data are presented thusly, F* is temperature independent, because the
temperature dependence is carried by Rw for the brine and B for the exchange ions.
When all the dilute region data are regressed, an empirical relationship for the
exponential rise of B is developed:

B 1 0.8 e 50Cw 3.83

Eq. (7)

at 77 F, for the Na+ .

Monroe Waxman continued to work on gathering laboratory data on water-bearing
shaly sands and extended his measurements to 200 C (392 F ) to satisfy the needs
of engineers solving field problems at elevated temperature. These elevated
temperature results were published in a joint article with E. C. Thomas in 1974 and
the resulting B vs. Rw vs. Temp for Na+ is shown below as Figure 2. (NOTE: Monroe
refit all his experimental data after he discovered an inconsistency and Figure 2
below is the reworked data set, published September 2008.)

Figure 2. Waxman-Thomas B vs. Rw vs. Temperature for the Na+

Notice the large range of B as a function of temperature it is almost one to one, i.e.,
Bmax increases 7-fold during an 8-fold Celsius temperature increase. Also notice
how rapidly Bmax falls off in the dilute region, where Na+ concentration is less than
+ 0.5 Normal ( thats roughly 28 kppm Na Cl ). Also notice that Monroe showed the
data in terms of Rw not Cw as recognition that the field engineers would have to
convert Cw to Rw for their use and he wanted to save them time. From time to time,
various individuals have developed a math model for the data in Figure 2 for use in
digital computing. Arthur Purpich fit the data and his equations are used in Techlog
as the default value.
Lastly note that the term B now contains all the statistical uncertainty of the directly
measured parameters: Cw , Co , T , and Qv . Qv also has statistical uncertainty of its
sub-components: grain density, , and cation exchange capacity. Monroe never
published the regression errors, but they are substantial. Do not be lulled into
believing that we know the value of B better than 20% at temperatures above room
temperature.
Monroe always felt he met his objectives in producing an equation with the minimum
number of parameters and that the parameters were verifiable in the lab. Once the
form of the equation was proven to work, then the linear portion of the curves ( high
salinity part ) could be refit with two independent parameters, i.e., since Cw and C0 are

measured, we can solve for two unknowns; the slope of the line is 1/F* and the
negative X-axis intercept is BQv, so we can determine B and Qv independently of any
laboratory wet chemistry method for CEC. Comparison of Qv by the two methods
showed excellent agreement in the Group II samples; this was not unexpected since
they were carefully chosen to be homogeneous. The agreement between methods for
the Group I samples was poorer due to the heterogeneous nature of many of the very
shaly samples.
Monroe then extended the above relationship in the same form as did Archie, and set

Cw B Qv
F * m*
C0

Eq. (8)

Since C0 and Cw do not have the same temperature dependence, when F* is presented
as Eq. (8) it is temperature dependent, and therefore so is m*. When one uses
laboratory core measurements to determine the value of m*, then optimally they
should be done at reservoir temperature using the same brine concentration and ionic
make-up as the reservoir.
Of course, since we are really interested in quantifying and producing oil reserves,
these water-bearing equations are not sufficient. They must be extended to include
the case of hydrocarbon-bearing reservoirs. Monroe reluctantly agreed that an
additional parameter would be needed to account for the effect of the oil saturation,
but he reasoned that he could just manipulate the existing set of parameters to do the
trick and not have to solve for another parameter other than oil saturation (water
saturation, actually). Monroes reasoning is as follows: The hydrated clay counterions
are in rapid exchange with the hydrated brine ions, thus the effective charge
distribution is spread out over the entire pore network, so Qv is defined as CEC per
unit pore volume. When oil in introduced into the pore network, it displaces the pore
brine (water and hydrated ions), but not the hydrated clay counterions; we must
maintain charge neutrality. Thus as oil volume increases, pore brine volume
decreases and there are fewer hydrated brine ions left which can participate in
exchange reactions with the hydrated clay counterions. Thus the effective
concentration of hydrated clay counterions goes up because the pore water volume
goes down, i.e., since Qv = CEC/PV, when PV decreases, Qv goes up. The question
then becomes, what is the functional form of the change in volume. Monroe in
classical form opted to try the simplest form, inversely proportional to water
saturation (as for relative permeability), and he proposed the following equation:
Qv'

Qv
Sw

Eq. (9)

where Qv is the new effective value of Qv when hydrocarbons are present, and Sw is
the fraction of pore volume value of water saturation.

Monroe then used the same formulation as Archie, so lets review Archies equations
for Clean Rocks:
F

R0
C
w
Rw
C0

Eq. (10)

where F is defined by the relationship in Eq. (10) and R is resistivity, C is

conductivity, the subscript 0 is for zero oil in the core or 100 % water saturation, and
w is for water (brine).
When we now allow for the case that there may be some hydrocarbon in the pore
space, then the expression becomes:
G

Rt
C
w
Rw
Ct

Eq. (11)

where the subscript t follows the field convention for true resistivity, and since Rt
and Ct can vary with hydrocarbon saturation, G is not a constant as is F.
Archie then defined the Resistivity Index:
I

Rt
C
0
R0
Ct

Eq. (12)

If we substitute Eq. (10) and Eq. (11) into Eq. (12), we get
I

GR w
G

FRw
F

Eq. (13)

Archie then related the Resistivity Index, I, to water saturation:

I S w n

Eq. (14)

Combining Eq. (13) and Eq. (14), we get the following:

I

G
S w n
F

and thus,

G FS w n

Eq. (15)

Monroe then reasoned by analogy for a shaly sand, the following:

G * F * S w n*

Eq. (16)

where the asterisk is added to each term to signify that there is some contribution
from hydrated clay counterions in the measurement.
Then Monroe rewrote Eq. (11) to account for hydrated clay counterion conductivity:

Ct

Qv
1
C

B
w

G*
Sw

Eq. (17)

Substituting Eq. (16) into Eq. (17) yields

Ct

1
F * S w n*

Cw B

Qv

Sw

Eq. (18)

If we now convert conductivity into resistivity and solve for Sw, we get

S wn*

Rt
Qv
1 Rw B
F * Rw
Sw

Eq. (19)

Then if we substitute Eq. (8) into Eq. (19) and simplify, we get:

S wn*

Rw

Q
m* Rt 1 Rw B v
Sw

Eq. (20)

This is the equation that field engineers need to be able to solve. There are a lot of
parameters that one needs to know: n* , m* , B , Qv , Rw, , and Rt. Standard Archie
analysis requires five of the seven listed parameters, and only B and Qv have been
added. The value of B is available if one knows temperature and brine resistivity
(Figure 2), thus only Qv must be deduced by other means. A Qv log does not
presently exist, thus numerous local correlations have been developed between clay
sensitive parameters and Qv ; none are universally applicable. It is very much a
reservoir-by-reservoir quest to be able to correlate Qv to some measurable log
parameter. This need for knowing Qv led to self-consistent methods for determining
BQv from logs alone, i.e., Thomas-Haley method, later re-invented by Juhasz as the
Normalized Qv method.

How do I know when it is important enough to go to all the trouble to make a

Waxman-Smits correction? One simple screening method is based upon Eq. (20).
The comparable Archie Clean Rock equation is
S wn

Rw
m Rt

Eq. (21)

Thus when one compares term for term Eq. (20) and Eq. (21) one sees that they are
the same except for a multiplier in the denominator:

Rw B Qv / S w

and when the term,

Rw B Qv / S w 0.1

then the value of Rt has been boosted by more than 10% and Sw is reduced by about
1%, assuming n* is near 2, or Sw is reduced by about 2% if n* is near 1.7. I do not
bother with a Waxman-Smits correction unless Rw B Qv/Sw > 0.1 .
Please notice the clay conduction correction effect is a function of saturation, so the
correction effect is larger the higher one goes up in the hydrocarbon column.
The fact that Eq. (20) is transcendental and one cannot solve for Sw in closed form
should not be a deterrent to anyone with the ability to program a simple iterative
loop; these programs usually converge in 4 or 5 iterative steps taking less than a
second on even the smallest computer. Excel spreadsheet is quite adequate.
References
1. Thomas, E.C. and Haley, R.A. (1977) Log Derived Shale Distribution in
Sandstone and Its Effect Upon Porosity, Water Saturation and Permeability," Trans.,
Canadian Well Logging Society (1977) Paper N.
2. Waxman, M. H. and Smits, L. J. M.:"Electrical Conductivities in Oil-Bearing Shaly
Sands," Soc. Pet. Eng. J. (June 1968) 107-22; Trans., AIME, 243.
3. Waxman, M. H. and Thomas, E. C.:"Electrical Conductivities in Shaly Sands - I.
The Relation Between Hydrocarbon Saturation and Resistivity Index; II. The
Temperature Coefficient of Electrical Conductivity," J. Pet. Tech. (Feb. 1974) 213-23;
Trans., AIME, 257.

4. Waxman, M. H. and Thomas, E. C.:"Electrical Conductivities in Shaly Sands - I.

The Relation Between Hydrocarbon Saturation and Resistivity Index; II. The
Temperature Coefficient of Electrical Conductivity," SPEJ. (Sept. 2007).
5. Jin, G., Torres-Verdin, C., Devarajan, S, Toumelin, E., and Thomas, E.C. (2007)
Pore-Scale Analysis of the Waxman-Smits Shaly-Sand Conductivity Model,
Petrophysics, 48; 2, P 104-120
Submitted by Dr. E. C. Thomas

March 2010