Society of Petroleum Engineers

SPE 26436
Enhanced Reservoir Description: Using Core and Log Data
to Identify Hydraulic (Flow) Units and Predict Permeability in
Uncored Intervals/Wells
Jude O. Amaefule* and Mehmet Altunbay*, Core Laboratories; Djebbar Tiab*, U. of Oklahoma; David G. Kersey
and Dare K. Keelan*, Core Laboratories

·SPE Member

Copyright 1993, Society at Petroleum Engineers, Inc.

This paper was prepared for presentation at the 68th Annual Technical Conterence and EXhibition of the Society of Petroleum Engineers held in Houston, Texas, 3-6 October 1993.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented. have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society
of Petroleum Engineers, Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment
of where and by whom the paper is presented. Write librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A. Telex, 163245 SPEUT.

Abstract technique has been successfully tested in clastic rocks from East
Texas. South America. West Africa. South East Asia and Far East
Understanding complex varcatlOns in pore geometry within Asia, as »'ell as carbonate sequences from West Texas and
different lithofacies is the key to improved reservoir description Canada, This paper documents the theoretical development.
and exploitation. Core data provide information on various validates and characterizes the hydraulic units, and presents
depositional and diagenetic controls on pore geometry, predicted versus actual permeability data to demonstrate the
Variations in pore geometrical attributes in turn. define the efficacy of the proposed technique.
existence ofdistinct zones (hydraulic units) with similar fluid-flow
characteristics. Classic discrimination of rock types has been
based on subjective geological observations and on empirical Introduction
relationships bern'een the log of permeability versus porosity.
One of the most important existing and emerging challenges of
However, for any porosity within a given rock type. permeability
geoscientists and engineers is to improve reservoir description
can vary by several orders of magnitude. which indicates the
techniques, It is well recognized that improvements in reservoir
existence of several flow units.
description will reduce the amount of hydrocarbon left behind
In this paper, a new. practical and theoretically correct pipe, Accurate detennination of pore-bodylthroat attributes and
methodology is proposed for identification and characteri:ation fluid distribution are central elements in improved reservoir
of hydraulic units within mappable geological units (facies). The description. Many reservoir description programs. though
detailed. have not included descriptions at the pore-throat scale.
technique is based on a modified Ko:eny-Carmen equation and
Yet. pore-throat attributes control initiaUresidual hydrocarbon
the concept ofmean hydraulic radius, The equation indicates that
distrihution and fluid flow. Because they are readily available.
for any hydraulic unit. a log-log plot of a "Reservoir Quality
continuous sources of data. logging tool responses are often used
Index." (RQI). which is equal to O_03l4~. versus a
to draw inferences ahout lithology. depositional and diagenetic
"Normaliud Porosity Index" (4),) which is equal to 4>/( 1-4» should
sequences. and fluid content. These inferences are based on
yield a straight line with a unit slope. The intercept of the unit empirical models utilizing cOlTciations among tool responses.
slope line with 4>, = I, designated as the "Flow Zone Indicator" rock and fluid properties. In many instances. unfortunately. the
(FZI), is a unique parameter for each hydraulic unit. RQI. 4>, and correlation models can not be used globally because of the
FZI are based on stressed porosity and permeability data influences of factors not fully considered by the models. Factors
measured on core samples. include (a) the presence of potassium-feldspar. zircon. etc.
causing erroneously high calculated Vsh from the gamma ray; (b)
FZ1 is then correlated to certain combinations of logging tool microporosity in kaolinite. chert. etc. leading to high apparent
responses to develop regression models for permeability water saturation calculations; and (c) siderite. pyrite. barite. and
predictions in cored and uncored intervals or wells. The proposed smectite influencing the resistivity. density and neutron log

205
2 ENHANCED RESERVOIR DESCRIPTION: USING CORE AND LOG DATA TO SPE 26436
IDENTIFY HYDRAULIC (FLOW) UNITS AND PREDICT PERMEABILITY IN UNCORED INTERVALSIWELLS

calculations. The key to enhanced reserves determination and normally distributed parameters does not necessarily establish
improved productivity is not based on the use of empirical causality. On the classical plot. the relationship between
correlations. It is based on the establishment of causal permeability and porosity is not causal. Whereas porosity is
relationships among core-derived, microscopic pore-throat generally independent of grain size. permeability is strongly
parameters and log-derived macroscopic attributes. These dependent on grain size. For exam pie. in a reservoir. porosity and
theoretically correct relationships can then be used as input permeability may. in general, be directly proportional. Yet, in the
variables to calibrate logs for improved reservoir deseription. same reservoir, there may be both high and low permeability
zones with equal porosity values (Fig. I). Therefore this
In this paper, a new, practical and theoretically sound traditional plot can not be used reliably to estimate accurate
methodology is introduced to identify and characterize hydraulic permeability from porosity data. Several investigators5.7.8.9.10
units within mappable geologic units (facies). This methodology have noted the inadequacy of this classical approach and have
uses core data to develop an understanding of the complex proposed alternative models for relating porosity to permeability.
variations in pore geometry within different lithofacies. Core data From the classical approach it can be concluded that for any given
provide information for various depositional and diagenetic rock type. the different .porosity/permeability relationships are
controls on pore geometry. Variations in pore geometrical evidence of the existence of different hydraulic units. Apparently
attributes, in turn, define the existence of distinct zones (hydraulic several investigators 3•5•11 had come to a similar conclusion about
units) with similar fluid-flow characteristics. A hydraulic (pore porosity/permeahility relationships.
geometrical) unit is defined as the representative elementary
volume l (REV) "of total reservoir rock within which geological In this paper. a theoretically correct. and fundamentally derived
and petrophysical properties that affect fluid flow are internally relationship hetween porosity and permeahility is proposed. The
consistent and predictably different from properties of other rock influence of various geological (depositional and diagenetic)
volumes.,,2 variables that control fluid flow are reflected in the proposed
methodology.
Hydraulic units are related to geologic facies distribution. hut do
not necessarily coincide with facies boundaries. 3 Therefore,
hydraulic units may not he vertically contiguous. Hydraulic units Fundamental Theory
are often defined by (a) geological attributes of texture,
The hydraulic quality of a rock is controlled hy pore geometry.
mineralogy, sedimentary structures, hedding contacts and nature
This. in turn, is a function of mineralogy (i.e., type. abundance,
of permeahility barriers and by (h) petrophysical properties of
morphology and location relative to pore throats) and texture (i.e..
porosity, permeability and capillary pressure.
grain size. grain shape. sorting. and packing). Various
permutations of these geological attributes often indicate the
Statement of the Problem existence of distinct rock units with similar pore throat attributes.
Determination of these pore throat attributes is central to accurate
zoning of reservoirs into units with similar hydraulic properties.
Classical Permeability-Porosity Relationships The mean hydraulic unit radius (rmh) concept 12 is the key to
Knowledge of permeability and permeahility distribution is unraveling the hydraulic units and rclating porosity. permeability
critical to effective reservoir description. Several authors4 .5. 6 and capillary pressure.
have noted the importance of these parameters for planning and
implementation of completion strategies for successful
waterflooding programs and for construction of representative
Cross Sectional Area Volume Open to Flow
simulation models for effective reservoir management. r",h .. (2)
Pelmeability and permeability distrihution are usually determined Wetted Perimeter Wetted S ur f ace A rea
from core data. However. most wells are often not cored. As a
result, permeability is estimated in uncored sections/wells from
For a circular. cylindrical capillary tube.
permeability versus porosity relationships that are often
developed from statistically insignificant data sets. In uncored r
wells or zones, empirical permeability is estimated from log- rill" = 2' (3)
derived porosity using Eq. I.
By invoking the concept of the mean hydraulic radius, KOl.eny l3
and Carmen 14 considered the reservoir rock to be composed of a
logk a<l>+b (I) bundle of capillary tubes. They then applied Poisseuille's and
Darcy's Laws to derive a relationship (Eq. 4) between porosity
and permeahility. The primary assumptions in their delivation are
that "the travel time of a fluid clement in a capillary tube is equal
There is apparently no theoretical hasis to support the traditional
to that in a REV:,) 5 and that porosity is effective.
crossplot of the logarithm of permeability versus porosity.
Permeability is plotted as a log function only hecause it appears
to be log-normally distributed. However, correlation of two

206
SPE 26436 1.0. AMAEFULE, M. ALTUNBA Y, D. TIAB, D.G. KERSEY AND D.K. KEELAN 3

<P 1'2
<P e I' 2
k = e -(-) . . . . . . . . . . . . . . .. (4) FZI (11m). designated as the Flow Zone Indicator. is given by
8,2 2,2 2

The mean hydraulic radius (rmh) can be related to the surface area
per unit grain volume (Sgv) and effective porosity (<Pe) as follows: RQI
FZI = ..................... (11)
fFs ,Sg\ <P.
Sg\' = ~(I~e~) = f-(I~e<p) '+'e mh e
(5)
Substituting these variables into Eq. 8 and taking the logarithm of
Substituting Eg. 5 for rmh in Eg. 4. Kozeny and Carmen obtained both sides results in
the following relationship
logRQI = log<p. + 10gFZI (12)

k = (l ~~A2"Is;J' (61
On a log-log plot of RQI versus <Pz' all samples with similar FZl
values Ifill lie on a straight line with unit slope. Samples with
different FZI values will lie on other parallel lines. The value of
where k is in 11m2 and <P e is a fraction. the FZI constant can be determined from the intercept of the unit-
slope straight line at $z =1. Samples that lie on the same straight
The generalized form of the Kozeny-Carmen relationship is given
line have similar pore throat attributes and. thereby. constitute a
by Eq. 7.
hvdraulic unit.
3

k = (1~~e)2lFs,~s~J (7)
Alternative relationships that yield FZI values similar to those
derived from Eg. 12 have also been developed from Eg. 7 as
follows.
where F, is the shape factor (2 is for a circular cylinder). The term
F,,2 has classically been referred to as the Kozeny constant. For If k (md). FZI (11m) and ¢c (fraction). then
ideaL uniform. and unconsolidated rocks. Carmen l4 and
Leverett l6 computed the value of this term to be about 5.
However. Rose and Bruce 17 showed that this term (F, ,2) could k = 1014 ( FZl) -o ( $~ 0
) •••••••••••••••••• (13)
(I - <De>-
vary from 5 to 100 in real reservoir rocks.

Many investigators"-l] have attempted to calculate permeability If $R is defined as

from porosity by using Eg. 7. These attempts generally have not ¢3
e
been successful due to the use of a constant value (typically 5) for ............................. (14)
F s'" and the lack of consideration of Sg} in these computations.
In reality. the Kozeny constant is a variable "constant." which
varies betwecn hydraulic units. but is constant within a given unit. then

The issue of the variability of the KOl.eny constant is addressed in

k = 1014(FZI")$R (15)
the following manner. Dividing both sidcs of Eq. 7 by porosity
(¢e) and taking the square root of both sides results in
Taking the logarithm of both sides of Eq. 15 then resulls in

. . . . . . . . . . . . . .. (8) logk = log ($R) +log[IOI4(FZI")] (16)

A log-log plot of k versus $R results in a straight line with a unit

slope and an intercept (at <PR = I) of !014 (FZIl".
However. if permeability is presented in millidarcies. then the
Some investigators 6. 18 have attempted to zone the reservoir into
following parameter can be defined:
different layers by using the parameter (k/$). Dividing both sides
RQI (11m) = ResevoirQuality Index of Eq. 13 by ¢ results in

= 0.0314 It (9)
-k = IOI4(FZl)-
$e
0 (
-$-
e )"
1-$e
= 1014(FZl)-($J-.
0

-
0
(17)

Taking the logalithm of both sides of Eg. 17 yields

¢z is defined as the pore volume-to-grain volume ratio. k

10"(-) 2Iog$~ + log [1014 (FZl) "1. (18)
<P~ = C~e¢J (lO)
e $e

207
4 ENHANCED RESERVOIR DESCRIPTION: USING CORE AND LOG DATA TO SPE 26436
IDENTIFY HYDRAULIC (FLOW) UNITS AND PREDICT PERMEABILITY IN UNCORED INTERVALSIWELLS

A log-log plot of (k/<1» versus <1>, results in a straight line with a any given homogeneous population are approximately normally
slope of 2 and an intercept of <1>, = 1 of 1014 (FZI)2. As noted distributed. A probability plot of a unimodally and normally
earlier, all samples with similar FZI values belong to the same distributed variable often results in a straight line. In contrast, the
hydraulic unit and will, therefore, lie on the same straight line. existence of multiple. homogeneous subgroups within a given
population often give rise to multimodal distributions and thus
results in multiple straight lines on a probability plot. 22
Geological Significance of FZI Determination of the specific number of hydraulic units is
Statistical techniques based only on variations in permeability constrained by random errors in the porosity and permeability
have been used by previous investigators 19 to zone the reservoir data used to compute FZI. The magnitude of the random errors
into layers. The problem is that these approaches ignored can be estimated by the root mean square technique. 23 Relative
geological attributes that control reservoir zonation. 20 random errors in FZI computed from Eq. 21 were used to
establish the uncertainty envelopes around FZI trend lines. All
The Flow Zone Indicator (FZI) is a unique parameter that samples with the coefficient of variance (~FZUFZI > 0.5) were
incorporates the geological attributes of texture and mineralogy in considered unreliable and therefore were not used in the hydraulic
the discrimination of distinct pore geometrical facies (hydraulic unit zonation process. The following parameters were used to
units). compute (~FZI/FZIhn Eq. 21: (~= 0.2; Me = 0.5% (if <1>1 is
used instead of <1>e. then ~<1> = 1%».
Data 21 displayed on Figs. 2 and 3 demonstrate the excellent
correlation between Swr (from capillary pressure), surface area
(from NMR), weight percent of grains with sizes less than 30
microns (from wet sieve), and FZI. As expected, Swr decreased
with increasing FZI (Fig. 2). The relationship between Swr and
FZI can be represented mathematically by the following
expression. Validation and Characterization of Hydraulic
Units: Case Histories
Swr = 1 - [a + b~ZICJ' (19)
Confirmation of the validity of the proposed FZI technique for
identification of distinct pore geometrical (hydraulic) units is
provided by using a combination of the following technologies:
where a = 1.12, b = 0.5634, c = 1.44, and ~ = 0.998. The
relationship between SgV" W (weight percent grains less than 30 • Mineralogical and textural characteristics of each unit as
microns) and FZI can be represented by the following equation: determined from petrographic data (XRD, FTIR mineralogy,
thin-section photography and SEM)
x. • RQI stress sensitivity
y -- y min + (y lIIax -ymin )~
FZI (20)
• Pore throat characterization as determined from mercury
injection and/or centrifuge capillary pressure.
Values of Ymin' Ymax' and X min are summarized in Table I.
The size distribution of pore throat radii. determined from
In general, rocks containing authigenic pore lining, pore filling.
capillary pressure data. provides an independent method to
and pore bridging clays as well as fine grained. poorly sorted
corroborate the hydraulic unit zonation. The ratio of macro (1' >
sands tend to exhibit high surface area and high tortuosity. hence
1.5 ~m) plus meso (0.5~m < l' < 1.5~m) to micro (1' < 0.5 ~m) is
low FZI (Table 3). In contrast, clean. coarse grained and well
an excellent delineator of hydraulic units.
sorted sands exhibit lower surface areas. lower shape factor. lower
tortuosity, and higher FZI values. In a related fashion. different Manipulation of Eqs. 3-7 yields the fundamental relationship
depositional environments and diagenetic processes control pore between FZI and the mean hydraulic radius (rmh):
geometry and FZI in carbonate rocks.

Hydraulic Unit Zonation Process log (r mh ) log (FZI) + log (Fs 1<1>:> (22)

A detailed flowchart of the hydraulic unitization process is shown

in Fig. 4. This process involves the application of classical
Thus.
statistical techniques including histogram/frequency diagrams.
normality tests, cluster analysis. and error analysis for
discriminating the hydraulic units. A one-dimensional FZI
frequency histogram coupled with a classical test for normal Y = mX (23)
distribution were used to discriminate the distinct family of
hydraulic units. As documented by standard statistical r",h r;;-
where X = FZI, Y = - . and m = slope =",; Fs 1.
textbooks. 22 intrinsic and unimodally distributed variables within <1>.

208
SPE 26436 J.O. AMAEFULE, M. ALTUNBAY, D. TIAB, D.G. KERSEY AND D.K. KEELAN 5 '

Fig. 5 provides the corroborative evidence of the soundness of the South East Asia
fundamental theory as depicted by the straight line in the center of Porosity-permeability data from a clastic sequence from South
the plot of rmh/ljlz versus FZI. The slope is the product of JF."C' East Asia were subjected to the hydraulic unitization process.
which is equal to 6.74. s Fig. 10 shows the existence of distinct hydraulic units within the
cored interval as determined from a log-log plot of RQI versus ljlz.
Four typical examples from clastic sequences in Texas. South The different curves depicted on this figure indicate the FZI
America, South East Asia. and West Africa as well as two average for each of the seven hydraulic units.
carbonate rock examples from West Texas and Canada are
presented to demonstrate the worldwide applicability of this Thin section photomicrographs 24 of samples from the various
proposed technique. units shown on Figs. II to 15 substantiate the effectiveness of the
proposed zonation process. Notiee the effect of texture and
mineralogy on the pore geometrical quality of the different units.
Field Examples of Clastic Reservoirs For instance. comparison of Figs. II A and 118 show the textural
and mineralogical similarity of hydraulic unit I taken from both
Travis Peak Formation, East Texas shallow and deep play samples. In contrast, hydraulic unit 2 (Fig.
The first step in the verification of this hydraulic unit zonation 12) is typified by a very fine to fine grained texture and a much
method is to compare the crossplot of the log of permeability higher clay content (10%) than hydraulic unit I (5%). Hydraulic
versus porosity (Fig. I). The log of permeability/porosity unit 3 (Fig. 13) from a deep play is characterized by a coarse
crossplot shows wide dispersion around the regression line (e.g., siltstone to very fine grained sandstone with a total clay content of
permeability varies up to four orders of magnitude for a given 21 %, while hydraulic unit 4 (Fig. 14) is a medium grained
porosity). The dispersion is decreased when depositional siltstone with a higher clay content of25%. The seventh (Fig. 15)
environments are superimposed onto the plot. Samples deposited and worst unit is a argillaceous siltstone with a total clay content
in the fluvial environment (lower interval) have higher of 34%.
permeability values for given porosities than samples deposited in
the delta plain environment (Fig. I). Replacement of the general West Africa
regression line with regression lines for the upper and lower Fig. 16 shows the classical log k versus ljl plot for a typical data set
intervals reduces the dispersion. from the Niger Delta. Seven distinct hydraulic units were
established within the cored interval by utilizing the proposed
Plotting the distribution of the major diagenetic clays highlights
techniques. As previously observed. the permeability-porosity
the impact of diagenesis on reservoir properties. Regardless of
relationship in the Niger Delta is also nonlinear and predictable
the depositional environment. the highest permeability values are
(Eg. 13). FZI ranged from 0.3 to well over 11 in these
associated with kaolinite, intermediate permeability values
depositionaVdiagenetic sequences. The variability of FZI in the
correlate with chlorite, and the lowcst permeability values
Niger Delta rocks appears to be both texturally and
correlate with illite.
mineralogically controlled.
A log-log plot of RQI versus ljlz for this data set (Fig. 6) shows the
Fig. 17 documents the intrinsic pore geometrical characteristics
existence of six distinct hydraulic units within the cored interval.
and reflects the effect of geological atllibutes (Figs. 18A to l8C)
Each hydraulic unit is characterized by a different average FZI
on hydraulic quality. For example. hydraulic unit 4 (FZI =4.83).
value. The influence of diagenesis has modified the original
which is a fine-grained. moderately well-rounded and well-sorted
depositional parameters to give these multiple hydraulic units.
sand with a low clay content (I % kaolinite) (Fig. 18A), had the
following distribution of pore throat sizes: macro =83%, meso =
South America 2%. and micro = 15%. In contrast. a sample from hydraulic unit
Porosity and permeability data generated on a typical South 5 (FZl =3.7). with a pore throat size distribution of macro =68%,
American clastic reservoir rock were used to compute RQI. ljlzo meso = 5%. and micro = 27%. is comprised of laminated. coarse
and FZI. A log-log plot of RQI versus ljlz (Fig. 7) shows five and fine grained sequences with angular. poorly sorted grains and
distinct hydraulic units within the cored interval. These units an intermediate clay content (5% kaolinite) (Fig. 188). The worst
were discriminated by the previously discussed statistical quality hydraulic unit (FZI = 0.4), which is comprised of
techniques with the theoretical unit slope constraint. laminated. fine grained, well rounded, moderately sorted
sandstone sequences. had a high clay content (12% kaolinite and
Additionally, Fig. 8 shows the classical log k versus ljl plot after chlorite) (Fig. 18C) and a pore throat size disllibution of macro =
the zonation process. The permeability response equations were 22. meso = 29. and micro = 49%.
derived from Eq. 13. It is evident from this plot that permeability
is a nonlinear function of porosity, texture and mineralogy. The Porosity and permeability data determined at multiple net
differences between the hydraulic units was further veri/led by overburden pressures were used to characterize the RQI stress
water-oil capillary pressure data and cation exchange capacity sensitivity of each hydraulic unit. RQI was correlated to net
(CEC) per unit pore volume (Qy). It is evident on Fig. 9 that Qy overburden stress (0) by the following relationship:25
decreased with increasing FZI. thus manifesting the effect of clay
minerals on the rock's hydraulic quality.
209
6 ENHANCED RESERVOIR DESCRIPTION: USING CORE AND LOG DATA TO SPE 26436
IDENTIFY HYDRAULIC (FLOW) UNITS AND PREDICT PERMEABILITY IN UNCORED INTERVALSIWELLS

y = RQI = EXP[-b •
RQ~ s
[I _EXP (_( 0-Oi))]]
c
... (24)
successfully applied to rocks from the UK North Sea and South
East Asia.

where b s =stress sensitivity factor. OJ =initial stress used for RQl j , Field Examples of Carbonate Reservoirs
and c =stress constant (2000-4000 psi).

The stress sensitivity factor (b s ) was further correlated to FZI, Canada

which was computed from porosity and permeability at initial Fig. 20 shows the distribution of FZI values versus class intervals
stress conditions. to arrive at the following predictive relationship: for a Canadian carbonate rock. The existence of three distinct.
uniformly distributed families of k versus $ (Fig. 21) are quite
apparent on the classical crossplot with median FZI values of 1.5.
6.8. and 12.5. This example clearly demonstrates the
applicability of this technique to all rock types.
y= . . . . . . . . . . . . . (25)

West Texas
The crossplot of log permeability versus porosity shows the
seatter common in many carbonate reservoirs (Fig. 22). Zoning
For the Niger Delta clastic rocks. the RQI stress sensitivity (b s )
the interval into hydraulic units shows six distinct FZI groups
was correlated to FZI with the following parameters: Al = 1.39. (Table 3). The grouping into six units is supported by
X = FZI. X o = FZImjn • B 1 = 1.31. C 1 =0.96. D l =5.2. nj = 2.09. independent geological and engineering data. In this example.
with a coefficient of determination (R 2 = .9999). In generaL it was petrology and mercury-injection capillary pressure were
established that rocks from the same hydraulic unit (similar FZI)
performed on selected samples from each hydraulic unit.
exhibited similar RQI stress sensitivities (b s )' b s decreased with
increasing FZI. Additionally. rocks with abundant macropore The effects of the depositional environment on pore throat
throats typically showed lower RQI stress sensitivities (b s < 0.05). attributes were examined. As expected in Paleozoic carbonate
in contrast to microporosity-dominated rocks that showed higher reservoirs. the overprint of diagenesis (Table 4) has significantly
b s values (b s > 0.05). altered the original depositional pore/grain relationships. The
slight correlation of mottling (i.e.. bioturbation) with FZI shows
The classical J function is not adequate for differentiating the
the residual influence of the depositional environment on
various hydraulic units. This is because the J function only
reservoir quality.
normalizes capillary pressure with respect to porosity.
permeability and tortuosity. but does not include the effect of In comparison. diagenetic effecl~ are directly related to FZI. An
surface area. Eq. 26 confirms this observation. inverse correlation (r2 = 0.9999) of dolomite/calcite with FZI
results from the initial diagenetic development and subsequent

~$
. partial dissolution of dolomite.
PC - 2~k RQI I
J = ocose =;. ~ = r = JF.1: (26)
Pore-throat size distributions were calculated from the capillary
lllh pressure data. The ratio of macropore to micropore throat sizes is
directly proportional to FZI.
It is for this reason that no global J function appears plausible for
normalization of capillary pressure. Hydraulic Units and Permeability Prediction
Therefore. a more generalized water saturation versus capillary from Core and Log Data
pressure model was developed based on the concept of the A typical Far East Asia example was used to demonstrate the
hydraulic unit. The proposed relationship (Eq. 27) is well integration of core and log data to identify hydraulic units and to
behaved. predict permeability. A flowchaIt of the process is shown on Fig.
23. Initially. the core depth was matched to the wireline log depth
by comparing the total/spectral gamma derived from log and core
data. Next the proposed zonation process was developed from
stressed core (porosity and Klinkenberg permeability) data from
the cored interval. Based on this process. six distinct hydraulic
As FZI tends to FZI min • Sw approaches I as expected. Similarly. units were identified within the cored interval. The result of this
as Pc = Pd' Sw also approaches I. However. as FZI or Pc zonation process is shown on the log k versus $ plot in Fig. 24. As
approaches large values. Sw approaches the irreducible saturation expected. each pore geometrical family had similar average rmh
Swr Fig. 19 shows a typical Niger Delta example on the and FZI values.
application of the proposed model. This model has also been

210
SPE 26436 J.O. AMAEFULE, M. ALTUNBA Y, D. TIAB, D.G. KERSEY AND D.K. KEELAN 7

Selection of the Wireline Data for the Derivation of The discrimination matlix created from the control welles) is used
Transform Equations as a reference set (a set of histograms for every log for each
hydraulic unit) for the neighhoring uncored welles) to determine
Environmentally corrected wire line log data were selected based
the prohability of having the same hydraulic unit in a given
on their ability to reflect the pore space attributes such as clays
prediction window. This then forms the basis for the qualitative
(type. abundance, location. morphology). pore-throat geometry
predictions of hydraulic uniLs in an uncored well.
and effective porosity. Individual wireline tool responses can not
be directly correlated with the pore-space/throat attributes. Fig. 25 shows the output of a case study using the proposed
Therefore, a statistical evaluation of the correlation between
technique on a Far East Asia well in a typical logging trace with
wireline data and pore-spacelthroat attrihutes is performed for eight tracks. The hydraulic unit numbers staning from the best to
each case. Selected log data are then used as variables in a the worst are assigned to the extreme right track (track 8). In the
transform equation to generate the permeability profiles. first well. the interval from X420 to X500 feet was cored. Data
from X420-X472 were used to develop the hydraulic unit tracks
A set of environmentally-corrected wireline logging tool
and regression models from wireline logs. Hydraulic uniLs were
responses (y. <PW<PD' <PD' L'lt. Rxo, R t ) were rank-correlated to FZI
suhsequently predicted for the section X472 to X500 and
using Spearman's Rho Statistical Technique. 25
permeahility determined from the regression models for those
units. Without hydraulic unitization. permeahility predicted from
Derivation of Regression Models for FZI and wireline logging tool response through the classical multilinear
Permeability correlation technique shows significant deviations from actual
A matrix of known core (FZI, RQI and k) values and r
measured values with ~ .22 (Fig. 26). In contrast. permeability
corresponding wircline log responses is ereated for each hydraulic predicted from the same logging tool responses after hydraulic
unit. A logarithmic filtering of the data is performed to normalize zonation (Fig. 27) exhibits excellent correlations (r > .999) with
the data distribution by calculating the logarithm of each variable. the actual measured values in the cored interval. Fig. 28 compares
Then. a matrix of the logarithms of normalized independent the pelmeability predicted from the classical log permeability vs.
variables and logarithm of the normalized dependent variable porosity crossplot to permeability derived from logging tool
(FZL RQI or k) are generated. Quadratic. multilinear. or linear responses after hydraulic unitization using actual measured
models26 with modifications are then selected to form the basis permeability values in another interval. As is evident on this plot.
for the transform equations. The transformation matrix is the classical log permeability versus porosity correlation
standardized by subtracting the mean of each column from the underestimates pcrmeahility for k > 50 md in contrast to the
row variable itself and dividing the result by the standard proposed hydraulic unit controlled log-derived permeahility.
deviation of that column.
In a second uncored well. the proposed process was used to
Transform equations for each hydraulic unit were derived from predict the hydraulic unit profiles and to assign the regression
the models and assigned to the corresponding depth poinLs based equations for permeability determination based on logging
on the hydraulic unit profile. parameters (Fig. 29). This hydraulic unit process also has been
successfully applied in the Norwegian Nonh Sea for permeability
prediction. 28 More extensive theoretical developments dealing
Qualitative Predictions of Hydraulic Units for the
with clastic. carbonate and naturally fractured reservoirs can be
Uncored Sections found in Tiab. 29
Qualitative predictions for the uncored sections of each well are
performed by using a probahilistic method with a deterministic
tool. The probabilistic method uses relationships implicit in the Conclusions
data to derive results. It docs not require any assumption or A new. practical and theoretically-based technique has been
predetermined equations. Results are obtained by statistical developed to identify and characterize uniLs with similar pore
inferences. The deterministic tool that complements this throat geometrical attrihutes (hydraulic units). This technique has
methodology is the hydraulic unitization process that has been a wide variety of practical field applications for both cored and
performed individually and independently on each well. uncored wells. These include:
Therefore. although it seems like a contradiction in terms. the
deterministic approach (i.e .. hydraulic units profile of the cored • Improved prediction of permeability and permeability
sections) enhances the reliability of the predictions of a distributions from wireline logs in partially coredJuncored
probabilistic tool. because it provides a reliable and intervals and adjacent wells
mathematically sound inference datahase. Probabilities are
• Improved well-to-well rock properties correlations for
computed from estimated distributions of each hydraulic unit hy
refinement of petrophysical models
application of the Bayes Theorem. 27 Level hy level. transition
matrices were created and solved to establish a vertical • Forecasts reservoir rock quality (and formation damage
association between hydraulic units. Details of the probabilistic potential) in panially coredJuncored wells for improved
techniques are well documented. 27 completions and enhanced recovery decisions

211
8 ENHANCED RESERVOIR DESCRIPTION: USING CORE AND LOG DATA TO SPE 26436
IDENTIFY HYDRAULIC (FLOW) UNITS AND PREDICT PERMEABILITY IN UNCORED INTERVALSIWELLS

• Provides a unique parameter, the Flow Zone Indicator (FZI) 5. Stiles, J.H., Jr. and Hutfilz, J.M.: "The use of Routine and
for delineating the number of layers (hydraulic units) Special Core Analysis in Characterizing Brent Group
required for assignment of geological and petrophysical Reservoirs. U.K. North Sea," SPE 18388 (1988).
parameters in numerical simulators.
6. Chopra. A.K., Stein. M.H. and Ader. J.e.: "Development of
The proposed hydraulic unit process has been successfully Reservoir Description to Aid in Design of EOR Project,"
applied worldwide for both clastic (East Texas, South America. SPE 16370. presented at the SPE California Regional
West Africa, South East Asia and Far East Asia) and carbonate meeting, Ventura. California (1987).
(Canada and West Texas) rocks.
7. Dorfman. M.H.. Newey. 1.1.. and Coates, G.R.: "New
Techniques in Lithofacies Determination and Permeability
Nomenclature Prediction in Carbonates using Well Logs" (1990) 113-120.
in Geological Applications ojWireline Logs, A. Hurst, M.A.
stress sensitivity factor Lovell and A.e. Morton, Eds., Geological Society Special
critical stress constant (2000-4000 psi)
Publication No. 48.
shape factor
Kozeny constant 8. Dubrule. O. and Haldorsen, H.H.: "Geostatistics for
permeability (11m2) Permeahility Estimation," Reservoir Characterization, L.W.
pore throat radi us (11m) Lake and H.B. Carroll. Jr. (cds.). Academic Press (1986)
rmh mean hydraulic radius (11m) 223-247.
RQI = Reservoir Quality Index (11m)
(J overburden stress (psi) 9. Timur. A.: "An Investigation of Permeability. Porosity and
(Jj initial overburden stress (psi) Residual Water Saturation Relationships," Trans. SPWLA
Sgy surface area per unit grain volume (I1m- l ) Ninth Annual Logging Symposium (1968) Paper K.
Sw water saturation (fractional pore volume)
Swr irreducible water saturation (fractional pore volume) 10. Wendt. W.A.. Sakurai. S. and Nelson. P.H.: "Permeability
lJle effective porosity (fraction bulk volume) Prediction from Well Logs Using Multiple Regression,"
lJlR reduced porosity index Reservoir Characterization, L.W. Lake and H.B. Carroll, Jr.
lJlz pore volume-to-grain volume ratio (cds.). Academic Press (1986) 181-221.
t tortuosity
II. Slatts. R.M. and Hopkins. G.L.: Scaling Geologic Reservoir
Description to Engineering Needs," JPT. v. 42. No. 2
Acknowledgments (Fehruary 1990) 202-211.

The authors thank Core Laboratories Division of Western Atlas 12. Bird. R.B .. Steware, W.E. and Lightfoot, E.N.: Transport
International for permission granted for the publication of this Phenomena. New York. Wiley (1960).
manuscript. Additional thanks are due to Cynthia Philipson for
technical editing and assistance in figure preparation and to 13. Kozeny, J.: "Uher Kapillare Lcitung des Wassers im Boden,
Shelley Barnett for dedicated efforts in the preparation of this Sitzungsherichte," Royal Academy of Science. Vienna. Proc.
manuscript. Class I (1927) v. 136.271-306.

14. Carmen. P.e.: "Fluid Flow through Granular Beds," Trans.

References AIChE (1937) v. 15. 150-166.

1. Bear, J.: Dynamics oj Fluids in Porous Media. New York. 15. Lake. L.W.: Enhanced Oil Recovery. Prentice Hall,
Elsevier (1972). Englewood Cliffs. New Jersey (1989).
2. Ebanks W.J.. Jr.: "Integrated Approach to Reservoir 16. Leverett. M.e.: "Capillary Behavior in Porous Solids,"
Description for Engineering Projects." (abstract only) AAPG Petroleum Technology. T. P. 1223 (1940) 152-169.
Abstract Flow Unit Concept.
17. Rose, W. and Bruce. W.A.: "Evaluation of Capillary
3. Hearn. e.L., Ebanks. w'J.• Tye. R.S. and Ranganathan. Y.: Character in Petroleum Reservoir Rock," Petroleum
"Geological Factors Influencing Reservoir Performance of Transactions. AIME (1949) 127-142.
the Hartzog Draw Field. Wyoming," JPT v. 36 No.9 (August
1984) 1335-1344. 18. Chopra. AX.: "Reservoir Descriptions Via Pulse," SPE
17568 Int'l Meeting on Petroleum Engineering, Tianjin.
4. Shirer. J.A., Langston, E.P., and Strong. R.B.: "Application China (1988) 171-87.
of Field-Wide Conventional Coring in the Jay-Little
Escambia Creek Unit," JPT(December 1978) 1774-1780. 19. Testerman. J.D.: "A Statistical Reservoir-Zonation Tech-
nique," JPT (August 1962) 889-893: Trans., AIME.

212
 SPE 26436 1.0. AMAEFULE, M. ALTUNBAY, D. TIAB, D.G. KERSEY AND D.K. KEELAN 9

20. Weber, K.J. and van Geuns, L.c.: "Framework for Table 1. Relationship between Surface
Constructing Clastic Reservoir Simulation Models." JPT v. Area, Weight Percent
42 No. 10 (October, 1990) 1248-1253 and 1296-1297. Grains < 30~ and FZI 21
21. Zemanek, J.: "Low-Resistivity Hydrocarbon-Bearing Sand Surface Area Weight %
Reservoirs," SPE 15713, SPE Middle East Oil Show (1987) Parameter
(NMR) (Wet Sieve)
I 85-97/SPEFE (Dec. 1989) 515-21.
Y Sgv W
22. Taylor, J.K.: Statistical Techniquesjor Data Analysis, Lewis
Publishers, Inc., Michigan (1990).
Y max 153.50 m2/g 19.79%
Y min 16.92 m2/g 13.68
23. Amaefule, J.O., and D.K. Keelan: "Stochastic Approach to
Computation of Uncertainties in Petrophysical Properties," X min 0.10 11m 0.80 11m
SCA Paper 8907 (1989) 1-28. 2
R 0.95 0.995

24. BaIT, D.C. and Altunbay, M.: "Identifying Hydraulic Units

as an Aid to Quantifying Depositional Environments and
Table 2. Effects of Geological Attributes on
Diagenetic Facies," Geological Soc. of Malaysia Symp. on
Hydraulic Unit Variables
Reservoir EvaluationIFormation Damage, Kuala Lumpur
(July II, 1992). Qualitative Effects On
Geologic Attribute
25. Amaefule, J.O., D.G. Kersey, D.M. Marshall, J.D. Powell, Fs FZI
't Sgv SWR rmh
L.E. Valencia and D.K. Keelan: "Reservoir Description: A
Practical Synergistic Engineering and Geological Approach Texture
based on Analysis of Core Data." SPE 18167 (1988) 1-30.
Coarse grained L L L H L H
26. Chaterjee, Samprit and Price, Bertram: Regression Analvsis Fine grained H H H L H L
by Example, New York. NY. John Wiley and Sons (1977).
Coarse grained. well sorted L L L H L H
27. Tetzlaff. D.M.. Rodriguez, E. and Anderson, R.L.: Coarse grained. poorly H H H L H L
"Estimating Facies and Petrophysical Parameters from sorted
Integrated Well Data." Paper *. Log Analysis Software
Fine grained. well sorted M M M M M M
Evaluation and Review (LASER) Symp. Trans., Soc. of Well
Log Analysts, London Chapter. London. England (Dec. 13- Mineralogy
15. 1989).
High clay content - smectite H H H L H L
tillite/chlorite
28. Cooper, R.D. and Oliveira, A.M.B.: "The Integration of
Core Permeability. Log-Derived Permeability and Measured High clay - kaolinite M M M M M M
Flow Profiles in the Study of Cut-Off Criteria," SPWLA
Low clay content at H H H L H L
Permeability Seminar (May 18, 1992). pore throat
- smectite
29. Tiab, D.: "Modern Core Analysis Manuscript." Internal
Core Laboratories publication (May. 1993). L = Low
M= Medium
FZI = ---,---
H = High
fFs'ts gv

'f
'Ia bl e 3 : D eoosllona 1Mo de Iiiora W es t 11exas C ar b ona e
HU Depth (ft) 4' (%) k (md) RQI FZI Features

I 6436 7.9 21.87 0.522 6.090 faintly mottled: small anhydrite patches
2 6390 8.8 6.38 0.267 2.774 faintly mottled: small to large anhydrite patches: faintly to moderately laminated
3 6417 10.1 2.23 0.148 1.313 mottled to massive: small anhydrite patches: faint allochem ghosts
4 6491 lOA 1.43 0.116 1.005 massive to faintly laminated; bird's eye anhydrite; fine stylolites

5 6454 8.3 0.37 0.066 0.732 massive to faintly mottled: small anhydrite patches; common allochem ghosts
6 6621 19.4 0.76 0.062 0.258 massive to faintly mottled: small anhydrite patches; faint allochem ghosts: healed fractures

213