You are on page 1of 27

Egy. J. Appl. Geophys., Vol. 12, No. 1, March 2013, 41-67.

THE OPTIMUM MODEL FOR PERMEABILITY


PREDICTION FROM MERCURY INJECTION
CAPILLARY PRESSURE DATA:
A CASE STUDY OF THE NUBIA SANDSTONE,
GULF OF SUEZ, EGYPT.
Mohamed S. El Sharawy
Geophysical Sciences Dept., National Research Centre, Dokki, Cairo, Egypt

Abstract
Permeability is one of the most important parameters that must be determined in a
reservoir evaluation. In a homogenous reservoir containing a simple pore system,
porosity can be possibly correlated with permeability. However, in a heterogeneous
reservoir with other than a simple pore system, parameters other than porosity are
needed to predict permeability such as pore throat size. In this work, several published
permeability models were reviewed for utility in the prediction of permeability, using
inputs obtained from the characteristics of the capillary pressure curve of the pore
system. We review these models and select the best one to apply in the Pre –
Cenomanian Nubia Sandstone reservoir, Gulf of Suez, Egypt. The Nubia Sandstone is
considered as one of the most prolific oil reservoirs in the Gulf of Suez. The Pre –
Cenomanian deposits consist mainly of sandstone with intercalations of shale belonging
to different depositional environments. The sand section thickness generally decreases
southward along the Gulf of Suez. This sandstone reservoir is considered to be
extremely heterogeneous based on the Dykstra-Parsons coefficient.
Among several permeability models evaluated; Timur, Thomeer, Kozeny -
Carman, Swanson, Jennings, Kamath, MRA, Hagiwara and Purcell, the study indicated
that the Thomeer model yielded the best results as indicated from the statistical
variables to validated core data. Other three models yielded close results, which are
Hagiwara, multiple regression analysis (MRA) and Kamath. However, the most
applicable model is the Timur equation. This work investigates the variations in the
quality of the model results, generates hypotheses for the varying permeability
prediction quality proposes. New equations to describe the relation between the
parameters used to estimate permeability and the petrophysical variables, such as
porosity are introduced.

1- Introduction: core geological material. Of course,


Permeability is the key reservoir cores are always rare, costly to measure
parameter, as it describes the flow of and sometimes suffering from
fluids in the formations and therefore, uncertainties. Furthermore, the
the deliverability of the accumulated measured cores are usually carried out
hydrocarbons. Permeability on a small scale that represents a part
measurements can only be obtained of the whole reservoir. So, the need to
using expensive and hard – to – obtain predict permeability is extremely

41
42 Mohamed S. El Sharawy

necessary. Nevertheless, permeability aeolian environment. The section


prediction is probably the most thickness generally decreases
challenging issue facing geologists, southward along the Gulf of Suez. This
petrophysicists and reservoir engineers. sandstone reservoir, according to
There have been several attempts to Gameel and Darwish, 1994, is ranged
correlate permeability with porosity. from poor to high quality reservoir
However, in the heterogeneous rock, based on the attributes of
reservoirs, as the studied reservoir, petrographic and petrophysical
such attempts will result in poor investigation of the Nubia section of
correlations. So, the need to other the Sidki oilfield. Applying Dykstra-
parameters other than porosity is Parsons (1950) coefficient (Vk),
arisen. In this paper, we use the indicated extremely heterogeneous
mercury injection capillary pressure reservoir with Vk more than 0.88 (Fig.
data obtained from core plugs to 2). Furthermore, the trimodal pore size
estimate permeability, using several distribution systems are the
mercury injection capillary pressure – predominant with some samples of
based empirical models in the prolific mono and bimodal (Fig. 3).
Pre – Cenomanian Nubia Sandstone oil The profound economic
reservoir, Gulf of Suez, Egypt. importance of the Nubia Sandstone
The Nubia sandstone of the leads to drive to improve the Nubian
Gulf of Suez Egypt is Pre- Cenomanian reservoir permeability characterization.
age consolidated sandstone generally In well A, two core plugs covering the
considered to be continental to shallow lowermost third part of the Nubia
marine with a wide range of grain sizes section (about 80 ft) were analyzed
and with low to high clay content. with 100 % recovery, using the
Commonly, the Nubia Sandstone conventional fiber glass coring type.
experiences various types of diagenesis The cored interval indicates that
and cementation. With this permeability ranges from 0.07 to 177
heterogeneity, the Nubia Sandstone md with average value of 14 md. In
exhibits monomodal, bimodal and this well, special core analysis (SCAL)
trimodal pore systems. The Nubia was performed on 15 core samples,
Sandstone has been prolific with about providing us with 15 mercury injection
one-fifth of the Gulf of Suez oil capillary pressure curves. The lithology
production of the samples are gray sand, very fine
to fine grained, with silt cement as well
2- Study area and the data: as traces of pyrite and anhydrite in
The Pre – Cenomanian Nubia some samples. In well B, routine core
Sandstone is considered as one of the analysis was carried out on three core
most prolific oil reservoirs in the Gulf plugs, which covered the whole Nubia
of Suez, Egypt. It produces oil from section. About 119 ft thick of mainly
many fields throughout the Gulf of sandstone were encountered in well B.
Suez (Fig.1). The Pre – Cenomanian The cored interval indicates that the
deposits consist mainly of sandstone permeability ranges from 0.004 to 1610
with intercalations of shale deposited in md, with average value of 169 md.
fluvial braided system graded to SCAL was performed on six core
THE OPTIMUM MODEL FOR PERMEABILITY 43

Fig. (1): Location map of the Gulf of Suez showing producing oilfields from
the Nubia sandstone and the studied well (modified after Alsharhan, and
Salah, 1997).
44 Mohamed S. El Sharawy

10000

Vk = k50-k84.1/k50
1000 Vk = 10.2-1.2/10.2 = 0.88

100
Permeability, md.

10

0.1

0.01

0.0010 .002 0.005 0.01 0.02 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95 0.98 0.99 0 .995 0.998

Cumulative frequency data

Fig. (2): Dykstra-Parsons Coefficient on a Log-normal probability graph for


well C showing a heterogeneous reservoir.
Vk = Dykstra-Parsons Coefficient.
k50 =permeability at 50% cumulative frequency.
k84.1 =permeability at 84.1% cumulative frequency.

samples, giving six mercury injection encountered in this well. The cored
capillary pressure curves (Fig. 4). In interval ranges from poor to good
well C, routine core analysis was reservoir quality. The recovery was
carried out on 13 cores plugs. These varied from 29% to 100%, using
core plugs covered the lower 600 ft conventional fiber glass coring type.
thick of 713 ft total thickness The cored interval indicates that
THE OPTIMUM MODEL FOR PERMEABILITY 45

(a)
10000 0.01

N ano-
Mercury capillary pressure, psi

1000 0.1

3
M icro-

100 1
M es o-

M acro-
10 10
1

M ega -

1 100
Pore throat
16 12 8 4 0 radius, microns
Incremental pore volume entered by mercury in %.

(b) (c)
10000 0.01 10000 0.01
Mercury capillary pressure, psi
Mercury capillary pressure, psi

1000 0.1 1000 0.1

100 1 100 1

10 10 10 10

1 100 1 100

20 16 12 8 4 0 20 16 12 8 4 0
Incremental pore volume entered by mercury in %.
Incremental pore volume entered by mercury in %.

Fig. (3): Pore size distribution system as illustrated from plotting of pore
throat size against incremental volume of mercury injection.
a) Trimodal pore size distribution system
b) Bimodal pore size distribution system
c) Monomodal pore size distribution system

permeability ranges from 0.01 to 11050 samples in well C, in which the air
md, with average value of 69 md. permeability, porosity and water
SCAL was performed on ten core saturation were measured.
46 Mohamed S. El Sharawy

The core analyses were 2.77 g/cc, with average value of 2.65
carried out based on the following g/cc. Mineralogy was dominantly
procedure: The core plugs used for quartz with traces of feldspar with 1%
tests were drilled, using a mineral oil as heavy minerals represented mainly by
the bit coolant. The one – inch – zircon, tourmaline and pyrite. The
diameter samples (well B), 2 5/8 inches sandstones are mainly quartzarenite
(well C) and 5– inches in (well A) for grading to lithicarenite at the
the routine core analysis were then lowermost part of well A. At this part,
placed in Dean – Stark type toluene volcanic rock fragments, chert and iron
distillation apparatus to determine the oxides are forming the main
core liquid saturation. The samples constituents of the rock. Consequently,
were then placed in a toluene permeability is reduced below 1 md.
extraction apparatus for extended The sandstone is usually light to dark
cleaning. Following cleaning, many of gray, fine to medium grains, poorly to
the samples appeared to be oil stained. well – sorted with siliceous cement.
These samples were then cleaned with Such composition resulted in the
chloroform and in a Co2/ toluene dominated bi and trimodal pore size
cleaner neither of these methods distributions. Clay content is generally
removed the oil staining. After drying, low, grading to be high in some
air permeability and helium porosity of intervals, especially at the base of the
the samples were measured. For MICP Nubia section (Fig. 5). This is due to
test, the samples were cleaned with the effect of authigenic kaolinite,
toluene and alcohol to remove which formed as a result of alteration
hydrocarbons and salts. Air of feldspars. The clays are mainly
permeability and porosity were kaolinite with rare illite, chlorite and
measured on the cleaned and dried core iron oxides. Figure (6) indicates that
plugs. Mercury was then injected into kaolinite represents the main clay
the cores, using pressure ranging from minerals, with a low percentage of
1 to 2000 psi. Pore size distributions heavy thorium – bearing minerals.
were calculated from these tests. Geological review (Gameel and
The cored intervals had well Darwish, 1994) tended to indicate the
logs and well logging data for the three porosity variation arises from siliceous
wells in the Southern Gulf of Suez that cement and the permeability variation
were utilized to examine the resulted from the filling of
permeability prediction models. The intergranular spaces by clays. These
well log data include Gamma ray, are highly inversely correlated. Above
spectral gamma ray, sonic, density, 10% clay content, low permeabilities
neutron and resistivity. The routine (k<1 md) have been occurred (Fig. 7).
core analyses indicated that the Nubia The problem resulting from
Sandstone reservoir permeability the occurrence of kaolinite is related to
ranges from poor to very good; from its structure. Patches of kaolinite are
0.01 to 1610 md, with average value of loosely attached to sand grains.
77 md. The porosity ranges from 0.012 Because of the large size of these
to 0.22, with average value of 0.131 patches, they can block the large pore
and grain density ranges from 2.55 to throats, resulting in a permanent,
THE OPTIMUM MODEL FOR PERMEABILITY 47

10000

1000
Mercury injection capillary pressure, psi

100

10
K = 853 PHI = 19.3
K = 318 PHI = 17.9
K = 342 PHI = 20.8
K = 120 PHI = 15.3
K = 0.462 PHI = 9.5
K = 174 PHI = 17.9
1

100 90 80 70 60 50 40 30 20 10 0
Mercury saturation, %

Fig. (4): Mercury injection capillary pressure curves for six samples in well B.

irreversible damage to the formation. According to Tiab and


permeability. The problem increased Donaldson (2004), the dispersed clays
with the dispersed mode of clay contain more bounded water and in
distribution, as in the study reservoir, core analysis, much of this bounded
which evolved from the in - situ water is lost during the drying process,
alteration and precipitation of various which results in an overestimation of
clay minerals. These authigenic clays the porosity of the core sample.
may adhere and coat the sand grains, or Relation between the core
they may partially fill the pore spaces porosity and core permeability for all
resulted in considerably reducing measured samples in the three wells
permeability and porosity of the indicates fair correlation with a
48 Mohamed S. El Sharawy

Fig. (5): Litho – saturation crossplot of well A showing increase of shale content
downward due to the effect of authigenic kaolinite.
THE OPTIMUM MODEL FOR PERMEABILITY 49

Fig. (6): Crossplot of thorium (ppm) versus potassium (%) in well C to


determine clay type. It showed that the clay is mostly kaolinite with rare illite
and the uppermost points indicate heavy thorium – bearing minerals.

coefficient of determination (r2) = 2009). Following is a review for the


0.548 (Fig. 8). So, permeability is most common models:
controlled by parameters other than
porosity. Investigation of such 3-1- kozeny – Carman (1937) model:
parameters is a subject of the following The most fundamental and popular
sections. In this study, we examine the correlation that relating permeability as
permeability data above 0.1 md a function of porosity and capillary
exclusively. pressure tube radius is the Kozeny –
Carman equation, which has the
3- Capillary pressure models for following form:
permeability prediction: Φr 2
Several classic models put in Thomeer K= ……………………….(1)
and Swanson references, here were 8
developed to correlate permeability Where: K is permeability in cm2, Ф is
with the characteristics of the capillary the porosity in fraction and r is the
pressure curve. Recently, efforts to capillary tube radius in cm. If the
revive these early attempts have been capillary tubes are tortuous, the above
revisited and improved (e.g. Huet et al., equation can be written as the
2005; Glover et al., 2006 and Clerke, following:
50 Mohamed S. El Sharawy

Φr 2
K= ……………………….(2)

Where: τ is the tortuosity, which can be
defined as (Tiab and Donaldson, 2004):
τ = ΦF ……………………………(3)
Where: F is the formation resistivity
factor that equals (Archie, 1942):
F = Φ − m ……….………………(4)
Where: m is the cementation exponent.
So, the Kozeny – Carman equation can
be taken the following form:
r2
K= ………………………..(5)
8F
With taking into consideration the
effect of surface area, the Kozeny –
Carman equation can be written as
follows:
Φ3
k= ……………(6)
2τ (1 − Φ ) Sg 2
Where: Sg is the surface area per grain
volume and can be expressed using the
following equation:

r= …………………..(7)
(1 − Φ ) Sg
Amaefule et al. (1993) used the
Kozeny-Carman (1937) model and
developed an expression for the mean
hydraulic radius to construct a
graphical technique for isolating the
flow units. The Kozeny-Carman
equation was re-arranged to yield:
k Φ 1
=( )( ) …...(8)
Φ 1− Φ FsτSg
Where: K is permeability (10-12 m2), Φ
is the effective porosity, Fs is the shape
Fig. (7): Crossplot of clay content
factor, usually ranged between 2 and 3
versus core permeability in well C. It
(Carman, 1956), τ is the tortuosity and
can be noted that at clay content
Sg is the surface area per unit grain
above 10 %, permeability begins to
volume in micron. The authors then
reduce.
THE OPTIMUM MODEL FOR PERMEABILITY 51

Fig. (8): Crossplot of horizontal permeability versus core porosity showing


fair relation as indicated from coefficient of determination.

defined the flow zone indicator as tortuosity, low surface area and high
follows: FZI. The surface area per unit grain
1 volume can be expressed as follows:
FZI = ………….……..(9) Φ 2Φ
FsτSg Sg = Sp = …(10)
The flow zone indicator (FZl) is a (1 − Φ ) r (1 − Φ )
unique parameter that includes the Where: Sp is the surface area per unit
geologic attributes of the texture and pore volume, and:
mineralogy in the structure of distinct τ = ΦF
pore geometrical facies (Tiab and Then, permeability in md can be
Donaldson, 2004). Poorly sorted sands formulated, from the previous four
tend to exhibit high tortuosity, high equations as follows:
surface area and low FZI, while well –
sorted sands tend to exhibit low
52 Mohamed S. El Sharawy

Φ3 dimensionless pore radius distribution


K=1014(FZI)2 ( ) ……(11) (pore throat impedance) and the
(1 − Φ ) 2
coordination number of the pore throats
Where: 1014 is the conversion factor per pore body, respectively.
for permeability that ranged between Blasingame and Ali (1995, unpublished
m2 and md. It is clear that equation (5) work presented in Wu, 2004) indicated
is the simple form of the equations 6 & that:
11. ω (1 − Swir )
=α ……………….(14)
n Φ
3- 2- Purcell (1949) model: Where: α is an empirical adjustment
Purcell (1949) introduced a semi – constant that is typically set equal to 1.
empirical equation for permeability, as Se is the normalized water saturation,
a function of porosity, non - wetting which according to Brooks and Corey
phase and capillary pressure. The (1966), equals:
equation was expressed as follows: −1
dSnw Sw − Swir Pc
K=10.66(σHg,Acosθ)2FpФ ∫ …
1
Se = = ( ) λ …….(15)
0 Pc 2 1 − Swir Pd
…………………………..……….(12) Where: Pd is the displacement pressure
in psi and λ is the pore throat size
Where: K is permeability in md, σHg,A
is the mercury-air interfacial tension in distribution index. According to
dynes/cm (480 dynes/cm), cosθ is the Brooks and Corey (1966), the porous
media having narrow pore throat size
contact angle of mercury in air (140ο) ,
Ф is the porosity in fraction, Snw is the distributions (well sorted) tend to have
non - wetting phase saturation in large values of λ, while porous media
having wide pore throat size
fraction, Pc is the capillary pressure in
psi and Fp is the Purcell lithology distributions (poorly-sorted) tend to
have small values of λ. Incorporating
factor, which represents the tortuosity
of the system. According to Calhoun et the previous three equations,
al. (1949), the Purcell lithology factor, Blasingame and Ali (1995) derived the
following equation:
Fp, was inversely related to the
formation tortuosity (τ). An empirical K = 10.66(σHg . A cos θ ) 2
value of 0.216 was found for it, in
Φ 2 λ .…....(16)
clean sandstone (Wu, 2004). Based on (1 − Swir ) 4 ( ) ( )
Purcell (1949) and Brooks and Corey Pd λ + 2
(1966), Nakornthap and Evans (1986)
derived the following equation: 3- 3- Timur (1968) model:
K = 10.66(σHg . A cos θ ) 2 Timur (1968) investigated the
permeability prediction for sandstone
ω 1 dSe ……...(13) reservoirs, using indirect in-situ
[Φ (1 − Swir ) ∫ Pc
3
2 measurements of porosity and residual
n 0 water saturation, using an early version
Where: Swir is the irreducible wetting of the nuclear magnetic resonance
phase, ω and n are the empirical pore (NMR) tool. Timur developed an
geometry terms that account for the empirical relation for the permeability
THE OPTIMUM MODEL FOR PERMEABILITY 53

in terms of porosity and irreducible


water saturation, Swir, using the core 3- 5-Swanson (1981) model:
and well log data for 155 sandstone Swanson (1981) developed an equation
samples. The Timur equation has the for the permeability as a function of the
following form: capillary pressure curve characteristics.
Φb The general form of the equation is:
K =a …………………..(17) Vb b
Swir c K = a( ) apex ………………...(21)
Where: K is the permeability in md; Ф Pc
and Swir are in %. He found that a = Where: K is the permeability (md), Vb
0.136, b = 4.4 and c = 2. is the bulk volume mercury saturation
(%) and Pc is the capillary pressure
3- 4- Winland (1980) equation: (psi) corresponding to the apex of a
Winland developed an empirical hyperbolic log-log mercury injection
relationship between porosity, plot. The coefficients a and b are
permeability and pore size determined from regression. The
corresponding to a mercury saturation Vb
of 35% (r35), based on multiple ( ) apex term is the inflection point
regression analysis (MRA). The
Pc
of the capillary pressure curve. The Pc
Winland equation was published by
apex reflects the dominant
Kolodzie in1980. It has the following
interconnected pores and pore throats
form:
controlling most of the fluid flow.
Log r35 = 0.732 +0.588 log K – 0.864
Swanson best fit equation, based on
logФ………………………………(18)
319 clean sandstone and carbonate
samples, is:
Where: r35 is the pore aperture radius
corresponding to the 35th percentile in Vb 1.691
K = 339( ) apex …………(22)
micron, K is the permeability in md Pc
and Ф is the porosity in %. In 1992,
Pittman carried out several analyses 3- 6- Thomeer (1983) model:
and found that the best fit between the Thomeer (1983) developed an equation
three parameters occurred at r36. Using for the permeability as a function of
the permeability as a dependent three parameters, pore geometrical
variable, Pittman derived the following factor (Fg), displacement pressure (Pd)
empirical relationships: and the fractional bulk volume
Log K = -1.221 + 1.415 log Ф + 1.512 occupied by mercury at infinite
log r25………………….……….…(19) pressure (Vb ∞ , assumed equal to the
Log K = -0.861 + 1.185 Log Ф + porosity). Thomeer’s (1960 and 1983)
1.627 Log rapex ……..………...…(20) pore geometrical factor (Fg) is a widely
Where: r25 is the pore aperture radius used parameter to delineate the pore
corresponding to the 25th percentile in throat distribution. According to
micron, rapex is the pore radius Thomeer (1960), well-sorted pore
corresponding to the apex (µm), k is throats tend to have smaller Fg values,
the permeability in md and Ф is the while poorly-sorted pore throats tend to
porosity in %.
54 Mohamed S. El Sharawy

have larger Fg values. The Thomeer 3- 8- Jennings (1987) model:


equation has the following form: Jennings (1987) found that the second
Vb∞ 2 quartile capillary pressure was well
K = 3.8068 Fg −1.3334 ( ) …..(23) correlated with permeability for the
Pd
well – sorted core samples, in which
Assuming, Vb ∞ = Ф, so
the permeability can be calculated by
100Φ 2 the following expression:
K = 3.8068Fg −1.3334 ( ) …
Pd K = exp(-2.5 ln(P2) + 11.9)………(29)
……………………….…………(24) Where: P2 is the second quartile
In shaly sand, Wu and Berg (2003) capillary pressure in psi. The second
assumed that: quartile non – wetting phase saturation
Vb ∞ = Ф (1-Swir)……………….(25) can be calculated as follows:
Therefore, SnwQ2 = 0.5 (1-Swir) ………….…..(30)
Φ (1 − Swir ) 2 Where: SnwQ2 is the second – quartile
K = 3.8068Fg −1.3334 ( ) non – wetting phase saturation. Then,
Pd the P2 can be read off the capillary
…………………………………….(26) pressure at mercury saturation equals
Exponent (2) can be treated as the SnwQ2.
cementation exponent "m", according
to Wu (2004) and therefore, 3- 9 -Kamath et al. (1992) model:
permeability can be calculated as Kamath et al. (1992) performed a
follows: comparison between the previous
Φ (1 − Swir ) m models and concluded that the best
K = 3.8068Fg −1.3334 ( ) result is obtained with new
Pd correlations, based on the Swanson
…………………………………….(27)
Vb
(1981) characteristic length ( ) apex .
3- 7 -Hagiwara (1984) model: Pc
In 1984, Hagiwara developed a semi- They proposed two power law
empirical correlation for the correlations, according to the
permeability as a function of porosity, permeability range:
cementation factor and the average K = 413 Lmax
1.85
if K <1 md
pore throat radius. Using 24 sandstone
samples, the best least-squares line fit …………………………………….(31)
1.6
for the air permeability was obtained
using the following expression: K = 347 Lmax if K >1 md
K = 19.8 Φ rm 2 …………………………………….(32)
avg . ……………….(28)
Where: Lmax is the characteristic length
Where: K is the permeability in µm2, Φ defined as follows:
is the porosity in fraction and ravg is the
ΦSnw Φλ (100 − Swir )
average pore throat radius in µm. Lmax= ( ) max = 1
Hagiwara stated that “the constant Pc +1
multiplier (19.8 for these data) pd (1 + λ ) λ
accounted for the effects of tortuosity …………………………………….(33)
and other pore geometrical factors”. The main contribution of the Kamath et
al. model is providing a formulation to
THE OPTIMUM MODEL FOR PERMEABILITY 55

predict the low permeability rocks ( k < Pd is correlated with FZI. This relation
1 md). Determination of the parameters yielded r2 = 0.924 (Fig. 9C). The
Vb relations can be formulated as follow:
(such as reff, ravg, λ, Fg, Pd, ( ) apex Pd = 0.065 Ф-2.62 r2 = 0.625
Pc …………………………………….(37)
and Swir) used in the previous
mentioned models was well explained Pd = 17.085(FZI) −0.324 r2 = 0.924
in several works e.g. Jennings, 1987; …………………………………..(38)
Pittman, 1992; Doublet, 2001; Holmes, Pore geometrical factor (Fg) that was
2002 and Wu, 2004. determined from Wu (2004) method,
yielded weak relations with porosity.
4- Prediction of the parameters used This relation is improved somehow,
to estimate permeability: when correlated with FZI (Fig. 9D&E).
The predicted permeability models The relations are given as:
contain several parameters that need to Fg = −0.4623 ln(Φ ) − 0.557
determine. In this section, we will try r2 = 0.481 ………………………..(39)
to correlate these parameters with a Fg = −0.1267 ln(FZI) + 0.4119
known one, such as porosity, as well as
r2 = 0.578 ………………………..(40)
flow zone indicator (FZI). The
The term Lmax contains four parameters
incorporated parameters usually are
that mostly determine the properties of
Swir, reff, ravg, Pd, λ, Fg, Lmax, (Vb/Pc)A
the rock pore system. It gives a good
and P2. Table (1) summarized such
correlation with porosity and strong
relations with the corresponding
one with FZI (Fig. 9F). The relations
coefficient of determination (r2) for
take the following forms:
each one. From this Table, we can
Lmax = 2365.37 Ф4.895 r2 = 0.784
conclude that there is no correlation
…………………………………..(41)
between Swir and Φ. The relation
Lmax = 0.0804 FZI1.285 r2 = 0.925
improved dramatically with FZI, which
………………………...…………(42)
yielded r2 = 0.628 (Fig. 9A). The
The pore throat size
relation is given as:
distribution index (λ) gives so weak
Swir = 0.1186( FZI ) −0..791 ….……(34) relation, especially with that extracted
Slightly good relation exists between from the type curve. The correlation is
the effective and average pore throat improved somehow with (λ) derived
radii and porosity, which yielded r2 = from plotting Pc vs. Se, which yielded
0.71 and 0.715, respectively. The r2 = 0.4, when correlated with porosity.
relations can be expressed as the The relation still weak when FZI is
following: used in correlation with r2 = 0.43. So, it
reff = 6091.6 Ф3.64 r2 = 0.71 can be concluded that, the pore throat
…………………………………….(35) size distribution index is not reflecting
ravg = 3659.2 Ф3.63 r2 = 0.715 the physical properties of the rock. On
…………………..………..………(36) the other hand, Swanson's term
Good relation exists between Vb
the porosity and the displacement ( ) apex yielded good correlation
pressure, Pd. On the other hand, an Pc
excellent relation can be obtained when with porosity, and improved strongly
56 Mohamed S. El Sharawy

Table (1): Correlation of the parameters used to predict permeability with


porosity, formation resistivity factor and flow zone indicator (T.C = type curve).

when correlated with FZI (Fig. 9G&H). Vb


The relations can be formulated as concluded that ravg and ( ) apex may
follow: Pc
reflect the same criteria on the capillary
Vb
( ) apex = 1478.55 (Φ)4.79 r2 = 0.81 pressure curve.
Pc
… ……………………………….(43) 5- Results and discussions:
Vb Twenty one core samples, obtained
( ) apex = 0.0637 (FZI)1.231 from wells A and B were used to
Pc evaluate the permeability prediction
r2 = 0.913 …………….………….(44) models, based on mercury injection
Correlation of the second quartile capillary pressure characteristics. In
capillary pressure (P2) with porosity addition to ten samples, from well C,
yielded r2 = 0.665. The relation has the were used to evaluate Timur model.
following form: Table (3) listed the core porosity, air
P2 = 0.04 Φ-3.68 …………………(45) permeability and the predicted
The relation becomes strong, when permeabilities, based on the different
correlated with FZI (Fig. 9I). It yielded discussed models. To evaluate the
r2 = 0.917 and has the following form: goodness of fit and the qualify
performance of the discussed models,
P2 = 91.49 FZI-1.082 …………….(46) log - log correlation plots were used
specifically for the perfect correlation
So, it can be concluded that, there are (1:1 trend) and 2.5 – factor trends. The
good relations between porosity and reff ranking of the permeability prediction
and ravg, and strong relation models was based on the accumulative
Vb rank of some statistical variables
with ( ) apex . However, both (Fg)
Pc (Table 4). These statistical variables
and (λ) give weak correlations with include the sum of squared error (SSE),
porosity and FZI. The parameters, residual mean square (MRE),
interrelations and the corresponding coefficient of determination (r2), F –
statistical variables are summarized in test and the standard error (σe).
Table 2 and graphically represented in Four forms for the Kozeny –
Figure (10). From this Table, it can be Carman model were discussed (Eqs. 1,
5, 6 & 11). Figure (11) indicated that
THE OPTIMUM MODEL FOR PERMEABILITY 57

Fig. (9): Collection chart showing the interrelations among the parameters
used to predict air permeability from mercury injection capillary pressure –
based models.

Equations (1, 5 and 6) give over – estimated predicted values were


estimated permeability, especially for obtained for permeability below 10 md.
Equation 1, where the error reached The Purcell model with (λ) calculated
two orders of magnitude. The from mercury injection capillary
correlation is improved by using pressure versus normalized water
Equation (11). However, under – saturation (Se) yielded good correlation
58 Mohamed S. El Sharawy

Table (2): showing the relations among the terms involved in the permeability
prediction models and corresponding r2.

No. Equation r2 SSE RMS


1 reff =96.78 Pd-1.237 0.902 1.681 0.0885
2 reff = 24.81 (Vb/Pc)0.759 0.935 1.107 0.0583
3 reff = 19.27 Lmax0.759 0.942 0.998 0.052
-0.8372
4 reff = 141.475 P2 0.764 4.033 0.212
5 ravg = 15.475 (Vb/Pc)0.7927 0.969 0.524 0.0276
6 ravg = 11.56 Lmax0.745 0.92 2.241 0.118
0.892
7 ravg = 11.105 exp (-0.093 Pd) 1.824 0.096
0.853
8 ravg = 101.781 P2-0.8783 2.48 0.1305

9 (Vb/Pc)A = 0.61 exp (-0.1113 Pd) 0.827 4.49 0.2363


10 (Vb/Pc)A = 0.6566 Lmax0.9075 0.885 2.985 0.157
11 (Vb/Pc)A = 11.05 P2-1.1147 0.891 2.827 0.1488
12 Pd = 0.09762 P2 – 5.3 0.753 429 22.58

(Fig. 12). The strength of the relation Using rapex instead of r20 yielded, the
decreases when using λ derived from following relation with r2 = 0.837:
other sources. Log k = 4.184 log Φ + 1.262 log rapex -
Timur equation contains two 4.124 …………………………….(49)
parameters to predict the permeability. Using the displacement pressure (Pd)
In this study, the Timur equation takes and the porosity as independent
the following form, with r2 = 0.779: variables and permeability as a
Φ 6.65 dependent variable, the following
k = 0.0000013 ….……(47) relation is obtained with r2 = 0.894:
Swir 0.43 Log k = 3.022 log Ф – 1.582 log Pd +
Applying the multiple regression 5.571 …………….………………(50)
analysis (MRA) yielded the best Applying the Swanson (1981) model
relation with r20, as an effective pore gives r2 = 0.86 with the following
size, when using the permeability as a form:
dependent variable. The relation has Vb 1.445
the following form with r2 = 0.906: K = 457.5 ( ) apex ………….…..(51)
Log K = 1.953 log Φ + 1.432 log r20 - pc
1.8806 ……………………………(48)
THE OPTIMUM MODEL FOR PERMEABILITY 59

Fig.(10): Comparison between measured and predicted permeabilities using


different models.
60 Mohamed S. El Sharawy

Table (3): Showing the study data listing of core porosity, air permeability and
the predicted permeabilities using various mercury injection capillary pressure –
based models.

Table (4): Ranking for the capillary pressure estimation models.

Thomeer (1983) model (Eq. 24) gives a porosity can be used instead of the
good correlation, when using (Fg), that fractional bulk volume occupied by
derived from the type curve. Using mercury at infinite pressure (Vb ∞ )
(Fg) derived from Wu (2004) method, with no real effect on the results.
the relation is improved considerably Using the second quartile
and recorded the highest correlation capillary pressure, as defined by
among the used models (Table 3). The
THE OPTIMUM MODEL FOR PERMEABILITY 61

Fig. (11): Comparison between measured and predicted permeabilities


using different forms of Kozeny – Carman equation.

Jennings (1987), the following relation Kamath (1992) model depending on


has been obtained: the Swanson approach gives good
K = exp (-1.58 ln (P2) + 9.485) correlation based on (λ) calculated
……….……………...…...…….….(52) from the mercury injection capillary
Applying the average pore throat radius pressure versus normalized water
and the porosity powered to saturation (Se). The relation takes the
cementation exponent (m), as following form:
introduced by Hagiwara (1984), K = 302.5 Lmax1.43 for k> 0.1 md
yielded the following equation: ……..……….…………………….(54)
k = 170.776 Φm r1.45avg ………..….(53) Consequently, nine mercury
injection capillary pressure – based
62 Mohamed S. El Sharawy

Fig. (12): Collection chart showing correlation of porosity and flow zone
indicator versus different parameters used to permeability prediction.

models were reviewed to predict controlled by three parameters; Fg, Pd


permeability. All the models contain and Vb∞ will weakness the end results.
porosity, except for Swanson and The latter parameter can be replaced by
Jennings models (Table 5). Another porosity, but Pd gives moderate
important parameter is the pore throat relation with porosity (Eq. 37).
size, which incorporates in some Furthermore, being Pd powered to
models either the effective or average. negative exponent, will increase the
According to the cumulative ranking, effect of uncertainty. Additionally, Fg
Thomeer (1983) model gives the gives weak correlation with porosity
highest rank (Table 3). Other three and moderate correlation with FZI.
models give good results. These Elimination of Fg from the Thomeer
models are Hagiwara, MRA and model and correlation, the term (Ф/Pd)
Kamath. with permeability yielded a result, as
Thomeer (1983) model gives similar to Equation (50).
the best correlation, but being
THE OPTIMUM MODEL FOR PERMEABILITY 63

Table (5): showing the used parameters in the permeability predictions models.

No. Model Used parameters

Modified kozeny –
1 Ф ,reff, τ, F, Sg, FZI, Fs
Carman (Eq. 11)

2 Purcell Ф , Pd , Swir, λ, σ, θ

3 Timur Ф, Swir

4 MRA Ф, reff, rapex, rPd

5 Swanson (Vb/Pc)A

6 Thomeer Ф , Pd, Vb ∞ , Fg

7 Hagiwara Ф, ravg, m

8 Jennings P2

9 Kamath et al. Ф, Pd, Swir, λ

Hagiwara et al. (1984) model gives 35 as a function of porosity, although


strong correlation, depending on the the relation is not strong. Kamath et al.,
porosity powered to cementation model based on the Lmax term gives
exponent and the average pore throat good and strong relations with Ф and
radius. The former parameter can be FZI, respectively. This is due to Lmax
easily obtained from a routine core term involved Ф and reff. So, Hagiwara,
analysis or well logging tools, such as MRA and Kamath models can be
neutron and density, when calibrated applied to predict the air permeability
with the core data. The average pore in uncored interval for the Nubia
throat size is the parameter needed to Sandstone reservoir.
predict. This can be done using the The strong correlation
porosity as indicated in Equation (36). between FZI and Swanson term
Using the MRA, two parameters are Vb
needed; Ф and reff. The effective pore ( ) apex urges to develop a relation
throat size can be predicted using Eq. Pc
64 Mohamed S. El Sharawy

Fig. (13): Comparison between measured and predicted permeabilities


using Thomeer and Timur models, when no capillary pressure is available.

Vb model, based on equation (47). The


with replaced ( ) apex by FZI. So, a goodness of the Timur model is close
Pc to the Thomeer model (Fig. 13). Such
new improved relation can be obtained. closeness makes the Timur equation
It has the following form with r2 = the most applicable model in the study
0.892: area with the least required parameters.
K = 8.6925 (FZI)1.7788 ……..…….(55)
Also, improving Jennings 6- CONCLUSIONS:
(1987) model can be made through • Thomeer model gives the best
replacing P2 by FZI. The new form of results to predict the air permeability
the relation yielded r2 = 0.892. It can be for the Nubia Sandstone reservoir in
written as the following: the Gulf of Suez, Egypt, based on the
K = exp (2.705 ln (FZI) +0.61) mercury injection capillary pressure
……………………..……………(56) analysis. As well as, Hagiwara, MRA
To explain the applicability of and Kamath models give reasonable
such models in a reservoir has no results. However, Timur equation can
capillary pressure tests, we used a be considered the most applicable
Nubian sandstone interval containing model to predict the air permeability in
air permeability, porosity and water an uncored interval for the Nubia
saturation. We tested the first four rank Sandstone reservoir.
models; Thomeer, Hagiwara, MRA and • The Thomeer (Fg) and the
Kamath as well as Timur model. The Brooks and Corey (λ) can't be
Thomeer model gives the best results, considered as indicators of the change
based on equations 26, 37 and 39. The of pore system properties of the rock,
second best result was with the Timur where they have weak correlations with
THE OPTIMUM MODEL FOR PERMEABILITY 65

porosity. In contrast, reff, ravg, P2 and Units and Predict Permeability in


Vb Uncored Intervals/Wells. SOC.
( ) apex are good indicators of the Petrol. Eng. Paper No. 26436.
Pc Archie G. E. 1942. The theoretical
change of the physical rock properties. resistivity log as an aid in
• Beside porosity, the effective determining some reservoir
pore throat size can be considered the characteristics. American Institute
main key in controlling permeability, of Mining and Metallurgical
which determines the magnitude of Engineers, Transactions146, 54-
some parameters on the capillary 61.
pressure curve, such as the Brooks R. H. and Corey A. T. 1966.
displacement pressure and pore Properties of porous media
geometrical factor. affecting fluid flow. Journal of
• The best relation controlled Irrigation and Drainage Division,
the tortuous nature of the pores must be Proceedings of the American
determined in order to obtain more Society of Civil Engineers 92, No.
accurate FZI. IR 2, 61-88.
• New modified forms for the Calhoun J.C., Lewis M. and Newman
Swanson and Jennings models are R.C 1949. Experiments on the
Vb Capillary Properties of Porous
introduced by replacing ( ) apex and Solids. Trans., AIME 186, 189-
Pc 196.
P2 by FZI. Such modifications resulted Carman P. C. 1956. Flow of gases
in an improvement of the permeability through porous media,
prediction results for the two models. Butterworth's Scientific
Publications, London, 128 p.
Acknowledgements Carman P.C. 1937. Fluid flow through
The Author is grateful to the Egyptian granular beds. American Institute
General Petroleum Corporation of Chemistrical Engineers
(EGPC) and the Gulf of Suez Transactions 15, 150-166.
Petroleum Company (GUPCO) for Clerke E.A. 2009. Permeability,
providing the data. relative permeability, microscopic
displacement efficiency, and pore
References: geometry of M_1 bimodal pore
Alsharhan A.S. and Salah M.G., 1997. systems in Arab D Limestone. SPE
Lithostratigraphy, sedimentology Journal, no. 105259, 8 p.
and hydrocarbon habitat of the Doublet L.E. 2001. An integrated
Pre-Cenomanian Nubian geologic and engineering reservoir
Sandstone in the Gulf of Suez oil characterization of the North
Province, Egypt. GeoArabia, 2, Robertson (Clear Folk) Unit,
no. 4, 385-400. Gaines County, Texas. PhD
Amaefule J. O., Altunbay M. H., Tiab Dissertation, Texas A&M
D., Kersey D. G. and Keelan D. K. University, v. I, pp. 327.
1993. Enhanced Reservoir Dykstra, H. and Parsons, R. L. 1950.
Description Using Core and Log The prediction of oil recovery in
Data to Identify Hydraulic Flow
66 Mohamed S. El Sharawy

waterflood. Secondary Recovery Kolodzie S. Jr. 1980. Analysis of pore


of Oil in the United States, 2nd ed. throat size and use of the
American Petroleum Institute Waxman-Smits equation to
(API), 1950, 160-174. determine OOIP in Spindle Field,
Gameel M. and Darwish M. 1994. Colorado. Society of Petroleum
Reservoir behavior of the Pre- Engineers, 55th Annual Fall
Turonian sandstones in south Gulf Technical Conference, Paper SPE-
of Suez province (Sidki field – 9382, 10 p.
case history). 12th EGPC Nakornthap K. and Evans R. D. 1986.
Exploration and Production Temperature-dependent relative
Conference 2, 449-471. permeability and its effect on oil
Glover P. W. J., Zadjali I. and Frew K. displacement by thermal methods.
A. 2006. Permeability prediction SPE Reservoir Engineering, May,
from MICP and NMR data using 230-242.
an electrokinetic approach. Pittman E. D. 1992. Relationship of
Geophysics 71, F49–F60. porosity and permeability to
Hagiwara T. 1984. Archie’s m for various parameters derived from
permeability. SPE paper 13100, mercury injection-capillary
Proceedings of the Society of pressure curves for sandstone.
Petroleum Engineers 59th Annual AAPG Bulletin 76, 191-198.
Technical Conference and Purcell W. R. 1949. Capillary
Exhibition, 9p. pressures--their measurement
Holmes M. 2002. System for using mercury and the calculation
evaluating fluid distributions of of permeability therefrom.
subsurface reservoirs. US Patent, American Institute of Mechanical
no. 6,484,102B1, 17P. Engineers, Petroleum
Huet C.C., Rushing J.A., Newsham Transactions, Feb., 39-48.
K.E. and Blasingame T.A. 2005. A Schlumberger, 1982. Essentials of
modified Purcell/ Burdine model NGS interpretation.
for estimating absolute Swanson B. F. 1981. A simple
permeability from mercury- correlation between permeabilities
injection capillary pressure data. and mercury capillary pressures.
IPTC, no. 10994,12 p. Journal of Petroleum Technology,
Jennings J. B. 1987. Capillary pressure Dec., 2488-2504.
techniques: application to Thomeer J. H. M. 1960. Introduction of
exploration and development a pore geometrical factor defined
geology. AAPG Bulletin 71, 1196- by the capillary pressure curve.
1209. Journal Petroleum Technology,
Kamath J., Boyer R. E. and Nakagawa March, 73-77.
F. M. 1992. Characterization of Thomeer J. H. M. 1983. Air
Core-scale Heterogeneities using permeability as a function of three
Laboratory Pressure Transients. pore network parameters. Journal
SPE Formation Evaluation, of Petroleum Technology, 809-
September, 219-227. 814.
THE OPTIMUM MODEL FOR PERMEABILITY 67

Tiab D. and Donaldson E.C. 2004.


Petrophysics. 2nd edition, Gulf
Professional Publishing.
Timur A. 1968. An investigation of
permeability, porosity, and
residual water saturation
relationships for sandstone
reservoirs. The Log Analyst 9, 8-
17.
Wu T. 2004. permeability prediction
and drainage capillary Pressure
simulation in sandstone reservoirs,
PhD Dissertation, Texas A&M
University, pp. 168.
Wu T. and Berg R. R. 2003. A method
for synthesizing and averaging
capillary pressure curves. AAPG
Annual Meeting, Salt Late City,
Extended Abstract, 3 p.

You might also like