You are on page 1of 14

Journal of Applied Geophysics, 30 ( 1993 ) 215-228 215

Elsevier Science Publishers B.V., Amsterdam

The uses and abuses of the Archie equations,


1" The formation factor-porosity relationship

Paul F. Worthington
Consultant, Ascot, SL5 9HP, UK
(Received March 16, 1992; accepted June 9, 1992)

ABSTRACT

Worthington, P.F., 1993. The uses and abuses of the Archie equations, 1: The formation factor-porosity relationship. J.
Appl. Geophys., 30:215-228.

Model studies of the first Archie equation, which relates the formation resistivity factor and porosity specifically for
effectively clean (shale-free) porous media, have quantified the changes to this relationship caused by latent shale effects.
The scales of these departures correspond to the ranges reported by experimental investigators on the basis of empirical
studies of supposedly clean formations, for which some of the experimental data would actually be physically untenable.
Thus, many of the published versions of this equation do not, as is claimed, definitively relate porosity to an intrinsic
formation factor but merely to an apparent formation factor, which shows an incidental relationship to porosity that is
salinity-dependent and can even become inverted.
The primary implication of departures from clean conditions, illustrated using field data from reservoirs with varying
degrees of shaliness, is that the relationship between (apparent) formation factor and porosity can remain well defined
even though the numerical form changes significantly with variations in salinity. The inadvertent use of such an arbitrary
relationship to estimate formation factor from porosity can transmit errors of over 30 saturation units to the subsequent
determination of water (and thence hydrocarbon ) saturation.
On this basis, the "standard" practice of determining the form of the first Archie equation from measurements of"tbr-
marion factor" that use simulated formation water must be viewed with extreme caution. There is no guarantee that a
well-defined linear trend on a bilogarithmic scale actually reflects effectively clean conditions. This is contrary to conven-
tional thinking, which predicts a curvilinear trend where shale effects are manifest. Apart from the demonstrable effect on
the determination of hydrocarbon saturation, these observations also have a direct bearing on the concept of reservoir
zonation and on evaluation approaches that use the Archie exponent as an indicator of lithology, pore geometry, degree of
induration, or permeability.

Introduction tionship provided a vital means of estimating


water-zone porosity from a downhole mea-
surement of an electrical parameter. More re-
The relationship between the electrical
cently this role has been reversed and the cur-
properties of a reservoir rock and its intercon-
rent practice is to use porosity-tool response to
nected porosity has always been fundamental
evaluate specified electrical parameters which
to quantitative formation evaluation in the pe-
are than used in the determination of water
troleum industry. Prior to the development of
saturation. In this reverse mode the relation-
radiometric and sonic logging tools, this rela-
ship is applicable to both water and hydrocar-
Correspondence to: Dr. P.F. Worthington, Consultant in
bon zones.
Reservoir Evaluation, Charfield House, Acot, Berkshire, Sundberg ( 1932 ) postulated a relationship
SL5 9HP, UK. between interconnected porosity ~ and a "re-

0926-9851/93/$06.00 © 1993 Elsevier Science Publishers B.V. All rights reserved.


216 P.F. WORTHINGTON

tOO0
sistivity factor", defined as the ratio of the re-
sistivity Ro of a fully-saturated granular reser-
voir rock to the resistivity Rw of the saturating
electrolyte. Archie (1942) presented strong
F= I
empirical evidence in support of this relation- ¥.7
ship but without reference to Sundberg. Archie
restyled the resistivity ratio as the "formation tO0

resistivity factor" or formation factor F be-

I
cause it did not vary with electrolyte salinity
and therefore was seen to be determined by
formation properties rather than by fluid char-
acteristics. Archie proposed that:
1
F = -¢m
- (1)

where m has been termed the "cementation


exponent" for the reservoir in question. Ar-
chie further noted that m = 2. It is Archie's ter-
minology that has been universally adopted.*
Equation ( 1 ) has been called the "first Archie 0.01 0 1 10
equation". Equation ( 1 ) has usually been es-
tablished through a regression analysis of an F
vs ~ bilogarithmic crossplot for a statistically Fig. 1. Bilogarithmic crossplot of formation factor vs po-
significant population of core data. An exam- rosity for data from a North Sea field, with a regression
fit according to eq. ( 1 ): data from Hook (1983).
ple of such a plot is shown in Fig. 1, where the
value of m departs from the Archie specifica-
tion. If the sample population is representative eq. ( 1 ) but it differs in that m is a definition
of a reservoir unit, and the degree of correla- rather than the product of a regression analysis.
tion is high, the resulting value of m can be ap- In establishing the empirical eq. ( 1 ) Archie
plied throughout that unit. I adopt the conven- assumed that his suite of samples were effec-
tion of a numerically positive m when the tively clean, i.e. virtually free of clay minerals
gradient on a log F vs. log ~ crossplot is nega- or shale (these terms are used synonymously
tive. Thus a "negative m" describes a positive here), and he used a high-salinity brine that
gradient. would suppress the electrical manifestation of
Some recent investigators have preferred a any minor shaly constituents. For these con-
variable-m approach whereby a single-point m ditions, eq. ( 1 ) has been supported theoreti-
is calculated from the equation: cally but only for unconsolidated sands (Ac-
cerboni, 1970), a paradoxical situation in view
log F
m. . . . (2) of the early physical interpretation of the ex-
log ponent m. The term "cementation exponent"
Equation (2) is mathematically equivalent to has subsequently lost much of its credibility,
with later investigators preferring to interpret
*In 1970, almost by way of a belated apology to Karl m as a grain-shape (Jackson et al., 1978) and/
Sundberg on behalf of the petrophysics community, the
Society of Professional Well Log Analysts reprinted Sund-
or pore-shape (Ransom, 1984) indicator.
berg's original paper in their house journal The Log Since 1942 the formation factor-porosity
Analyst. relationship ha been confirmed empirically in
USES AND ABUSESOF ARCHIE EQUATIONS 217

numerous formations. However, in many cases mental conditions. Therefore, unless shale ef-
the reported equation has taken the form fects can be properly accounted for, no mean-
ingful assessment is possible of the other
a
F- (3) causative factors.
~b'n
This paper re-examines the formation fac-
where the coefficient a is regarded as a reser- tor-porosity relationship with particular ref-
voir constant that can depart from unity (Car- erence to the influence of clay-mineral content
upon the quantities a and m. The object is to
others, 1968; Porter and Carothers, 1970). Al-
identify in the context of shale effects possible
though eq. (3) does not satisfy the obvious
reasons for the marked variations in these pa-
boundary condition that F ~ 1 as ~ 1, it is ar-
rameters that have been reported in the litera-
gued to be an empirical relationship that is
valid only over the range of porosity repre- ture. It is hoped that this treatment will lead to
a better appreciation of the physical signifi-
sented by the data points. Thus the boundary
cance of this most fundamental interpretative
condition can be seen as irrelevant. On this ba-
equation.
sis eq. (3) has come to be regarded as a gener-
alised form of eq. ( 1 ).
The quantities a and m have been reported Model studies
to vary widely for different formations. The re-
ported ranges are exemplified in Table 1, which
is based upon separate compilations of differ- In order that a consistent analysis might be
ent investigators. A summary of the reasons for made of the influence of shaliness upon the pa-
such variations has been given by T i m u r et al. rameters a and m, an idealised petrophysical
model has been specified for a hypothetical
(1972) as follows: degree of cementation;
reservoir rock. In this way the related "data"
shape, sorting and packing of grains; type of
porosity; tortuosity; pressure; wettability of are free of observational errors and can be rep-
resented by exact functions.
rock surface; pore geometry; and clay content.
The last item is especially significant because
it directly infringes Archie's original experi- The model
TABLE1
(i) Formation samples are fully saturated
Reported ranges of the Archie exponent a and coefficient m with an NaCl-electrolyte of resistivity Rw
Lithology a m Investigator (s)
and maintained at a temperature of
25 °C.
Sandstones 0.47 -1.8 1.64-2.23 Hilland Milburn
(ii) Where the Archie assumptions are sat-
(1956)
0.62 -1.65 1.3 -2.15 Carothers(1968) isfied and the formation behaves as a
1.0 -4.0 0.57-1.85 Porter and Carothers clean sand, the ratio Ro/Rw defines the
(1970) intrinsic formation factor F, which is re-
0.48 -4.31 1.2 -2.21 Timuretal.(1972)
0.004-17.7 0.02-5.67 Gornez-Rivero lated to effective porosity ~ through eq.
(1976) (1).
(iii) Where the Archie assumptions are not
Carbonates 0.73 -2.3 1.64-2.10 Hill and Milburn
(1956)
satisfied and the formation does not be-
0.45 -1.25 1.78-2.38 Carothers (1968) have as a clean sand, the ratio Ro/Rw
0.33-78.0 0.39-2.63 Gornez-Rivero merely defines an apparent formation
(1976)
factor Fa, which varies with Rw and is
0.35 -0.8 1.7 -2.3 Schon (1983)
not related to porosity through eq. ( 1 ).
218 P.F. WORTHINGTON

(iv) Where the formation shows shale ef- Computations


fects, the quantities Fa and F are related
through the equation of Waxman and Values of F and Qv have been calculated for
Smits ( 1968): each of 14 logarithmically-incremental levels of
effective porosity over the stated range of 0,
Fa=Ro/Rw=F(I+BQvRw) -I (4)
through eqs. ( 1 ) and (6), respectively. The re-
where Q~ is the cation-exchange capac- suiting formation factors vary from 8.16 at high
ity (CEC) per unit pore volume of the porosity to 816 at the lower limit of 0. The
rock in units of equiv litre- l, an electri- quantity Qv occupies the range 0.005-2.69
cal measure of shaliness, and B is the equiv litre- ~. Thus, in accordance with the de-
equivalent conductance of (sodium) scriptive classification of Waxman and Smits
clay-exchange cations expressed (Wax- ( 1968 ), the hypothetical formation is clean at
man and Thomas, 1974) as a function high porosities and very shaly (high clay-min-
of Rw by the equation: eral content) at low porosities. It is therefore
ideal for studying the effect of shaliness upon
B=3.83[1-O.83exp(-O.5/Rw)] (5)
the formation factor-porosity relationship.
The units of B are mho m - ~litre equiv- l For each pair of values o f F and Q~ the ratio
so that the product BQ~has units of con- Ro/Rw has been computed for each of seven
ductivity. The composite term BQJF is selected values of electrolyte resistivity within
sometimes called the excess conductiv- the range 0 . 0 5 < R w < 5 . 0 0 m . This range
ity of the rock, because it represents the broadly corresponds to that adopted by Wax-
effects of surface conduction, primarily man and Smits (1968). From earlier com-
due to clay minerals, which imparts to ments it might be expected that the ratio Ro/
the rock a conductivity over and above Rw will correspond to F at high porosities where
that due to bulk electrolytic conduction Qv is small, but will represent merely an appar-
within the free pore space. ent formation factor F~ at low porosities where
(v) The formation shows a porosity range of high Q~ values are indicative of shaly
0.035-0.35. conditions.
(vi) The quantity Qv varies with porosity ac- These expectations are validated in Fig. 2,
cording to the equation: which compares the bilogarithmic plot o f F vs
0 with those of Ro/Rw vs 0 for three specified
log Q~ = - 3 . 5 6 - 2.74 log 0 (6)
levels of Rw in accordance with our model. It
Equation (6) was established through is apparent that all four plots converge at higher
the realistic specifications that porosities. This means, in this case, that pro-
Qv=0.005, 1.0 equiv litre -1 when vided Rw < 5 I2rn, the hypothetical formation
0=0.35, 0.05, respectively. The postu- will be satisfactorily represented at high poros-
late of an inverse relationship between ities (i.e. greater than about 0.2) by the
Q~ and 0 would be valid even for a con- equation:
stant cation exchange capacity within the
Ro 1
hypothetical reservoir since the refer-
Rw - 0 2 (7)
ence of CEC to porosity within Q~ means
that Q~ would still increase as 0 de- Equation (7) defines a "clean-sand line", of
creased. Equation (6) is broadly com- gradient 2, on the bilogarithmic plot of Fig. 2.
patible with reported data trends (Lav- As porosity is decreased the Ro/Rw curves de-
ers et al., 1974; Juhasz, 1979; Hook, part from the clean-sand line: the ratio Ro/Rw
1983). now represents an apparent formation factor
USES AND ABUSES OF ARCHIE EQUATIONS 219

Qv
1000
° e0,,ro'
O.05(~m F vs 0

0.5C~m ! 10

100-

t
5.0/~m
\
0
12O" 01 I z, 'o;
2

uS
Fig. 3, Variation ofFa/Fwith Rwfor four specified values
10- of Qv (in equiv litre- ~), according to the hypothetical pe-
\ trophysical model.
\
\\
\
\
\ 20
xx
\
\
\
x
\
I 1
0.01 011 10

Fig. 2. Bilogarithmic plots of (i) F vs ~ and (ii) Ro/R, vs


1.0
¢~for three specified values of Rw (in ~2m), according to
the hypothetical petrophysical model.

Fa. The departures are greater for higher val-


ues of Rw. Furthermore, the curves are evi-
dently not monotonic. The variable gradient of
an Fa vs ~ bilogarithmic plot is here designated 0.0

ma. This quantity is equal to m only where ef-


fectively clean conditions prevail, i.e. on the
clean-sand line which, for our particular model,
is attained only at higher porosities.
The features of Fig. 2 can be more fully ap-
~1.0
preciated by reference to Fig. 3 which provides 0.01
I
0.1
q
1.0
a graphical representation of eq. (4). This chart
depicts four plots of the ratio Fa/F as ordinate q l l l I
against Rw as abscissa, each of which corre- 100 I 0 01 001 0001

sponds to a specified Qv. When Fa/F<< 1 the Qv (equiv litre ~)

formation shows shale effects. For practical Fig. 4. Variation of apparent Archie exponent rna With
purposes the limiting requirement for clean and Qv for Rw = 5.012m, according to the hypothetical pe-
trophysical model.
conditions might be specified as Fa/F> 0.90.
Figure 3 primarily illustrates that for a given
Q , the manifestation of shale effects is con- is determined by Qv. Using these complemen-
trolled by Rw. This observation complements tary figures it is possible to ascertain whether
Fig. 2 which principally indicated that for a any given pair of values of Rw and Qv will give
constant Rw, the manifestation of shale effects rise to shale effects in the electrical sense. These
220 P.F. wORTHINGTON

exact form of this function and by the resistiv-


ity of the fully saturating electrolyte. This phe-
nomenon is not an analytical artefact. Indeed,
the same pattern emerges using alternative
forms of the Q~ vs ~ relationship, e.g.:
oo c1 lO t30

Qv=~-q (8)
Fig. 5. Variation of apparent Archie exponent m~ with Rw
for 0= 0.085 and Qv= 0.24 equiv litre- ~, according to the when p and q are reservoir constants (Juhasz,
hypothetical petrophysical model. 1981).
The following physical explanation of Fig. 2
comments do, of course, relate specifically to relates to a fixed value of Rw. For sufficiently
conditions of full water saturation. high porosities and therefore, according to our
Thus far, it is apparent that the quantity m~ model, sufficiently low values of Q~, the plot of
can assume both positive and negative values Ro/Rw conforms to the Archie relationship to
that correspond to negative and positive gra- within the specified tolerance f 10%. As q~de-
dients, respectively, in Fig. 2. The variation of creases and Q~ increases, shale effects become
analytically-determined values of m~ with ef- increasingly significant and the ratio Ro/Rw
fective porosity, and therefore (for this model) gradually departs from its asymptotic clean-
with Q~, is shown in Fig. 4 for the particular sand level (F). As these departures increase,
value ofRw= 5 ~ n . It can be seen that m~ var- the negative gradient becomes shallower. Dur-
ies from - 0 . 8 to 1.94 for the adopted range of ing this phase the predominant influence upon
O and that m~ = 0 when ~ = 0.061, the so-called Fa is still the clean-sand term ofeq. (4). While
iso-conductance point at which the resistivity these conditions prevail, the gradient remains
of the fully-electrolyte-saturated (model) po- negative. However, at some sufficiently low
rous system is equal to that of the saturating porosity, and therefore sufficiently high Qv, the
electrolyte (e.g. Sauer et al., 1955). Figure 5 shale term starts to predominate. This causes
depicts the variation of ma with Rw for the the gradient to become even shallower and then
specified porosity of q~=0.085 which, through to reverse. In the region of positive gradient the
our model, corresponds to Q~=0.24 equiv conductivity is associated primarily with the
litre- 1. It is apparent that m~ varies from 0.57 ionic double layer at the mineral-electrolyte
to 1.89 over the range 0.05 <R5 E2rn and, for contact rather than with the free electrolyte of
this example, will not attain negative values resistivity Rw. The amount of conduction
until Rw attains some value above 10 g2m. through the clay-exchange counterions in the
double layer is a function of Q~. As Q~ in-
Discussion creases the double-layer conductivity in-
creases and this manifests itself as a reduction
Previous illustrations of hypothetical F~ vs in Ro. Because this effect now predominates
plots have taken the form of monotonic curves, over the opposing consequence of a decreasing
which progressively depart from an appropri- and thence an increasing F giving rise to a
ate form of the Archie relationship as Rw is in- higher formation resistivity, the overall effect
creased (e.g. Patchett and Rausch, 1967; Bus- is to reduce Fa.
sian, 1983 ). However, by specifying Q~ to be a With the exception of the negative values of
realistic function of porosity, Fa vs 0 plots have ma, which are given separate consideration, the
been generated which pass through a maxi- range of ma (Figs. 4 and 5) is consistent with
mum at some low value of ~ determined by the the range of values of " m " compiled by other
USES AND ABUSES OF ARCHIE EQUATIONS 221

investigators (Table 1 ) notwithstanding the normally be subjected to a detailed petrophys-


cut-off at m = 2 imposed by the idealised pe- ical evaluation unless secondary permeability
trophysical model. Since the theoretical lower was present. These observations are supported
limit of m is unity for a minimum-tortuosity by theoretical studies of a cubic fracture model
system of parallel capillaries, and since both ma which predicts that m > l . 1 (Towle, 1962;
and reported values o f m (Table 1 ) have taken Aguilera, 1974), by the modelling of conduc-
values less than unity, this raises the obvious tor networks which furnished m-~ 2 (Green-
question as to what extent those reportedly low berg and Brace, 1969 ), and by the verification
m-values reflect shale effects that were not re- of Archie's law for fractured systems (Brace et
cognised at the time. In this respect it is worth al., 1965) and for intergranular porosities as
noting that formation samples for which low as 0.001 (Brace and Orange, 1968).
Q,,>0.5 equiv litre-l might never behave as
electrically clean under ambient conditions. Field studies
Even when saturated with an electrolyte of sal-
inity in excess of 250,000 ppm (Rw=0.04 £2m Figure 2 has illustrated that an Fa vs ~ plot
at 25 °C) there can still be insufficient ions in shows progressively greater departures from
the free pore water to suppress the ionic dou- the F vs ~ (clean-sand) relationship as Rw in-
ble-layer conduction at the mineral surface (cf. creases. When using idealized data that repre-
Fig. 3). It is also worth noting that if Qv=0.1 sent wide ranges of ~ and Q~,and that can be
equiv litre -~, effectively clean conditions can described exactly by simple functions, it is easy
only be assured at 25°C ifRw<0.1 g-2m (cf. Fig. to diagnose the electrical behaviour of a given
3 ). Thus, even for such a slightly shaly forma- reservoir from single-salinity measurements of
tion, insignificant shale effects can only be pre- Ro/Rw. However, in a real situation the ranges
sumed if the NaC1 salinity is greater than of ~ and Qv are comparatively low, significant
70,000 ppm. In the light of these examples it is observational errors are associated with the ra-
perhaps inevitable that the widespread prac- tio Ro/Rw, and F and Qv are not exact func-
tice of using simulated formation water for the tions of ~ because of lithological and pore-ge-
laboratory measurement of "formation fac- ometry effects. Under these conditions it can
tor" has given rise to departures from Archie's be extremely difficult to decide whether a given
conditions and that these are reflected to some bilogarithmic plot of Ro/Rw vs ~ conforms to
extent in Table 1. The implications of these de- a truly linear trend as per the Archie equation
partures will be examined later in the context or whether the scatter of data points is con-
of real data. cealing a gentle curve induced by shale effects.
Negative values o f " m " reported for low-po- The implications of this ambiguity are now ex-
rosity sands have been attributed to fracturing amined with reference to selected field
(Gomez-Rivero, 1977, 1980). However, Fig. examples.
2 indicates that negative values of m~ can be
generated for low porosity, shaly sands with- Example I
out introducing fractures. In the light of the
earlier comment that m-- 1 for a parallel cap- Figure 6a depicts the variation of intrinsic
illary model, it is possible that the reportedly formation factor F with porosity for 31 sam-
negative values of " m " are, in fact, apparent ples of diverse sandstones from England and
quantities m~ and are therefore caused by shale Saudi Arabia. Values of F have been calculated
effects. It is presumably coincidental that they from multiple-salinity determinations of F,
have been identified with fracturing since tight (Zaafran, 1981 ) through eq. (4). Regression
formations with low ~ and high Q~ would not analysis has furnished the relationship:
222 P.F. WORTHINGTON

100 50- Rw = 0.18~m is shown in Fig. 6c. Regression analysis has


furnished the expression:
50- 0.93
Fa--01.51 (12)
F Fa
"L
and a correlation coefficient of -0.93. Again,
10-
this is marginally better than that for the F vs.
e, • correlation.
10 5 Figure 6d depicts the plot of Fa vs ~ for
01 0'.s 01 o'5 Rw= 4.5 g?m. It is evident that there is no well-
a) *~ b) *
defined relationship between F~ and ~ at this
50- 10-
electrolyte resistivity.
R w = 1.50[~.m R w = 4.50 A m

• Example 2
5-
° o e** •

Figure 7a shows the variation of F with q~ for


Fa Fa
15 samples of shaly sandstone of the (Eocene)
10- lower Wilcox sand, Sheridan, Texas, as re-
II
ported by Waxman and Smits ( 1968 ). As with
Example 1, the values of F have been calcu-
5-
01 01s

01 015 lated from multiple-salinity measurements of
¢ -----~ 0 > conductivity. The data can be represented by
c) d)
the equation:
Fig. 6. Plots o f ( a ) F v s 0 a n d ( b ) - ( d ) F , vs 0, for sand-
stones from England a n d Saudi Arabia ( d a t a from Zaaf- 0.5
ran, 1981). F:--02.31 (13)

and they show a correlation coefficient of


1.04 -0.99.
F=~2.,--5 (9)
The relationship between Q~ and 0 has taken
with log F and log ~ showing a correlation coef- the form:
ficient of - 0 . 9 2 . The relationship of Qv to 0 log Q~= - 2 . 9 6 - 2 . 7 5 log ~ (14)
has taken the form
with a correlation coefficient of -0.87.
log Qv= - 0 . 6 7 - 0 . 8 8 log 0 (10) Figure 7b depicts an Fa vs ~ plot for Rw = 0.12
g2rn. Regression analysis has resulted in the
with a correlation coefficient of - 0.49.
expression:
Figure 6b shows the variation of F~ with 0
for Rw = 0.18 g2m. The data can be represented 0.59
by the equation: Fa--0zl6 (15)

1.11 with a correlation coefficient of - 0 . 9 9 .


Fa-ol.81 (11) Figure 7c shows the Fa vs 0 plot for Rw=0.47
12m. The data can be represented by the
and show a correlation coefficient of - 0 . 9 4 , equation:
which is numerically greater than that for the
F vs ¢ plot. 0.87
Fa--q~l.89 (16)
The variation of F~ with 0 for Rw= 1.5 g2m
USESAND ABUSESOF ARCHIEEQUATIONS 223

100- 100- Rw = 0 . 1 2 ~ m 100-


Rw =0.47A m

50- 50- 50-

f
F
•%
• m
t
Fa ,%°
t
Fa
% #•

10- 1 10 5"
o.1 0.2 o.1 o'.~ o.~ o's
a) ~ > b) ~~ c) o

Fig. 7. Plots of (a) F v s 0 and (b), (c) Fa vs ¢, for Eocene sandstones from Texas (data from Waxman and Smits, 1968).

with a correlation coefficient of - 0.97. (Triassic) Sherwood Sandstone aquifer of


northwest England. This formation has been
Example 3 reported to be only slightly shaly with a range
of clay content from 0.5-5.2% by weight
The plots of Fig. 8 relate to data from the (Worthington, 1977b). Analysis of multiple-
salinity data for some 350 core samples (Wor-
thington, 1976 ) has resulted in the following F
vs O relationship:
1.1
30-~ " F= ~.161 F--01.61 (17)

By assuming that the Waxman-Smits equa-


tion applies to this formation it can be further
shown through the manipulation of published
relationships (Worthington, 1976) that the
expression relating Qv and 0 takes the form:
lo- B--'ff"~'s'la,
log Qv= - 3 . 5 4 - 2 . 8 5 log ~ ( 18 )
olo %'% \ This expression is similar to eq. ( 6 ) which was
established on the basis of data from elsewhere.
5- 2
It is interesting to compare eq. (17) with
o,/o i those established for individual well sites
~, "-o1~,o
~-O 3 within the study area. For populations of at
least 13 data points (Table 2) the ranges of the
3
0.1
n
0.3 01.5
Archie coefficient and exponent are, respec-
tively, 0 . 2 6 < a < 1.62 and 1.30< m<2.30.
Figure 8 compares the graphical representa-
Fig. 8. Comparisonof F vs ~ relationshipwith plots of F~
vs 0 for values of Rw of (1) 3.8, (2) 7.0, and (3) 13.5 tion ofeq. ( 17 ) with F, vs q~plots for Rw= 3.8,
Qm, for Triassicsandstonesfrom northwestEngland (data 7.0 and 13.5 g2rrl (Worthington, 1977a). Be-
from Worthington, 1977a). cause the samples are only slightly shaly, higher
224 P.F. WORTHINGTON

TABLE 2 There are, however, significant changes in the


Values of a and m for core samples of Sherwood Sandstone
quantities "a" and " m " with Rw for each of
from seven well sites in northwest England Examples 1 and 2. This variation is a conse-
quence of shale effects and it has been ob-
Well No. of a m Correlation
served even for the slightly shaly formation of
samples coefficient
Example 3 with Rw sufficiently high.
1 13 0.66__+0.20 1.81__+0.20 -0.94 A plot of Fa vs ~ might therefore be mistak-
2 26 1.04-+0.21 1.50__+0.12 -0.93
3 14 1.61 -+0.80 1.30__+0.31 -0.78 enly interpreted as an F vs ~ trend. In earlier
4 14 1.62-+0.60 1.41__+0.21 -0.89 years when eq. (3) was applied in the "for-
5 22 0.53-+0.26 1.99__+0.39 -0.77 ward mode", i.e. with F being used to predict
6 16 1.29__+0.46 1.44__+0.46 -0.89
7 13 0.26_+0.15 2.30__+0.33 -0.89
in the water zone, the unwitting deployment
of an Fa vs ~ relationship made little difference
provided that Rw was constant and that the
water resistivities are needed for shale effects correlation remained strong. In that case the
to be manifest. In this aquifer, typical ground- use of F~ in conjunction with pseudo-values of
water resistivities are 20 £2rn (Worthington, "a" and " m " was virtually self-compensating
1972 ). Each data point in Fig. 8 represents an and the predicted ~ was therefore no less use-
average for samples whose porosity lies within ful. However, the current practice of using eq.
a specified incremental range. Because poros- (3) in the "reverse mode", whereby ~ deter-
ity is broadly related to Qv, this approach has mined from porosity-tool response is used as
the effect of ordering the influence of Qv upon input to estimate F, can be invalidated if "a"
Fa. This, in turn, renders the manifestation of and " m " have not been determined for effec-
shale effects more systematic and thereby tively clean conditions. In many cases "a" and
makes it easier to identify data trends. This ap- " m " are calculated from a formation factor-
proach to data refinement has been adopted by porosity plot based upon core analysis involv-
other investigators to good effect (e.g. Caroth- ing a single determination of Ro at full satura-
ers, 1968). tion with simulated formation water of resis-
Figure 8 clearly shows the progressive tivity Rw. It has already been demonstrated that
emergence of a non-monotonic relationship an experimental approach of this kind can fur-
between F~ and 0 as Rw is increased. This is nish merely an F~ vs 0 relationship which, in
particularly interesting because the formation turn, provides estimated values of F, from
is only slightly shaly. Indeed, according to eq. measured porosities. These values of F, might
(18), Q,<0.15 equiv litre -1 for the entire then be used inadvertently as intrinsic forma-
range of porosity represented in Fig. 8. Since it tion factors in subsequent determinations of
is apparent that the linear trend is breaking water saturation Sw, a substitution that can give
down, no regression analysis has been made of rise to substantial error.
the Fa vs 0 plots. As an illustration of the potential pitfalls as-
sociated with this practice, Tables 3 and 4 draw
Discussion upon the results of Examples 1 and 2, respec-
tively, to compare Sw values calculated by us-
The preceding examples show that a bilogar- ing an F v s ~ equation correctly on the one hand
ithmic plot of apparent formation factor and an F~ vs ~ relationship "inadvertently" on
against porosity for sandstone samples can ap- the other. In each case the formation water re-
pear linear with a correlation coefficient which sistivity is assumed to be that to which the Fa
for all practical purposes is numerically as high values relate. In the "belief" that clean condi-
as those for corresponding F vs 0 correlations. tions prevail, the formation factors are input
USES AND ABUSESOF ARCHIE EQUATIONS 225

TABLE3 100-
Example 1 (Zaafran, 1981)
Example2 (Waxman & Smits, 1968)
Influence of latent shale effects upon the predicted water sat-
I ....
uration Swa obtained using Fa with Rw= 1.50m from eq. (12)
relative to the reference water saturation Sw obtained using
80-
the intrinsic eq. (9), from Example 1

Input Parameters Intrinsic System Apparent System

0 Rt 1.04 Sw 0.93 Swa • 60-


(.Qm) F ' - - ¢ 2.13 F a = ~ TM

0.1 585 140 0.6 30.1 0.28


r
0.2 134 32.1 0.6 10.6 0.34 ~,
0.3 56 13.5 0.6 5.7 0.39 =g 40-
g

TABLE 4

Influence of latent shale effects upon the predicted water sat- 20-

i---J
uration Swa obtained using Fa with R,---0.47 g2m from eq.
(16) relative to the reference water saturation Sw obtained
using the intrinsic eq. (13), from Example 2 I .... ]
I
Input Intrinsic Apparent ,
Parameters System System 000 0.03 0.06 0.125 0.25 0.5 1.0 20
Qv ( equiv litre 1 )

0 R, 0.5 Sw 0.87 Swa


(~n) F=(~2.3~ Fa=¢1.89 Fig. 9. Comparison of ranges of Qv for data of Examples 1
and 2.
0.1 133 102 0.6 67.5 0.49
0.2 27 20.6 0.6 18.2 0.56
0.3 11 8.1 0.6 8.5 0.61 erally decreases as porosity increases. This, too,
is a consequence of shale effects since Qv de-
creases as ¢ increases. Note that these dispari-
into the basic (second) Archie equation for ties would have been greater if we had fol-
partially saturated conditions, namely: lowed the practice of allowing n to assume the
=[FR,~] '/" value o f " m " in the prediction of Sw.
Sw L R, d (19) Figure 8 has provided further evidence for a
gradient reversal in Fa vs ¢ plots as was pre-
where n is the saturation exponent. For sim- dicted from the model studies. This trend
plicity R, has been specified to yield Sw= 0.60 might have been especially difficult to recog-
when the intrinsic formation factor is used with nize had these data not been refined, because
the conventional exponent of n = 2. they do not include low porosities which would
It can be seen from Example 1 in Table 3 that presumably correspond to high Q~ values that
the use of an Fa vs ¢ relationship for Rw= 1.5 would have emphasized the negative gradient.
g-2m renders the resulting values of Sw ex- For this same reason high values of Rw have
tremely optimistic. In contrast Example 2 been used in order to accentuate the manifes-
(Table 4) shows a much better agreement for tation of shale effects. Since all these samples
Rw=0.47 g2m. This is a consequence of the were unfractured, the postulate of a fractured
lower electrolyte resistivity and of the lower m e d i u m in order to explain negative values of
range of values ofQv (cf. Fig. 9 ), both of which " m " is again seen to be unnecessary.
lead to reduced shale effects. Tables 3 and 4 There is considerable overlap between the
show that the disparity between Sw values gen- ranges of m and ma reported here. Although
226 P.F. WORTHINGTON

Table 2 indicates that the quantity m can vary first Archie equation, is valid only for effec-
significantly for different sample populations tively clean (i.e. effectively shale-free) forma-
from the same formation, this intrinsic varia- tions and has usually taken the general form of
tion might be confused with that in ma which eq. (3). Widely varying values of a and m have
is due to shale effects. For this reason the pres- been reported in the technical literature. Some
ence of significant shale effects might go un- of these are physically untenable and this sug-
recognized. Clearly, in any correlation exercise gests that the basic assumption of clean condi-
where m is to be interpreted quantitatively in tions has not always been satisfied.
terms of the degree of cementation (Archie, A realistic petrophysical model has been de-
1942; Keller and Frischknecht, 1966), grain signed for a hypothetical reservoir rock in or-
shape (Atkins and Smith, 1961; Jackson et al., der to investigate the effect of shaliness upon
1978), porosity (Sethi, 1979), permeability the first Archie equation. Guided by the tech-
(Raiga-Clemenceau, 1977) and tortuosity nical literature, this model separately specifies
(Jackson et al., 1978), any influence of shale that ~ is an exact function of F o n the one hand
effects must be eliminated for if it is not, " m " and Qv, the cation exchange capacity per unit
becomes a function of salinity (Fig. 5 ). pore volume, on the other. The quantity Q~ is
The r e c o m m e n d e d procedure for deciding an electrical measure of shaliness. Considera-
whether a particular formation has to be ble variations in "a" and " m " can be gener-
treated as shaly in an electrical sense is the ated solely by taking account of shale effects.
multiple-salinity determination of excess con- These ranges correspond to those reported by
ductivity (BQJF) using preserved core. Reli- other investigators on the basis of empirical
able values of excess conductivity can then be studies. The variation in " m " extends below
used to estimate shale effects using a chart such the theoretical m i n i m u m of unity for a clean
as Fig. 3. This estimate will relate specifically sand. This finding supports the belief that
to the water zone. Extension to the hydrocar- many of the published empirical equations do
bon zone follows directly (Worthington, not definitively relate porosity to an intrinsic
1982). formation factor F but merely to an apparent
On a cautionary note, comparison of eq. quantity Fa which shows an arbitrary relation-
( 17 ) with Table 2 suggests that some of the re- ship to porosity.
ported variations of a and m might be due The implications of these departures from
solely to a non-representative sample popula- clean conditions have been illustrated for data
tion. It is, however, noteworthy that this same from three groups of reservoirs with widely
comparison reveals site-specific values of rn varying degrees of shaliness. In these cases the
that are never less than the theoretical value of Fa vs ~ relationship can remain well-defined
m proposed by Accerboni (1970) for uncon- but the associated values of "a" and " m " are
solidated media. This is in contrast to Table 1, not definitive in that they vary with electrolyte
which incorporates values of m that are even salinity. The "inadvertent" use of such a rela-
below the physically-tenable lower limit of tionship to estimate " F " from ~ can result in
unity. erratic errors of over 100% when this value of
" F " is used in the subsequent determination
Conclusions of water saturation.
On this basis the standard practice of deter-
The relationship between formation resis- mining a and m from service measurements of
tivity factor F and interconnected porosity ~ is porosity and "formation factor" using simu-
most fundamental to formation evaluation. lated formation water must be viewed with
The mathematical expression, the so-called caution. There is no guarantee that any result-
USES AND ABUSES OF ARCHIE EQUATIONS 227

ing relationship is independent of shale effects. Research for their support in bringing this for-
This independence can only be verified by mulation of ideas to fruition.
more extensive measurements preferably in the
form of multiple-salinity determinations of References
formation factor.
Attempts to use the quantity m as an indi- Accerboni, E., 1970. Sur la correlation existant entre po-
cator of lithology or pore geometry strictly re- rosit6 et facteur de formation dans les sediments non
quire that this parameter be intrinsic. These consolidrs. Geophys. Prospect., 18:505-515.
exercises are futile if "m" has been obtained Aguilera, R., 1974. Analysis of naturally fractured reser-
voirs from sonic and resistivity logs. J. Pet. Tech., 26:
from an Favs q~plot and can therefore vary with
764-772.
electrolyte composition. Archie, G.E., 1942. The electrical resistivity log as an aid
In particular, it has been proposed by others in determining some reservoir characteristics. Trans.
that negative values of "m" serve to indicate AIME, 146: 54-62.
fracturing. However, data from both field and Atkins, E.R. and Smith, G.H., 1961. The significance of
particle shape in formation resistivity factor-porosity
model studies have shown that negative "m- relationships. J. Pet. Tech., 13: 285-291.
values" can be encountered for unfractured Brace, W.F. and Orange, A.S., 1968. Further studies of
formations when high shale effects prevail. It the effects of pressure on electrical resistivity of rocks.
is considered much more likely that it is inter- J. Geophys. Res., 73: 5407-5420.
Brace, W.F., Orange, A.S. and Madden, T.R., 1965. The
granular shale effects which have caused these effect of pressure on the electrical resistivity of water-
reportedly negative quantities and that the for- saturated crystalline rocks. J. Geophys. Res., 70, 5669-
mations in question just happened to show a 5678.
predominantly secondary permeability. Bussian, A.E., 1983. Electrical conductance in a porous
medium. Geophysics, 48:1258-1268.
It has been demonstrated that a failure to re- Carothers, J.E., 1968. A statistical study of the formation
cognise and compensate for shale effects upon factor relation. Log Anal., 9 (5): 13-20.
the first Archie equation can result in a misuse Gomez-Rivero, O., 1977. Some considerations about the
of petrophysical parameters which can, in turn, possible use of the parameters a and m as a formation
evaluation tool through well logs. Trans. SPWLA 18th
lead to a gross misinterpretation of reservoir Ann. Logging Symp., pp. J 1-24.
characteristics. The key to improvement lies in Gomez-Rivero, O. and Tovar, V.M., 1980. Comporta-
a greater understanding of the underlying rock miento electrico de rocas de baja a muy baja porosi-
physics. This objective seeks to remove the dad. Pet. Int., 134: 53-56.
Greenberg, R.J. and Brace, W.F., 1969. Archie's law for
empiricism that has formed the cornerstone of
rocks modelled by simple networks. J. Geophys. Res.,
reservoir evaluation over the past 50 years. 74: 2099-2102.
These comments apply equally to the second Hill, H.J. and Milburn, J.D., 1956. Effect of clay and water
Archie equation which will be considered in a salinity on electrochemical behaviour of reservoir
subsequent paper. rocks. Trans. AIME, 207: 65-72.
Hook, J.R., 1983. The precision of core analysis data and
some implications for reservoir evaluation. Trans.
SPWLA 24th Ann. Logging Syrup., pp. Y 1-18.
Acknowledgements Jackson, P.D., Taylor-Smith, D. and Stanford, P.N., 1978.
Resistivity-porosity-particle shape relationships for
marine sands. Geophysics, 43:1250-1268.
This paper is based on presentations by the Juhasz, I., 1979. The central role of Qv and formation-
author to various chapters of the Society of water salinity in the evaluation of shaly formations.
Professional Well Log Analysts over the past Trans. SPWLA 20th Ann. Logging Syrup., pp. AA 1-
few years. It also formed part of the author's 26.
Juhasz, I., 1981. Normalised Qv--the key to shaly sand
annual presentation to the London student evaluation using the Waxrnan-Smits equation in the
section of the Society of Petroleum Engineers absence of core data. Trans. SPWLA 22nd Ann. Log-
during the 1980s. The author acknowledges BP ging Symp., pp. Z 1-36.
228 P.F. WORTHINGTON

Keller, G.V. and Frischknecht, F.C., 1966. Electrical Porosity and pressure dependence of formation resis-
Methods in Geophysical Prospecting. Pergamon, Ox- tivity factor for sandstones. Trans. CWLS 4th Forma-
ford, 517 pp. tion Evaluation Syrup., 30 pp.
Lavers, B.A., Smits, L.J.M. and van Baaren, C., 1974. Towle, G., 1962. An analysis of the formation resistivity
Some fundamental problems of formation evaluation factor-porosity relationship of some assumed pore ge-
in the North Sea. Trans. 3rd Eur. Formation Evalua- ometries. Trans. SPWLA 3rd Ann. Logging Symp., Pap.
tion Symp., SPWLA London Chapter, Pap. G, 3.
Patchett, J.G. and Rausch, R.W., 1967. An approach to Waxman, M.H. and Stairs, L.J.M., 1968. Electrical con-
determining water saturation in shaly sands. J. Pet. ductivities in oil-bearing shaly sands. Soc. Pet. Eng. J.,
Tech., 19: 1395-1402. 8: 107-122.
Porter, C.R. and Carothers, J.E., 1970. Formation factor- Waxman, M.H. and Thomas, E.C., 1974. Electrical con-
porosity relation derived from well log data. Trans. ductivities in oil-bearing shaly sands: I. The relation
SPWLA 11 th Ann. Logging Syrup., pp. A 1- 19. between hydrocarbon saturation and resistivity index.
Raiga-Clemenceau, J., 1977. The cementation exponent II. The temperature coefficient of electrical conductiv-
in the formation factor-porosity relation: the effect of ity. Soc. Pet. Eng. J., 14: 213-225.
permeability. Trans. SPWLA 18th Ann. Logging Worthington, P.F., 1972. The use of resistivity logging to
Syrup., pp. R 1-13. estimate borehole yield from a matrix-conducting
Ransom, R.C., 1984. A contribution towards a better un- sandstone. Q. J. Eng. Geol., 4: 263-279.
derstanding of the modified Archie formation resistiv- Worthington, P.F., 1976. Hydrogeophysical properties of
ity factor relationship. Log Anal., 25 (2): 7-12. parts of the British Trias. Geophys. Prospect., 24: 672-
Sauer, M.C., Jr., Southwick, P.F., Spiegler, K.S. and Wyl- 695.
lie, M.R.J., 1955. Electrical conductance of porous Worthington, P.F., 1977a. Influence of matrix conduc-
plugs-ion exchange resin-solution systems. Ind. Eng. tion upon hydrogeophysical relationships in arena-
Chem., 47:2187-2193. ceous aquifers. Water Resour. Res., 13: 87-92.
Schon, J., 1983. Petrophysik. Akademie-Verlag, Berlin, Worthington, P.F., 1977b. Reply to discussion by W.E.
405 pp. Kelly. Water Resour. Res., 13:1024.
Sethi, D.K., 1979. Some considerations about the forma- Worthington, P.F., 1982. The influence of shale effects
tion resistivity factor-porosity relations. Trans. upon the electrical resistivity of reservoir rocks. Geo-
SPWLA 20th Ann. Logging Syrup., pp. L 1-12. phys. Prospect., 30: 673-687.
Sundberg, K., 1932. Effect of impregnating waters on Zaafran, Z.M., 1981. Studying the effect of sandstone an-
electrical conductivity of soils and rocks. Trans. AIME isotropy on the quantitative interpretation of resistiv-
97: 367-391. ity sounding and logging. Trans. SPWLA 22nd Ann.
Timur, A., Hemkins, W.B. and Worthington, A.E., 1972. Logging Symp., pp. BB 1-13.

You might also like