SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

USING FRACTALS TO DETERMINE

A RESERVOIR’S HYDROCARBON DISTRIBUTION
Steve Cuddy
Copyright 2020 held jointly by the Society of Petrophysicists and Well Log INTRODUCTION
Analysts (SPWLA) and the submitting authors.
This paper was prepared for the SPWLA 61st Annual Logging Symposium held
online over 6 sessions, every Wednesday June 24 to July 29, 2020. Water saturation vs. height (SwH) functions are required
ABSTRACT to initialise a 3D static and dynamic reservoir model with
the water and hydrocarbon volumes, to pick fluid
To determine a field’s hydrocarbon in place, it is contacts and the net reservoir cut-off.
necessary to model the distribution of hydrocarbon and
water throughout the reservoir. A water saturation vs. Figure 1 shows a typical reservoir model. It has cells
height (SwH) function provides this for the reservoir where the colour represents porosity, permeability and/or
model. A good SwH function ensures the three facies type. The lines are wells where electrical log and
independent sources of fluid distribution data are core data are collected. These limited data are used to
consistent. These being the core, formation pressure and populate the reservoir model.
electrical log data. The SwH function must be simple to
apply, especially in reservoirs where it is difficult to map
permeability or where there appears to be multiple
contacts. It must accurately upscale the log and core
derived water saturations to the reservoir model cell
sizes.

This paper clarifies the, often misunderstood, definitions

for the free-water-level, transition zone and irreducible
water saturation. Using capillary pressure theory and the
concept of fractals, a practical SwH function is derived. Figure 1: Typical 3D Reservoir Model
Logs and core data from eleven fields, with very different
porosity and permeability characteristics, depositional WATER SATURATION VS. HEIGHT FUNCTION
environments and geological age, are compared. This
study demonstrates how this SwH function is The SwH function describes how water saturation varies
independent of permeability and litho-facies type and with height above the free water level (FWL).
accurately describes the reservoir fluid distribution.
Water saturation (Sw) determined from interpretation of
The shape of the SwH function shows that of the
log and core data can only represent the reservoir within
transition zone is related more to the fractal pore
a few feet surrounding the well bore. Sw cannot be
geometry rather than porosity or permeability alone.
mapped as it depends on numerous factors including
Consequently, this SwH function gives insights into a
porosity and the height above the local FWL.
reservoir’s quality as determined by its pore architecture.
Several case studies are presented showing the excellent
SwH functions are used in a field’s reservoir model to
match between the function and well data. The function
determine Sw away from well locations so that
makes an accurate prediction of water saturations, even
hydrocarbons initially in place can be calculated. The
in wells where the resistivity log was not run due to well
error in reserves resulting from an equation that poorly
conditions.
describes the reservoir can be significant.
The function defines the free water level, the The Saturation Height Function, as shown by Figure 2,
hydrocarbon to water contact, net reservoir cut-off, the tells us how water saturation varies as a function of the
irreducible water saturation and the shape of the height above the Free Water Level. It also tells us how
transition zone for the reservoir model. The function the formation porosity is split between hydrocarbon and
provides a simple way to quality control electrical log water; and the shape of the transition zone. It is used to
and core data and justifies using core plug sized samples populate and initialize the 3D reservoir model.
to model water saturations on the reservoir scale.

1
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Figure 3: The Free Water Level

THE HYDROCARBON WATER CONTACT

The Hydrocarbon Water Contact (HWC), as shown by

Figure 4, is the height where the pore entry pressure is
sufficient to allow hydrocarbon to start invading the
formation pores. This depends on the local porosity &
permeability. It is a surface of variable height.
Figure 2: Water Saturation vs. Height Function

A good SwH function requires that three independent

sources of fluid distribution data are consistent.

These are:
• Formation pressure data
• Electrical log data
• Core data

The function must account for varying permeability and

fluid contacts throughout the field. It must upscale
correctly from the core plug scale and ½ foot logging
scale to the reservoir model cells scale and should be
easy to apply.
Figure 4: The Hydrocarbon Water Contact
THE FREE WATER LEVEL
THE BULK VOLUME OF WATER
The Free Water Level (FWL) is the horizontal surface of
zero capillary pressure as shown in Figure 3. The FWL The Bulk Volume of Water (BVW) is the proportion of
is the level formation fluids would separate out in a very water in a unit volume of reservoir rock as shown by
wide borehole. It is the intersection point of hydrocarbon Figure 5. The blue shows the proportion of pore space
and water pressures on a formation vs. true vertical depth filled with water. BVW is simply the product of porosity
plot. The formation fluid pressures form linear lines, and water saturation. BVW is what is measured with
even in the transition zone as the formation pressure tool resistivity tools in clean formation, not Sw but the
only responds to the mobile fluid phase, as explained conductivity of the water volume. This is what is
later. The FWL is the start point for SwH plot, but only measured by core analysis, not Sw but the volume of
for very high porosities. water displaced.

2
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

The statistics of sediment distribution is controlled by the

natural processes which created them. Since these
processes have been proven to behave fractally, it is
expected that the distribution of sediments will show
fractal behaviour too. (Al-Zainaldin, Glover and
Lorinczi 2016).

A fractal image is shown by Figure 6. It may look

complex but is based on a very simple repeating pattern.
Figure 5: The Bulk Volume of Water (BVW)

FRACTALS AND RESERVOIR DESCRIPTION

Fractals are mathematical objects for which their parts

are identical to the whole set, except for a change of
scale. Other names for fractals are self-similarity or scale
invariance.

Detailed studies of the fractal modelling of reservoirs

were undertaken by Al-Zainaldin, Glover and Lorinczi,
who kindly discussed with the author their understanding
of the fractal modelling of reservoirs. It has been shown
that sandstones have fractal geometric pore spaces (Katz Figure 6: Fractal Image
and Thompson 1985). Fractals are very useful since they
can describe the broad range of variability which exists Fractals are found throughout nature, in the Cosmic
in reservoir properties including grain-size, porosity and Microwave Background (CMB) shown by Figure 7.
permeability (Perez and Chopra 1997). These patterns of slight variations in temperature, from
the early universe, go on to give rise to the galaxies and
The fractal concept as suggested by Mandelbrot (1977) galactic superclusters each with similar structures.
has found various applications throughout the
geosciences. This is because many physical systems in
nature produce a variation of properties that can be
described by fractals (Lozada-Zumaeta et al. 2012).
Several studies have been made relating the fractal
theory to the distribution of reservoir properties. Turcotte
(1997) modelled the sedimentation process by the
Devil’s Staircase, which is an exact fractal, and showed
that the rate of sedimentation can be related to the time
interval of deposition occurrence by a fractal power law
relationship. Figure 7: The Cosmic Microwave Background

Since the formation of porosity is closely linked to both Figure 8 shows the Himalayas mountain range as seen
sedimentation processes and to fragmentation, both of from space. The patterns seen in the mountain range are
which show fractal behaviour, it follows that porosity repeated as you zoom in on the main valleys and the
would be expected to follow a fractal behaviour too. In valleys that branch off them.
other words, if the grain size distribution is fractal, then
the pore size would also be expected to be distributed
fractally. Laboratory measurements have confirmed that
the porosity of sandstones is indeed fractal, exhibiting a
non-integer power law scaling behaviour, as will be
explained later.

3
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Figure 8: The Himalayas as seen from Space

The snowflake, shown in Figure 9 is a fractal object. Figure 10: Fractals shown in a Tree

From a distance the overall pattern of a tree can be seen

to consists of a trunk and the main branches. On closer
inspection, the smaller branches show the identical
pattern. Even the twigs show the same pattern.

This is why nature can create many complex organisms

through a simple fractal repeating process. As a
consequence, many complex objects may be described
by fractals.

FRACTAL DIMENSIONS

The origin of the term Fractal is due to the fact that they
have a fractional dimension, not a whole number value.
Whereas classical geometry deals with objects of integer
dimension, fractal geometry describes non-integer
dimension. Points have zero dimension, lines and curves
Figure 9: Snowflake have one dimension, squares and circles have two
dimensions and cubes and spheres have three
Fractals are never-ending patterns. Fractals are infinitely dimensions.
complex patterns that look the same at every scale, that
are created by a simple repeating process. Benoit If we increase the side of a cube by a factor of 3 the 2D
B. Mandelbrot (1977) coined the word Fractal. The area of the cube’s side increases by a factor of 9 and
Mandelbrot pattern shown in Figure 6 is created by the cube’s volume increases by a factor of 27. This
recursive formula: relationship is given by the Equation 2.

𝟐 𝑁 = 𝑟𝐷 Equation 2
𝒁𝒏 = 𝒁𝒏−𝟏 + Constant Equation 1
Where D is the dimension, r the length of the side of the
Fractals are objects where their parts are identical to the
object and N is the number of the units (with a side of r)
whole, except for scale as shown by the tree in Figure 10.
that will fill entire object.

4
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

If we measure the increase in a reservoir property, such

as porosity with the decrease of the measuring metric,
this will give the dimension D. The dimension D need
not be an integer, as it is in Euclidean geometry, as it
could be a fraction. This is why it is known as fractal
geometry.

The fractal dimension is a representation of the

heterogeneity of the fractal object. The greater the
dimension, the more heterogeneous the fractal object (Li
2004).

FRACTAL GEOMETRY OF PORE SPACE

We assume that the pore network of a rock sample is

made up of N pore tubes with the same length but
different radii. This capillary tube model is used in the
Washburn equation to calculate the pore size distribution
Figure 11: The Coastline of Great Britain
of a rock sample from mercury injection capillary
pressure data (Mandal 2006).

Therefore, from the capillary tube model, unit area (A) of

rock sample can be represented by:

𝐴 = 𝑁(𝑟). 𝑟 2 Equation 3

Substituting Equation 2 into Equation 3:

𝐴 = 𝑟 (2+𝐷) Equation 4 Figure 12: The Coastline of Great Britain at using

different rulers
The volume of the model is:
When we use a large ruler (r=1, a small magnification
𝑉 = 𝑟 (3+𝐷) Equation 5 factor), we get a very poor approximation, shown
in purple, and a value for the coastline of N=9. As the
QUANTIFYING FRACTAL GEOMETRY ruler length shrinks, the magnification r increases, and
the value of the coastline N increases. We are interested
We use the example of a coastline, as shown in Figure in the rate at which the coastline changes as a function
11, to mathematically confirm and qualify fractal of the ruler length. The curvier the coastline is, the more
behaviour. the coastline will increase as the ruler shrinks.

If we ask, "How long is the coast of Great Britain," the To understand the relationship, we plot the coastline
answer is that it depends on how closely you look at it, versus the magnification factor (or the inverse of the
or how long your measuring stick is. As shown by Figure ruler length) using logarithmic scales, as shown by
12, the length Great Britain’s coastline (N) depends on Figure 13. As the ruler shrinks the measured coastline
the length of your ruler (r) where r is the magnification increases. If the coastline is fractal the relationship
of the ruler.1 between r and N is linear when plotted using log scales,
which is the case for the coastline of Great Britain.

1
Fractalfoundation.org
5
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Thin sections were then imaged with a scanning electron

microscope (SEM). The method involves counting the
number of pixel units representing porosity at different
magnifications.

The fractal nature of Berea Sandstone is shown by Figure

15 which shows a linear relationship between the pixel
size and the number of pixels representing porosity in the
thin sections.

Figure 13: Plot of Ruler Length vs. Coastline Length

D is the fractal dimension and the colour of the points

referred to Figure 12.

FRACTAL BEHAVIOUR IN RESERVOIR ROCKS

Reservoir rocks can be shown to be fractal by using a box

counting method analogous to the measuring method we
used for the coastlines. But in this case, we cover the
image with a grid, and then count how many boxes of the
grid are covering part of the image. Then we do the same Figure 15: Fractal Nature of Berea Sandstone
thing but using a finer grid with smaller boxes. By
shrinking the size of the grid (i.e. by increasing the The smaller the pixel size the more porosity is identified
magnification) we can more accurately measure the in the smaller pores and in the pore throats. The linear
porosity as shown by Figure 14. relationship (on logarithmic scales) shows the Berea
sandstone is fractal in nature.

BUOYANCY FORCES IN RESERVOIR FLUIDS

When a field is originally deposited, the structure usually

contains water. When hydrocarbons migrate into a trap,
the buoyancy force exerted by the lighter oil (or gas) will
push the water that was previously in the pore space
downward. However, not all of the water is displaced;
some of it will be held by capillary (electrostatic) forces
within the pores. Narrower capillaries, pores with
smaller pore throats, with the larger surface area, hold
onto the water the strongest.

As shown by Figure 16 , the water at a given height in a

reservoir is determined by the balance between the
Figure 14: Pore Structure at Increased Magnification capillary forces holding the water up and the force of
gravity pulling the water down.
To study the pore structure of the rock samples,
specimens are first saturated with a blue-dyed epoxy.

6
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

the two immiscible phases present, the contact angle

between the wetting phase, the rock surface and the
density difference between the fluids.

Figure 16: The division of pore space as a function of

height for a fixed porosity.
Figure 17: Capillary pressure draws up and holds the
The oil (or gas) is the mobile phase and only enters the water
space not occupied by water in the reservoir pores.
Therefore, a given part of the pore space within the Capillary pressure curves can be defined for any two-
reservoir will contain both oil and water. The percentage phase system in a given rock. All that will vary is the
of water in the pore space is known as the water interfacial tension and the contact angle. It is therefore
saturation (Sw). possible to convert one capillary pressure curve to
another, provided the relevant values of interfacial
CAPILLARY PRESSURE tension and contact angle are known.

When two fluids meet in a capillary tube there is a The height of the water in a capillary depends on the
difference in pressure across their interface. This capillary pressure, which is determined by the radius of
"Capillary Pressure" is caused by the preferential wetting the capillary and the fluid types. The relationship
of the capillary walls by the water and gives rise to the between capillary pressure and pore size is the Young-
familiar curved meniscus and causes the water to rise up Laplace equation shown by Equation 6.
the capillary as shown by Figure 17.
2σ cos(𝜃)
The capillary pressure characteristics of reservoir rocks 𝑃𝐶 = Equation 6
𝑟
affects the distribution of fluids within the reservoir. It is
one of the most important measurements that can be Where:
made because it relates reservoir rock and reservoir fluid Pc capillary pressure
properties. The magnitude of capillary pressure reported r capillary radius
in laboratory measurements relates to the height above  interfacial tension
the free water level in the reservoir.  contact angle

The relationship between capillary pressure and water Consequently, smallest pores hold on to the most water
saturation is dependent upon grain size, grain shape, as shown by Figure 18.
packing, sorting and cementation (environment of
deposition and diagenesis). These all affect the pore
throat diameter distribution, often referred to as the pore
size distribution (PSD) within the rock. The relationship
is also dependent upon the interfacial tension between
7
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Figure 18: Smallest pores hold the most water

Substituting Equation 6 into Equation 5

𝑉 = 𝑎𝑃𝐶 −(𝐷+3) Equation 7

Where a = constant

As the capillary pressure only acts on the water phase, Figure 19 The capillary and gravitational forces acting
the volume V can be replaced by the Bulk Volume of on the reservoir fluids
Water (BVW)
The capillary-bound water comprises a continuous
𝑏 column of water within the oil leg, with a hydrostatic
𝐵𝑉𝑊 = 𝑎𝑃𝐶 Equation 8
pressure gradient. The oil located in the remaining pore
Where b = -(D +3) space also as a continuous phase but will have a lower
pressure gradient.
Remember the dimension D is fractal and not necessarily
an integer. Although oil and water can coexist in the same localised
volume of rock, the pressures acting on the two fluids are
THE FORCES ACTING ON RESERVOIR FLUIDS very different. The dotted line represents the water
gradient which continues into the water leg. The solid
The force of gravity on the column of water is line above the FWL is the oil gradient. The formation
determined by the difference between the water and oil pressure tester tool measures the mobile phase shown by
densities and is called the buoyancy pressure Pb and is the solid line. The intersection of the pressure gradients
given by indicates the free water level (FWL), as shown by the
dashed line.
𝑃𝑏 = (𝑤 − 𝑜 )𝑔𝐻 Equation 9
The buoyancy pressure (the difference in pressure
Where: between the oil and water phases) increases with height
Pb buoyancy pressure due to gravity above the FWL. As the buoyancy pressure increases the
𝑤 water density oil phase will displace more water from increasingly
𝑜 oil density smaller pore volumes. Most water is held in the smallest
pores closest to the FWL. Therefore, Sw will tend to
g acceleration of gravity
decrease with height above the FWL, but is not always
𝐻 height above the free water level (FWL)
the case as will be shown later.
Notice that the greater the density difference, the greater
the gravity force. The buoyancy pressure is shown in The volume of water remaining at a given height in a
Figure 19. reservoir is a function of the balance of capillary forces

8
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

holding up the water up and the force of gravity acting

together with the density contrast between the reservoir
fluids, acting to pull the water down.

The balance between capillary pressure and the

buoyancy pressure is:

Pc = Pb Equation 10

From Equation 9 above

𝑃𝑐 = 𝑃𝑏 = (𝑤 − 𝐻 )𝑔𝐻 Equation 11

Figure 20: Water Saturation vs. Height above the FWL
Rearranging this Equation, we get for a Southern North Sea Gas Field

Pc
H= Equation 12 There is considerable scatter in the data. It is normal
( W - H )• g practice to divide the data into porosity bands and fit
SwH lines.
Pc and H are interchangeable as a function of the
reservoir fluids Deriving the functions for these porosity bands can be
very difficult. The SwH curves, by porosity band,
From Fractals Equation 7 derived from this dataset, are shown by Figure 21. The
highest porosity band is the bottom left. The lowest
BVW = aPC b Equation 13 porosity band is upper right. These curves are
mathematically and visually unconvincing as they cross.
As Pc can be replaced by H Also, there is insufficient data in some of the porosity
bands to fit the lines. To fit the lines, it is necessary to
𝐵𝑉𝑊 = 𝑎𝐻 𝑏 Equation 14 know the pore entry pressure, also known as the
threshold height, which represents the point where the
The constant ‘b’ is dimensionless. Consequently, the porosity band line meets the right-hand y-axis. Clearly it
equation is independent of scale and applies to core plugs is not easy to determine this intercept point.
as well as the entire reservoir.

This equation has been called the Fractal (or FOIL)

Function and describes the variation of BVW as a
function of the Height above the Free Water Level
(Cuddy 1993).

SOUTHERN NORTH SEA GAS FIELD STUDY

Figure 20 shows the water saturation vs. height above the

FWL data from a large Southern North Sea field (Cuddy
1993). The excellent quality dune sands are shown in red,
the medium quality sandy sabkhas in blue and the poor
fluvial sands in green. The highest porosities, to the left
of the plot, give the lowest water saturations, as you Figure 21: Classical Sw-height Curves by Porosity Band
would expect. Where the rock is less than 12 p.u. the
formation is fully water saturated for hundreds of feet If we replace Sw in the x-axis by BVW the data collapses
about FWL as shown on the right of the plot. as shown by Figure 22.

9
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

NET RESERVOIR

‘Net Reservoir’ are intervals of rock that is capable of

holding hydrocarbon. This should not be confused with
‘Net Pay’ which is the ability of the reservoir to produce
hydrocarbons. Knowledge of what is net reservoir is
essential for upscaling parameters, including porosity
and water saturation and for calculating hydrocarbon in
place from the reservoir model. The net reservoir cut-off
varies as a function of height above the FWL as shown
by Figure 24.

Figure 22: Bulk Volume of Water vs. Height above the

FWL for a Southern North Sea Gas Field

This collapse demonstrates the bulk volume of water is

independent of porosity, as the porosity points show no
separate banding as seen in Figure 20. It can easily be
shown that the data clusters are independent of other rock
parameters, such as permeability or facies type by simply
plotting them, in colour, on the z-axis on the cross-plot.

Figure 22 is telling us something very important about

the reservoir. The formation water which was there first
and is now capillary bound and claims part of the
porosity space. At a particular depth in the reservoir the
BVW is determined by its height above the FWL. If the
porosity is 10 p.u. the hydrocarbon enters the remaining
space. If the porosity is 20 p.u., the BVW is the same, but Figure 24: Net Reservoir varying as a function of height
extra available porosity is filled with hydrocarbon. This above the FWL
is shown by Figure 23.
Reservoir high above the FWL has low saturations of
capillary bound water and hydrocarbon enters the
smaller pores. Reservoir just above the FWL contains
high saturations of capillary bound water and there is a
little room left for hydrocarbons.

This is shown in Figure 25 for a well from the Heather

oil field in the North Sea. Net = 1 at 150’ whereas Net =
0 at 280’, just above the FWL, even though the porosity
is higher at 280’.
Figure 23: Bulk Volume of Water as a function of height

In this case, with BVW is the same at 9 p.u., but Sw

increases from 40% to 80% as the porosity decreases
from 20 p.u. to 10 p.u.

10
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

off is replaced by the Bulk Volume of Water. Note that

the function is field specific to be determined by
electrical log and core data, as explained later.

Figure 26: Fractal Function of BVW vs. Height above

the FWL

The function has only two parameters and is used in the

3D reservoir model. It has several other important uses
as will be described next.

PICKING THE FREE WATER LEVEL

The fractal SwH function points to the Free Water Level

for a North Sea oil field as shown by Figure 27 (from
Kay 2002). The BVW is plotted against the True Vertical
Depth Subsea for two wells which don’t intercept the
FWL. The BVW trend identifies the FWL at 10,730
ftTVDss and confirms the wells are probably in the same
reservoir compartment.
Figure 25: Net Reservoir Case Study
One of a field’s main uncertainties is depth or more
THE FRACTAL FUNCTION precisely the true vertical depth below datum in the
reservoir model. The combination of logging depth and
From above the Fractal Function describes the variation deviation survey errors can give an error bar to the TVD
of BVW with height above the FWL and is given by: depth of +/- 30 feet. This error seriously influences the
volumetric computation of the hydrocarbon in place and
𝐵𝑉𝑊=𝑎𝐻𝑏 Equation 15 may even suggest wells are in different fluid/pressure
compartments, which would affect the field’s
Where: development plan. The fractal function can be used to
𝐵𝑉𝑊 = Bulk Volume Water normalise the TVD depths between wells. If the field’s
𝐻 = Height above FWL formation pressures, geochemical fluid analysis and/or
𝑎, 𝑏 = Constants geophysical mapping suggests that wells are in the same
compartment - the wells when plotted as shown in Figure
This function is shown by Figure 26. It is no surprise that 27 can be shifted to a common FWL.
this is similar to Figure 24 where the Net Reservoir Cut-

11
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Figure 27: Fractal Function points to the Free Water Figure 28: Fractal Function derived from Core Data
Level
A similar argument is used later for electrical log data,
CALCULATING THE FRACTAL FUNCTION where only the best data points from the centre of thick
beds are required to determine the fractal function.
The BVW function shown in Equation 15 is a straight Consequently, it is not necessary to correct the resistivity
line when plotted on log scales. log for thin beds, bed boundary effects or conductive
shales.
log BVW= log a + b log H Equation 16
DERIVING WATER SATURATION AND THE
which is the form of the straight-line equation y = mx + HYDROCARBON TO WATER CONTACT
c, where ‘c’ is the intercept of the line with the y-axis and
‘m’ is the line’s gradient. The line is determined by least By definition.
squares regression, where the predicted variable (BVW)
is x-axis (i.e. XonY), rather than the y-axis which is the 𝐵𝑢𝑙𝑘 𝑉𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 𝑊𝑎𝑡𝑒𝑟
usual case. Only two valid core or electrical log data 𝑆𝑤 = Equation 17
𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦
points are required to calculate the constants ‘a’ (from
10^c) and ‘b’ (gradient m which is negative). From Equation 15

Figure 28 shows the Fractal Function derived from core 𝑎𝐻 𝑏

data for six North Sea fields discussed later in this paper. 𝑆𝑤 = Equation 18
Notice how the gradient is constant for each of these 𝑃𝑜𝑟𝑜𝑠𝑖𝑡𝑦
fields. Consequently, only 2 data points are required to
compute the intercept and gradient for each of these core Where:
data sets. The author is not recommending that only 2 H = Height above FWL
data points are taken in a field, rather that only the best a, b = Constants
core data, unaffected by measurement error or core
sample fractures are used. The other good data points By simply dividing the fractal function by the porosity,
should be used to confirm the regression line. the water saturation can be computed in individual wells
or in the reservoir model. The Fractal SwH function
gives the Hydrocarbon Water Contact (HWC) as a
function of porosity as shown by Figure 29.

12
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Figure 29: Deriving Sw and the HWC

Figure 30: Irreducible Water Saturation
The HWC depends on the local porosity. In the example
shown in by Figure 29 the reservoir rock is fully water HETEROGENEOUS RESERVOIR CASE STUDY
saturated for a least 180 feet above the FWL when the
porosity is 5 p.u. or less. Note that the Free Water Level Figure 31 shows a highly heterogeneous reservoir. The
(FWL) is the same irrespective of the porosity. logs and core show large variations in porosity and
permeabilities. Consequently, the computed water
IRREDUCIBLE WATER SATURATION saturation also shows large variations. Unusually water
saturation reduces with height in this well. It would be
The Irreducible Water Saturation (Swirr) is the lowest very difficult to derive SwH Function for these data
Sw that can be achieved in a core plug by displacing the using conventional methods of porosity banding.
water. This is achieved by flowing hydrocarbon through
a sample or spinning the sample in a centrifuge which Notice that BVW, shown in white in Track 4, is a simple
depends on the drive pressure or the centrifuge speed. function of height. The log derived Sw, from the
resistivity log, is shown in black in Track 1. The SwH
This is equivalent to moving higher above the FWL. Sw function is shown in red, in Track 1, which is an excellent
therefore depends on the height above the free water match. Permeability is not required for this function as
level and the transition zone therefore extends the SwH simple function of just two variables; height and
indefinitely. Capillary pressure theory tells us that a porosity.
minimum irreducible Sw does not exist, as Sw depends
on the height and local porosity in the reservoir, as shown Only two data points, taken from the thick intervals in
by Figure 30. the best wells in the field, are required to compute the
water saturation. Consequently, the fractal SwH function
downgrades resistivity logs in later wells from being
‘essential’ to ‘nice to have’. In addition, only the very
best resistivity log is required to derive the function.
Consequently, thin bed effects, bed boundary effects and
shale effects, can be ignored.

13
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

Figure 32: Core Water Saturations

Consequently, it is not necessary to correct the resistivity

Figure 31: Fractal Case Study log for the thin bed effects.
CORE WATER SATURATIONS SWEPT AND BY-PASSED HYDROCARBON
In order to confirm that the Fractal SwH Function Figure 33 shows a well from the Heather oil field located
correctly predicts Sw, an independent source of Sw is in the Northern North Sea, where oil is produced from
required for comparison. Core derived Sw is an excellent sandstones of the Middle Jurassic Brent Group (Kay
independent source of Sw. Accurate water saturations 2002).
can be derived from core if taken from wells drilled with
oil based mud which has been ‘doped’ to identify any Track 1 shows the water saturation determined from the
mud filtrate contamination. Only the centre of cores resistivity log (black) and the fractal function (purple).
taken above the FWL, where the capillary bound water Tracks 2 and 3 show the computer processed well
is immobile, are sampled. interpretation using the resistivity and fractal function
respectively. The water saturations agree in the thick
Figure 32 shows the comparison between water beds except between 100 and 120 ft. As the fractal
saturations determined from the resistivity log (black function gives the initial oil in place and the resistivity
line), the fractal function (red line) and core (blue dots). log the water saturations at the time of logging, the
The core confirms the water saturations determined by difference shows a depleted zone and the value for the
fractal function. residual Sw.

14
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

COMPARISON BETWEEN CORE AND LOG

DERIVED FRACTAL SWH FUNCTIONS

Data from eleven UK North Sea fields (250 wells) were

used in this study (Gagnon 2008). Electrical logs and
conventional core (porosity and permeability) data were
available from all eleven fields together with thin section
and capillary pressure data.

The fields with electrical log data used in the study are
listed in Table 1. They were selected as they represent a
range of reservoir fluids and depositional environments.
The fields included both gas and oil accumulations in
different types of clastic reservoirs from different
depositional environments. The broad spectra of fields
were chosen to assess the robustness of the fractal
function. Table 1 lists the fields, their fluid type and
depositional environment.

Figure 33: Hydrocarbon swept and by-passed zones

In contrast the interval between 160 and 175 ft shows

similar Sw. Consequently, this has been interpreted as a
by-passed oil zone requiring further well intervention. If
two 3D reservoir models are created, using the fractal
prediction and current day resistivity logs, the difference
model will create a swept and by-passed map of the
reservoir.

The thin bed at 145 ft shows the correct Sw as interpreted

from the fractal function where the resistivity log over
estimates Sw. There are bed boundary effects shown in Table 1: Fields used in the Study
the Sw computed from the resistivity log at 180 ft that
are not shown on the Sw computed from the fractal
function.
15
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

The poroperm distribution for the eleven fields with log

and core data is shown in Figure 34, where average
permeability increases with average porosity as
expected. Average permeability spans four logarithmic
cycles: from 0.1 mD to 2 Darcies. Porosities range from
8 to 32 Porosity Units (p.u.).

Figure 35: Fractal Functions for the study fields

Figure 34: Average Porosity and Permeability for the

Study Wells

A Fractal Function was determined separately from

electrical log and the core data. The BVW for each well
was calculated as the product of the water saturation and
porosity curves. This was plotted against the height
above the FWL. Only data away from conductive bed
boundaries were included to minimise the effect of
shoulder bed effects on the resistivity logs.

The Fractal Functions were calculated by plotting the

logarithm (base 10) of BVW (x-axis) against the
logarithm of the true vertical height (y-axis) above the
FWL. Then a free linear regression (XonY) was used to
compute the Fractal Function parameters. These fractal Figure 36: Fractal Functions on log-log Scales
functions are shown by Figure 35.
The ‘quality’ of a reservoir is given by its value of water
The logarithmic scales plot of BVW against height above saturation at a certain height above FWL for a given
the FWL is shown in Figure 36. It is noticeable that all porosity: with a lower Sw being considered the better-
the fields share a similar ‘b’ parameter (slope) and the quality reservoir. The quality of a reservoir can be
main difference between the SwH Functions is due to the defined by the value of its ‘a’ parameter. Figure 35 shows
variation of ‘a’ (intercept) between the fields. reservoir quality increasing towards the bottom-left
corner of the cross-plot. Notice that the parameter ‘a’
varies much more between these fields compared to the
parameter ‘b’.

16
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

The Fractal Functions derived from the core data from

Sw =
( Sw +  Sw )
( +  )
1 1 2 2
the same fields is shown by Figure 37. The core data are
plotted against capillary pressure which is 1 2
Equation 19
interchangeable with the height above the FWL as shown
by Equation 12. Where:

Average porosity is determined by the sum of the

porosities together divided by their number. However,
Sw must be pore volume weighted, which is the same as
averaging BVW, the product of porosity and water
saturation. Integration works for BVW functions but not
for Sw functions. Worthington (2002) recommends
using BVW functions as BVW is implicit in this
equation.

CONCLUSIONS

We have determined from petrophysical first principles

a fractal derived water saturation vs. height function to
be used in reservoir modelling. The function has been
Figure 37: Core Derived Fractal Functions shown to accurately describe the hydrocarbon and water
distribution throughout the reservoir. This function can
Figure 36 shows the fractal function derived from the be derived from electrical log and/or core data using
electrical logs using data between the FWL and the top linear regression and is simply applied in the reservoir
of the reservoir for each field. Figure 37 shows the fractal 3D model.
function derived from the core data on the scale of a core
plug (a few inches). As the functions agree this supports The fractal function is independent of rock
the fractal nature of pore geometry of these reservoirs. characteristics such as facies type, porosity and
permeability. This function can be used to determine a
UPSCALING WATER SATURATIONS reservoir’s hydrocarbon distribution, the field’s free
water level, local hydrocarbon water contacts, the net
As Sw-Height functions (SwH) are used to initialize the reservoir cut-off, the shape of the transition zone and the
3D reservoir model, it is essential that the SwH predicted irreducible water saturation. The function can be derived
water saturations upscale accurately from ½ foot log or from just two good electrical log or core data points and
core scale to the cell size of the reservoir model. This is doesn’t require the resistivity log to be corrected for bed
done by integrating the Sw-Height function. boundaries, thin beds or conductive shales. As the
function is based on BVW, it upscales correctly when
Unlike other parameters, such as porosity, water integrated.
saturation must be pore volume averaged as shown by
equation 19. The electrical logs and core data give the same function
which confirms the fractal nature of hydrocarbon bearing
reservoirs. Fractal SwH Functions can therefore be used
to quality control core and electrical logs against each
other, and justifies using small core plugs to derive a
water saturation vs. height function on the scale of a
reservoir.

17
SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

ACKNOWLEDGEMENTS MANDAL, D., et al. 2006. Use of fractal geometry for

determination of pore scale rock
The author would like to thank Baker Hughes for the use heterogeneity. International Conference & Exposition on
of their data and resources, and the University of Leeds, Petroleum Geophysics, Kolkata.
School of Earth and Environment, for their guidance.
MANDELBROT, B.B., 1977. Fractals: Form, Chance,
REFERENCES and Dimension. W.H. Freeman, San Francisco.

AL-ZAINALDIN, S., GLOVER, P.W.J., LORINCZI, P. TURCOTTE, D.L., 1997. Fractals and Chaos in Geology
2016. Synthetic Fractal Modelling of Heterogeneous and and Geophysics. Cambridge University Press,
Anisotropic Reservoirs for Use in Simulation Studies: Cambridge.
Implications on Their Hydrocarbon Recovery Prediction
Transp Porous Med DOI 10.1007/s11242-016-0770-3 WORTHINGTON, P.F., LOVELL, M. and
PARKINSON, N., 2002. Application of saturation-
ANGULO, R., V. ALVARADO, V., GONZALEZ, H., height functions in integrated reservoir description:
1992. Fractal Dimensions from Mercury Intrusion AAPG Methods in Exploration Series, 13, pp. 89.
Capillary Tests R.F. SPE 23695
ABOUT THE AUTHOR
CUDDY, S., 1993. The Fractal function - a simple,
convincing model for calculating water saturations in Steve Cuddy is a retired Petrophysicist, having worked
Southern North Sea gas fields: Transactions of the 34th with Schlumberger, BP, and Baker Hughes. He holds a
Annual Logging Symposium of the SPWLA, H1-17, PhD in petrophysics at Aberdeen University. He also
Calgary, Canada., 1993, BP Exploration. holds a BSc in physics and a BSc in astrophysics and
philosophy.
GAGNON, D., CUDDY, S., CONTI, F., LINDSAY, C.,
2008. The effect of pore geometry on the distribution of He is the inventor of the Fractal FOIL Function that
reservoir fluids in U.K. North Sea oil and gas fields. describes the distribution of fluids in the reservoir model.
Transaction of the 49th Annual Logging Symposium of He writes AI software and has 45 years industry
the SPWLA, May 25-28, 2008. experience in petrophysics. In recognition of outstanding
service to the SPWLA, Steve was awarded the
KATZ, A.J., THOMPSON, A.H. 1985 Fractal sandstone Distinguished Service Award in 2018.
pores: implications for conductivity and pore formation.
Phys. Rev. Lett. Phys. Rev. Lett. 54, 1325–1328

KAY, S., CUDDY, S., 2002. Innovative Use of

Petrophysics in Field Rehabilitation, with Examples
from the Heather Field. Petroleum Geoscience, v.8, no 4,
pp. 317-325.

LEVERETT, M.C., 1941. Capillary behaviour in porous

solids: Trans AIME, Vol. 142.

LI, K., HORN, R., 2004. Universal Capillary Pressure

and Relative Permeability Model from Fractal
Characterization of Rock. Proceedings, Twenty-Ninth
Workshop on Geothermal Reservoir Engineering
Stanford University, Stanford, California, January 26-
28, 2004

LOZADA-ZUMAETA et al., 2012. Distribution of

petrophysical properties for sandy-clayey reservoirs by
fractal interpolation. Nonlinear Process. Geophys. 19,
239–250

18
 SPWLA 58 th Annual Logging Symposium, June 17-21, 2017

NOMENCLATURE AND DEFINITIONS

𝑎 Fractal Function constant (Line Intercept)

𝑏 Fractal Function constant (Line Gradient)
BVW Bulk volume of water (v/v)
The product of Sw and Phi
D Fractal dimension
FWL Free water level (feet)
Depth of zero capillary pressure
g Acceleration of gravity
H Height above the FWL (feet)
HWC Hydrocarbon water contact
N The number of the units (with a side of r)
that will fill entire object
Pc Capillary pressure (psi)
Phi Effective porosity (PU)
Sw Water saturation (%)
SwH Water saturation vs. height function
Swirr Irreducible water saturation
r Capillary radius
 Interfacial tension
 Contact angle
𝑤 Water density
𝑜 Oil density

19