What is resistivity?

Resistivity relates current density to electrical field strength (from Maxwell’s equations):
E = RJ (E in V/m, J in A/m2)

R has units of Ωm.

Resistivity can be thought of as the resistance of a wire times the cross-sectional

area of the wire divided by the length of the wire:

L
For this reason, resistivity is sometimes given units of Ωm 2/m.

Electrical conductivity is the reciprocal of resistivity and has units of S/m. Represented by ρ.

Sometimes, especially in old wireline logs, you see the unit “mho” (or “mmho”, which would
be the millimho) used for conductivity. While clever, this is not the SI standard.
What is the resistivity of a composite material, like a rock?
Think about the availability of conductive pathways for
flow of electrical current.

Brine: low R Dilute suspension Simple cubic Face-centered

of nonconducting packing: even cubic packing:
grains: higher R higher R even higher R

Resistivity increases as porosity decreases

(Conductivity decreases as porosity decreases)
What about a rock that contains nonconductive fluid
(e.g., hydrocarbon)?

No oil Add some oil More oil

Initial resistivity Higher resistivity Even higher resistivity

Resistivity increases as water saturation decreases

(Conductivity decreases as water saturation decreases)
Archie formalized these observations as

aRw
S  m
n
w
 RT

Rw = resistivity of formation water

Rt = measured resistivity of the formation (measured downhole)
a, m, n = empirical constants

Archie arrived at this equation by plotting things on log-log paper

However, the equation can be derived
Quantifying the dependence of resistivity/conductivity on porosity, saturation

2 methods: effective medium theory, percolation theory

Effective medium theory states that any property of a multicomponent mixture may
be expressed as a function of the volume fraction of each component, the individual
properties of the components, and their spatial arrangement.

Conceptually, the presence of each component causes a local perturbation to the bulk
property. EMT states that the volume average of all these perturbations must be zero.

p r   peff  0
 Non-self-consistent approach: spherical inclusions of one phase embedded in
a background phase

We’ll start with the general solution and the talk about the special case of zero
frequency.
Consider the electric field due to a dipole source with moment d at the center of a
sphere of radius r:
The field at the surface of the sphere r is given by
3rˆ  rˆ  d   d 4
Ed  r   3
   r d
r 3
The second term on the RHS is the depolarizing field (the difference between the
fields inside and outside the sphere) integrated over the volume of the sphere. This
term does not appear for point dipoles, only for dipoles of finite volume.
Now consider a bunch of these spheres immersed in a background medium.
Assume there are N spheres and the total volume of the medium is V.
The volume per sphere is v = V/N.
Now apply an external field Eext.

The total dipole moment of the medium, dtot, is simply Nd where d is the dipole
moment of each sphere. If α is the static polarizability of each sphere, we can write
dtot = NαEext. 1 
Similarly, we can express dtot as V E where E is the macroscopic electric field
4 inside the medium and ε is the
dielectric permittivity.
What is the relationship between E and Eext?
Use superposition. The internal field will be the sum of the external field and the
volume-average of the fields generated by each sphere:

E  Eext   E r
n
n  Eext   En  r 
n

If d is sufficiently small, we can represent En(r) simply as the depolarizing field

resulting from each sphere (recall that the other part of the dipole field decreases
as 1/r3). The volume average is thus
1 4 4 d
En  r    
V V 3
  r  rn dr 3
 
3 V
where rn is the surface of the nth sphere.
So we have
4 d  4  
E  Eext  N  1  Eext
3 V  3 v
Where v = V/N and α = d/Eext
 1 
Recall now that d tot  NEext V E
4

so
E N 4 4 4 
   1
Eext V   1 v   1 3 v

Which yields
8 
1
 3 v
4 
1
3 v

Which is the Lorentz formula for the dielectric permittivity of a non-polar gas.
 Where do we go from here? This equation is not quite useful.

The polarizability (α) of a sphere with radius a and dielectric permittivity ε s is

 s 1
  a3
s  2

Recalling that the specific volume of the spherical inclusions is v = V/N, we can
express the volume fraction of inclusions f as
NVsphere 1 4 a 3 4 a 3
f   Vsphere  v
V v 3 v 3 f
The dielectric permittivity of the medium can now be expressed as

8  1  8 a 3  s  1 3 f 1  2 f
 s 1
1
3 v  3  s  2 4a 3 s  2
 
4  4 3  s  1 3 f  s 1
1 1 a 1  f
3 v 3  s  2 4a 3 s  2
What if the background material has a nonzero dielectric permittivity?

Since electric fields are additive, we can take the previous expression and replace the
absolute dielectric permittivities with the ratio of each permittivity to the background
material permittivity εBG.

s
1

1  2 f BG
s    BG
2 1 2 f s
  BG  s  2 BG
     BG
 BG s    BG
1 1 f s
  s  2 BG
1  f BG
s
2
 BG

This is the famous Maxwell-Garnett equation.

Now let’s express this in terms of DC conductivity σ.

The amplitude of the displacement field D0 is related to the amplitude of the applied
field E0 by
D0 e  it      E0 e  it

The dielectric permittivity can be expressed as a complex number:


  i


In the limit of small ω, the imaginary part of the permittivity dominates. At ω = 0,

the behavior can be described entirely in terms of the DC conductivity σ. This
means that the Maxwell-Garnett equation can be expressed in terms of DC
conductivity as
 s   BG
1 2 f
 s  2 BG
   BG
   BG
1 f s
 s  2 BG
Let’s imagine the special case of nonconducting spheres (σ s = 0) immersed in water (σBG =
σw). The Maxwell-Garnett equation becomes

1 f  2
 w w w
f 1 3 
1 1
2 2
Since f = 1 – φ. Now using RT = 1/σ and Rw = 1/σw, we get an Archie-like expression:

RT 3  

Rw 2
 RT
  1.31
Rw

RT 3  

Rw 2
Self-consistent approximation (SCA): spheres are treated as noninteracting
individuals, and their interactions are included in the behavior of the
background matrix

To obtain an expression for conductivity using SCA, modify the Maxwell-Garnett

equation by considering the host and the inclusions both as introducing
perturbations about the overall conductivity:

 s   BG  
1 2 f  21  f 
 s  2  BG  2
 
 s   BG  
1 f 
 1 f 
 s  2  BG  2
With some algebraic manipulation this becomes
 BG    
  1    s 0
 BG  2  s  2

Again setting σs = 0 and σBG = σw, we get

 w  1 3  1 RT 2
  1     0     w 
 w  2 2 2 and Rw 3  1
 RT 2

Rw 3  1

RT RT
  1.5   2
Rw Rw
Differential effective medium model (Sen et al., 1981)
Consider a medium composed of spheres of different sizes surrounded by a
shell of water:
If we progressively add grains in increments of Δvg,
the conductivity at step k+1 is related to the conductivity
at step k by
 k 1   k v g , k  g   k

 k 1  2 k vw  vg  g  2 k

For an infinitesimal volume increment dvg, this can

be expressed as
d dvg  g  

3 vw  vg  g  2
d vw  vg  vg vw 1 1
If we let ψ = vg/(vw + vg), then     1  
 vw  vg  vw  vg vw  vg vw  vg
2
dvg
since vw = 1 – vg.
d d  g  
Therefore, d  dv g and we can write 
1  vw  vg 3 1   g  2
Rearrange and integrate:

 g  2 1
1
 3   g  
d  0
1 
d

 1/3 1
ln  C  ln  
  g  
Boundary condition:   1
 w

 w1/3  1/3  w1/3   1/3  w   g  1

C   ln so ln  ln  ln     ln  
 w  g   g w  g   w     g   

Take exp of both sides and invert:

 w    g
1/3

  
   w  g
This is the Hanai-Bruggeman equation.
 Consider the limit where σg = 0:
2/3
w  Rw
        w 3/2  RT 
   3/2

This is equivalent to Archie’s equation with Sw = 1, a = 1, and m = 1.5.

KEY OBSERVATION: Archie’s equation assumes σg = 0.

What about nonspherical particles?

Mendelson and Cohen, Geophysics, 1982:

Assuming σg = 0, we can write
d dvg d
 m m
 vw  vg 
where m = 1.5 in the case of spheres. For spheroids, m is given by
5  3L
m   f  L
3  1  L2 
where L is the depolarization factor and f(L) is the probability distribution for each L.
If all grains are the same size and shape, they will all have the same L and f(L) = 1.
 b1 Oblate spheroid (shaped like M&Ms)
b2 with semiaxes b1 < b2 = b3:

If aspect ratio a = b2/b1,

L
a2
 a 2  1
3/2  a 2  1  tan 1  a2  1 

The Hanai-Bruggeman equation is now

1
1
w  m   g
  
   w  g

Which reduces to Archie’s equation (Rt = Rw/φm) when σg = 0.

 10

5
m

0
1 10 100

Aspect ratio
Details about the depolarization factors
(from Mendelson and Cohen, 1982)
Consider an ellipsoidal grain with conductivity
tensor γi = σi + αi exposed to an external electric field E0.
(σ is the symmetric part of the conductivity tensor.)
σi The principal axes of E0, γi, and the grain itself are not
E0 necessarily aligned.

What is the magnitude of the field Ei in the grain?

What is the equivalent conductivity γ of the medium?
Let B be a symmetric tensor with eigenvalues given by bk-2, where bk are the principal
semiaxes of the ellipsoid. The surface of the ellipsoid is given by Bjkxjxk = 1.

We can perform a coordinate transform to1/2

ˆ  σ align
B Bσ
the principal axes of σi and B.
1/2
In doing so, we define a new tensor with eigenvalues lk-2 where lk are
the principal semiaxes of the transformed ellipsoid. (Note that this uses σ, not σi.)

The coordinate transform is defined by xj = (σi1/2)jkRklyl where xj are the old coordinates
and yl are the new ones. The matrix
l11/2 0 R is0orthogonal
 (its transpose is equal to its inverse)
ˆ   0 l 1/2
and is defined by R 1BR 0

 2 
 0 0 1/2
l 
 3 
 It can be shown that the electric field in the ellipsoid Ei is given by

1
Ei  1  σ 1/2 Λσ 1/2  γ i  γ   E0

Where Λ = RLR-1 and L is a diagonal matrix whose elements are equal to the
three depolarization factors Lp (p = 1,2,3):

1 d
Lp  D  0   2
0  lp    D   
2

where
1/2
 
D       lk2    
 k 

The conductivity of the medium γ can be obtained from EMT by

1
 γ i  γ  1  σ 1/2 Λσ 1/2  γ i  γ   0

where the brackets represent an average over the entire volume.

Spheroids (ellipsoids with two equal semiaxes) may be represented by a single
depolarization factor.
Chatzis et al., SPE Journal, 1983

Core was flooded with brine and styrene monomer with 1 wt% benzoyl peroxide
Trapped styrene was polymerized by heating to 80°C
Effective medium wrap-up
Limitations:
-Valid when conductivity difference between matrix and pore fluid is not large
-Valid at high porosity (>30-40%)

Problem: Archie’s equation seems to work even at low porosities

Solution: percolation theory

 Percolation theory
Consider a square lattice. Each line represents a bond.
Let p be the probability that a bond is occupied.
Start with p = 0.
p = 0.1
p = 0.3
p = 0.5
 p = 0.5

Sample-spanning pathway of occupied bonds = percolation

The percolation threshold pc is the occupation probability at which a sample-spanning
cluster forms.

pc is related to the coordination number (Z, average number of neighbors a particular

bond is connected to) by pc = 2/Z in 2 dimensions and pc = 1.5/Z in 3 dimensions.

As we saw in previous demonstration, Z = 4 for square lattice so pc = 2/4 = 0.5.

In a 3-D cubic lattice, Z = 6 so pc = 1.5/6 = 0.25.

We will explore this more when we talk about capillary pressure and relative permeability.
 Threshold porosity φc

Song, Vadose Zone Journal, 2010

Porosity below which percolation cannot occur

Varies from rock to rock – depends on grain size, grain shape,
grain sorting, etc. etc.
When φ is close to φc, transport properties (electrical conductivity, permeability,
diffusion coefficient, thermal conductivity, etc.) scale as (φ – φc)t where t is a
universal exponent.

In 2 dimensions, t = 1.6
In 3 dimensions, t = 2

(there are deviations from universality, but don’t worry about them for now)

We can use this scaling to describe electrical conductivities at low porosity.

Consider: when φ ≤ φc, σ = 0

When φ = 1, σ = σw
    c  2
Therefore, we can write   
1  c 
2
w
 What does “close to the percolation threshold” mean?

Kirkpatrick (1973) showed data suggesting percolation scaling is

not valid for p > 70%. But there is no evidence that this value is
universal.
Data from natural porous media suggest percolation
scaling is valid to high porosities (up to 70%)


w

   c  2
1  c  2
Mudstone data from Daigle et al. (2015)
Sandstone data from Berg (2012)
(note reciprocal y-axis)
General wrap-up
Archie-like relationships can be derived from EMT, percolation theory

EMT works best at high porosities; percolation theory appears to work

fine at any porosity

EMT can be used to describe porosity- and saturation-dependence of

conductivity

This capability has not really been demonstrated with percolation theory
Let’s ignore these complexities and come back to Archie’s equation.

aRw
S wn 
 m RT
When in doubt, the default values for the Archie parameters are:
a=1
m=2
n=2

We already discussed m in great detail

 Resistivity index = IR

Rt  m Rt
IR    S w n
R0 aRw

Why would oil-wet rock have

higher value of n?

HINT: Think about morphology of

oil phase in water-wet versus
oil-wet rock

Keller (1953)
Archie’s a is sometimes called the tortuosity factor. Why?

In the absence of surface conductivity, the formation factor F may be defined as

F = RT/Rw = a/φm.

Tortuosity (ratio of path length to straight-line distance) is related to porosity

and F by τ = Fφ = aφ1-m. The parameter a therefore is a constant in the porosity-
tortuosity relationship.

However, consider the case of φ approaching 1. In this case, RT will tend towards
Rw, and when φ = 1, by definition RT = Rw. But this will not happen unless a = 1.

Cases in which a ≠ 1 only arise due to improper application of Archie’s equation

caused by (1) failure to account for the percolation threshold, and (2) failure to
account for surface conduction.
 Classic example: The Humble formula (Winsauer et al., 1952)
100 100

F  0.62 2.15

F = RT/Rw
F = RT/Rw

10 10

F     0.06
1.48

1 1
0.01 0.1 1 0.01 0.1 1
Porosity Porosity – 0.06

Note that this is not exactly consistent with percolation theory – BUT it shows that a ≠ 1
could be caused by failure to account for a threshold porosity.
(This is a diverse group of rocks and shouldn’t have the same φ c anyway.)
 What about surface conduction?

Consider the Hanai-Bruggeman equation:

1 1 1 m
1 
1
w  m   g R  m 1 RT  1 Rg R  m 1  RT Rg R  1  RT Rg   m
      T      T    T   
   w  g  Rw  1 Rw  1 Rg  Rw  1  Rw Rg Rw  1  Rw Rg 


If we compare to Archie’s law,

m
RT  1  RT Rg 
 a  m a 
Rw 1 R R 
 w g 

When Rg →∞ (i.e., 0 surface conductivity), a = 1.

If Rg is finite, a is not constant.
Phyllosilicate crystal structures
Revil and Florsch, 2010
 Clavier et al., 1977
High salinity: Double layer has constant thickness
(equal to Helmholtz thickness)

Low salinity: Double layer thickness varies with salinity

1
x~
salinity
Thickness of the Outer Helmholtz Plane and diffuse layer

OHP has thickness approximately equal to 2x the van der Waals radius of water
(2.8-3.2 Å, depending whom you ask) yielding 5.6-6.4 Å.

The total thickness of the EDL is usually taken as the Debye length:

 r  0 k BT
x
2 N Ae 2 I
εr = relative dielectric permittivity of fluid (80 for water at 20°C)
ε0 = permittivity of vacuum (8.854 x 10-12 F/m)
kB = Boltzmann’s constant (1.381 x 10-23 J/K)
T = absolute temperature
NA = Avogadro’s number (6.022 x 1023 1/mol)
e = elementary charge (1.602 x 10-19 C)
I = ionic strength (mol/m3) (note that 1 mol/m3 = 1 mM… very clever, those chemists)
1 n
I   ci zi2 ci = molar concentration of the ith ion
2 i 1 zi = charge number of the ith ion
What is the Debye length for seawater at 20°C?

Typical seawater composition:

Na+ 479 mM
Ca2+ 10.5 mM
Mg2+ 54.3 mM
K+ 10.4 mM
Cl- 558 mM
SO42- 28.9 mM

I = 0.5*(479 + 4*10.5 + 4*54.3 + 10.4 + 558 + 4*28.9) = 711 mM

 r  0 k BT 10 T 10 293 10

x 2
 5 .63  10  5 .63  10  3 .6  10 m
2 N Ae I I 711
 Debye length versus salinity at 20°C
100

10
Debye length (nm)

0.1
1 10 100 1000 10000 100000 1000000

NaCl equivalent salinity (ppm)

 Effect of surface conductivity
on apparent formation
factor, resistivity index

Key points:
1. Surface conductivity
means that RT/Rw ≠ F.
The effect is less
pronounced at high
porosity or with high-
conductivity pore fluid.

2. Not accounting for

surface conductivity will
result in higher Sw
values – less
hydrocarbon!

(BQv = conductivity due to

clay counterions)
Zinszner and Pellerin, 2007
Derivation from Revil and Florsch (2010)
Let’s pretend that the surface conductivity and pore fluid conductivity act as resistors
in parallel:
   f  1    s
Here σf is the pore fluid conductivity and σs is the surface conductivity of the grains.
Since this is an effective medium model, it is strictly valid only at high porosity (and as
long as σf and σw are not too different)

Since 1/F is a measure of the effective interconnected porosity, let’s use it in place of
porosity:
1  1 f
   f  1   s  1   F  1 Du 
F  F F
where Du is the Dukhin number, defined as σs/σf.
 Now, it can be shown (Bolève et al. 2007) that
2  g 2      Qv
Du  CEC   
3 f 3 1   f
where Qv is the concentration of counterions per unit pore volume, β + is the mobility
of the counterions in the diffuse layer (in m2/s/V), and CEC is the cation exchange
capacity (excess charge per unit mass of solid material in C/kg).

Now consider a Taylor series expansion of F at high porosity (i.e. porosity = 1):
1. 5
1 1  3 1 
F     1  1.5  1  ...  1     ...
    2  
Substituting these expressions for F and Du into our original expression yields
f f  3 1  2    Q  1
 1   F  1 Du   1      v 
     Qv 
F F  2    3  1    f   F f
  
Now consider the case where there is an additional nonconducting phase (i.e. oil):
1 n  1
  S w f  1    s  S w f  1   s
F  F
By analog with the previous derivation,

   S w f    Qv    we
1 n S wn
F F
where σwe = σf + β+Qv/Swn. This is known as the Waxman-Smits or Dual Water
equation (see e.g. Clavier et al. 1984).

Important points:
-This is only valid at high porosity
-This is only valid at high water saturation

To date no one has derived appropriate expressions for electrical conductivity

at low porosity and/or low water saturation
(who wants a PhD?)
(hint: effective medium theory cannot be used in these cases – need to invoke
percolation theory)
A word on computing β+

 s zi ci
 i

  i

z c i
j j

Revil et al., 1998

Clays with more surface area have higher CEC

Woodruff and Revil, 2011

1 meq/g = 9.632 x 104 C/kg

Example
Consider the following rock:
Porosity = 0.30
Resistivity = 2.985 ohm.m
Rw = 0.184 ohm.m
aspect ratio = 10

What is the formation factor?

What is the surface conductivity?
What is the Dukhin number?
What is Archie’s a?
What would the resistivity of the rock be if it had 25% oil saturation?
What would we compute the water saturation to be in this case if we
blindly applied Archie’s equation without considering surface conduction?
Formation factor
F  m

5  3L
m

3 1  L2 
L
a 2
a2
 1
32
 a 2  1  tan 1 a 2  1 

With a = 10, we get L = 0.861, m = 3.111, and F = 42.35.

 Surface conductivity and Dukhin
number
1 3 f 1      f  31    
   f    Qv     f  Du
1
  1  Du 
F F 2  F  2 

 F  2 
Du    1 
  31    
 f 

With F = 42.35, σ = 1/2.985 = 0.335, σf = 1/0.184 = 5.43, φ = 0.30, we get

Du = 0.46.

Since Du = σs/σf, σs = Duσf = 2.5 S/m.

Archie’s a
Archie says
f
 aF


f  31    
Waxman-Smits says  1  Du 
F  2 

f F 1
so  a
 1  Du 31    31   
and
1  Du
2 2

With Du = 0.46 and φ = 0.30, we get a = 0.383.

 What if our rock had 25% oil
saturation?
   S w f    Qv 
1 n
Waxman-Smits:
F
3 f 1    1 n 3 f 1    
since   Qv  Du we have    S w f  Du 
2 F 2 
With F = 42.35, Sw = 0.75, n = 2, σf = 5.43, Du = 0.46, φ = 0.30, we get σ = 0.279 S/m
(which corresponds to Rt = 3.59 ohm.m)
What if we just used Archie’s equation
without considering surface conduction?

Porosity = 0.30
Resistivity = 3.59 ohm.m
Rw = 0.184 ohm.m
a=1
m = 3.111
n=2
Sw = 147%